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 656, Fri Jun 9 03:39:04 2000 UTC revision 1620, Mon Sep 13 20:10:38 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: John Reppy
17    Date: 2004/09/13
18    Tag: jhr-20040913-signals
19    Description:
20    
21    Fixed the signal masking code to properly nest mask/unmask operations
22    on a per-signal basis.
23    
24    ----------------------------------------------------------------------
25    Name: Matthias Blume (blume (at) tti - c (dot) org)
26    Date: 2004/09/08 13:20:00 CDT
27    Tag: blume-20040908-heap-magic
28    Description:
29    
30    Bumped the heap macig number to 0x09082004 to account for the changed
31    layout of the ML frame under MacOS X.
32    
33    ----------------------------------------------------------------------
34    Name: Allen Leung (leunga (at) reservoir (dot) com)
35    Date: 2004/09/03 11:26:00 EST
36    Tag: leunga-20040903-cygwin-install
37    Description:
38    
39       Added a patch to _arch-n-opsys to enable the Cygwin runtime.
40    The Cygwin runtime is turned on by setting the environment
41    variable SMLNJ_CYGWIN_RUNTIME to 1.
42    
43    ----------------------------------------------------------------------
44    Name: Matthias Blume (blume (at) tti - c (dot) org)
45    Date: 2004/08/31 17:15:00 CDT
46    Tag: blume-20040831-core
47    Description:
48    
49    Added some exports to src/compiler/core.cm upon request by J. Joemann.
50    
51    ----------------------------------------------------------------------
52    Name: Matthias Blume (blume (at) tti - c (dot) org)
53    Date: 2004/08/30 17:55:00 CDT
54    Tag: blume-20040830-installer
55    Description:
56    
57    Upon request by Johannes Joemann:
58    
59    - improved ML code of installer to fall back to coping when renaming
60      fails (i.e., when source and target are on different file systems);
61      the code compiles but has yet to be tested in anger
62    - removed mlrisc from list of default targets (config/targets)
63    
64    ----------------------------------------------------------------------
65    Name: Matthias Blume (blume (at) tti - c (dot) org)
66    Date: 2004/08/27 17:20:00 CDT
67    Tag: blume-20040827-ptreql
68    Description:
69    
70    Added ptreql primop to structure InlineT (upon request from Larry
71    Paulson).
72    
73    ----------------------------------------------------------------------
74    Name: Allen Leung (leunga (at) reservoir (dot) com)
75    Date: 2004/08/15 21:21:00 EST
76    Tag: leunga-110_48-udgraph
77    Description:
78    
79    Another bug fix from Carl Hauser:
80    
81    diff /net/niflab/smlnj48/src/MLRISC/graphs/udgraph.sml udgraph.sml
82    > 48c48
83    > <              | rmv((e as (k,_))::es,L) = rmv(es,if k = i then es else
84    > e::L)
85    > ---
86    > >              | rmv((e as (k,_))::es,L) = rmv(es,if k = i then L else e::L)
87    > Without this, any deletion of an edge in an undirected graph does severe
88    > violence to the graph.
89    
90    ----------------------------------------------------------------------
91    Name: Allen Leung (leunga (at) reservoir (dot) com)
92    Date: 2004/08/10 23:35:00 EST
93    Tag: leunga-110_48-ppc
94    Description:
95    
96       The IBM/MacOS syntax switch on PPC was incorrectly swapped.
97    
98    ----------------------------------------------------------------------
99    Name: Matthias Blume (blume (at) tti - c (dot) org)
100    Date: 2004/08/10 12:00:00 CDT
101    Tag: Release_110_48
102    Description:
103    
104    New working version (110.48).  NEW BOOTFILES!
105    
106    ----------------------------------------------------------------------
107    Name: Allen Leung (leunga (at) reservoir (dot) com)
108    Date: 2004/08/09 12:21:00 EST
109    Tag: leunga-110_47-dijsktra
110    Description:
111    
112    Bug fix from Carl Hauser:
113    
114    single_source_shortest_paths in dijkstra.sml was observed to get wrong
115    answers (by comparing to single_source_shortest_paths in bellman-ford.sml).
116    
117    The problem is that following the expression A.update(dist,s,Num.zero)
118    it is necessary to update the priority queue using Q.decreaseWeight(Q,s).
119    
120    ----------------------------------------------------------------------
121    Name: Matthias Blume (blume (at) tti - c (dot) org)
122    Date: 2004/08/06 18:10:00 CDT
123    Tag: blume-20040806-cmdline
124    Description:
125    
126    Fiddled with handling of command-line options:
127    
128      * sml now quits after processing the command line
129        if -H, -S, -h<n>, or -s<n> appears as the last
130        command-line argument
131      * a new option -q terminates the session when encountered on
132        the command line; subsequent arguments will be ignored
133      * bug fixes: short (erroneous) arguments are no longer ignored
134        completely
135    
136    ----------------------------------------------------------------------
137    Name: Allen Leung (leunga (at) reservoir (dot) com)
138    Date: 2004/08/04 18:17:00 EST
139    Tag: leunga-110_47-ppc-ibm-asm
140    Description:
141    
142      - Added minimal IBM assembly syntax support for PowerPC.
143    
144      - Cygwin: manually changed the file cygwin.def.  Some exported symbols have
145        been altered in the runtime.  We need an automatic way to keep the file
146        in sync.
147    
148    ----------------------------------------------------------------------
149    Name: Matthias Blume (blume (at) tti - c (dot) org)
150    Date: 2004/08/04 14:00:00 CDT
151    Tag: Release_110_47
152    Description:
153    
154    New working version (110.47).  NEW BOOTFILES!
155    
156    ----------------------------------------------------------------------
157    Name: Matthias Blume (blume (at) tti - c (dot) org)
158    Date: 2004/08/03 14:25:00 CDT
159    Tag: blume-20040803-callingconv
160    Description:
161    
162    Added low-level support for choosing C calling conventions by
163    twiddling the type of rawccall.  (See
164    src/compiler/Semant/types/cproto.sml for details.)
165    
166    ----------------------------------------------------------------------
167    Name: Matthias Blume (blume (at) tti - c (dot) org)
168    Date: 2004/08/02 15:55:00 CDT
169    Tag: blume-20040802-backout
170    Description:
171    
172    Backed out of change to win32-filesys.c.  The earlier patch to
173    get_file_time caused CM to produce files with the wrong time stamp.
174    
175    ----------------------------------------------------------------------
176    Name: Matthias Blume (blume (at) tti - c (dot) org)
177    Date: 2004/08/02 14:45:00 CDT
178    Tag: blume-20040802-nlffi-win32
179    Description:
180    
181    Added NLFFI support for Win32, adapted from a patch provided by David
182    Hansel.  This is currently completely untested.  Also, the issue
183    concerning stdcall vs. ccall is still unresolved.
184    
185    ----------------------------------------------------------------------
186    Name: Matthias Blume (blume (at) tti - c (dot) org)
187    Date: 2004/07/30 17:55:00 CDT
188    Tag: blume-20040730-various
189    Description:
190    
191    Gearing up towards 110.47...
192    
193    - various minor bugfixes to ml-nlffigen
194    - a beginning of a manual for nlffi
195    
196    - eliminated 'export name=value' in config/install.sh as this does
197      not work with certain versions of /bin/sh
198      (Thanks to David King at Motorola for catching this.)
199    
200    - several bugfixes provided or suggested by David Hansel at Reactive Systems:
201      - added a test for tm==NULL to gmtime.c and localtime.c
202      - applied patch for incorrect GetFileTime under win32
203      - toSeconds -> toMilliseconds in Win32/win32-process.sml
204    
205    ----------------------------------------------------------------------
206    Name: Matthias Blume (blume (at) tti - c (dot) org)
207    Date: 2004/07/21 18:20:00 CDT
208    Tag: blume-20040721-nlffigen
209    Description:
210    
211    - Fixed minor issue in ml-nlffigen: Now generate structure T_foo
212      for a typedef to an incomplete type, but leave out the "typ" member.
213      (This is just for consistency.)
214    - Started to produce what is supposed to become better (i.e., comprehensive)
215      documentation of what ml-nlffigen does and produces.
216    
217    ----------------------------------------------------------------------
218    Name: Matthias Blume (blume (at) tti - c (dot) org)
219    Date: 2004/07/14 16:25:00 CDT
220    Tag: blume-20040714-union
221    Description:
222    
223    Added C_UNION to c-calls/c-types.sml and updated the machinery
224    (ml-nlffigen, cproto.sml) that conveys C function interface
225    information to the code generator.
226    
227    However, the actual architecture-specific implementation of function
228    arguments and results that are C unions is still not implemented.
229    
230    ----------------------------------------------------------------------
231    Name: Allen Leung (leunga (at) reservoir (dot) com)
232    Date: 2004/07/14 14:38:00 EST
233    Tag: leunga-110_46_1-ppc-lwzu
234    Description:
235    
236        Added these instructions to the PowerPC architecture:
237        LBZU(X), LHZU(X), LWZU(X),
238        STWU(X), STFDU, STFSU
239    
240        etc...
241    
242        Note: I haven't added their instruction encoding into the description.
243    
244    ----------------------------------------------------------------------
245    Name: Allen Leung (leunga (at) reservoir (dot) com)
246    Date: 2004/07/13 15:04:00 EST
247    Tag: leunga-110_46_1-ppc-lwarx
248    Description:
249    
250       Added the two instructions LWARX and STWCX to the PowerPC
251    instruction set.
252    
253       A (untested) rewrite of loop-structure.sml.  The old version
254    is completely broken.
255    
256    ----------------------------------------------------------------------
257    Name: Matthias Blume (blume (at) tti - c (dot) org)
258    Date: 2004/07/13 13:50:00 CDT
259    Tag: blume-20040713-nlffi
260    Description:
261    
262    - use paramAlloc to report c-calls with too many arguments
263      (for PPC version where parameter area is pre-allocated)
264    - added ccall_maxargspace to machspec (to implement the above)
265    - made "make" commend in CM's "make" tool configurable
266    - added option (default: on) for passing the name of the SML/NJ's "bin"
267      directory to "make"; the call looks like this:
268    
269         make <options> SMLNJ_BINDIR=<dir> <target>
270    
271      This can be used by the Makefile to, e.g., pick the "right" version
272      of ml-nlffigen.
273    - minor code tweaks
274    
275    ----------------------------------------------------------------------
276    Name: Matthias Blume (blume (at) tti - c (dot) org)
277    Date: 2004/07/12 22:50:00 CDT
278    Tag: blume-110_46_1-macosx-nlffi
279    Description:
280    
281    NLFFI under Mac OS X now working (sort of).  This is largely untested,
282    though.
283    
284    Note:  1. You have to make a new, clean build of the runtime system.
285           2. There are new BOOTFILES, you have to use them!
286              (Doing the bootstrap process yourself would be *very* painful!
287               If you absolutely have to do it, build the system under
288               a different architecture and then cross-compile.)
289    
290    Version bumped to 110.46.1 to account for runtime data format changes.
291    
292    ----------------------------------------------------------------------
293    Name: Matthias Blume (blume (at) tti - c (dot) org)
294    Date: 2004/06/18 14:30:00 CDT
295    Tag: blume-20040618-unix
296    Description:
297    
298    Changed the implementation of structure Unix so that the same stream
299    is returned every time one of the {text,bin}{In,Out}streamOf functions
300    is invoked on the same proc.  This is not what the spec currently
301    says -- although IMO it arguably should.  (See discussion below.)
302    
303    ----------------------------------------------------------------------
304    Name: Matthias Blume (blume (at) tti - c (dot) org)
305    Date: 2004/06/17 18:15:00 CDT
306    Tag: Release_110_46
307    Description:
308    
309    New working version (110.46).  NEW BOOTFILES!
310    
311    ----------------------------------------------------------------------
312    Name: Matthias Blume (blume (at) tti - c (dot) org)
313    Date: 2004/06/17 17:20:00 CDT
314    Tag: blume-20040617-timer-unix
315    Description:
316    
317    Changed the interface of structures Timer and Unix to match the most
318    recent Basis spec.
319    
320    In the case of Unix there still seems to be an open/weird issue:
321    
322       The {text,bin}{In,Out}streamOf functions are supposed to create
323       fresh streams whenever they are called -- as opposed to have them
324       return the same stream every time.  This design is supposed to
325       prevent space leaks caused by proc values hanging on to streams.
326    
327       The reap function, on the other hand, is supposed to close the
328       streams.  This cannot be done without having a handle on the
329       stream in proc after all...
330    
331    I took the liberty to implement the following stopgap solution:
332    
333      The proc value hangs on to the most recently created stream(s).
334      Reap closes those.  If either or both of the two streams hadn't
335      been created at all yet, then reap will close the corresponding
336      file descriptors directly.
337    
338    PS: I don't understand the original space leak argument anymore.  If
339    a proc hangs on to the imperative stream, then I/O operations on those
340    will advance the state of the cached stream and avoid the space leak.
341    
342    ----------------------------------------------------------------------
343    Name: Matthias Blume (blume (at) tti - c (dot) org)
344    Date: 2004/05/28 16:45:00 CDT
345    Tag: blume-20040528-basis
346    Description:
347    
348    Added signature PACK_REAL and exported functor PrimIO.
349    
350    ----------------------------------------------------------------------
351    Name: Matthias Blume (blume (at) tti - c (dot) org)
352    Date: 2004/05/25 16:00:00 CDT
353    Tag: blume-20040525-group-owner
354    Description:
355    
356    CM now ignores (but still accepts) the "owner" information in group
357    descriptions.  The owner of a group is its next enclosing
358    library. Each group must have a unique owner.  (There is a virtual
359    "toplevel" library that owns groups which are not nested within a real
360    library.)  Previously, each group had to explicitly declare its owner,
361    and CM would check that such a declaration is correct.  The new scheme
362    is to have CM check that for each group there is precisely one owning
363    library.
364    
365    The advantage of the new scheme is that the programmer no longer needs
366    to maintain the somewhat annoying owner information.  The downside is
367    that CM cannot enforce the ownership rule across multiple runs of
368    CM.make.  Fortunately, enclosing the same group in two different
369    libraries A and B which are not part of the same program does not
370    cause real problems.
371    
372    ----------------------------------------------------------------------
373    Name: Matthias Blume (blume (at) tti - c (dot) org)
374    Date: 2004/05/20 16:00:00 CDT
375    Tag: blume-20040520-win32
376    Description:
377    
378    Made the win32 version work again.  (Strangely, a misplaced comma had
379    slipped into win32-process.c which prevented the runtime from being
380    compiled correctly.)
381    
382    Also, included a minor addition to ml-build.bat analogous to what was
383    done in blume-20040519-ml-build.
384    
385    ----------------------------------------------------------------------
386    Name: Matthias Blume (blume (at) tti - c (dot) org)
387    Date: 2004/05/19 22:10:00 CDT
388    Tag: blume-20040519-ml-build
389    Description:
390    
391    Arranged for ml-build to clean up after itself a little bit better.
392    The script generates a temporary SML source file and compiles it using
393    CM, so CM generates metadata (GUID, SKEL, objectfile) for it.  It now
394    gets rid of those at the end, so they don't accumulate under .cm.
395    
396    This required a minor change to install.sh because the name of the
397    metadata directory (default: .cm) is actually configurable at
398    installation time.
399    
400    ----------------------------------------------------------------------
401    Name: Matthias Blume (blume (at) tti - c (dot) org)
402    Date: 2004/05/18 15:50:00 CDT
403    Tag: blume-20040518-mkreader
404    Description:
405    
406    Added Posix.IO.mk{Bin,Text}{Reader,Writer} by lifting their respective
407    implementations from internal modules PosixBinPrimIO and PosixTextPrimIO.
408    
409    ----------------------------------------------------------------------
410    Name: Matthias Blume (blume (at) tti - c (dot) org)
411    Date: 2004/05/11 14:35:00 CDT
412    Tag: blume-20040511-win32sock
413    Description:
414    
415    Added previously missing support for many socket-related functions
416    under win32.  Thanks to David Hansel <hansel@reactive-systems.com>
417    for the voluminous patch!
418    
419    (I have not tested this patch under win32 yet.)
420    
421    Here is David's e-mail:
422    
423    Hi,
424    
425    Attached to this email you find a diff against sml/nj 110.45
426    that will enable socket support under Windows.
427    
428    To apply the patch (using unix or cygwin)
429    1) gunzip runtime.diff.gz
430    2) "cd" into "src/runtime" in the source tree of a fresh
431       110.45 installation.
432    3) patch -p 1 < [your/path/to]runtime.diff
433    
434    The code compiles fine but has NOT yet been extensively tested.
435    I only ran a few tests for basic socket client functionality
436    (which worked fine).  Especially the functions that use ioctl
437    are not tested at all and might not work (see below).
