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 606, Sun Apr 9 23:10:59 2000 UTC revision 1631, Fri Sep 24 21:21:31 2004 UTC
# Line 8  Line 8 
8  The form of an entry should be:  The form of an entry should be:
9    
10  Name:  Name:
11  Date:  Date: yyyy/mm/dd
12  Tag: <post-commit CVS tag>  Tag: <post-commit CVS tag>
13  Description:  Description:
14    
15    ----------------------------------------------------------------------
16    Name: Matthias Blume (blume (at) tti - c (dot) org)
17    Date: 2004/09/24 16:22:00 CDT
18    Tag: blume-20040924-ppc-long-branch
19    Description:
20    
21    Turned message about "emiting long form of branch" off by default.
22    Added a control flag to turn it back on when desired.
23    
24    ----------------------------------------------------------------------
25    Name: Matthias Blume (blume (at) tti - c (dot) org)
26    Date: 2004/09/24 16:05:00 CDT
27    Tag: blume-20040924-rounding
28    Description:
29    
30    Applied patch for setting rounding modes under Mac OS X.  Thanks to
31    Melissa O'Neill for providing the code!
32    
33    ----------------------------------------------------------------------
34    Name: Matthias Blume (blume (at) tti - c (dot) org)
35    Date: 2004/09/23 17:30:00 CDT
36    Tag: blume-20040923-envvars
37    Description:
38    
39    1. Changed definition of type ControlRegistry.registry_tree to
40       include control_info (i.e., the name of the controlling environment
41       variable).
42    
43    2. Added command-line flags -e and -E to print the names of environment
44       variables that can be used to control internal settings.
45       (This uses the new API mentioned in 1.)
46    
47    ----------------------------------------------------------------------
48    Name: Matthias Blume (blume (at) tti - c (dot) org)
49    Date: 2004/09/13 16:50:00 CDT
50    Tag: Release_110_49
51    Description:
52    
53    New working version (110.49).  NEW BOOTFILES!
54    
55    ----------------------------------------------------------------------
56    Name: Matthias Blume
57    Date: 2004/09/13 16:20:00 CDT
58    Tag: blume-20040913-config-mlrisc
59    Description:
60    
61    Put target "mlrisc" back into the default list.
62    (There is no harm in having it, and some users have expressed their
63    wish to have "mlrisc" included by default.)
64    
65    ----------------------------------------------------------------------
66    Name: John Reppy
67    Date: 2004/09/13
68    Tag: jhr-20040913-signals
69    Description:
70    
71    Fixed the signal masking code to properly nest mask/unmask operations
72    on a per-signal basis.
73    
74    ----------------------------------------------------------------------
75    Name: Matthias Blume (blume (at) tti - c (dot) org)
76    Date: 2004/09/08 13:20:00 CDT
77    Tag: blume-20040908-heap-magic
78    Description:
79    
80    Bumped the heap macig number to 0x09082004 to account for the changed
81    layout of the ML frame under MacOS X.
82    
83    ----------------------------------------------------------------------
84    Name: Allen Leung (leunga (at) reservoir (dot) com)
85    Date: 2004/09/03 11:26:00 EST
86    Tag: leunga-20040903-cygwin-install
87    Description:
88    
89       Added a patch to _arch-n-opsys to enable the Cygwin runtime.
90    The Cygwin runtime is turned on by setting the environment
91    variable SMLNJ_CYGWIN_RUNTIME to 1.
92    
93    ----------------------------------------------------------------------
94    Name: Matthias Blume (blume (at) tti - c (dot) org)
95    Date: 2004/08/31 17:15:00 CDT
96    Tag: blume-20040831-core
97    Description:
98    
99    Added some exports to src/compiler/core.cm upon request by J. Joemann.
100    
101    ----------------------------------------------------------------------
102    Name: Matthias Blume (blume (at) tti - c (dot) org)
103    Date: 2004/08/30 17:55:00 CDT
104    Tag: blume-20040830-installer
105    Description:
106    
107    Upon request by Johannes Joemann:
108    
109    - improved ML code of installer to fall back to coping when renaming
110      fails (i.e., when source and target are on different file systems);
111      the code compiles but has yet to be tested in anger
112    - removed mlrisc from list of default targets (config/targets)
113    
114    ----------------------------------------------------------------------
115    Name: Matthias Blume (blume (at) tti - c (dot) org)
116    Date: 2004/08/27 17:20:00 CDT
117    Tag: blume-20040827-ptreql
118    Description:
119    
120    Added ptreql primop to structure InlineT (upon request from Larry
121    Paulson).
122    
123    ----------------------------------------------------------------------
124    Name: Allen Leung (leunga (at) reservoir (dot) com)
125    Date: 2004/08/15 21:21:00 EST
126    Tag: leunga-110_48-udgraph
127    Description:
128    
129    Another bug fix from Carl Hauser:
130    
131    diff /net/niflab/smlnj48/src/MLRISC/graphs/udgraph.sml udgraph.sml
132    > 48c48
133    > <              | rmv((e as (k,_))::es,L) = rmv(es,if k = i then es else
134    > e::L)
135    > ---
136    > >              | rmv((e as (k,_))::es,L) = rmv(es,if k = i then L else e::L)
137    > Without this, any deletion of an edge in an undirected graph does severe
138    > violence to the graph.