438    
439    I implemented this since we want to move to a newer version of sml/nj
440    but need socket support in order to use it.  This is the first time I
441    even had a look at the sml/nj source,  so please review my changes
442    before making this part of the distribution!  Here are a few issues
443    that I think might be better for someone to solve who is more
444    familiar with the sml/nj source (and socket programming):
445    
446    - getnetbyaddr.c and getnetbyname.c will raise a "not implemented"
447      exception since I could not figure out what the windows equivalent
448      of these functions is
449    
450    - In sockets-osdep.h there are a some #include statements that are
451      only used in a few files that include sockets-osdep.h
452    
453    - In smlnj-sock-lib.c, function init_fn() calls WSAStartup() but
454      does not process its return value since I don't know how to
455      report an error upwards.
456    
457    - It would probably be good to have a call to WSACleanup() when
458      the library is unloaded (if there is such a possibility).
459      Otherwise I think Windows will take care of this automatically
460      when the process finishes.
461    
462    - I used ioctlsocket() as a replacement for ioctl() but I have
463      no idea if that is actually the proper replacement on Windows.
464    
465    - All these issues are marked in the code by "FIXME" comments.
466    
467    We use sml/nj extensively in our products and are quite happy
468    with it.  I hope this contribution will help you.
469    
470    Keep up the good work!
471    
472    David
473    
474    ----------------------------------------------------------------------
475    Name: Matthias Blume (blume (at) tti - c (dot) org)
476    Date: 2004/05/11 14:20:00 CDT
477    Tag: blume-20040511-installml
478    Description:
479    
480    Fixed two bugs in installml script.  (Thanks to Vesa A. Norrman for
481    the patch.)
482    
483    ----------------------------------------------------------------------
484    Name: Matthias Blume (blume (at) tti - c (dot) org)
485    Date: 2004/05/11 14:05:00 CDT
486    Tag: blume-20040511-nlffi-netbsd
487    Description:
488    
489    Added support for nlffi under netbsd.  (Thanks to Vesa A. Norrman for
490    the patch.)
491    
492    ----------------------------------------------------------------------
493    Name: Matthias Blume (blume (at) tti - c (dot) org)
494    Date: 2004/05/11 12:05:00 CDT
495    Tag: blume-20040511-exports
496    Description:
497    
498    As per request by Adam Chlipala <adam@hcoop.net>, extended various
499    export lists in compiler-related .cm-files.
500    
501    ----------------------------------------------------------------------
502    Name: Matthias Blume (blume (at) tti - c (dot) org)
503    Date: 2004/05/11 11:35:00 CDT
504    Tag: blume-20040511-allsource
505    Description:
506    
507    The installer now honors the "src-smlnj" target again, although its meaning
508    has changed from "all sources required for the compiler" to "all sources
509    the installer knows about".  In other words, if you enable "src-smlnj"
510    in the "targets" file, then the installer will pull in sources for
511    everything.  (Notice that this refers to source code only.  Compiled
512    code is still only installed for modules that were requested explicitly
513    or which are required for other modules that were requested explicitly.)
514    
515    ----------------------------------------------------------------------
516    Name: Matthias Blume (blume (at) tti - c (dot) org)
517    Date: 2004/04/23 17:40:00 CDT
518    Tag: blume-20040423-ieee-scan
519    Description:
520    
521    Fixed IEEEReal.scan (and .fromString) so that if there is an overflow
522    in the exponent calculation we get INF or ZERO (depending on the mantissa
523    and the sign of the exponent).
524    
525    ----------------------------------------------------------------------
526    Name: Matthias Blume (blume (at) tti - c (dot) org)
527    Date: 2004/04/23 10:40:00 CDT
528    Tag: blume-20040423-ml-build
529    Description:
530    
531    The ml-build script now terminates with a non-0 status when something
532    goes wrong.
533    
534    ----------------------------------------------------------------------
535    Name: Matthias Blume (blume (at) tti - c (dot) org)
536    Date: 2004/04/22 16:35:00 CDT
537    Tag: blume-20040422-Option
538    Description:
539    
540    Made exception Option to be the same as exception Option.Option
541    (as it should be).
542    
543    ----------------------------------------------------------------------
544    Name: Allen Leung (leunga (at) reservoir (dot) com)
545    Date: 2004/03/19 14:40:00 EST
546    Tag: leunga-20040319-cygwin-nlffi
547    Description:
548    
549    Fixed the runtime so that ml-nlffi-lib runs on the cygwin version
550    of SML/NJ.  The problem is that
551    
552       lib = dlopen(NULL, ...)
553       f   = dlsym(lib, "malloc");
554    
555    does not work on Windows unless we explicitly export symbols
556    such as 'malloc' during linking.  We fixed this by explicitly
557    exporting the required symbols with the magic gcc incantation:
558    
559        -Wl,--export-all cygwin.def
560    
561    where cygwin.def is a file containing all the symbols that
562    we wish to export.
563    
564    I suspect this is a Windows problem and we'll have to
565    do the same (somehow with windows compilers) when
566    we build the native win32 version with the system
567    calls LoadLibrary/GetProcAddress.
568    
569    ----------------------------------------------------------------------
570    Name: Matthias Blume (blume (at) tti - c (dot) org)
571    Date: 2004/03/04 16:35:00 CST
572    Tag: blume-20040304-intinf-fmt
573    Description:
574    
575    Fixed problem with IntInf.fmt (sign would show up on the right instead
576    of on the left for BIN, OCT, and HEX).
577    
578    ----------------------------------------------------------------------
579    Name: Matthias Blume (blume (at) tti - c (dot) org)
580    Date: 2004/03/04 11:25:00 CST
581    Tag: blume-20040304-symlinks
582    Description:
583    
584    Fixed problem with installer script (unix only) where bin/ml-yacc and
585    friends pointed (via symlinks) to absolute locations instead of just
586    .run-sml.  This was reported by Vesa A Norrman.
587    
588    ----------------------------------------------------------------------
589    Name: Matthias Blume (blume (at) tti - c (dot) org)
590    Date: 2004/02/13 14:50:00 CST
591    Tag: Release_110_45
592    Description:
593    
594    New working version (110.45). New bootfiles.
595    
596    ----------------------------------------------------------------------
597    Name: Matthias Blume (blume (at) tti - c (dot) org)
598    Date: 2004/01/26 15:15:15 CST
599    Tag: blume-20040126-toplevel
600    Description:
601    
602    Improved handling of exceptions at the interactive toplevel.
603    
604    ----------------------------------------------------------------------
605    Name: Matthias Blume (blume (at) tti - c (dot) org)
606    Date: 2004/01/26 11:25:00
607    Tag: blume-20040126-app
608    Description:
609    
610    Type of top-level "app" corrected.
611    Added code for setting vp_limitPtrMask to Win32-specific runtime.
612    
613    ----------------------------------------------------------------------
614    Name: Matthias Blume (blume (at) tti - c (dot) org)
615    Date: 2003/11/18 17:10 CST
616    Tag: blume-20031118-basis-fiddle
617    Description:
618    
619    - changed Timer interface to what might become the spec
620    - POSIX_FLAGS -> BIT_FLAGS according to spec
621    - some other minor discrepancies wrt. spec eliminated
622    
623    ----------------------------------------------------------------------
624    Name: Matthias Blume (blume (at) tti - c (dot) org)
625    Date: 2003/11/06 12:00:00 CST
626    Tag: Release_110_44
627    Description:
628    
629    New working version (110.44). New bootfiles.
630    
631    ----------------------------------------------------------------------
632    Name: Matthias Blume (blume (at) tti - c (dot) org)
633    Date: 2003/11/04 11:50:00 CST
634    Tag: blume-20031104-move-libraries
635    Description:
636    
637    Eliminated the "dont_move_libraries" directive in config/targets.
638    (The mechanism was broken and could not be fixed easily.  Moreover,
639    there does not seem to be any reason not to move all libraries into
640    lib during installation.  I originally implemented this directive as a
641    backward-compatibility feature when I first introduced the new CM.
642    Now that things have been stable for a long time and going back to the
643    old CM is not an option, there is no reason to keep it around.)
644    
645    ----------------------------------------------------------------------
646    Name: Matthias Blume (blume (at) tti - c (dot) org)
647    Date: 2003/11/03 16:00:00 CST
648    Tag: blume-20031103-installdir
649    Description:
650    
651    Made installer honor INSTALLDIR variable again.  (Thanks to Chris
652    Richards for pointing out the problem and providing the solution.)
653    
654    ----------------------------------------------------------------------
655    Name: Matthias Blume (blume (at) tti - c (dot) org)
656    Date: 2003/10/01 17:05:00 CDT
657    Tag: blume-20031001-lal-mlrisc
658    Description:
659    
660    MLRISC bug fix from Lal.
661    
662    ----------------------------------------------------------------------
663    Name: Matthias Blume (blume (at) tti - c (dot) org)
664    Date: 2003/09/30 16:10:00 CDT
665    Tag: blume-20030930-primio-bat
666    Description:
667    
668    1. Added openVector, nullRd, and nullWr to PRIM_IO.
669    2. Improved .bat files (for Win32 port) to make things work under Win95.
670       (thanks to Aaron S. Hawley for this one)
671    
672    ----------------------------------------------------------------------
673    Name: Matthias Blume (blume (at) tti - c (dot) org)
674    Date: 2003/09/26 16:05:00 CDT
675    Tag: blume-20030926-wrappriv
676    Description:
677    
678    Added missing wrapper for privilege "primitive" in $smlnj/viscomp/core.cm.
679    
680    ----------------------------------------------------------------------
681    Name: Matthias Blume (blume (at) tti - c (dot) org)
682    Date: 2003/09/26 15:00:00 CDT
683    Tag: blume-20030926-110_43_3
684    Description:
685    
686    - additional cleanup
687    - version number bump, NEW BOOTFILES
688    
689    ----------------------------------------------------------------------
690    Name: Matthias Blume (blume (at) tti - c (dot) org)
691    Date: 2003/09/26 12:00:00 CDT
692    Tag: blume-20030926-ppautoload
693    Description:
694    
695    I modified the read-eval-print loop so that the autoloader gets
696    invoked whenever the prettyprinter tries to look up a symbol that
697    is not currently defined in the toplevel environment but which
698    appears in CM's autoload registry.  As a result, we see far fewer of
699    those ?.Foo.Bar.xxx names in the prettyprinter's output.
700    
701    In addition to this I tried to clean up some pieces of the Basis
702    implementation (e.g., Socket, Word8Array) in order to prevent other
703    instances of these ?.Foo.Bar.xxx names from being printed.
704    
705    The mechanism that picks names for types still needs some work, though.
706    (Right now it seems that if there is a type A.t which is defined to
707    be B.u, but B is unavailable at toplevel, then A.t gets printed as
708    "?.B.u" although the perhaps more sensible solution would be to use
709    "A.t" in this case.  In other words, the prettyprinter should follow
710    a chain of DEFtycs not farther than there are corresponding toplevel
711    names in the current environment.)
712    
713    ----------------------------------------------------------------------
714    Name: Matthias Blume (blume (at) tti - c (dot) org)
715    Date: 2003/09/24 16:31:00 CDT
716    Tag: blume-20030924-installer
717    Description:
718    
719    Another installer tweak:  All the ML code for the installer is now
720    compiled during CMB.make and put into a little library called
721    $smlnj/installer.cm.  The installation then simply invokes
722    
723       sml -m $smlnj/installer.cm
724    
725    and everything happens automagically.
726    
727    Win32:  ML code senses value of environment variable SMLNJ_HOME.
728    Unix:   ML code senses values of environment variables ROOT, CONFIGDIR,
729            and BINDIR.
730    
731    The new scheme guarantees that the ML code responsible for the installation
732    is in sync with the APIs of the main system.  Also, the installer is
733    somewhat faster because the installer script is precompiled.
734    
735    ----------------------------------------------------------------------
736    Name: Matthias Blume (blume (at) tti - c (dot) org)
737    Date: 2003/09/24 15:35:00 CDT
738    Tag: blume-20030924-synsock
739    Description:
740    
741    Added a signature SYNCHRONOUS_SOCKET to basis.cm.  This is like SOCKET
742    but excludes all non-blocking operations.  Defined SOCKET (in Basis)
743    and CML_SOCKET in terms of SYNCHRONOUS_SOCKET.  Removed superfluous
744    implementations of non-blocking operations from CML's Socket
745    structure.
746    
747    ----------------------------------------------------------------------
748    Name: Matthias Blume (blume (at) tti - c (dot) org)
749    Date: 2003/09/24 15:10:05 CDT
750    Tag: blume-20030924-sockets
751    Description:
752    
753    1. Fixed SOCKET API and implementation to match Basis spec.
754       This required changing the internal representation of sockets to one
755       that remembers (for each socket file descriptor) whether it is currently
756       blocking or non-blocking.  This state is maintained lazily (i.e., a system
757       call is made only if the state actually needs to change).
758    
759    2. OS-specific details of sockets were moved into separate files, thus
760       making it possible to unify the bulk of the socket implementations
761       between Unix and Win32.
762    
763    3. CML's socket API changed accordingly.
764       (Note that we need to remove non-blocking functions from this API
765       since they are redundant in the case of CML!)
766    
767    4. CML's socket implementation now makes use of non-blocking functions
768       provided by Basis, thus removing all OS-dependent code from this part
769       of CML.
770    
771    5. Changed Real64.precision from 52 to 53.  Minor cleanup in Real64 code.
772    
773    ----------------------------------------------------------------------
774    Name: Matthias Blume (blume (at) tti - c (dot) org)
775    Date: 2003/09/22 12:10:00 CDT
776    Tag: blume-20030922-110_43_2
777    Description:
778    
779    Made a new interim version and bootfiles for developer's bootstrapping
780    convenience.
781    
782       110.43.2 -- NEW BOOTFILES
783    
784    ----------------------------------------------------------------------
785    Name: Matthias Blume (blume (at) tti - c (dot) org)
786    Date: 2003/09/19 15:55:00 CDT
787    Tag: blume-20030919-cmdir
788    Description:
789    
790    1. new-install.sh -> install.sh
791    2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
792    3. tweaked installer so that another name instead of .cm can be chosen
793       at install time (by setting the CM_DIR_ARC environment variable
794       during installation); once installation is complete, the name is
795       fixed
796    
797    ----------------------------------------------------------------------
798    Name: Matthias Blume (blume (at) tti - c (dot) org)
799    Date: 2003/09/18 16:00:00 CDT
800    Tag: blume-20030918-110_43_1
801    Description:
802    
803    Made a new interim version and bootfiles for developer's bootstrapping
804    convenience.
805    
806       110.43.1 -- NEW BOOTFILES
807    
808    ----------------------------------------------------------------------
809    Name: Matthias Blume (blume (at) tti - c (dot) org)
810    Date: 2003/09/18 15:20:00 CDT
811    Tag: blume-20030918-misc
812    Description:
813    
814    1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
815       this seems to be controversial at the moment) and used those in
816       Posix.ProcEnv.times.
817    
818    2. Added Time.{from,to}Nanoseconds to Time.
819    
820    3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
821       For example, fromLargeInt never needs to look at more than 3 "big
822       digits" to get its 53 bits of precision.
823    
824    ----------------------------------------------------------------------
825    Name: Matthias Blume (blume (at) tti - c (dot) org)
826    Date: 2003/09/17 16:30:00 CDT
827    Tag: blume-20030917-real32-slices
828    Description:
829    
830    Added an entry to the primitive environment
831    (compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
832    added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
833    
834    Removed some of the "magic" constants in real64.sml and replaced them
835    with code that generates these values from their corresponding
836    integer counterparts.
837    
838    Made all(?) the slice-related changes to the Basis and made everything
839    compile again...
840    
841    ----------------------------------------------------------------------
842    Name: Matthias Blume (blume (at) tti - c (dot) org)
843    Date: 2003/09/15 17:45:00 CDT
844    Tag: blume-20030915-rbase
845    Description:
846    
847    Fixed bug in Real.fromLargeInt.
848    
849    ----------------------------------------------------------------------
850    Name: Matthias Blume (blume (at) tti - c (dot) org)
851    Date: 2003/09/13 18:11:00 CDT
852    Tag: blume-20030913-libinstall
853    Description:
854    
855    Minor bugfix in config/libinstall (set anchor with path to
856    standalone tool after installing it, otherwise libraries that
857    need ml-lex or ml-yacc won't compile the first time the installer
858    runs).
859    
860    ----------------------------------------------------------------------
861    Name: Matthias Blume (blume (at) tti - c (dot) org)
862    Date: 2003/09/12 11:45:00 CDT
863    Tag: blume-20030912-various
864    Description:
865    
866     - fixed bug in Real.toLargeInt
867     - fixed bug in Posix.ProcEnv.times
868     - changed inputLine functions to return an option
869     - minor installer improvements / bugfixes
870     - changed default @SMLalloc parameter for x86/celeron to 64k
871    
872    ----------------------------------------------------------------------
873    Name: Matthias Blume (blume (at) tti - c (dot) org)
874    Date: 2003/09/09 22:00:00 CDT
875    Tag: Release_110_43
876    Description:
877    
878        New working release 110.43.  New bootfiles.
879    
880    ----------------------------------------------------------------------
881    Name: Matthias Blume (blume (at) tti - c (dot) org)
882    Date: 2003/09/09 19:20:00 CDT
883    Tag: blume-20030909-installer
884    Description:
885    
886    Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
887    Modified config/install.bat to take advantage of it.  Also modified
888    config/install.sh (and called it config/new-install.sh) to take advantage
889    of it on Unix systems.  (The SML code is (supposed to be) platform-
890    independent.)
891    
892    The installer can now install everything under Win32
893    as well as under *nix as long as it compiles.
894    
895    Other changes:
896    
897      - made CML compile again under Win32
898      - made eXene compile under Win32 (by providing a fake structure UnixSock
899        and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
900      - fixed a bug in nowhere: it assumed that type OS.Process.status is the
901        same as type int;  under Win32 it isn't
902      - fixed some slice-related problems in the win32-specific parts of CML
903      - added a functor argument "sameVol" to os-path-fn.sml in the Basis
904        (under Win32, the volume name is case-insensitive, and the
905        OS.Path code compares volume names for equality)
906    
907    ----------------------------------------------------------------------
908    Name: Matthias Blume (blume (at) tti - c (dot) org)
909    Date: 2003/09/08 11:55:00 CDT
910    Tag: blume-20030908-fullpath
911    Description:
912    
913    Made Win32 version of OS.FileSys.fullPath return current directory
914    when given an empty string.  This is what the spec says, and incidentally,
915    CM depends on it.  (CM otherwise goes into an infinite loop in certain
916    cases when presented with the name of a non-existing .cm file.)
917    
918    ----------------------------------------------------------------------
919    Name: Matthias Blume (blume (at) tti - c (dot) org)
920    Date: 2003/09/04 16:30:00 CDT
921    Tag: blume-20030905-slices-etc
922    Description:
923    
924      1. Changed interface to vectors and arrays in Basis to match
925         (draft) Basis spec.
926      2. Added signatures and implementations of slices according to
927         Basis spec.
928      3. Edited source code throughout the system to make it compile again
929         under 1. and 2.  (In some cases code had to be added to have it
930         match the new signatures.)
931      4. MLRISC should be backward-compatible: the copies of the originals
932         of files that needed to change under 3. were retained, the .cm files
933         check the compiler version number and use old versions when
934         appropriate.
935      5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
936         dirstream -> string option (in accordance with Basis spec).
937      6. When generating code that counts lines, ml-lex used function
938         CharVector.foldli, taking advantage of its old interface.
939         This has been replaced with the corresponding code from
940         CharVectorSlice.  (html-lex must be re-lexed!)
941      7. BitArray in smlnj-lib/Util has been extended/modified to match the
942         new MONO_ARRAY signature.  (Do we need BitArraySlice?)
943      8. Removed temporary additions (fromInternal, toInternal) from the
944         (now obsolete) IntInf in smlnj-lib/Util.
945      9. Cleaned up structure Byte.
946     10. Added localOffset, scan, and fromString to Date (according to spec).
947         Cleaned/corrected implementation of Date.
948         (Still need to check for correctness; implement better canonicalizeDate.)
949     11. Added "scan" to signature IEEE_REAL.
950     12. Some improvements to IntInf [in particular: efficiency-hack for
951         mod and rem when second operand is 2 (for parity checks).]
952     13. Changed representation of type Time.time, using a single IntInf.int
953         value counting microseconds.  This considerably simplified the
954         implementation of structure Time.  We now support negative time
955         values; scan and fromString handle signs.
956     14. Functor PrimIO now takes two additional arguments (VectorSlice and
957         ArraySlice).
958    
959    ----------------------------------------------------------------------
960    Name: Matthias Blume (blume (at) tti - c (dot) org)
961    Date: 2003/08/28 17:00:00 CDT
962    Tag: blume-20030828-intinf
963    Description:
964    
965    This is a major update which comes with a version number bump
966    (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
967    and an implementation of IntInf in the Basis.
968    
969    There are a fairly large number of related changes and updates throughout
970    the system:
971    
972    Basis:
973            - Implemented IntInf.
974            - Made LargeInt a projection of IntInf (by filtering through INTEGER).
975            - Added some missing Real64 operations, most notably Real.toLargeInt.
976            - Added FixedInt as a synonym for Int32.
977    
978    compiler:
979            * Added support for a built-in intinf type.
980              - literals
981              - pattern matching
982              - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
983              - overloading on literals and operations
984    
985            This required adding a primitive type intinf, some additional
986            primops, and implementations for several non-trivial intinf
987            operations in Core.  (The intinf type is completely abstract
988            to the compiler; all operations get delegated back to the Core.)
989    
990            * Intinf equality is handled by polyequal.  However, the compiler
991            does not print its usual warning in this case (since polyequal
992            is the right thing to do there).
993    
994            * Improved the organization of structure InlineT.
995    
996            * A word about conversion primops:
997            If conversions involving intinf do not cancel out during
998            CPS contract, then the compiler must insert calls to Core functions.
999            Since all core access must be resolved already during the FLINT
1000            translate phase, it would be too late a the time of CPS contract
1001            to add new Core calls.  For this reason, conversion primops
1002            for intinf carry two arguments:  1. the numeric argument that
1003            they are supposed to convert, and 2. the Core function that
1004            can help with this conversion if necessary.  If CPS contract
1005            eliminates a primop, then the associated Core function becomes
1006            dead and goes away.  Intinf conversion primops that do not get
1007            eliminated by CPS contract get rewritten into calls of their
1008            core functions by a separate, new phase.
1009    
1010    interactive system:
1011            - Control.Print.intinfDepth controls max length of intinf constants
1012            being printed. (Analogous to Control.Print.stringDepth.)
1013            - Cleanup in printutil and pputil: got rid of unused stuff and
1014            duplicates; replaced some of the code with code that makes better
1015            use of library functionality.
1016    
1017    CM:
1018            Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
1019            are no longer silent.
1020    
1021    CKIT:
1022            Fixed mismatched uses of Int32 and LargeInt.  I always decided
1023            in favor of LargeInt -- which is now the same as IntInf.
1024            CKIT-knowledgable people should check whether this is what's
1025            intended and otherwise change things back to using Int32 or
1026            FixedInt.
1027    
1028    Throughout the code:
1029            Started using IntInf.int literals and built-in operations
1030            (e.g., comparison with 0) where this seems appropriate.
1031    
1032    
1033    ----------------------------------------------------------------------
1034    Name: Dave MacQueen (dbm@cs.uchicago.edu)
1035    Date: 2003/08/13 11:36:00 CDT
1036    Tag: dbm-20030813-mcz-merge1
1037    Description:
1038    
1039    Merging changes from the mcz-branch development branch into trunk.
1040    These changes involve replacement of the emulated old prettyprinter
1041    interface with direct use of the SML/NJ Lib PP library, and fixing
1042    of a couple of bugs (895, 1186) relating to error messages.  A new
1043    prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
1044    has been added.
1045    
1046    ----------------------------------------------------------------------
1047    Name: Matthias Blume (blume (at) tti - c (dot) org)
1048    Date: 2003/08/11 15:45:00 CDT
1049    Tag: blume-20030811-windows
1050    Description:
1051    
1052    Version number bumped to 110.42.9.  NEW BOOTFILES!!!
1053    
1054        http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
1055    
1056    This patch restores SML/NJ's ability to run under win32.  There are a
1057    number of changes, including fixes for several bugs that had gone
1058    unnoticed until now:
1059    
1060      - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
1061        trying to run the win32 version from within cygwin.)
1062    
1063      - There are a number of simple .bat scripts that substitute for their
1064        corresponding Unix shell-scripts.  (See below.)
1065    
1066      - The internals of ml-build have been modified slightly.  The main
1067        difference is that instead of calling ".link-sml" (or link-sml.bat)
1068        using OS.Process.system, the ML process delegates this task back
1069        to the script.  Otherwise problems arise in mixed environments such
1070        as Cygwin where scripts look and work like Unix scripts, but
1071        where OS.Process.system cannot run them.
1072    
1073      - In CM, the srcpath pickler used native pathname syntax -- which
1074        is incorrect in the case of cross-compilation.  The new pickle format
1075        is independent of platform-specific naming conventions.
1076    
1077      - Path configuration files (such as lib/pathconfig) can now choose
1078        between native and standard syntax.  Placing a line of the form
1079    
1080            standard!
1081    
1082        into the file causes all subsequent paths to be interpreted using
1083        CM standard pathname syntax (= Unix conventions); a line
1084    
1085            native!
1086    
1087        switches back to native style.  This was needed so that
1088        path config files can be written portably, see src/system/pathconfig.
1089    
1090      - Runtime system:
1091    
1092          - win32-filesys.c:  get_file_time and set_file_time now
1093            access modification time, not creation time.
1094    
1095          - I/O code made aware of new array representation.
1096    
1097          - Bug fixes in X86.prim.masm.
1098    
1099       - src/system/makeml made aware of win32. (For use under cygwin
1100            and other Unix-environments for windows.)
1101    
1102       - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
1103         which caused BinIO.inputAll to fail consistently.
1104    
1105    .bat scripts:
1106    
1107       Windows .bat scripts assume that SMLNJ_HOME is defined.
1108    
1109       - sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
1110         applications (sml, ml-yacc, ml-lex).
1111       - ml-build.bat: analogous to ml-build.
1112       - config\install.bat:  Analogous to config/install.sh.  This requires
1113         that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
1114         (nmake etc. must be on the path, and vcvars32 must have been run.)
1115         Moreover, sources for ml-lex and ml-yacc need to exist under src,
1116         and the bootfile hierarchy must have been unpacked under
1117         sml.boot.x86-win32.
1118         The script is very primitive and does a poor job at error checking.
1119         It only installs the base system, ml-lex, and ml-yacc.  No other
1120         libraries are being installed (i.e., you get only those that
1121         are part of the compiler.)
1122       - link-sml.bat:  analogous to .link-sml, but not currently used
1123    
1124    Unrelated bug fixes:
1125    
1126      - ml-nlffigen now exports structures ST_* corresponding to incomplete
1127        types.
1128      - Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
1129        otherwise.)
1130    
1131    ----------------------------------------------------------------------
1132    Name: Dave MacQueen (macqueen@cs.uchicago.edu)
1133    Date: 2003/06/17
1134    Tag: macqueen-20030617-bug895
1135    Description:
1136    
1137    Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
1138    Tag will be used for new development branch (mcz-branch) for
1139    use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
1140    
1141    ----------------------------------------------------------------------
1142    Name: Matthias Blume (blume (at) tti - c (dot) org)
1143    Date: 2003/05/27 16:55:00 CDT
1144    Tag: blume-20030527-polyeq
1145    Description:
1146    
1147    Tried to eliminated most cases of polymorphic equality.
1148    
1149    ----------------------------------------------------------------------
1150    Name: Matthias Blume (blume (at) tti - c (dot) org)
1151    Date: 2003/05/21 17:45:00 CDT
1152    Tag: blume-20030517-complete
1153    Description:
1154    
1155    Two changes:
1156    
1157      1. Added a flag for controlling whether non-exhaustive bindings will
1158         be treated as errors (default is false).
1159      2. Cleaned up the *entire* source tree so that CMB.make goes through
1160         without a single non-exhaustive match- or bind warning.
1161    
1162    ----------------------------------------------------------------------
1163    Name: Matthias Blume (blume (at) tti - c (dot) org)
1164    Date: 2003/05/17 10:20:00 CDT
1165    Tag: blume-20030517-absyn
1166    Description:
1167    
1168    1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
1169    
1170       This mainly affects the quality of error messages.  However, some
1171       of the code is now more straightforward than before. (Treatment of
1172       the above four constructs in translate.sml is much simpler than
1173       the "macro-expansion" that was going on before.  Plus, the mach-
1174       compiler no longer gets invoked just to be able to compile an
1175       if-expression.)
1176    
1177    2. The ErrorMsg.Error exception is now caught and absorbed by the
1178       interactive loop.
1179    
1180    ----------------------------------------------------------------------
1181    Name: Allen Leung
1182    Date: 2003/05/16 13:05:00 CDT
1183    Tag: leunga-20030516-cygwin-runtime
1184    Description:
1185    
1186        Ported the runtime system to cygwin, which uses the unix
1187    x86-unix bin files.  Missing/buggy features:
1188    
1189       o getnetbyname, getnetbyaddr: these functions seem to be missing in
1190         the Cygwin library.
1191       o Ctrl-C handling may be flaky.
1192       o Windows system calls and Windows I/O are not supported.
1193    
1194       A new set of binfiles is located at:
1195    
1196         http://www.dorsai.org/~leunga/boot.x86-unix.tgz
1197    
1198       This is only needed for bootstrapping the cygwin version of smlnj.
1199       Other x86 versions can use the existing binfiles.
1200    
1201    ----------------------------------------------------------------------
1202    Name: Matthias Blume
1203    Date: 2003/04/08 15:42:00 CDT
1204    Tag: blume-20030408-listpair
1205    Description:
1206    
1207    1. Added a target 'mlrisc' to installer.
1208    
1209    2. Added missing elements to structure ListPair.
1210    
1211    ----------------------------------------------------------------------
1212    Name: Allen Leung
1213    Date: 2003/01/07 10:40:00 EST
1214    Tag: leunga-20030107-int-rem
1215    Description:
1216    
1217    Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
1218    The arguments to the SUBL instruction were swapped.
1219    
1220    ----------------------------------------------------------------------
1221    Name: Matthias Blume
1222    Date: 2002/12/12 16:25:00 EST
1223    Tag: blume-20021212-risc-ra
1224    Description:
1225    
1226    Fixed a serious bug in the rewrite code for FP spilling/reloading that
1227    sent the RA into an infinite loop when floating point registers get
1228    spilled.  (Because of this bug, e.g., nucleic stopped compiling between
1229    110.37 and 110.38.)
1230    There was another set of potential problems related to the handling of
1231    MLRISC annotations (but those did not yet cause real problems, apparently).
1232    
1233    ----------------------------------------------------------------------
1234    Name: Matthias Blume
1235    Date: 2002/12/06 22:40:00 EST
1236    Tag: blume-20021206-cm-fileid
1237    Description:
1238    
1239    Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
1240    This fixes the problem of CM getting confused by files that suddenly
1241    change their identity (e.g., by getting unlinked and recreated by some
1242    text editor such as vi).  There might be a better/cheaper/cleaner way
1243    of doing this, but for now this will have to do.
1244    
1245    ----------------------------------------------------------------------
1246    Name: Matthias Blume
1247    Date: 2002/10/28 09:50:00 EST
1248    Tag: blume-20021028-typecheck
1249    Description:
1250    
1251    Exported structure Typecheck from $smlnj/viscomp/core.cm.
1252    
1253    ----------------------------------------------------------------------
1254    Name: Matthias Blume
1255    Date: 2002/10/17 09:10:00 EDT
1256    Tag: Release_110_42
1257    Description:
1258    
1259    In good old tradition, there has been a slight hiccup so that we have
1260    to patch 110.42 after the fact.   The old release tag has been replaced
1261    (see below).
1262    
1263    The change solves a problem with two competing approaches the
1264    configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
1265    each other's way.
1266    
1267    This change only affects the runtime system code and the installer script.
1268    (No new bootfiles.)
1269    
1270    ----------------------------------------------------------------------
1271    Name: Matthias Blume
1272    Date: 2002/10/16 12:00:00 EDT
1273    Tag: Release_110_42_removed
1274    Description:
1275    
1276    New working release.  New bootfiles.
1277    
1278    ----------------------------------------------------------------------
1279    Name: Matthias Blume
1280    Date: 2002/10/10 13:10:00 EDT
1281    Tag: blume-20021010-ppc-divs
1282    Description:
1283    
1284    The mltree operator DIVS must be implemented with an overflow check on
1285    the PPC because the hardware indicates divide-by-zero using "overflow" as
1286    well.
1287    
1288    ----------------------------------------------------------------------
1289    Name: Matthias Blume
1290    Date: 2002/07/23 11:45:00
1291    Tag: blume-20020723-smlnj-home
1292    Description:
1293    
1294    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
1295    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
1296    CM_PATHCONFIG is also set), the path configuration file is assumed
1297    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
1298    the entire tree to some other place and everything will "just work".
1299    
1300    (Companion commands such as ml-build and ml-makedepend also sense this
1301    variable.)