139    
140    ----------------------------------------------------------------------
141    Name: Allen Leung (leunga (at) reservoir (dot) com)
142    Date: 2004/08/10 23:35:00 EST
143    Tag: leunga-110_48-ppc
144    Description:
145    
146       The IBM/MacOS syntax switch on PPC was incorrectly swapped.
147    
148    ----------------------------------------------------------------------
149    Name: Matthias Blume (blume (at) tti - c (dot) org)
150    Date: 2004/08/10 12:00:00 CDT
151    Tag: Release_110_48
152    Description:
153    
154    New working version (110.48).  NEW BOOTFILES!
155    
156    ----------------------------------------------------------------------
157    Name: Allen Leung (leunga (at) reservoir (dot) com)
158    Date: 2004/08/09 12:21:00 EST
159    Tag: leunga-110_47-dijsktra
160    Description:
161    
162    Bug fix from Carl Hauser:
163    
164    single_source_shortest_paths in dijkstra.sml was observed to get wrong
165    answers (by comparing to single_source_shortest_paths in bellman-ford.sml).
166    
167    The problem is that following the expression A.update(dist,s,Num.zero)
168    it is necessary to update the priority queue using Q.decreaseWeight(Q,s).
169    
170    ----------------------------------------------------------------------
171    Name: Matthias Blume (blume (at) tti - c (dot) org)
172    Date: 2004/08/06 18:10:00 CDT
173    Tag: blume-20040806-cmdline
174    Description:
175    
176    Fiddled with handling of command-line options:
177    
178      * sml now quits after processing the command line
179        if -H, -S, -h<n>, or -s<n> appears as the last
180        command-line argument
181      * a new option -q terminates the session when encountered on
182        the command line; subsequent arguments will be ignored
183      * bug fixes: short (erroneous) arguments are no longer ignored
184        completely
185    
186    ----------------------------------------------------------------------
187    Name: Allen Leung (leunga (at) reservoir (dot) com)
188    Date: 2004/08/04 18:17:00 EST
189    Tag: leunga-110_47-ppc-ibm-asm
190    Description:
191    
192      - Added minimal IBM assembly syntax support for PowerPC.
193    
194      - Cygwin: manually changed the file cygwin.def.  Some exported symbols have
195        been altered in the runtime.  We need an automatic way to keep the file
196        in sync.
197    
198    ----------------------------------------------------------------------
199    Name: Matthias Blume (blume (at) tti - c (dot) org)
200    Date: 2004/08/04 14:00:00 CDT
201    Tag: Release_110_47
202    Description:
203    
204    New working version (110.47).  NEW BOOTFILES!
205    
206    ----------------------------------------------------------------------
207    Name: Matthias Blume (blume (at) tti - c (dot) org)
208    Date: 2004/08/03 14:25:00 CDT
209    Tag: blume-20040803-callingconv
210    Description:
211    
212    Added low-level support for choosing C calling conventions by
213    twiddling the type of rawccall.  (See
214    src/compiler/Semant/types/cproto.sml for details.)
215    
216    ----------------------------------------------------------------------
217    Name: Matthias Blume (blume (at) tti - c (dot) org)
218    Date: 2004/08/02 15:55:00 CDT
219    Tag: blume-20040802-backout
220    Description:
221    
222    Backed out of change to win32-filesys.c.  The earlier patch to
223    get_file_time caused CM to produce files with the wrong time stamp.
224    
225    ----------------------------------------------------------------------
226    Name: Matthias Blume (blume (at) tti - c (dot) org)
227    Date: 2004/08/02 14:45:00 CDT
228    Tag: blume-20040802-nlffi-win32
229    Description:
230    
231    Added NLFFI support for Win32, adapted from a patch provided by David
232    Hansel.  This is currently completely untested.  Also, the issue
233    concerning stdcall vs. ccall is still unresolved.
234    
235    ----------------------------------------------------------------------
236    Name: Matthias Blume (blume (at) tti - c (dot) org)
237    Date: 2004/07/30 17:55:00 CDT
238    Tag: blume-20040730-various
239    Description:
240    
241    Gearing up towards 110.47...
242    
243    - various minor bugfixes to ml-nlffigen
244    - a beginning of a manual for nlffi
245    
246    - eliminated 'export name=value' in config/install.sh as this does
247      not work with certain versions of /bin/sh
248      (Thanks to David King at Motorola for catching this.)
249    
250    - several bugfixes provided or suggested by David Hansel at Reactive Systems:
251      - added a test for tm==NULL to gmtime.c and localtime.c
252      - applied patch for incorrect GetFileTime under win32
253      - toSeconds -> toMilliseconds in Win32/win32-process.sml
254    
255    ----------------------------------------------------------------------
256    Name: Matthias Blume (blume (at) tti - c (dot) org)
257    Date: 2004/07/21 18:20:00 CDT
258    Tag: blume-20040721-nlffigen
259    Description:
260    
261    - Fixed minor issue in ml-nlffigen: Now generate structure T_foo
262      for a typedef to an incomplete type, but leave out the "typ" member.
263      (This is just for consistency.)
264    - Started to produce what is supposed to become better (i.e., comprehensive)
265      documentation of what ml-nlffigen does and produces.
266    
267    ----------------------------------------------------------------------
268    Name: Matthias Blume (blume (at) tti - c (dot) org)
269    Date: 2004/07/14 16:25:00 CDT
270    Tag: blume-20040714-union
271    Description:
272    
273    Added C_UNION to c-calls/c-types.sml and updated the machinery
274    (ml-nlffigen, cproto.sml) that conveys C function interface
275    information to the code generator.
276    
277    However, the actual architecture-specific implementation of function
278    arguments and results that are C unions is still not implemented.
279    
280    ----------------------------------------------------------------------
281    Name: Allen Leung (leunga (at) reservoir (dot) com)
282    Date: 2004/07/14 14:38:00 EST
283    Tag: leunga-110_46_1-ppc-lwzu
284    Description:
285    
286        Added these instructions to the PowerPC architecture:
287        LBZU(X), LHZU(X), LWZU(X),
288        STWU(X), STFDU, STFSU
289    
290        etc...
291    
292        Note: I haven't added their instruction encoding into the description.
293    
294    ----------------------------------------------------------------------
295    Name: Allen Leung (leunga (at) reservoir (dot) com)
296    Date: 2004/07/13 15:04:00 EST
297    Tag: leunga-110_46_1-ppc-lwarx
298    Description:
299    
300       Added the two instructions LWARX and STWCX to the PowerPC
301    instruction set.
302    
303       A (untested) rewrite of loop-structure.sml.  The old version
304    is completely broken.
305    
306    ----------------------------------------------------------------------
307    Name: Matthias Blume (blume (at) tti - c (dot) org)
308    Date: 2004/07/13 13:50:00 CDT
309    Tag: blume-20040713-nlffi
310    Description:
311    
312    - use paramAlloc to report c-calls with too many arguments
313      (for PPC version where parameter area is pre-allocated)
314    - added ccall_maxargspace to machspec (to implement the above)
315    - made "make" commend in CM's "make" tool configurable
316    - added option (default: on) for passing the name of the SML/NJ's "bin"
317      directory to "make"; the call looks like this:
318    
319         make <options> SMLNJ_BINDIR=<dir> <target>
320    
321      This can be used by the Makefile to, e.g., pick the "right" version
322      of ml-nlffigen.
323    - minor code tweaks
324    
325    ----------------------------------------------------------------------
326    Name: Matthias Blume (blume (at) tti - c (dot) org)
327    Date: 2004/07/12 22:50:00 CDT
328    Tag: blume-110_46_1-macosx-nlffi
329    Description:
330    
331    NLFFI under Mac OS X now working (sort of).  This is largely untested,
332    though.
333    
334    Note:  1. You have to make a new, clean build of the runtime system.
335           2. There are new BOOTFILES, you have to use them!
336              (Doing the bootstrap process yourself would be *very* painful!
337               If you absolutely have to do it, build the system under
338               a different architecture and then cross-compile.)
339    
340    Version bumped to 110.46.1 to account for runtime data format changes.
341    
342    ----------------------------------------------------------------------
343    Name: Matthias Blume (blume (at) tti - c (dot) org)
344    Date: 2004/06/18 14:30:00 CDT
345    Tag: blume-20040618-unix
346    Description:
347    
348    Changed the implementation of structure Unix so that the same stream
349    is returned every time one of the {text,bin}{In,Out}streamOf functions
350    is invoked on the same proc.  This is not what the spec currently
351    says -- although IMO it arguably should.  (See discussion below.)
352    
353    ----------------------------------------------------------------------
354    Name: Matthias Blume (blume (at) tti - c (dot) org)
355    Date: 2004/06/17 18:15:00 CDT
356    Tag: Release_110_46
357    Description:
358    
359    New working version (110.46).  NEW BOOTFILES!
360    
361    ----------------------------------------------------------------------
362    Name: Matthias Blume (blume (at) tti - c (dot) org)
363    Date: 2004/06/17 17:20:00 CDT
364    Tag: blume-20040617-timer-unix
365    Description:
366    
367    Changed the interface of structures Timer and Unix to match the most
368    recent Basis spec.
369    
370    In the case of Unix there still seems to be an open/weird issue:
371    
372       The {text,bin}{In,Out}streamOf functions are supposed to create
373       fresh streams whenever they are called -- as opposed to have them
374       return the same stream every time.  This design is supposed to
375       prevent space leaks caused by proc values hanging on to streams.
376    
377       The reap function, on the other hand, is supposed to close the
378       streams.  This cannot be done without having a handle on the
379       stream in proc after all...
380    
381    I took the liberty to implement the following stopgap solution:
382    
383      The proc value hangs on to the most recently created stream(s).
384      Reap closes those.  If either or both of the two streams hadn't
385      been created at all yet, then reap will close the corresponding
386      file descriptors directly.