1302    
1303    ----------------------------------------------------------------------
1304    Name: Matthias Blume
1305    Date: 2002/07/12 21:19:00 EDT
1306    Tag: blume-20020712-liveness
1307    Description:
1308    
1309    Exported two useful "step" functions from liveness module (MLRISC).
1310    
1311    ----------------------------------------------------------------------
1312    Name: Matthias Blume
1313    Date: 2002/07/05 16:00 EDT
1314    Tag: Release_110_41
1315    Description:
1316    
1317    New working release.  New bootfiles.
1318    
1319    ----------------------------------------------------------------------
1320    Name: Matthias Blume
1321    Date: 2002/07/05 10:25:00 EDT
1322    Tag: blume-20020705-btimp
1323    Description:
1324    
1325    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
1326    other clients can set up backtracing support.
1327    
1328    ----------------------------------------------------------------------
1329    Name: Matthias Blume
1330    Date: 2002/06/25 17:23:00 EDT
1331    Tag: blume-20020625-fpmax
1332    Description:
1333    
1334    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
1335    case.  (The sense of the isNaN test was reversed -- which made min and
1336    max always return their first argument.)
1337    
1338    ----------------------------------------------------------------------
1339    Name: Matthias Blume
1340    Date: 2002/06/11
1341    Tag: blume-20020611-unixpath
1342    Description:
1343    
1344    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
1345    
1346    ----------------------------------------------------------------------
1347    Name: Matthias Blume
1348    Date: 2002/06/10 16:35:00 EDT
1349    Tag: blume-20020610-ieeereal
1350    Description:
1351    
1352    I back-ported my implementation of IEEEReal.fromString from the
1353    idlbasis-devel branch so that we can test it.
1354    
1355    Another small change is that ppDec tries to give more information
1356    than just "<sig>" in the case of functors.  However, this code is
1357    broken in some mysterious way if the functor's body's signature
1358    has not been declared by ascription but gets inferred from the
1359    implementation.  This needs fixing...
1360    
1361    ----------------------------------------------------------------------
1362    Name: Matthias Blume
1363    Date: 2002/05/31
1364    Tag: blume-20020531-btrace-mode
1365    Description:
1366    
1367    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
1368    the wayside when I switched over to using Controls everywhere.)
1369    
1370    ----------------------------------------------------------------------
1371    Name: Lal George
1372    Date: 2002/05/23 12:21:40 EDT
1373    Tag: george-20020523-visual-labels
1374    Description:
1375    
1376    Labels are now displayed in the graphical output to make
1377    the fall-through and target blocks obvious.
1378    
1379    ----------------------------------------------------------------------
1380    Name: Matthias Blume
1381    Date: 2002/05/22 11:03:00 EDT
1382    Tag: blume-20020522-shrink
1383    Description:
1384    
1385    John tweaked yesterday's fix for 1131 to handle an out-of-memory
1386    situation that comes up when allocating huge arrays.
1387    
1388    ----------------------------------------------------------------------
1389    Name: Matthias Blume
1390    Date: 2002/05/21 16:00:00 EDT
1391    Tag: Release_110_40
1392    Description:
1393    
1394    New working release (110.40).  New bootfiles.
1395    
1396    [Also: John Reppy fixed GC bug 1131.]
1397    
1398    ----------------------------------------------------------------------
1399    Name: Matthias Blume
1400    Date: 2002/05/21 12:35:00 EDT
1401    Tag: blume-20020521-cmdoc
1402    Description:
1403    
1404    CM documentation update.
1405    
1406    ----------------------------------------------------------------------
1407    Name: Matthias Blume
1408    Date: 2002/05/21 10:55:00 EDT
1409    Tag: blume-20020521-misc
1410    Description:
1411    
1412    - John tweaked runtime to be silent on heap export (except when
1413      GC messages are on).
1414    - I added a few more things (cross-compiling versions of CMB) to
1415      config/preloads (as suggestions).
1416    
1417    ----------------------------------------------------------------------
1418    Name: Matthias Blume
1419    Date: 2002/05/20 22:25:00 EDT
1420    Tag: blume-20020520-controls
1421    Description:
1422    
1423    - Added ControlUtil structure to control-lib.cm.
1424    - Use it throughout.
1425    - Used Controls facility to define MLRISC controls (as opposed to
1426      registering MLRISC control ref cells with Controls after the fact)
1427    - Fixed messed-up controls priorities.
1428    
1429    * Removed again all the stuff from config/preloads that one wouldn't
1430      be able to preload at the time the initial heap image is built.
1431      (Many libraries, e.g., CML, do not exist yet at this time.  The
1432       only libraries that can be preloaded via config/preloads are those
1433       that come bundled with the bootfiles.)
1434    
1435    ----------------------------------------------------------------------
1436    Name: Matthias Blume
1437    Date: 2002/05/20 10:59:00 EDT
1438    Tag: blume-20020520-preloads
1439    Description:
1440    
1441    Added a lot of commented-out suggestions for things to be included
1442    in config/preloads.
1443    
1444    ----------------------------------------------------------------------
1445    Name: Allen Leung
1446    Date: 2002/05/18 14:20:00 EDT
1447    Tag: leunga-20020518-mdl
1448    Description:
1449    
1450    o Made the mdl tool stuff compile and run again.
1451    o I've disabled all the stuff that depends on RTL specifications; they
1452      are all badly broken anyway.
1453    
1454    ----------------------------------------------------------------------
1455    Name: Matthias Blume
1456    Date: 2002/05/17 16:49:00 EDT
1457    Tag: blume-20020517-controls
1458    Description:
1459    
1460    0. John Reppy made several modifications to the SML/NJ library.
1461       In particular, there is a shiny new controls-lib.cm.
1462    
1463    1. Pushed new controls interface through compiler so that everything
1464       compiles again.
1465    
1466    2. Added FormatComb and FORMAT_COMB to the CML version of the
1467       SML/NJ library (so that CML compiles again).
1468    
1469    3. Modified init scripts because XXX_DEFAULT environment variables
1470       are no longer with us.  (Boot-time initialization is now done
1471       using the same environment variables that are also used for
1472       startup-time initialization of controls.)
1473    
1474    ----------------------------------------------------------------------
1475    Name: Lal George
1476    Date: 2002/05/15 09:20:10 EDT
1477    Tag: george-20020515-pseudo-op-decls
1478    Description:
1479    
1480            All pseudo-ops emitted before the first segment declaration
1481    such as TEXT, DATA, and BSS directives are assumed to be global
1482    declarations and are emitted first in the assembly file. This is
1483    useful in a number of situations where one has pseudo-ops that are not
1484    specific to any segment, and also works around the constraint that one
1485    cannot have client pseudo-ops in the TEXT segment.
1486    
1487            Because no segment is associated with these declarations it is
1488    an error to allocate any space or objects before the first segment
1489    directive and an exception will be raised. However, we cannot make
1490    this check for client pseudo-ops.
1491    
1492            These top level declarations are a field in the CFG graph_info.
1493    In theory you can continue to add to this field after the CFG has been
1494    built -- provided you know what you are doing;-)
1495    
1496    ----------------------------------------------------------------------
1497    Name: Matthias Blume
1498    Date: 2002/05/13 16:40:00 EDT
1499    Tag: blume-20020513-pp-etc
1500    Description:
1501    
1502    A few minor bugfixes:
1503    
1504      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
1505        (Bogus printouts for redefined bindings still occur.  Compiler
1506        bug should no longer occur now.  We need to redo the prettyprinter
1507        from scratch.)
1508    
1509      - CM pathname printer now also adds escape sequences for ( and )
1510    
1511      - commend and docu fixes for ml-nlffi
1512    
1513    ----------------------------------------------------------------------
1514    Name: Matthias Blume
1515    Date: 2002/05/10 16:40:00 EDT
1516    Tag: blume-20020510-erg-textio
1517    Description:
1518    
1519    Applied the following bugfix provided by Emden Gansner:
1520    
1521        Output is corrupted when outputSubstr is used rather than output.
1522    
1523        The problem occurs when a substring
1524    
1525            ss = (s, dataStart, dataLen)
1526    
1527        where dataStart > 0, fills a stream buffer with avail bytes left.
1528        avail bytes of s, starting at index dataStart, are copied into the
1529        buffer, the buffer is flushed, and then the remaining dataLen-avail
1530        bytes of ss are copied into the beginning of the buffer. Instead of
1531        starting this copy at index dataStart+avail in s, the current code
1532        starts the copy at index avail.
1533    
1534        Fix:
1535        In text-io-fn.sml, change line 695 from
1536             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
1537        to
1538             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
1539    
1540    ----------------------------------------------------------------------
1541    Name: Matthias Blume
1542    Date: 2002/04/12 13:55:00 EDT
1543    Tag: blume-20020412-assyntax
1544    Description:
1545    
1546    1. Grabbed newer assyntax.h from the XFree86 project.
1547    2. Fiddled with how to compile X86.prim.asm without warnings.
1548    3. (Very) Minor cleanup in CM.
1549    
1550    ----------------------------------------------------------------------
1551    Name: Matthias Blume
1552    Date: 2002/04/01 (no joke!) 17:07:00 EST
1553    Tag: blume-20020401-x86div
1554    Description:
1555    
1556    Added full support for div/mod/rem/quot on the x86, using the machine
1557    instruction's two results (without clumsily recomputing the remainder)
1558    directly where appropriate.
1559    
1560    Some more extensive power-of-two support was added to the x86 instruction
1561    selector (avoiding expensive divs, mods, and muls where they can be
1562    replaced with cheaper shifts and masks).  However, this sort of thing
1563    ought to be done earlier, e.g., within the CPS optimizer so that
1564    all architectures benefit from it.
1565    
1566    The compiler compiles to a fixed point, but changes might be somewhat
1567    fragile nevertheless.  Please, report any strange things that you might
1568    see wrt. div/mod/quot/rem...
1569    
1570    ----------------------------------------------------------------------
1571    Name: Matthias Blume
1572    Date: 2002/03/29 17:22:00
1573    Tag: blume-20020329-div
1574    Description:
1575    
1576    Fixed my broken div/mod logic.  Unfortunately, this means that the
1577    inline code for div/mod now has one more comparison than before.
1578    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
1579    The problem was with quotient = 0, because that alone does not tell
1580    us which way the rounding went.  One then has to look at whether
1581    remainder and divisor have the same sign...  :(
1582    
1583    Anyway, I replaced the bootfiles with fresh ones...
1584    
1585    ----------------------------------------------------------------------
1586    Name: Matthias Blume
1587    Date: 2002/03/29 14:10:00 EST
1588    Tag: blume-20020329-inlprims
1589    Description:
1590    
1591    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
1592    
1593    Primops have changed. This means that the bin/boot-file formats have
1594    changed as well.
1595    
1596    To make sure that there is no confusion, I made a new version.
1597    
1598    
1599    CHANGES:
1600    
1601    * removed REMT from mltree (remainder should never overflow).
1602    
1603    * added primops to deal with divisions of all flavors to the frontend
1604    
1605    * handled these primops all the way through so they map to their respective
1606      MLRISC support
1607    
1608    * used these primops in the implementation of Int, Int32, Word, Word32
1609    
1610    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
1611    
1612    * parameterized INLMIN, INLMAX, and INLABS by a numkind
1613    
1614    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
1615      floating point
1616    
1617    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
1618      and Real (but Real.abs maps to a separate floating-point-only primop)
1619    
1620    
1621    TODO items:
1622    
1623    * Hacked Alpha32 instruction selection, disabling the selection of REMx
1624      instructions because the machine instruction encoder cannot handle
1625      them.  (Hppa, PPC, and Sparc instruction selection did not handle
1626      REM in the first place, and REM is supported by the x86 machine coder.)
1627    
1628    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
1629      selection phase.  (The two can be streamlined because the hardware
1630      delivers both quotient and remainder at the same time anyway.)
1631    
1632    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
1633      (Currently the behavior is inconsistent both across architectures and
1634      wrt. the draft Basis spec.)
1635    
1636    * Word8 should eventually be handled natively, too.
1637    
1638    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
1639      as if there currently is no execution path that could trigger it in
1640      SML/NJ.  (The assumptions underlying functions arith and promotable do not
1641      hold for things like multiplication and division.)
1642    
1643    ----------------------------------------------------------------------
1644    Name: Matthias Blume
1645    Date: 2002/03/27 16:27:00 EST
1646    Tag: blume-20020327-mlrisc-divisions
1647    Description:
1648    
1649    Added support for all four division operations (ML's div, mod, quot,
1650    and rem) to MLRISC.  In the course of doing so, I also rationalized
1651    the naming (no more annoying switch-around of DIV and QUOT), by
1652    parameterizing the operation by div_rounding_mode (which can be either
1653    DIV_TO_ZERO or DIV_TO_NEGINF).
1654    
1655    The generic MLTreeGen functor takes care of compiling all four
1656    operations down to only round-to-zero div.
1657    
1658    Missing pieces:
1659    
1660      * Doing something smarter than relying on MLTreeGen on architectures
1661        like, e.g., the x86 where hardware division delivers both quotient and
1662        remainder at the same time.  With this, the implementation of the
1663        round-to-neginf operations could be further streamlined.
1664    
1665      * Remove inlining support for div/mod/rem from the frontend and replace it
1666        with primops that get carried through to the backend.  Do this for all
1667        int and word types.
1668    
1669    ----------------------------------------------------------------------
1670    Name: Matthias Blume
1671    Date: 2002/03/25 17:25:00 EST
1672    Tag: blume-20020325-divmod
1673    Description:
1674    
1675    I improved (hopefully without breaking them) the implementation of Int.div,
1676    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
1677    advantage of the following observations:
1678    
1679      Let  q = x quot y      r = x rem y
1680           d = x div  y      m = x mod y
1681    
1682    where "quot" is the round-to-zero version of integer division that
1683    hardware usually provides.  Then we have:
1684    
1685         r = x - q * y        where neither the * nor the - will overflow
1686         d = if q >= 0 orelse x = q * y then q else q - 1
1687                              where neither the * nor the - will overflow
1688         m = if q >= 0 orelse r = 0 then r else r + y
1689                              where the + will not overflow
1690    
1691    This results in substantial simplification of the generated code.
1692    The following table shows the number of CFG nodes and edges generated
1693    for
1694            fun f (x, y) = x OPER y
1695            (* with OPER \in div, mod, quot, rem *)
1696    
1697    
1698        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
1699        --------------------------------------------------------
1700         div |         24 |         39 |         12 |         16
1701         mod |         41 |         71 |         12 |         16
1702        quot |          8 |         10 |          8 |         10
1703         rem |         10 |         14 |          8 |         10
1704    
1705    
1706    ----------------------------------------------------------------------
1707    Name: Matthias Blume
1708    Date: 2002/03/25 22:06:00 EST
1709    Tag: blume-20020325-cprotobug
1710    Description:
1711    
1712    Fixed a bug in cproto (c prototype decoder).
1713    
1714    ----------------------------------------------------------------------
1715    Name: Matthias Blume
1716    Date: 2002/03/25 16:00:00 EST
1717    Tag: blume-20020325-raw-primops
1718    Description:
1719    
1720    I did some cleanup to Allen's new primop code and
1721    replaced yesterday's bootfiles with new ones.
1722    (But they are stored in the same place.)
1723    
1724    ----------------------------------------------------------------------
1725    Name: Matthias Blume
1726    Date: 2002/03/24 22:40:00 EST
1727    Tag: blume-20020324-bootfiles
1728    Description:
1729    
1730    Made the bootfiles that Allen asked for.
1731    
1732    ----------------------------------------------------------------------
1733    Name: Allen Leung
1734    Date: 2002/03/23 15:50:00 EST
1735    Tag: leunga-20020323-flint-cps-rcc-primops
1736    Description:
1737    
1738      1. Changes to FLINT primops:
1739    
1740        (* make a call to a C-function;
1741         * The primop carries C function prototype information and specifies
1742         * which of its (ML-) arguments are floating point. C prototype
1743         * information is for use by the backend, ML information is for
1744         * use by the CPS converter. *)
1745      | RAW_CCALL of { c_proto: CTypes.c_proto,
1746                       ml_args: ccall_type list,
1747                       ml_res_opt: ccall_type option,
1748                       reentrant : bool
1749                     } option
1750       (* Allocate uninitialized storage on the heap.
1751        * The record is meant to hold short-lived C objects, i.e., they
1752        * are not ML pointers.  With the tag, the representation is
1753        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
1754        *)
1755      | RAW_RECORD of {tag:bool,sz:int}
1756      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
1757    
1758      2.  These CPS primops are now overloaded:
1759    
1760           rawload of {kind:numkind}
1761           rawstore of {kind:numkind}
1762    
1763          The one argument form is:
1764    
1765             rawload {kind} address
1766    
1767          The two argument form is:
1768    
1769             rawload {kind} [ml object, byte-offset]
1770    
1771      3. RAW_CCALL/RCC now takes two extra arguments:
1772    
1773         a. The first is whether the C call is reentrant, i.e., whether
1774            ML state should be saved and restored.
1775         b. The second argument is a string argument specifying the name of
1776            library and the C function.