387    
388    PS: I don't understand the original space leak argument anymore.  If
389    a proc hangs on to the imperative stream, then I/O operations on those
390    will advance the state of the cached stream and avoid the space leak.
391    
392    ----------------------------------------------------------------------
393    Name: Matthias Blume (blume (at) tti - c (dot) org)
394    Date: 2004/05/28 16:45:00 CDT
395    Tag: blume-20040528-basis
396    Description:
397    
398    Added signature PACK_REAL and exported functor PrimIO.
399    
400    ----------------------------------------------------------------------
401    Name: Matthias Blume (blume (at) tti - c (dot) org)
402    Date: 2004/05/25 16:00:00 CDT
403    Tag: blume-20040525-group-owner
404    Description:
405    
406    CM now ignores (but still accepts) the "owner" information in group
407    descriptions.  The owner of a group is its next enclosing
408    library. Each group must have a unique owner.  (There is a virtual
409    "toplevel" library that owns groups which are not nested within a real
410    library.)  Previously, each group had to explicitly declare its owner,
411    and CM would check that such a declaration is correct.  The new scheme
412    is to have CM check that for each group there is precisely one owning
413    library.
414    
415    The advantage of the new scheme is that the programmer no longer needs
416    to maintain the somewhat annoying owner information.  The downside is
417    that CM cannot enforce the ownership rule across multiple runs of
418    CM.make.  Fortunately, enclosing the same group in two different
419    libraries A and B which are not part of the same program does not
420    cause real problems.
421    
422    ----------------------------------------------------------------------
423    Name: Matthias Blume (blume (at) tti - c (dot) org)
424    Date: 2004/05/20 16:00:00 CDT
425    Tag: blume-20040520-win32
426    Description:
427    
428    Made the win32 version work again.  (Strangely, a misplaced comma had
429    slipped into win32-process.c which prevented the runtime from being
430    compiled correctly.)
431    
432    Also, included a minor addition to ml-build.bat analogous to what was
433    done in blume-20040519-ml-build.
434    
435    ----------------------------------------------------------------------
436    Name: Matthias Blume (blume (at) tti - c (dot) org)
437    Date: 2004/05/19 22:10:00 CDT
438    Tag: blume-20040519-ml-build
439    Description:
440    
441    Arranged for ml-build to clean up after itself a little bit better.
442    The script generates a temporary SML source file and compiles it using
443    CM, so CM generates metadata (GUID, SKEL, objectfile) for it.  It now
444    gets rid of those at the end, so they don't accumulate under .cm.
445    
446    This required a minor change to install.sh because the name of the
447    metadata directory (default: .cm) is actually configurable at
448    installation time.
449    
450    ----------------------------------------------------------------------
451    Name: Matthias Blume (blume (at) tti - c (dot) org)
452    Date: 2004/05/18 15:50:00 CDT
453    Tag: blume-20040518-mkreader
454    Description:
455    
456    Added Posix.IO.mk{Bin,Text}{Reader,Writer} by lifting their respective
457    implementations from internal modules PosixBinPrimIO and PosixTextPrimIO.
458    
459    ----------------------------------------------------------------------
460    Name: Matthias Blume (blume (at) tti - c (dot) org)
461    Date: 2004/05/11 14:35:00 CDT
462    Tag: blume-20040511-win32sock
463    Description:
464    
465    Added previously missing support for many socket-related functions
466    under win32.  Thanks to David Hansel <hansel@reactive-systems.com>
467    for the voluminous patch!
468    
469    (I have not tested this patch under win32 yet.)
470    
471    Here is David's e-mail:
472    
473    Hi,
474    
475    Attached to this email you find a diff against sml/nj 110.45
476    that will enable socket support under Windows.
477    
478    To apply the patch (using unix or cygwin)
479    1) gunzip runtime.diff.gz
480    2) "cd" into "src/runtime" in the source tree of a fresh
481       110.45 installation.
482    3) patch -p 1 < [your/path/to]runtime.diff
483    
484    The code compiles fine but has NOT yet been extensively tested.
485    I only ran a few tests for basic socket client functionality
486    (which worked fine).  Especially the functions that use ioctl
487    are not tested at all and might not work (see below).
488    
489    I implemented this since we want to move to a newer version of sml/nj
490    but need socket support in order to use it.  This is the first time I
491    even had a look at the sml/nj source,  so please review my changes
492    before making this part of the distribution!  Here are a few issues
493    that I think might be better for someone to solve who is more
494    familiar with the sml/nj source (and socket programming):
495    
496    - getnetbyaddr.c and getnetbyname.c will raise a "not implemented"
497      exception since I could not figure out what the windows equivalent
498      of these functions is
499    
500    - In sockets-osdep.h there are a some #include statements that are
501      only used in a few files that include sockets-osdep.h
502    
503    - In smlnj-sock-lib.c, function init_fn() calls WSAStartup() but
504      does not process its return value since I don't know how to
505      report an error upwards.
506    
507    - It would probably be good to have a call to WSACleanup() when
508      the library is unloaded (if there is such a possibility).
509      Otherwise I think Windows will take care of this automatically
510      when the process finishes.