1777    
1778         These things are currently not handled in the code generator, yet.
1779    
1780      4. In CProto,
1781    
1782         An encoding type of "bool" means "ml object" and is mapped into
1783         C prototype of PTR.  Note that "bool" is different than "string",
1784         even though "string" is also mapped into PTR, because "bool"
1785         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
1786    
1787      5. Pickler/unpicker
1788    
1789         Changed to handle RAW_RECORD and newest RAW_CCALL
1790    
1791      6. MLRiscGen,
1792    
1793         1. Changed to handle the new rawload/rawstore/rawrecord operators.
1794         2. Code for handling C Calls has been moved to a new module CPSCCalls,
1795            in the file CodeGen/cpscompile/cps-c-calls.sml
1796    
1797      7. Added the conditional move operator
1798    
1799             condmove of branch
1800    
1801         to cps.  Generation of this is still buggy so it is currently
1802         disabled.
1803    
1804    ----------------------------------------------------------------------
1805    Name: Lal George
1806    Date: 2002/03/22 14:18:25 EST
1807    Tag: george-20020322-cps-branch-prob
1808    Description:
1809    
1810    Implemented the Ball-Larus branch prediction-heuristics, and
1811    incorporated graphical viewers for control flow graphs.
1812    
1813    Ball-Larus Heuristics:
1814    ---------------------
1815    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
1816    
1817    By design it uses the Dempster-Shafer theory for combining
1818    probabilities.  For example, in the function:
1819    
1820        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
1821    
1822    the ball-larus heuristics predicts that the n=0 is unlikely
1823    (OH-heuristic), and the 'then' branch is unlikely because of the
1824    RH-heuristic -- giving the 'then' branch an even lower combined
1825    probability using the Dempster-Shafer theory.
1826    
1827    Finally, John Reppy's loop analysis in MLRISC, further lowers the
1828    probability of the 'then' branch because of the loop in the else
1829    branch.
1830    
1831    
1832    Graphical Viewing:
1833    ------------------
1834    I merely plugged in Allen's graphical viewers into the compiler. The
1835    additional code is not much. At the top level, saying:
1836    
1837            Control.MLRISC.getFlag "cfg-graphical-view" := true;
1838    
1839    will display the graphical view of the control flow graph just before
1840    back-patching.  daVinci must be in your path for this to work. If
1841    daVinci is not available, then the default viewer can be changed
1842    using:
1843    
1844            Control.MLRISC.getString "viewer"
1845    
1846    which can be set to "dot" or "vcg" for the corresponding viewers. Of
1847    course, these viewers must be in your path.
1848    
1849    The above will display the compilation unit at the level of clusters,
1850    many of which are small, boring, and un-interesting. Also setting:
1851    
1852            Control.MLRISC.getInt "cfg-graphical-view_size"
1853    
1854    will display clusters that are larger than the value set by the above.
1855    
1856    
1857    ----------------------------------------------------------------------
1858    Name: Matthias Blume
1859    Date: 2002/03/21 22:20:00 EST
1860    Tag: blume-20020321-kmp-bugfix
1861    Description:
1862    
1863    Changed the interface to the KMP routine in PreString and fixed
1864    a minor bug in one place where it was used.
1865    
1866    ----------------------------------------------------------------------
1867    Name: Allen Leung
1868    Date: 2002/03/21 20:30:00 EST
1869    Tag: leunga-20020321-cfg
1870    Description:
1871    
1872      Fixed a potential problem in cfg edge splitting.
1873    
1874    ----------------------------------------------------------------------
1875    Name: Allen Leung
1876    Date: 2002/03/21 17:15:00 EST
1877    Tag: leunga-20020321-x86-fp-cfg
1878    Description:
1879    
1880      1. Recoded the buggy parts of x86-fp.
1881    
1882         a. All the block reordering code has been removed.
1883            We now depend on the block placement phases to do this work.
1884    
1885         b. Critical edge splitting code has been simplified and moved into the
1886            CFG modules, as where they belong.
1887    
1888         Both of these were quite buggy and complex.  The code is now much, much
1889         simpler.
1890    
1891      2. X86 backend.
1892    
1893         a. Added instructions for 64-bit support.  Instruction selection for
1894            64-bit has not been committed, however, since that
1895            requires changes to MLTREE which haven't been approved by
1896            Lal and John.
1897    
1898         b. Added support for FUCOMI and FUCOMIP when generating code for
1899            PentiumPro and above.  We only generate these instructions in
1900            the fast-fp mode.
1901    
1902         c. Added cases for JP and JNP in X86FreqProps.
1903    
1904      3. CFG
1905    
1906         CFG now has a bunch of methods for edge splitting and merging.
1907    
1908      4. Machine description.
1909    
1910         John's simplification of MLTREE_BASIS.fcond broke a few machine
1911         description things:
1912    
1913         rtl-build.{sig,sml} and hppa.mdl fixed.
1914    
1915         NOTE: the machine description stuff in the repository is still broken.
1916               Again, I can't put my fixes in because that involves
1917               changes to MLTREE.
1918    
1919    ----------------------------------------------------------------------
1920    Name: Matthias Blume
1921    Date: 2002/03/20 15:55:00 EST
1922    Tag: blume-20020320-kmp
1923    Description:
1924    
1925    Implemented Knuth-Morris-Pratt string matching in PreString and used
1926    it for String.isSubstring, Substring.isSubstring, and
1927    Substring.position.
1928    
1929    (Might need some stress-testing.  Simple examples worked fine.)
1930    
1931    ----------------------------------------------------------------------
1932    Name: Matthias Blume
1933    Date: 2002/03/19 16:37:00 EST
1934    Tag: blume-20020319-witnesses
1935    Description:
1936    
1937    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
1938    
1939    This implements a generic mechanism for changing constness qualifiers
1940    anywhere within big C types without resorting to outright "casts".
1941    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
1942    modify the constness at the outermost level.)
1943    The implementation of "convert" is based on the idea of "witness"
1944    values -- values that are not used by the operation but whose types
1945    "testify" to their applicability.  On the implementation side, "convert"
1946    is simply a projection (returning its second curried argument).  With
1947    cross-module inlining, it should not result in any machine code being
1948    generated.
1949    
1950    ----------------------------------------------------------------------
1951    Name: Matthias Blume
1952    Date: 2002/03/15 16:40:00 EST
1953    Tag: blume-20020315-basis
1954    Description:
1955    
1956    Provided (preliminary?) implementations for
1957    
1958      {String,Substring}.{concatWith,isSuffix,isSubstring}
1959    
1960    and
1961    
1962      Substring.full
1963    
1964    Those are in the Basis spec but they were missing in SML/NJ.
1965    
1966    ----------------------------------------------------------------------
1967    Name: Matthias Blume
1968    Date: 2002/03/14 21:30:00 EST
1969    Tag: blume-20020314-controls
1970    Description:
1971    
1972    Controls:
1973    ---------
1974    
1975    1. Factored out the recently-added Controls : CONTROLS stuff and put
1976       it into its own library $/controls-lib.cm.  The source tree for
1977       this is under src/smlnj-lib/Controls.
1978    
1979    2. Changed the names of types and functions in this interface, so they
1980       make a bit more "sense":
1981    
1982          module -> registry
1983          'a registry -> 'a group
1984    
1985    3. The interface now deals in ref cells only.  The getter/setter interface
1986       is (mostly) gone.
1987    
1988    4. Added a function that lets one register an already-existing ref cell.
1989    
1990    5. Made the corresponding modifications to the rest of the code so that
1991       everything compiles again.
1992    
1993    6. Changed the implementation of Controls.MLRISC back to something closer
1994       to the original.  In particular, this module (and therefore MLRISC)
1995       does not depend on Controls.  There now is some link-time code in
1996       int-sys.sml that registers the MLRISC controls with the Controls
1997       module.
1998    
1999    CM:
2000    ---
2001    
2002      * One can now specify the lambda-split aggressiveness in init.cmi.
2003    
2004    ----------------------------------------------------------------------
2005    Name: Allen Leung
2006    Date: 2002/03/13 17:30:00 EST
2007    Tag: leunga-20020313-x86-fp-unary
2008    Description:
2009    
2010    Bug fix for:
2011    
2012    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
2013    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
2014    > - fun f(x,(y,z)) = Real.~ y;
2015    > [autoloading]
2016    > [autoloading done]
2017    >       fchsl   (%eax), 184(%esp)
2018    > Error: MLRisc bug: X86MCEmitter.emitInstr
2019    >
2020    > uncaught exception Error
2021    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
2022    
2023    The problem was that the code generator did not generate any fp registers
2024    in this case, and the ra didn't know that it needed to run the X86FP phase to
2025    translate the pseudo fp instruction.   This only happened with unary fp
2026    operators in certain situations.
2027    
2028    ----------------------------------------------------------------------
2029    Name: Matthias Blume
2030    Date: 2002/03/13 14:00:00 EST
2031    Tag: blume-20020313-overload-etc
2032    Description:
2033    
2034    1. Added _overload as a synonym for overload for backward compatibility.
2035       (Control.overloadKW must be true for either version to be accepted.)
2036    
2037    2. Fixed bug in install script that caused more things to be installed
2038       than what was requested in config/targets.
2039    
2040    3. Made CM aware of the (_)overload construct so that autoloading
2041       works.
2042    
2043    ----------------------------------------------------------------------
2044    Name: Matthias Blume
2045    Date: 2002/03/12 22:03:00 EST
2046    Tag: blume-20020312-url
2047    Description:
2048    
2049    Forgot to update BOOT and srcarchiveurl.
2050    
2051    ----------------------------------------------------------------------
2052    Name: Matthias Blume
2053    Date: 2002/03/12 17:30:00 EST
2054    Tag: blume-20020312-version110392
2055    Description:
2056    
2057    Yet another version number bump (because of small changes to the
2058    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
2059    
2060    Changes:
2061    
2062      The new pid generation scheme described a few weeks ago was overly
2063      complicated.  I implemented a new mechanism that is simpler and
2064      provides a bit more "stability":  Once CM has seen a compilation
2065      unit, it keeps its identity constant (as long as you do not delete
2066      those crucial CM/GUID/* files).  This means that when you change
2067      an interface, compile, then go back to the old interface, and
2068      compile again, you arrive at the original pid.
2069    
2070      There now also is a mechanism that instructs CM to use the plain
2071      environment hash as a module's pid (effectively making its GUID
2072      the empty string).  For this, "noguid" must be specified as an
2073      option to the .sml file in question within its .cm file.
2074      This is most useful for code that is being generated by tools such
2075      as ml-nlffigen (because during development programmers tend to
2076      erase the tool's entire output directory tree including CM's cached
2077      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
2078      revert to the old, broken behavior of SML/NJ, but in specific cases
2079      where there is no danger of interface confusion, its use is ok
2080      (I think).
2081    
2082      ml-nlffigen by default generates "noguid" annotations.  They can be
2083      turned off by specifying -guid in its command line.
2084    
2085    ----------------------------------------------------------------------
2086    Name: Lal George
2087    Date: 2002/03/12 12 14:42:36 EST
2088    Tag: george-20020312-frequency-computation
2089    Description:
2090    
2091    Integrated jump chaining and static block frequency into the
2092    compiler. More details and numbers later.
2093    
2094    ----------------------------------------------------------------------
2095    Name: Lal George
2096    Date: 2002/03/11 11 22:38:53 EST
2097    Tag: george-20020311-jump-chain-elim
2098    Description:
2099    
2100    Tested the jump chain elimination on all architectures (except the
2101    hppa).  This is on by default right now and is profitable for the
2102    alpha and x86, however, it may not be profitable for the sparc and ppc
2103    when compiling the compiler.
2104    
2105    The gc test will typically jump to a label at the end of the cluster,
2106    where there is another jump to an external cluster containing the actual
2107    code to invoke gc. This is to allow factoring of common gc invocation
2108    sequences. That is to say, we generate:
2109    
2110            f:
2111               testgc
2112               ja   L1      % jump if above to L1
2113    
2114            L1:
2115               jmp L2
2116    
2117    
2118    After jump chain elimination the 'ja L1' instructions is converted to
2119    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
2120    up being implemented in their long form (if L2 is far away) using:
2121    
2122            jbe     L3      % jump if below or equal to L3
2123            jmp     L2
2124         L3:
2125            ...
2126    
2127    
2128    For large compilation units L2  may be far away.
2129    
2130    
2131    ----------------------------------------------------------------------
2132    Name: Matthias Blume
2133    Date: 2002/03/11 13:30:00 EST
2134    Tag: blume-20020311-mltreeeval
2135    Description:
2136    
2137    A functor parameter was missing.
2138    
2139    ----------------------------------------------------------------------
2140    Name: Allen Leung
2141    Date: 2002/03/11 10:30:00 EST
2142    Tag: leunga-20020311-runtime-string0
2143    Description:
2144    
2145       The representation of the empty string now points to a
2146    legal null terminated C string instead of unit.  It is now possible
2147    to convert an ML string into C string with InlineT.CharVector.getData.
2148    This compiles into one single machine instruction.
2149    
2150    ----------------------------------------------------------------------
2151    Name: Allen Leung
2152    Date: 2002/03/10 23:55:00 EST
2153    Tag: leunga-20020310-x86-call
2154    Description:
2155    
2156       Added machine generation for CALL instruction (relative displacement mode)
2157    
2158    ----------------------------------------------------------------------
2159    Name: Matthias Blume
2160    Date: 2002/03/08 16:05:00
2161    Tag: blume-20020308-entrypoints
2162    Description:
2163    
2164    Version number bumped to 110.39.1.  NEW BOOTFILES!
2165    
2166    Entrypoints: non-zero offset into a code object where execution should begin.
2167    
2168    - Added the notion of an entrypoint to CodeObj.
2169    - Added reading/writing of entrypoint info to Binfile.
2170    - Made runtime system bootloader aware of entrypoints.
2171    - Use the address of the label of the first function given to mlriscGen
2172      as the entrypoint.  This address is currently always 0, but it will
2173      not be 0 once we turn on block placement.
2174    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
2175      with entry points) from mlriscGen.
2176    
2177    ----------------------------------------------------------------------
2178    Name: Allen Leung
2179    Date: 2002/03/07 20:45:00 EST
2180    Tag: leunga-20020307-x86-cmov
2181    Description:
2182    
2183       Bug fixes for CMOVcc on x86.
2184    
2185       1. Added machine code generation for CMOVcc
2186       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
2187       3. CMOVcc cannot have an immediate operand as argument.
2188    
2189    ----------------------------------------------------------------------
2190    Name: Matthias Blume
2191    Date: 2002/03/07 16:15:00 EST
2192    Tag: blume-20020307-controls
2193    Description:
2194    
2195    This is a very large but mostly boring patch which makes (almost)
2196    every tuneable compiler knob (i.e., pretty much everything under
2197    Control.* plus a few other things) configurable via both the command
2198    line and environment variables in the style CM did its configuration
2199    until now.
2200    
2201    Try starting sml with '-h' (or, if you are brave, '-H')
2202    
2203    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
2204    implements the underlying generic mechanism.
2205    
2206    The interface to some of the existing such facilities has changed somewhat.
2207    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
2208    (The getFoo interface is still there for backward-compatibility, but its
2209    use is deprecated.)
2210    
2211    The ml-build script passes -Cxxx=yyy command-line arguments through so
2212    that one can now twiddle the compiler settings when using this "batch"
2213    compiler.
2214    
2215    TODO items:
2216    
2217    We should go through and throw out all controls that are no longer
2218    connected to anything.  Moreover, we should go through and provide
2219    meaningful (and correct!) documentation strings for those controls
2220    that still are connected.
2221    
2222    Currently, multiple calls to Controls.new are accepted (only the first
2223    has any effect).  Eventually we should make sure that every control
2224    is being made (via Controls.new) exactly once.  Future access can then
2225    be done using Controls.acc.
2226    
2227    Finally, it would probably be a good idea to use the getter-setter
2228    interface to controls rather than ref cells.  For the time being, both
2229    styles are provided by the Controls module, but getter-setter pairs are
2230    better if thread-safety is of any concern because they can be wrapped.
2231    
2232    *****************************************
2233    
2234    One bug fix: The function blockPlacement in three of the MLRISC
2235    backpatch files used to be hard-wired to one of two possibilities at
2236    link time (according to the value of the placementFlag).  But (I
2237    think) it should rather sense the flag every time.
2238    
2239    *****************************************
2240    
2241    Other assorted changes (by other people who did not supply a HISTORY entry):
2242    
2243    1. the cross-module inliner now works much better (Monnier)
2244    2. representation of weights, frequencies, and probabilities in MLRISC
2245       changed in preparation of using those for weighted block placement
2246       (Reppy, George)
2247    
2248    ----------------------------------------------------------------------
2249    Name: Lal George
2250    Date: 2002/03/07 14:44:24 EST 2002
2251    Tag: george-20020307-weighted-block-placement
2252    
2253    Tested the weighted block placement optimization on all architectures
2254    (except the hppa) using AMPL to generate the block and edge frequencies.
2255    Changes were required in the machine properties to correctly
2256    categorize trap instructions. There is an MLRISC flag
2257    "weighted-block-placement" that can be used to enable weighted block
2258    placement, but this will be ineffective without block/edge
2259    frequencies (coming soon).
2260    
2261    
2262    ----------------------------------------------------------------------
2263    Name: Lal George
2264    Date: 2002/03/05 17:24:48 EST
2265    Tag: george-20020305-linkage-cluster
2266    
2267    In order to support the block placement optimization, a new cluster
2268    is generated as the very first cluster (called the linkage cluster).
2269    It contains a single jump to the 'real' entry point for the compilation
2270    unit. Block placement has no effect on the linkage cluster itself, but
2271    all the other clusters  have full freedom in the manner in which they
2272    reorder blocks or functions.
2273    
2274    On the x86 the typical linkage code that is generated is:
2275       ----------------------
2276            .align 2
2277       L0:
2278            addl    $L1-L0, 72(%esp)
2279            jmp     L1
2280    
2281    
2282            .align  2
2283       L1:
2284       ----------------------
2285    
2286    72(%esp) is the memory location for the stdlink register. This
2287    must contain the address of the CPS function being called. In the
2288    above example, it contains the address of  L0; before
2289    calling L1 (the real entry point for the compilation unit), it
2290    must contain the address for L1, and hence
2291    
2292            addl $L1-L0, 72(%esp)
2293    
2294    I have tested this on all architectures except the hppa.The increase
2295    in code size is of course negligible
2296    
2297    ----------------------------------------------------------------------
2298    Name: Allen Leung
2299    Date: 2002/03/03 13:20:00 EST
2300    Tag: leunga-20020303-mlrisc-tools
2301    
2302      Added #[ ... ] expressions to mlrisc tools
2303    
2304    ----------------------------------------------------------------------
2305    Name: Matthias Blume
2306    Date: 2002/02/27 12:29:00 EST
2307    Tag: blume-20020227-cdebug
2308    Description:
2309    
2310    - made types in structure C and C_Debug to be equal
2311    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
2312    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
2313    
2314    ----------------------------------------------------------------------
2315    Name: Matthias Blume
2316    Date: 2002/02/26 12:00:00 EST
2317    Tag: blume-20020226-ffi
2318    Description:
2319    
2320    1. Fixed a minor bug in CM's "noweb" tool:
2321       If numbering is turned off, then truly don't number (i.e., do not
2322       supply the -L option to noweb).  The previous behavior was to supply
2323       -L'' -- which caused noweb to use the "default" line numbering scheme.
2324       Thanks to Chris Richards for pointing this out (and supplying the fix).
2325    
2326    2. Once again, I reworked some aspects of the FFI:
2327    
2328       A. The incomplete/complete type business:
2329    
2330       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
2331         gone!
2332       - ML types representing an incomplete type are now *equal* to
2333         ML types representing their corresponding complete types (just like
2334         in C).  This is still safe because ml-nlffigen will not generate
2335         RTTI for incomplete types, nor will it generate functions that
2336         require access to such RTTI.   But when ML code generated from both
2337         incomplete and complete versions of the C type meet, the ML types
2338         are trivially interoperable.
2339    
2340         NOTE:  These changes restore the full generality of the translation
2341         (which was previously lost when I eliminated functorization)!
2342    
2343       B. Enum types:
2344    
2345       - Structure C now has a type constructor "enum" that is similar to
2346         how the "su" constructor works.  However, "enum" is not a phantom
2347         type because each "T enum" has values (and is isomorphic to
2348         MLRep.Signed.int).
2349       - There are generic access operations for enum objects (using
2350         MLRep.Signed.int).
2351       - ml-nlffigen will generate a structure E_foo for each "enum foo".
2352         * The structure contains the definition of type "mlrep" (the ML-side
2353         representation type of the enum).  Normally, mlrep is the same
2354         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
2355         then mlrep will be defined as a datatype -- thus facilitating
2356         pattern matching on mlrep values.
2357         ("-ec" will be suppressed if there are duplicate values in an
2358          enumeration.)
2359         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
2360         will be generated for each C enum constant xxx.
2361         * Conversion functions m2i and i2m convert between mlrep and
2362         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
2363         * Coversion functions c and ml convert between mlrep and "tag enum".
2364         * Access functions (get/set) fetch and store mlrep values.
2365       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
2366         enumerations are merged into one single enumeration represented by
2367         structure E_'.
2368    
2369    ----------------------------------------------------------------------
2370    Name: Allen Leung
2371    Date: 2002/02/25 04:45:00 EST
2372    Tag: leunga-20020225-cps-spill
2373    
2374    This is a new implementation of the CPS spill phase.
2375    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
2376    In case of problems, replace it with the old file spill.sml
2377    
2378    The current compiler runs into some serious performance problems when
2379    constructing a large record.  This can happen when we try to compile a
2380    structure with many items.  Even a very simple structure like the following
2381    makes the compiler slow down.
2382    
2383        structure Foo = struct
2384           val x_1 = 0w1 : Word32.int
2385           val x_2 = 0w2 : Word32.int
2386           val x_3 = 0w3 : Word32.int
2387           ...
2388           val x_N = 0wN : Word32.int
2389        end
2390    
2391    The following table shows the compile time, from N=1000 to N=4000,
2392    with the old compiler:
2393    
2394    N
2395    1000   CPS 100 spill                           0.04u  0.00s  0.00g
2396           MLRISC ra                               0.06u  0.00s  0.05g
2397              (spills = 0 reloads = 0)
2398           TOTAL                                   0.63u  0.07s  0.21g
2399    
2400    1100   CPS 100 spill                           8.25u  0.32s  0.64g
2401           MLRISC ra                               5.68u  0.59s  3.93g
2402              (spills = 0 reloads = 0)
2403           TOTAL                                   14.71u  0.99s  4.81g
2404    
2405    1500   CPS 100 spill                           58.55u  2.34s  1.74g
2406           MLRISC ra                               5.54u  0.65s  3.91g
2407              (spills = 543 reloads = 1082)
2408           TOTAL                                   65.40u  3.13s  6.00g
2409    
2410    2000   CPS 100 spill                           126.69u  4.84s  3.08g
2411           MLRISC ra                               0.80u  0.10s  0.55g
2412              (spills = 42 reloads = 84)
2413           TOTAL                                   129.42u  5.10s  4.13g
2414    
2415    3000   CPS 100 spill                           675.59u  19.03s  11.64g
2416           MLRISC ra                               2.69u  0.27s  1.38g
2417              (spills = 62 reloads = 124)
2418           TOTAL                                   682.48u  19.61s  13.99g
2419    
2420    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
2421           MLRISC ra                               4.96u  0.27s  2.72g
2422              (spills = 85 reloads = 170)
2423           TOTAL                                   2375.26u  57.21s  48.00g
2424    
2425    As you can see the old cps spill module suffers from some serious
2426    performance problem.  But since I cannot decipher the old code fully,
2427    instead of patching the problems up, I'm reimplementing it
2428    with a different algorithm.  The new code is more modular,
2429    smaller when compiled, and substantially faster
2430    (O(n log n) time and O(n) space).  Timing of the new spill module:
2431    
2432    4000  CPS 100 spill                           0.02u  0.00s  0.00g
2433          MLRISC ra                               0.25u  0.02s  0.15g
2434             (spills=1 reloads=3)
2435          TOTAL                                   7.74u  0.34s  1.62g
2436    
2437    Implementation details:
2438    
2439    As far as I can tell, the purpose of the CPS spill module is to make sure the
2440    number of live variables at any program point (the bandwidth)
2441    does not exceed a certain limit, which is determined by the
2442    size of the spill area.
2443    
2444    When the bandwidth is too large, we decrease the register pressure by
2445    packing live variables into spill records.  How we achieve this is
2446    completely different than what we did in the old code.
2447    
2448    First, there is something about the MLRiscGen code generator
2449    that we should be aware of:
2450    
2451    o MLRiscGen performs code motion!
2452    
2453       In particular, it will move floating point computations and
2454       address computations involving only the heap pointer to
2455       their use sites (if there is only a single use).
2456       What this means is that if we have a CPS record construction
2457       statement
2458    
2459           RECORD(k,vl,w,e)
2460    
2461       we should never count the new record address w as live if w
2462       has only one use (which is often the case).
2463    
2464       We should do something similar to floating point, but the transformation
2465       there is much more complex, so I won't deal with that.
2466    
2467    Secondly, there are now two new cps primops at our disposal:
2468    
2469     1. rawrecord of record_kind option
2470        This pure operator allocates some uninitialized storage from the heap.
2471        There are two forms:
2472    
2473         rawrecord NONE [INT n]  allocates a tagless record of length n
2474         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
2475                                     and initializes the tag.
2476    
2477     2. rawupdate of cty
2478          rawupdate cty (v,i,x)
2479          Assigns to x to the ith component of record v.
2480          The storelist is not updated.
2481    
2482    We use these new primops for both spilling and increment record construction.
2483    
2484     1. Spilling.
2485    
2486        This is implemented with a linear scan algorithm (but generalized
2487        to trees).  The algorithm will create a single spill record at the
2488        beginning of the cps function and use rawupdate to spill to it,
2489        and SELECT or SELp to reload from it.  So both spills and reloads
2490        are fine-grain operations.  In contrast, in the old algorithm
2491        "spills" have to be bundled together in records.
2492    
2493        Ideally, we should sink the spill record construction to where
2494        it is needed.  We can even split the spill record into multiple ones
2495        at the places where they are needed.  But CPS is not a good
2496        representation for global code motion, so I'll keep it simple and
2497        am not attempting this.
2498    
2499     2. Incremental record construction (aka record splitting).
2500    
2501        Long records with many component values which are simulatenously live
2502        (recall that single use record addresses are not considered to
2503         be live) are constructed with rawrecord and rawupdate.
2504        We allocate space on the heap with rawrecord first, then gradually
2505        fill it in with rawupdate.  This is the technique suggested to me
2506        by Matthias.
2507    
2508        Some restrictions on when this is applicable:
2509        1. It is not a VECTOR record.  The code generator currently does not handle
2510           this case. VECTOR record uses double indirection like arrays.
2511        2. All the record component values are defined in the same "basic block"
2512           as the record constructor.  This is to prevent speculative
2513           record construction.
2514    
2515    ----------------------------------------------------------------------
2516    Name: Allen Leung
2517    Date: 2002/02/22 01:02:00 EST
2518    Tag: leunga-20020222-mlrisc-tools
2519    
2520    Minor bug fixes in the parser and rewriter
2521    
2522    ----------------------------------------------------------------------
2523    Name: Allen Leung
2524    Date: 2002/02/21 20:20:00 EST
2525    Tag: leunga-20020221-peephole
2526    
2527    Regenerated the peephole files.  Some contained typos in the specification
2528    and some didn't compile because of pretty printing bugs in the old version
2529    of 'nowhere'.
2530    
2531    ----------------------------------------------------------------------
2532    Name: Allen Leung
2533    Date: 2002/02/19 20:20:00 EST
2534    Tag: leunga-20020219-mlrisc-tools
2535    Description:
2536    
2537       Minor bug fixes to the mlrisc-tools library:
2538    
2539       1.  Fixed up parsing colon suffixed keywords
2540       2.  Added the ability to shut the error messages up
2541       3.  Reimplemented the pretty printer and fixed up/improved
2542           the pretty printing of handle and -> types.
2543       4.  Fixed up generation of literal symbols in the nowhere tool.
2544       5.  Added some SML keywords to to sml.sty
2545    
2546    ----------------------------------------------------------------------
2547    Name: Matthias Blume
2548    Date: 2002/02/19 16:20:00 EST
2549    Tag: blume-20020219-cmffi
2550    Description:
2551    
2552    A wild mix of changes, some minor, some major:
2553    
2554    * All C FFI-related libraries are now anchored under $c:
2555        $/c.cm      --> $c/c.cm
2556        $/c-int.cm  --> $c/internals/c-int.cm
2557        $/memory.cm --> $c/memory/memory.cm
2558    
2559    * "make" tool (in CM) now treats its argument pathname slightly
2560      differently:
2561        1. If the native expansion is an absolute name, then before invoking
2562           the "make" command on it, CM will apply OS.Path.mkRelative
2563           (with relativeTo = OS.FileSys.getDir()) to it.
2564        2. The argument will be passed through to subsequent phases of CM
2565           processing without "going native".  In particular, if the argument
2566           was an anchored path, then "make" will not lose track of that anchor.
2567    
2568    * Compiler backends now "know" their respective C calling conventions
2569      instead of having to be told about it by ml-nlffigen.  This relieves
2570      ml-nlffigen from one of its burdens.
2571    
2572    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
2573    
2574    * Export C_DEBUG and C_Debug from $c/c.cm.
2575    
2576    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
2577      subtyping relationship between incomplete pointers and their complete
2578      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
2579      with the convention of instantiating 'o with ('t, 'c) obj whenever
2580      the pointer target type is complete.  In the incomplete case, 'o
2581      will be instantiated with some "'c iobj" -- a type obtained by
2582      using one of the functors PointerToIncompleteType or PointerToCompleteType.
2583    
2584      Operations that work on both incomplete and complete pointer types are
2585      typed as taking an 'o ptr while operations that require the target to
2586      be known are typed as taking some ('t, 'c) obj ptr.
2587    
2588      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
2589      where void is an eqtype without any values.  This makes it possible
2590      to work on voidptr values using functions meant to operate on light
2591      incomplete pointers.
2592    
2593    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
2594      been vastly simplified.
2595    
2596    ----------------------------------------------------------------------
2597    Name: Matthias Blume
2598    Date: 2002/02/19 10:48:00 EST
2599    Tag: blume-20020219-pqfix
2600    Description:
2601    
2602    Applied Chris Okasaki's bug fix for priority queues.
2603    
2604    ----------------------------------------------------------------------
2605    Name: Matthias Blume
2606    Date: 2002/02/15 17:05:00
2607    Tag: Release_110_39
2608    Description:
2609    
2610    Last-minute retagging is becoming a tradition... :-(
2611    
2612    This is the working release 110.39.
2613    
2614    ----------------------------------------------------------------------
2615    Name: Matthias Blume
2616    Date: 2002/02/15 16:00:00 EST
2617    Tag: Release_110_39-orig
2618    Description:
2619    
2620    Working release 110.39.  New bootfiles.
2621    
2622    (Update: There was a small bug in the installer so it wouldn't work
2623    with all shells.  So I retagged. -Matthias)
2624    
2625    ----------------------------------------------------------------------
2626    Name: Matthias Blume
2627    Date: 2002/02/15 14:17:00 EST
2628    Tag: blume-20020215-showbindings
2629    Description:
2630    
2631    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
2632    the latter can be useful for exploring what bindings are available at
2633    the interactive prompt.  (The first function returns only the list
2634    of symbols that are really bound, the second prints those but also the
2635    ones that CM's autoloading mechanism knows about.)
2636    
2637    ----------------------------------------------------------------------
2638    Name: Matthias Blume
2639    Date: 2002/02/15 12:08:00 EST
2640    Tag: blume-20020215-iptrs
2641    Description:
2642    
2643    Two improvements to ml-nlffigen:
2644    
2645      1. Write files only if they do not exist or if their current contents
2646         do not coincide with what's being written.  (That is, avoid messing
2647         with the time stamps unless absolutely necessary.)
2648    
2649      2. Implement a "repository" mechanism for generated files related
2650         to "incomplete pointer types".   See the README file for details.
2651    
2652    ----------------------------------------------------------------------
2653    Name: Matthias Blume
2654    Date: 2002/02/14 11:50:00 EST
2655    Tag: blume-20020214-quote
2656    Description:
2657    
2658    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
2659    because of the new and improved tag generation scheme.  (Thanks to Allen
2660    Leung for pointing it out.)
2661    
2662    ----------------------------------------------------------------------
2663    Name: Lal George
2664    Date: 2002/02/14 09:55:27 EST 2002
2665    Tag: george-20020214-isabelle-bug
2666    Description:
2667    
2668    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
2669    of Isabelle on the x86.
2670    
2671    From Allen:
2672    -----------
2673     I've found the problem:
2674    
2675         in ra-core.sml, I use the counter "blocked" to keep track of the
2676         true number of elements in the freeze queue.  When the counter goes
2677         to zero, I skip examining the queue.  But I've messed up the
2678         bookkeeping in combine():
2679    
2680             else ();
2681             case !ucol of
2682               PSEUDO => (if !cntv > 0 then
2683                     (if !cntu > 0 then blocked := !blocked - 1 else ();
2684                                        ^^^^^^^^^^^^^^^^^^^^^^^
2685                      moveu := mergeMoveList(!movev, !moveu)
2686                     )
2687                  else ();
2688    
2689         combine() is called to coalesce two nodes u and v.