511    
512    - I used ioctlsocket() as a replacement for ioctl() but I have
513      no idea if that is actually the proper replacement on Windows.
514    
515    - All these issues are marked in the code by "FIXME" comments.
516    
517    We use sml/nj extensively in our products and are quite happy
518    with it.  I hope this contribution will help you.
519    
520    Keep up the good work!
521    
522    David
523    
524    ----------------------------------------------------------------------
525    Name: Matthias Blume (blume (at) tti - c (dot) org)
526    Date: 2004/05/11 14:20:00 CDT
527    Tag: blume-20040511-installml
528    Description:
529    
530    Fixed two bugs in installml script.  (Thanks to Vesa A. Norrman for
531    the patch.)
532    
533    ----------------------------------------------------------------------
534    Name: Matthias Blume (blume (at) tti - c (dot) org)
535    Date: 2004/05/11 14:05:00 CDT
536    Tag: blume-20040511-nlffi-netbsd
537    Description:
538    
539    Added support for nlffi under netbsd.  (Thanks to Vesa A. Norrman for
540    the patch.)
541    
542    ----------------------------------------------------------------------
543    Name: Matthias Blume (blume (at) tti - c (dot) org)
544    Date: 2004/05/11 12:05:00 CDT
545    Tag: blume-20040511-exports
546    Description:
547    
548    As per request by Adam Chlipala <adam@hcoop.net>, extended various
549    export lists in compiler-related .cm-files.
550    
551    ----------------------------------------------------------------------
552    Name: Matthias Blume (blume (at) tti - c (dot) org)
553    Date: 2004/05/11 11:35:00 CDT
554    Tag: blume-20040511-allsource
555    Description:
556    
557    The installer now honors the "src-smlnj" target again, although its meaning
558    has changed from "all sources required for the compiler" to "all sources
559    the installer knows about".  In other words, if you enable "src-smlnj"
560    in the "targets" file, then the installer will pull in sources for
561    everything.  (Notice that this refers to source code only.  Compiled
562    code is still only installed for modules that were requested explicitly
563    or which are required for other modules that were requested explicitly.)
564    
565    ----------------------------------------------------------------------
566    Name: Matthias Blume (blume (at) tti - c (dot) org)
567    Date: 2004/04/23 17:40:00 CDT
568    Tag: blume-20040423-ieee-scan
569    Description:
570    
571    Fixed IEEEReal.scan (and .fromString) so that if there is an overflow
572    in the exponent calculation we get INF or ZERO (depending on the mantissa
573    and the sign of the exponent).
574    
575    ----------------------------------------------------------------------
576    Name: Matthias Blume (blume (at) tti - c (dot) org)
577    Date: 2004/04/23 10:40:00 CDT
578    Tag: blume-20040423-ml-build
579    Description:
580    
581    The ml-build script now terminates with a non-0 status when something
582    goes wrong.
583    
584    ----------------------------------------------------------------------
585    Name: Matthias Blume (blume (at) tti - c (dot) org)
586    Date: 2004/04/22 16:35:00 CDT
587    Tag: blume-20040422-Option
588    Description:
589    
590    Made exception Option to be the same as exception Option.Option
591    (as it should be).
592    
593    ----------------------------------------------------------------------
594    Name: Allen Leung (leunga (at) reservoir (dot) com)
595    Date: 2004/03/19 14:40:00 EST
596    Tag: leunga-20040319-cygwin-nlffi
597    Description:
598    
599    Fixed the runtime so that ml-nlffi-lib runs on the cygwin version
600    of SML/NJ.  The problem is that
601    
602       lib = dlopen(NULL, ...)
603       f   = dlsym(lib, "malloc");
604    
605    does not work on Windows unless we explicitly export symbols
606    such as 'malloc' during linking.  We fixed this by explicitly
607    exporting the required symbols with the magic gcc incantation:
608    
609        -Wl,--export-all cygwin.def
610    
611    where cygwin.def is a file containing all the symbols that
612    we wish to export.
613    
614    I suspect this is a Windows problem and we'll have to
615    do the same (somehow with windows compilers) when
616    we build the native win32 version with the system
617    calls LoadLibrary/GetProcAddress.
618    
619    ----------------------------------------------------------------------
620    Name: Matthias Blume (blume (at) tti - c (dot) org)
621    Date: 2004/03/04 16:35:00 CST
622    Tag: blume-20040304-intinf-fmt
623    Description:
624    
625    Fixed problem with IntInf.fmt (sign would show up on the right instead
626    of on the left for BIN, OCT, and HEX).
627    
628    ----------------------------------------------------------------------
629    Name: Matthias Blume (blume (at) tti - c (dot) org)
630    Date: 2004/03/04 11:25:00 CST
631    Tag: blume-20040304-symlinks
632    Description:
633    
634    Fixed problem with installer script (unix only) where bin/ml-yacc and
635    friends pointed (via symlinks) to absolute locations instead of just
636    .run-sml.  This was reported by Vesa A Norrman.
637    
638    ----------------------------------------------------------------------
639    Name: Matthias Blume (blume (at) tti - c (dot) org)
640    Date: 2004/02/13 14:50:00 CST
641    Tag: Release_110_45
642    Description:
643    
644    New working version (110.45). New bootfiles.
645    
646    ----------------------------------------------------------------------
647    Name: Matthias Blume (blume (at) tti - c (dot) org)
648    Date: 2004/01/26 15:15:15 CST
649    Tag: blume-20040126-toplevel
650    Description:
651    
652    Improved handling of exceptions at the interactive toplevel.
653    
654    ----------------------------------------------------------------------
655    Name: Matthias Blume (blume (at) tti - c (dot) org)
656    Date: 2004/01/26 11:25:00
657    Tag: blume-20040126-app
658    Description:
659    
660    Type of top-level "app" corrected.
661    Added code for setting vp_limitPtrMask to Win32-specific runtime.
662    
663    ----------------------------------------------------------------------
664    Name: Matthias Blume (blume (at) tti - c (dot) org)
665    Date: 2003/11/18 17:10 CST
666    Tag: blume-20031118-basis-fiddle
667    Description:
668    
669    - changed Timer interface to what might become the spec
670    - POSIX_FLAGS -> BIT_FLAGS according to spec
671    - some other minor discrepancies wrt. spec eliminated
672    
673    ----------------------------------------------------------------------
674    Name: Matthias Blume (blume (at) tti - c (dot) org)
675    Date: 2003/11/06 12:00:00 CST
676    Tag: Release_110_44
677    Description:
678    
679    New working version (110.44). New bootfiles.
680    
681    ----------------------------------------------------------------------
682    Name: Matthias Blume (blume (at) tti - c (dot) org)
683    Date: 2003/11/04 11:50:00 CST
684    Tag: blume-20031104-move-libraries
685    Description:
686    
687    Eliminated the "dont_move_libraries" directive in config/targets.
688    (The mechanism was broken and could not be fixed easily.  Moreover,
689    there does not seem to be any reason not to move all libraries into
690    lib during installation.  I originally implemented this directive as a
691    backward-compatibility feature when I first introduced the new CM.
692    Now that things have been stable for a long time and going back to the
693    old CM is not an option, there is no reason to keep it around.)
694    
695    ----------------------------------------------------------------------
696    Name: Matthias Blume (blume (at) tti - c (dot) org)
697    Date: 2003/11/03 16:00:00 CST
698    Tag: blume-20031103-installdir
699    Description:
700    
701    Made installer honor INSTALLDIR variable again.  (Thanks to Chris
702    Richards for pointing out the problem and providing the solution.)
703    
704    ----------------------------------------------------------------------
705    Name: Matthias Blume (blume (at) tti - c (dot) org)
706    Date: 2003/10/01 17:05:00 CDT
707    Tag: blume-20031001-lal-mlrisc
708    Description:
709    
710    MLRISC bug fix from Lal.
711    
712    ----------------------------------------------------------------------
713    Name: Matthias Blume (blume (at) tti - c (dot) org)
714    Date: 2003/09/30 16:10:00 CDT
715    Tag: blume-20030930-primio-bat
716    Description:
717    
718    1. Added openVector, nullRd, and nullWr to PRIM_IO.
719    2. Improved .bat files (for Win32 port) to make things work under Win95.
720       (thanks to Aaron S. Hawley for this one)
721    
722    ----------------------------------------------------------------------
723    Name: Matthias Blume (blume (at) tti - c (dot) org)
724    Date: 2003/09/26 16:05:00 CDT
725    Tag: blume-20030926-wrappriv
726    Description:
727    
728    Added missing wrapper for privilege "primitive" in $smlnj/viscomp/core.cm.
729    
730    ----------------------------------------------------------------------
731    Name: Matthias Blume (blume (at) tti - c (dot) org)
732    Date: 2003/09/26 15:00:00 CDT
733    Tag: blume-20030926-110_43_3
734    Description:
735    
736    - additional cleanup
737    - version number bump, NEW BOOTFILES
738    
739    ----------------------------------------------------------------------
740    Name: Matthias Blume (blume (at) tti - c (dot) org)
741    Date: 2003/09/26 12:00:00 CDT
742    Tag: blume-20030926-ppautoload
743    Description:
744    
745    I modified the read-eval-print loop so that the autoloader gets
746    invoked whenever the prettyprinter tries to look up a symbol that
747    is not currently defined in the toplevel environment but which
748    appears in CM's autoload registry.  As a result, we see far fewer of
749    those ?.Foo.Bar.xxx names in the prettyprinter's output.
750    
751    In addition to this I tried to clean up some pieces of the Basis
752    implementation (e.g., Socket, Word8Array) in order to prevent other
753    instances of these ?.Foo.Bar.xxx names from being printed.
754    
755    The mechanism that picks names for types still needs some work, though.
756    (Right now it seems that if there is a type A.t which is defined to
757    be B.u, but B is unavailable at toplevel, then A.t gets printed as
758    "?.B.u" although the perhaps more sensible solution would be to use
759    "A.t" in this case.  In other words, the prettyprinter should follow
760    a chain of DEFtycs not farther than there are corresponding toplevel
761    names in the current environment.)