2690         I think I was thinking that if the move counts of u and v are both
2691         greater than zero then after they are coalesced then one node is
2692         removed from the freeze queue.  Apparently I was thinking that
2693         both u and v are of low degree, but that's clearly not necessarily true.
2694    
2695    
2696    02/12/2002:
2697        Here's the patch.  HOL now compiles.
2698    
2699        I don't know how this impact on performance (compile
2700        time or runtime).  This bug caused the RA (especially on the x86)
2701        to go thru the potential spill phase when there are still nodes on the
2702        freeze queue.
2703    
2704    
2705    
2706    
2707    ----------------------------------------------------------------------
2708    Name: Matthias Blume
2709    Date: 2002/02/13 22:40:00 EST
2710    Tag: blume-20020213-fptr-rtti
2711    Description:
2712    
2713    Fixed a bug in ml-nlffigen that was introduced with one of the previous
2714    updates.
2715    
2716    ----------------------------------------------------------------------
2717    Name: Matthias Blume
2718    Date: 2002/02/13 16:41:00 EST
2719    Tag: blume-20020213-cmlpq
2720    Description:
2721    
2722    Added new priority queue export symbols (which have just been added to
2723    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
2724    not compile and the installer would choke.)
2725    
2726    ----------------------------------------------------------------------
2727    Name: Matthias Blume
2728    Date: 2002/02/13 16:15:00 EST
2729    Tag: blume-20020213-various
2730    Description:
2731    
2732    1. More tweaks to ml-nlffigen:
2733    
2734       - better internal datastructures (resulting in slight speedup)
2735       - "-match" option requires exact match
2736       - "localized" gensym counters (untagged structs/unions nested within
2737         other structs/unions or within typedefs get a fresh counter; their
2738         tag will be prefixed by a concatenation of their parents' tags)
2739       - bug fixes (related to calculation of transitive closure of types
2740         to be included in the output)
2741    
2742    2. Minor Basis updates:
2743    
2744       - added implementations for List.collate and Option.app
2745    
2746    ----------------------------------------------------------------------
2747    Name: Matthias Blume
2748    Date: 2002/02/11 15:55:00 EST
2749    Tag: blume-20020211-gensym
2750    Description:
2751    
2752    Added a "-gensym" option to command line of ml-nlffigen.  This can be
2753    used to specify a "stem" -- a string that is inserted in all "gensym'd"
2754    names (ML structure names that correspond to unnamed C structs, unions,
2755    and enums), so that separate runs of ml-nlffigen do not clash.
2756    
2757    ----------------------------------------------------------------------
2758    Name: Matthias Blume
2759    Date: 2002/02/11 12:05:00 EST
2760    Tag: blume-20020211-gensml
2761    Description:
2762    
2763    A quick fix for a problem with GenSML (in the pgraph-util library):
2764    Make generation of toplevel "local" optional.  (Strictly speaking,
2765    signature definitions within "local" are not legal SML.)
2766    
2767    Other than that: updates to INSTALL and cm/TODO.
2768    
2769    ----------------------------------------------------------------------
2770    Name: Matthias Blume
2771    Date: 2002/02/08 15:00:00 EST
2772    Tag: blume-20020208-uniquepid
2773    Description:
2774    
2775    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
2776    
2777    1. The installer (config/install.sh) has gotten smarter:
2778    
2779         - Configuration options are a bit easier to specify now
2780           (in config/targets).
2781         - Bug in recognizing .tar.bz2 files fixed.
2782         - Installer automatically resolves dependencies between
2783           configuration options (e.g., if you ask for eXene, you will
2784           also get cml -- regardless whether you asked for it or not).
2785         - Installer can run in "quieter mode" by setting the environment
2786           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
2787           "completely silent", though.
2788         - Build HashCons library as part of smlnj-lib.
2789    
2790    2. A new scheme for assigning persistent identifiers to compilation
2791       units (and, by extension, to types etc.) has been put into place.
2792       This fixes a long-standing bug where types and even dynamic values
2793       can get internally confused, thereby compromising type safety
2794       (abstraction) and dynamic correctness.  See
2795    
2796         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
2797    
2798       for an example of how things could go wrong until now.
2799    
2800       The downside of the new scheme is that pids are not quite as
2801       persistent as they used to be: CM will generate a fresh pid
2802       for every compilation unit that it thinks it sees for the first
2803       time.  That means that if you compile starting from a clean, fresh
2804       source tree at two different times, you end up with different
2805       binaries.
2806    
2807       Cutoff recompilation, however, has not been compromised because
2808       CM keeps pid information in special caches between runs.
2809    
2810    ----------------------------------------------------------------------
2811    Name: Lal George
2812    Date: 2002/02/07 15:34:13 EST 2002
2813    Tag: <none>
2814    Description:
2815    
2816    Compilers that generate assembly code may produce  global labels
2817    whose value is resolved at link time. The various peephole optimization
2818    modules did not take this in account.
2819    
2820    TODO. The Labels.addrOf function should really return an option
2821    type so that clients are forced to deal with this issue, rather
2822    than an exception being raised.
2823    
2824    ----------------------------------------------------------------------
2825    Name: Lal George
2826    Date: 2002/02/06 13:55:02 EST
2827    Tag: george-20020206-ra-breakup
2828    Description:
2829    
2830    1. A bug fix from Allen.
2831    
2832        A typo causes extra fstp %st(0)'s to be generated at compensation
2833        edges, which might cause stack underflow traps at runtime.  This
2834        occurs in fft where there are extraneous fstps right before the 'into'
2835        trap instruction (in this case they are harmless since none of the
2836        integers overflow.)
2837    
2838    2. Pulled out various utility modules that were embedded in the modules
2839       of the register allocator. I need these modules for other purposes, but
2840       they are not complete enough to put into a library (just yet).
2841    ----------------------------------------------------------------------
2842    Name: Matthias Blume
2843    Date: 2002/01/31 16:05:00 EST
2844    Tag: blume-20020131-sparc-ccalls
2845    Description:
2846    
2847    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
2848       of extra stack space by mistake.  Fixed.
2849    
2850    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
2851    
2852    ----------------------------------------------------------------------
2853    Name: Allen Leung
2854    Date: 2002/01/30
2855    Tag: leunga-20020130-nowhere-bug-fix
2856    Description:
2857    
2858       MLRISC bug fixes:
2859       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
2860       2. MachineInt.fromString was negating its value.
2861    
2862    ----------------------------------------------------------------------
2863    Name: Matthias Blume
2864    Date: 2002/01/29
2865    Tag: blume-20020129-INSTALL
2866    Description:
2867    
2868    - Added somewhat detailed installation instructions (file INSTALL).
2869    - Fixed curl-detection bug in config/install.sh.
2870    - It is now possible to select the URL getter using the URLGETTER
2871      environment variable:
2872    
2873          not set / "unknown"      --> automatic detection (script tries wget,
2874                                       curl, and lynx)
2875          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
2876                                       how to properly invoke them)
2877          other                    --> use $URLGETTER directly, it must take
2878                                       precisely two command-line arguments
2879                                       (source URL and destination file name)
2880    
2881    ----------------------------------------------------------------------
2882    Name: Matthias Blume
2883    Date: 2002/01/28
2884    Tag: blume-20020128-sparc-ccalls
2885    Description:
2886    
2887    - Fixed problem with calculation of "used" registers in sparc-c-calls.
2888    - Make use of the allocParam argument in sparc-c-calls.
2889    
2890    ----------------------------------------------------------------------
2891    Name: Matthias Blume
2892    Date: 2002/01/28
2893    Tag: blume-20020128-allocParam
2894    Description:
2895    
2896    John Reppy:  Changes c-calls API to accept client-callback for
2897    allocating extra stack space.
2898    me: Corresponding changes to mlriscGen (using a dummy argument that
2899        does not change the current behavior).
2900    
2901    ----------------------------------------------------------------------
2902    Name: Matthias Blume
2903    Date: 2002/01/28 12:00:00
2904    Tag: Release_110_38
2905    Description:
2906    
2907    This time for real!!!
2908    
2909    ----------------------------------------------------------------------
2910    Name: Matthias Blume
2911    Date: 2002/01/28 10:56:00 EST
2912    Tag: blume-20020128-retraction
2913    Description:
2914    
2915    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
2916       with blume-Release_110_38-retracted.)
2917    
2918    1. Fixed a problem with incorrect rounding modes in real64.sml.
2919       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
2920    
2921    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
2922       and enums fixed.  The naming of corresponding ML identifiers should
2923       now be consistent again.
2924    
2925    ----------------------------------------------------------------------
2926    Name: Allen Leung
2927    Date: 2002/01/27
2928    Tag: leunga-20020127-nowhere
2929    Description:
2930    
2931       Added a target called nowhere in the configuration scripts.
2932       Enabling this will build the MLRISC 'nowhere' tool (for translating
2933       programs with where-clauses into legal SML code) during installation.
2934    
2935    ----------------------------------------------------------------------
2936    Name: Matthias Blume
2937    Date: 2002/01/25 21:27:00 EST
2938    Tag: blume-Release_110_38-retracted
2939    Description:
2940    
2941    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
2942    
2943    README will be added later.
2944    
2945    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
2946    (Release_110_38) removed.  Reason: Last-minute bug fixes.
2947    
2948    ----------------------------------------------------------------------
2949    Name: Matthias Blume
2950    Date: 2002/01/25
2951    Tag: blume-20020125-ffi
2952    Description:
2953    
2954    A large number of tweaks and improvements to ml-nlffi-lib and
2955    ml-nlffigen:
2956    
2957       - ML represenation types have been streamlined
2958       - getter and setter functions work with concrete values, not abstract
2959         ones where possible
2960       - ml-nlffigen command line more flexible (see README file there)
2961       - some bugs have been fixed (hopefully)
2962    
2963    ----------------------------------------------------------------------
2964    Name: Lal George
2965    Date: 2002/01/24
2966    Tag: george-20020124-risc-ra-interface
2967    Description:
2968    
2969       There is a dramatic simplification in the interface to the
2970       register allocator for RISC architectures as a result of making
2971       parallel copy instructions explicit.
2972    
2973    ----------------------------------------------------------------------
2974    Name: Matthias Blume
2975    Date: 2002/01/22
2976    Tag: blume-20020122-x86-ccalls
2977    Description:
2978    
2979    Bug fix for c-calls on x86 (having to do with how char- and
2980    short-arguments are being handled).
2981    
2982    ----------------------------------------------------------------------
2983    Name: Matthias Blume
2984    Date: 2002/01/21
2985    Tag: blume-20020121-ff
2986    Description:
2987    
2988    Another day of fiddling with the FFI...
2989    
2990    1. Bug fix/workaround:  CKIT does not complain about negative array
2991       dimensions, so ml-nlffigen has to guard itself against this possibility.
2992       (Otherwise a negative dimension would send it into an infinite loop.)
2993    
2994    2. Some of the abstract types (light objects, light pointers, most "base"
2995       types) in structure C are now eqtypes.
2996    
2997    3. Added constructors and test functions for NULL function pointers.
2998    
2999    ----------------------------------------------------------------------
3000    Name: Matthias Blume
3001    Date: 2002/01/18
3002    Tag: blume-20020118-ready-for-new-release
3003    Description:
3004    
3005    Made config/srcarchiveurl point to a new place.  (Will provide boot
3006    files shortly.)
3007    
3008    Maybe we christen this to be 110.38?
3009    
3010    ----------------------------------------------------------------------
3011    Name: Matthias Blume
3012    Date: 2002/01/18
3013    Tag: blume-20020118-more-ffifiddle
3014    Description:
3015    
3016    Today's FFI fiddling:
3017    
3018      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
3019        C.Set.* that use concrete (MLRep.*) arguments and results instead
3020        of abstract ones.
3021    
3022      - Provided word-style bit operations etc. for "int" representation
3023        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
3024        and Long.
3025    
3026    ----------------------------------------------------------------------
3027    Name: Matthias Blume
3028    Date: 2002/01/18
3029    Tag: blume-20020118-use-x86-fp
3030    Description:
3031    
3032    Now that x86-fast-fp seems to be working, I turned it back on again
3033    by default.  (Seems to work fine now, even with the FFI.)
3034    
3035    Other than that, I added some documentation about the FFI to
3036    src/ml-nlffigen/README and updated the FFI test examples in
3037    src/ml-nlffi-lib/Tests/*.
3038    
3039    ----------------------------------------------------------------------
3040    Name: Allen Leung
3041    Date: 2002/01/17
3042    Tag: leunga-20020117-x86-fast-fp-call
3043    Description:
3044    
3045       1. Fixed a problem with handling return fp values when x86's fast fp
3046          mode is turned on.
3047    
3048       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
3049          of %f32.
3050    
3051       3. Added a constructor INT32lit to the ast of MLRISC tools.
3052    
3053    ----------------------------------------------------------------------
3054    Name: Matthias Blume
3055    Date: 2002/01/16
3056    Tag: blume-20020116-ffifiddle
3057    Description:
3058    
3059    More fiddling with the FFI interface:
3060    
3061     - Make constness 'c instead of rw wherever possible.  This eliminates
3062       the need for certain explicit coercions.  (However, due to ML's
3063       value polymorphism, there will still be many cases where explicit
3064       coercions are necessary.  Phantom types are not the whole answer
3065       to modeling a subtyping relationship in ML.)
3066    
3067     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
3068    
3069     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
3070       clumsy workaround for varargs problem.)
3071    
3072    ----------------------------------------------------------------------
3073    Name: Lal George
3074    Date: 2002/01/15
3075    Tag: <none>
3076    Description:
3077    
3078    1. Since COPY instructions are no longer native to the architecture,
3079       a generic functor can be used to implement the expandCopies function.
3080    
3081    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
3082       TEXT segment.
3083    
3084    ----------------------------------------------------------------------
3085    Name: Matthias Blume
3086    Date: 2002/01/15
3087    Tag: blume-20020115-ffiupdates
3088    Description:
3089    
3090    1. Fix for bug resulting in single-precision float values being returned
3091       incorrectly from FFI calls.
3092    
3093    2. Small modifications to C FFI API:
3094    
3095        - memory-allocation routines return straight objects (no options)
3096          and raise an exception in out-of-memory situations
3097        - unsafe extensions to cast between function pointers and pointers
3098          from/to ints
3099        - added structure C_Debug as an alternative to structure C where
3100          pointer-dereferencing (|*| and |*!) always check for null-pointers
3101        - added open_lib' to DynLinkage;  open_lib' works like open_lib
3102          but also takes a (possibly empty) list of existing library handles
3103          that the current library depends on
3104    
3105    ----------------------------------------------------------------------
3106    Name: Matthias Blume
3107    Date: 2002/01/10
3108    Tag: blume-20020110-newffigen
3109    Description:
3110    
3111    1. Updates to portable graph code.
3112    
3113    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
3114       more scalable now so that even huge interfaces such as the one
3115       for GTK compile in finite time and space. :-)
3116       See src/ml-nlffigen/README for details on what's new.
3117    
3118    ----------------------------------------------------------------------
3119    Name: Lal George
3120    Date: 2001/01/09 14:31:35 EST 2002
3121    Tag: george-20011206-rm-native-copy
3122    Description:
3123    
3124            Removed the native COPY and FCOPY instructions
3125            from all the architectures and replaced it with the
3126            explicit COPY instruction from the previous commit.
3127    
3128            It is now possible to simplify many of the optimizations
3129            modules that manipulate copies. This has not been
3130            done in this change.
3131    
3132    ----------------------------------------------------------------------
3133    Name: Lal George
3134    Date: 2001/12/06 16:50:13 EST 2001
3135    Tag: george-20011206-mlrisc-instruction
3136    Description:
3137    
3138    Changed the representation of instructions from being fully abstract
3139    to being partially concrete. That is to say:
3140    
3141      from
3142            type instruction
3143    
3144      to
3145            type instr                              (* machine instruction *)
3146    
3147            datatype instruction =
3148                LIVE of {regs: C.cellset, spilled: C.cellset}
3149              | KILL of {regs: C.cellset, spilled: C.cellset}
3150              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
3151              | ANNOTATION of {i: instruction, a: Annotations.annotation}
3152              | INSTR of instr
3153    
3154    This makes the handling of certain special instructions that appear on
3155    all architectures easier and uniform.
3156    
3157    LIVE and KILL say that a list of registers are live or killed at the
3158    program point where they appear. No spill code is generated when an
3159    element of the 'regs' field is spilled, but the register is moved to
3160    the 'spilled' (which is present, more for debugging than anything else).
3161    
3162    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
3163    We used to generate:
3164    
3165            DEFFREG f1
3166            f1 := f2 + f3
3167            trapb
3168    
3169    but now generate:
3170    
3171            f1 := f2 + f3
3172            trapb
3173            LIVE {regs=[f1,f2,f3], spilled=[]}
3174    
3175    Furthermore, the DEFFREG (hack) required that all floating point instruction
3176    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
3177    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
3178    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
3179    this architecture).
3180    
3181    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
3182    all the architectures. This will result in further simplification of the
3183    register allocator that must be aware of them for coalescing purposes, and
3184    will also simplify certain aspects of the machine description that provides
3185    callbacks related to parallel copies.
3186    
3187    ANNOTATION should be obvious, and now INSTR represents the honest to God
3188    machine instruction set!
3189    
3190    The <arch>/instructions/<arch>Instr.sml files define certain utility
3191    functions for making porting easier -- essentially converting upper case
3192    to lower case. All machine instructions (of type instr) are in upper case,
3193    and the lower case form generates an MLRISC instruction. For example on
3194    the alpha we have:
3195    
3196      datatype instr =
3197         LDA of {r:cell, b:cell, d:operand}
3198       | ...
3199    
3200      val lda : {r:cell, b:cell, d:operand} -> instruction
3201        ...
3202    
3203    where lda is just (INSTR o LDA), etc.
3204    
3205    ----------------------------------------------------------------------
3206    Name: Matthias Blume
3207    Date: 2001/11/22 21:40:00 EST
3208    Tag: Release_110_37
3209    Description:
3210    
3211    Release 110.37.  This time for real.
3212    
3213    ----------------------------------------------------------------------
3214    Name: Matthias Blume
3215    Date: 2001/11/21 16:35:00 EST
3216    Tag: blume-20011121-foot-in-mouth
3217    Description:
3218    
3219    Removed the "Release_110_37" tag because of a serious bug.
3220    This will be re-tagged once the bug is fixed.
3221    
3222    ----------------------------------------------------------------------
3223    Name: Matthias Blume
3224    Date: 2001/11/21 16:14:00 EST
3225    Tag: blume-20011121-forgottenfile
3226    Description:
3227    
3228    Forgot to add a file.  (Just a .tex-file -- part of
3229    the CM manual source.)
3230    
3231    ----------------------------------------------------------------------
3232    Name: Matthias Blume
3233    Date: 2001/11/21 16:10:00 EST
3234    Tag: blume-20011121-invalid_110_37
3235    Description:
3236    
3237    Note: I removed the original tag "Release_110_37" from this commit
3238          because we found a serious bug in all non-x86 backends.
3239          - Matthias
3240    
3241    1. Modifications to the SML/NJ code generator and to the runtime system
3242       so that code object name strings are directly inserted into code
3243       objects at code generation time.  The only business the runtime system
3244       has with this is now to read the name strings on occasions.
3245       (The encoding of the name string has also changed somewhat.)
3246    
3247    2. CM now implements a simple "set calculus" for specifying export lists.
3248       In particular, it is now possible to refer to the export lists of
3249       other libraries/groups/sources and form unions as well as differences.
3250       See the latest CM manual for details.
3251    
3252    3. An separate notion of "proxy" libraries has again be eliminated from
3253       CM's model.  (Proxy libraries are now simply a special case of using
3254       the export list calculus.)
3255    
3256    4. Some of the existing libraries now take advantage of the new set
3257       calculus.
3258       (Notice that not all libraries have been converted because some
3259       of the existing .cm-files are supposed to be backward compatible
3260       with 110.0.x.)
3261    
3262    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
3263       "General.exnMessage"!  The former relies on a certain hook to be
3264       initialized, and that often does not happen in the stand-alone case.)
3265    
3266    ----------------------------------------------------------------------
3267    Name: Lal George
3268    Date: 2001/11/21  13:56:18 EST
3269    Tag: george-2001121-pseudo-ops
3270    Description:
3271    
3272      Implemented a complete redesign of MLRISC pseudo-ops. Now there
3273      ought to never be any question of incompatabilities with
3274      pseudo-op syntax expected by host assemblers.
3275    
3276      For now, only modules supporting GAS syntax are implemented
3277      but more should follow, such as MASM, and vendor assembler
3278      syntax, e.g. IBM as, Sun as, etc.
3279    
3280    ----------------------------------------------------------------------
3281    Name: Matthias Blume
3282    Date: 2001/11/14 11:52:00 EST
3283    Tag: blume-20011114-srcname
3284    Description:
3285    
3286    1. Routed the name of the current source file to mlriscgen where it
3287       should be directly emitted into the code object.  (This last part
3288       is yet to be done.)
3289    
3290    2. Some cleanup of the pgraph code to make it match the proposal that
3291       I put out the other day.  (The proposal notwithstanding, things are
3292       still in flux here.)
3293    
3294    ----------------------------------------------------------------------
3295    Name: Lal George
3296    Date: 2001/11/14 09:44:04 EST
3297    Tag:
3298    Description:
3299    
3300      Fix for a backpatching bug reported by Allen.
3301    
3302      Because the boundary between short and long span-dependent
3303      instructions is +/- 128, there are an astounding number of
3304      span-dependent instructions whose size is over estimated.
3305    
3306      Allen came up with the idea of letting the size of span
3307      dependent instructions be non-monotonic, for a maxIter
3308      number of times, after which the size must be monotonically
3309      increasing.
3310    
3311      This table shows the number of span-dependent instructions
3312      whose size was over-estimated as a function of maxIter, for the
3313      file Parse/parse/ml.grm.sml:
3314    
3315         maxIter            # of instructions:
3316            10                      687
3317            20                      438
3318            30                      198
3319            40                        0
3320    
3321      In compiling the compiler, there is no significant difference in
3322      compilation speed between maxIter=10 and maxIter=40. Actually,
3323      my measurements showed that maxIter=40 was a tad faster than
3324      maxIter=10! Also 96% of the  files in the compiler reach a fix
3325      point within 13 iterations, so fixing maxIter at 40, while high,
3326      is okay.
3327    
3328    ----------------------------------------------------------------------
3329    Name: Matthias Blume
3330    Date: 2001/10/31 15:25:00 EST
3331    Tag: blume-20011031-pgraph
3332    Description:
3333    
3334    CKIT:
3335    * Changed the "Function" constructor of type Ast.ctype to carry optional
3336      argument identifiers.
3337    * Changed the return type of TypeUtil.getFunction accordingly.
3338    * Type equality ignores the argument names.
3339    * TypeUtil.composite tries to preserve argument names but gives up quickly
3340      if there is a mismatch.
3341    
3342    installation script:
3343    * attempts to use "curl" if available (unless "wget" is available as well)
3344    
3345    CM:
3346    * has an experimental implementation of "portable graphs" which I will
3347      soon propose as an implementation-independent library format
3348    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
3349    
3350    NLFFI-LIB:
3351    * some cleanup (all cosmetic)
3352    
3353    NLFFIGEN:
3354    * temporarily disabled the mechanism that suppresses ML output for
3355      C definitions whose identifiers start with an underscore character
3356    * generate val bindings for enum constants
3357    * user can request that only one style (light or heavy) is being used;
3358      default is to use both (command-line arguments: -heavy and -light)
3359    * fixed bug in handling of function types involving incomplete pointers
3360    * generate ML entry points that take record arguments (i.e., using
3361      named arguments) for C functions that have a prototype with named
3362      arguments
3363      (see changes to CKIT)
3364    
3365    ----------------------------------------------------------------------
3366    Name: Allen Leung
3367    Date: 2001/10/27 20:34:00 EDT
3368    Tag: leunga-20011027-x86-fast-fp-call
3369    Description:
3370    
3371       Fixed the bug described in blume-20010920-slowfp.
3372    
3373       The fix involves
3374          1. generating FCOPYs in FSTP in ia32-svid
3375          2. marking a CALL with the appropriate annotation
3376    
3377    ----------------------------------------------------------------------
3378    Name: Matthias Blume
3379    Date: 2001/10/16 11:32:00 EDT
3380    Tag: blume-20011016-netbsd
3381    Description:
3382    
3383    Underscore patch from Chris Richards (fixing problem with compiling
3384    runtime system under recent NetBSD).
3385    
3386    ----------------------------------------------------------------------
3387    Name: Allen Leung
3388    Date: 2001/10/12 17:18:32 EDT 2001
3389    Tag: leung-20011012-x86-printflowgraph
3390    Description:
3391    
3392    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
3393    
3394    ----------------------------------------------------------------------
3395    Name: Lal George
3396    Date: 2001/10/11 23:51:34 EDT
3397    Tag: george-20011011-too-many-instrs
3398    Description:
3399    
3400    The representation of a program point never expected to see more
3401    than 65536 instructions in a basic block!
3402    
3403    ----------------------------------------------------------------------
3404    Name: Lal George
3405    Date: 2001/10/09 09:41:37 EDT
3406    Tag: george-20011008-mlrisc-labels
3407    Description:
3408    
3409    Changed the machine description files to support printing of
3410    local and global labels in assembly code, based on host assembler
3411    conventions.
3412    
3413    ----------------------------------------------------------------------
3414    Name: Matthias Blume
3415    Date: 2001/09/25 15:25:00 EDT
3416    Tag: blume-20010925-exninfo
3417    Description:
3418    
3419    I provided a non-hook implementation of exnName (at the toplevel) and
3420    made the "dummy" implementation of exnMessage (at the toplevel) more
3421    useful: if nothing gets "hooked in", then at least you are going to
3422    see the exception name and a message indicating why you don't see more.
3423    
3424    [For the time being, programs that need exnMessage and want to use
3425    ml-build should either use General.exnMessage (strongly recommended) or
3426    refer to structure General at some other point so that CM sees a
3427    static dependency.]
3428    
3429    [Similar remarks go for "print" and "use":  If you want to use their
3430    functionality in stand-alone programs generated by ml-build, then use
3431    TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
3432    
3433    ----------------------------------------------------------------------
3434    Name: Matthias Blume
3435    Date: 2001/09/20 17:28:00 EDT
3436    Tag: blume-20010920-slowfp
3437    Description:
3438    
3439    Allen says that x86-fast-fp is not safe yet, so I turned it off again...
3440    
3441    ----------------------------------------------------------------------
3442    Name: Matthias Blume
3443    Date: 2001/09/20 17:20:00 EDT
3444    Tag: blume-20010920-canonicalpaths
3445    Description:
3446    
3447    0. Updated the BOOT file (something that I forgot to do earlier).
3448    
3449    1. Small internal change to CM so that it avoids "/../" in filenames
3450       as much as possible (but only where it is safe).
3451    
3452    2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
3453       that arguments that contain delimiters are passed through correctly.
3454       This change also means that all "special" arguments of the form
3455       @SMLxxx... must come first.
3456    
3457    3. Changed install script to put relative anchor names for tool commands
3458       into pathconfig.
3459    
3460    ----------------------------------------------------------------------
3461    Name: Matthias Blume
3462    Date: 2001/09/18 15:35:00 EDT
3463    Tag: blume-20010918-readme11036
3464    Description:
3465    
3466    Added README files.
3467    
3468    ----------------------------------------------------------------------
3469    Name: Matthias Blume
3470    Date: 2001/09/18 11:45:00 EDT
3471    Tag: Release_110_36 (retag)
3472    Description:
3473    
3474    Fixed mistake in config/preloads. Retagged as 110.36.
3475    
3476    ----------------------------------------------------------------------
3477    Name: Matthias Blume
3478    Date: 2001/09/18 09:40:00 EDT
3479    Tag: Release_110_36_orig (tag changed)
3480    Description:
3481    
3482    New version (110.36).  New bootfiles.
3483    
3484    ----------------------------------------------------------------------
3485    Name: Matthias Blume
3486    Date: 2001/09/14 16:15:00 EDT
3487    Tag: blume-20010914-x86fastfp
3488    Description:
3489    
3490    John committed some changes that Allen made, in particular a (hopefully)
3491    correctly working version of the x86-fp module.
3492    
3493    I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
3494    flag to "true".  Everything seems to compile to a fixpoint ok, and
3495    "mandelbrot" speeds up by about 15%.
3496    
3497    ----------------------------------------------------------------------
3498    Name: Matthias Blume
3499    Date: 2001/09/13 11:20:00 EDT
3500    Tag: blume-20010913-minimal
3501    Description:
3502    
3503    1. Stefan Monnier's patch to fix a miscompilation problem that
3504       was brought to light by John Reppy's work on Moby.
3505    
3506    2. Implemented a minimal "structure Compiler" that contains just
3507       "version" and "architecture".  The minimal version will be
3508       available when the full version is not.  This is for backward-
3509       compatibility with code that wants to test Compiler.version.
3510    
3511    ----------------------------------------------------------------------
3512    Name: Matthias Blume
3513    Date: 2001/08/28 14:03:00 EDT
3514    Tag: blume-20010828-ml-lex
3515    Description:
3516    
3517    Fix for bug 1581, received from Neophytos Michael.
3518    
3519    ----------------------------------------------------------------------
3520    Name: Matthias Blume
3521    Date: 2001/08/27 11:20:00 EDT
3522    Tag: blume-20010827-readme11035
3523    Description:
3524    
3525    Fleshed out the README file for 110.35.
3526    
3527    ----------------------------------------------------------------------
3528    Name: Matthias Blume
3529    Date: 2001/08/24 17:10:00 EDT
3530    Tag: Release_110_35
3531    Description:
3532    
3533    New version number (110.35).  New bootfiles.
3534    
3535    ----------------------------------------------------------------------
3536    Name: Lal George
3537    Date: 2001/08/24 13:47:18 EDT 2001
3538    Tag: george-20010824-MLRISC-graphs
3539    Description:
3540    
3541     removed clusters from MLRISC completely and replaced with graphs.
3542    
3543    ----------------------------------------------------------------------
3544    Name: Matthias Blume
3545    Date: 2001/08/23 17:50:00 EDT
3546    Tag: blume-20010823-toplevel
3547    Description:
3548    
3549    - some reorganization of the code that implements various kinds of
3550      environments in the compiler (static, dynamic, symbolic, combined)
3551    - re-implemented the EnvRef module so that evalStream works properly
3552      (if the stream contains references to "use", "CM.make", etc.)
3553    - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
3554    
3555    ----------------------------------------------------------------------
3556    Name: Matthias Blume
3557    Date: 2001/08/20 15:50 EDT
3558    Tag: blume20010820-slipup
3559    Description:
3560    
3561    I forgot to commit a few files.  Here they are...
3562    
3563    ----------------------------------------------------------------------
3564    Name: Matthias Blume
3565    Date: 2001/08/20 15:35:00 EDT
3566    Tag: blume-20010820-debugprof
3567    Description:
3568    
3569    !!!! NEW BOOTFILES !!!!
3570    
3571    This is another round of reorganizing the compiler sources.  This
3572    time the main goal was to factor out all the "instrumentation"
3573    passes (for profiling and backtracing) into their own library.
3574    The difficulty was to do it in such a way that it does not depend
3575    on elaborate.cm but only on elabdata.cm.
3576    
3577    Therefore there have been further changes to both elaborate.cm and
3578    elabdata.cm -- more "generic" things have been moved from the former
3579    to the latter.  As a result, I was forced to split the assignment
3580    of numbers indicating "primtyc"s into two portions: SML-generic and
3581    SML/NJ-specific.  Since it would have been awkward to maintain,
3582    I bit the bullet and actually _changed_ the mapping between these
3583    numbers and primtycs.  The bottom line of this is that you need
3584    a new set of bin- and bootfiles.
3585    
3586    I have built new bootfiles for all architectures, so doing a fresh
3587    checkout and config/install.sh should be all you need.
3588    
3589    The newly created library's name is
3590    
3591        $smlnj/viscomp/debugprof.cm
3592    
3593    and its sources live under
3594    
3595        src/compiler/DebugProf
3596    
3597    ----------------------------------------------------------------------
3598    Name: Matthias Blume
3599    Date: 2001/08/15 17:15:00 EDT
3600    Tag: blume-20010815-compreorg
3601    Description:
3602    
3603    This is a first cut at reorganizing the CM libraries that make up the
3604    core of the compiler.  The idea is to separate out pieces that could
3605    be used independently by tools, e.g., the parser, the typechecker, etc.
3606    
3607    The current status is a step in this direction, but it is not quite
3608    satisfactory yet.  Expect more changes in the future.
3609    
3610    Here is the current (new) organization...
3611    
3612        What used to be $smlnj/viscomp/core.cm is now divided into
3613        six CM libraries:
3614    
3615             $smlnj/viscomp/basics.cm
3616                           /parser.cm
3617                           /elabdata.cm
3618                           /elaborate.cm
3619                           /execute.cm
3620                           /core.cm
3621    
3622        The CM files for these libraries live under src/system/smlnj/viscomp.
3623     &