762    
763    ----------------------------------------------------------------------
764    Name: Matthias Blume (blume (at) tti - c (dot) org)
765    Date: 2003/09/24 16:31:00 CDT
766    Tag: blume-20030924-installer
767    Description:
768    
769    Another installer tweak:  All the ML code for the installer is now
770    compiled during CMB.make and put into a little library called
771    $smlnj/installer.cm.  The installation then simply invokes
772    
773       sml -m $smlnj/installer.cm
774    
775    and everything happens automagically.
776    
777    Win32:  ML code senses value of environment variable SMLNJ_HOME.
778    Unix:   ML code senses values of environment variables ROOT, CONFIGDIR,
779            and BINDIR.
780    
781    The new scheme guarantees that the ML code responsible for the installation
782    is in sync with the APIs of the main system.  Also, the installer is
783    somewhat faster because the installer script is precompiled.
784    
785    ----------------------------------------------------------------------
786    Name: Matthias Blume (blume (at) tti - c (dot) org)
787    Date: 2003/09/24 15:35:00 CDT
788    Tag: blume-20030924-synsock
789    Description:
790    
791    Added a signature SYNCHRONOUS_SOCKET to basis.cm.  This is like SOCKET
792    but excludes all non-blocking operations.  Defined SOCKET (in Basis)
793    and CML_SOCKET in terms of SYNCHRONOUS_SOCKET.  Removed superfluous
794    implementations of non-blocking operations from CML's Socket
795    structure.
796    
797    ----------------------------------------------------------------------
798    Name: Matthias Blume (blume (at) tti - c (dot) org)
799    Date: 2003/09/24 15:10:05 CDT
800    Tag: blume-20030924-sockets
801    Description:
802    
803    1. Fixed SOCKET API and implementation to match Basis spec.
804       This required changing the internal representation of sockets to one
805       that remembers (for each socket file descriptor) whether it is currently
806       blocking or non-blocking.  This state is maintained lazily (i.e., a system
807       call is made only if the state actually needs to change).
808    
809    2. OS-specific details of sockets were moved into separate files, thus
810       making it possible to unify the bulk of the socket implementations
811       between Unix and Win32.
812    
813    3. CML's socket API changed accordingly.
814       (Note that we need to remove non-blocking functions from this API
815       since they are redundant in the case of CML!)
816    
817    4. CML's socket implementation now makes use of non-blocking functions
818       provided by Basis, thus removing all OS-dependent code from this part
819       of CML.
820    
821    5. Changed Real64.precision from 52 to 53.  Minor cleanup in Real64 code.
822    
823    ----------------------------------------------------------------------
824    Name: Matthias Blume (blume (at) tti - c (dot) org)
825    Date: 2003/09/22 12:10:00 CDT
826    Tag: blume-20030922-110_43_2
827    Description:
828    
829    Made a new interim version and bootfiles for developer's bootstrapping
830    convenience.
831    
832       110.43.2 -- NEW BOOTFILES
833    
834    ----------------------------------------------------------------------
835    Name: Matthias Blume (blume (at) tti - c (dot) org)
836    Date: 2003/09/19 15:55:00 CDT
837    Tag: blume-20030919-cmdir
838    Description:
839    
840    1. new-install.sh -> install.sh
841    2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
842    3. tweaked installer so that another name instead of .cm can be chosen
843       at install time (by setting the CM_DIR_ARC environment variable
844       during installation); once installation is complete, the name is
845       fixed
846    
847    ----------------------------------------------------------------------
848    Name: Matthias Blume (blume (at) tti - c (dot) org)
849    Date: 2003/09/18 16:00:00 CDT
850    Tag: blume-20030918-110_43_1
851    Description:
852    
853    Made a new interim version and bootfiles for developer's bootstrapping
854    convenience.
855    
856       110.43.1 -- NEW BOOTFILES
857    
858    ----------------------------------------------------------------------
859    Name: Matthias Blume (blume (at) tti - c (dot) org)
860    Date: 2003/09/18 15:20:00 CDT
861    Tag: blume-20030918-misc
862    Description:
863    
864    1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
865       this seems to be controversial at the moment) and used those in
866       Posix.ProcEnv.times.
867    
868    2. Added Time.{from,to}Nanoseconds to Time.
869    
870    3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
871       For example, fromLargeInt never needs to look at more than 3 "big
872       digits" to get its 53 bits of precision.
873    
874    ----------------------------------------------------------------------
875    Name: Matthias Blume (blume (at) tti - c (dot) org)
876    Date: 2003/09/17 16:30:00 CDT
877    Tag: blume-20030917-real32-slices
878    Description:
879    
880    Added an entry to the primitive environment
881    (compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
882    added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
883    
884    Removed some of the "magic" constants in real64.sml and replaced them
885    with code that generates these values from their corresponding
886    integer counterparts.
887    
888    Made all(?) the slice-related changes to the Basis and made everything
889    compile again...
890    
891    ----------------------------------------------------------------------
892    Name: Matthias Blume (blume (at) tti - c (dot) org)
893    Date: 2003/09/15 17:45:00 CDT
894    Tag: blume-20030915-rbase
895    Description:
896    
897    Fixed bug in Real.fromLargeInt.
898    
899    ----------------------------------------------------------------------
900    Name: Matthias Blume (blume (at) tti - c (dot) org)
901    Date: 2003/09/13 18:11:00 CDT
902    Tag: blume-20030913-libinstall
903    Description:
904    
905    Minor bugfix in config/libinstall (set anchor with path to
906    standalone tool after installing it, otherwise libraries that
907    need ml-lex or ml-yacc won't compile the first time the installer
908    runs).
909    
910    ----------------------------------------------------------------------
911    Name: Matthias Blume (blume (at) tti - c (dot) org)
912    Date: 2003/09/12 11:45:00 CDT
913    Tag: blume-20030912-various
914    Description:
915    
916     - fixed bug in Real.toLargeInt
917     - fixed bug in Posix.ProcEnv.times
918     - changed inputLine functions to return an option
919     - minor installer improvements / bugfixes
920     - changed default @SMLalloc parameter for x86/celeron to 64k
921    
922    ----------------------------------------------------------------------
923    Name: Matthias Blume (blume (at) tti - c (dot) org)
924    Date: 2003/09/09 22:00:00 CDT
925    Tag: Release_110_43
926    Description:
927    
928        New working release 110.43.  New bootfiles.
929    
930    ----------------------------------------------------------------------
931    Name: Matthias Blume (blume (at) tti - c (dot) org)
932    Date: 2003/09/09 19:20:00 CDT
933    Tag: blume-20030909-installer
934    Description:
935    
936    Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
937    Modified config/install.bat to take advantage of it.  Also modified
938    config/install.sh (and called it config/new-install.sh) to take advantage
939    of it on Unix systems.  (The SML code is (supposed to be) platform-
940    independent.)
941    
942    The installer can now install everything under Win32
943    as well as under *nix as long as it compiles.
944    
945    Other changes:
946    
947      - made CML compile again under Win32
948      - made eXene compile under Win32 (by providing a fake structure UnixSock
949        and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
950      - fixed a bug in nowhere: it assumed that type OS.Process.status is the
951        same as type int;  under Win32 it isn't
952      - fixed some slice-related problems in the win32-specific parts of CML
953      - added a functor argument "sameVol" to os-path-fn.sml in the Basis
954        (under Win32, the volume name is case-insensitive, and the
955        OS.Path code compares volume names for equality)
956    
957    ----------------------------------------------------------------------
958    Name: Matthias Blume (blume (at) tti - c (dot) org)
959    Date: 2003/09/08 11:55:00 CDT
960    Tag: blume-20030908-fullpath
961    Description:
962    
963    Made Win32 version of OS.FileSys.fullPath return current directory
964    when given an empty string.  This is what the spec says, and incidentally,
965    CM depends on it.  (CM otherwise goes into an infinite loop in certain
966    cases when presented with the name of a non-existing .cm file.)
967    
968    ----------------------------------------------------------------------
969    Name: Matthias Blume (blume (at) tti - c (dot) org)
970    Date: 2003/09/04 16:30:00 CDT
971    Tag: blume-20030905-slices-etc
972    Description:
973    
974      1. Changed interface to vectors and arrays in Basis to match
975         (draft) Basis spec.
976      2. Added signatures and implementations of slices according to
977         Basis spec.
978      3. Edited source code throughout the system to make it compile again
979         under 1. and 2.  (In some cases code had to be added to have it
980         match the new signatures.)
981      4. MLRISC should be backward-compatible: the copies of the originals
982         of files that needed to change under 3. were retained, the .cm files
983         check the compiler version number and use old versions when
984         appropriate.
985      5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
986         dirstream -> string option (in accordance with Basis spec).
987      6. When generating code that counts lines, ml-lex used function
988         CharVector.foldli, taking advantage of its old interface.
989         This has been replaced with the corresponding code from
990         CharVectorSlice.  (html-lex must be re-lexed!)
991      7. BitArray in smlnj-lib/Util has been extended/modified to match the
992         new MONO_ARRAY signature.  (Do we need BitArraySlice?)
993      8. Removed temporary additions (fromInternal, toInternal) from the
994         (now obsolete) IntInf in smlnj-lib/Util.
995      9. Cleaned up structure Byte.
996     10. Added localOffset, scan, and fromString to Date (according to spec).
997         Cleaned/corrected implementation of Date.
998         (Still need to check for correctness; implement better canonicalizeDate.)
999     11. Added "scan" to signature IEEE_REAL.
1000     12. Some improvements to IntInf [in particular: efficiency-hack for
1001         mod and rem when second operand is 2 (for parity checks).]
1002     13. Changed representation of type Time.time, using a single IntInf.int
1003         value counting microseconds.  This considerably simplified the
1004         implementation of structure Time.  We now support negative time
1005         values; scan and fromString handle signs.
1006     14. Functor PrimIO now takes two additional arguments (VectorSlice and
1007         ArraySlice).
1008    
1009    ----------------------------------------------------------------------
1010    Name: Matthias Blume (blume (at) tti - c (dot) org)
1011    Date: 2003/08/28 17:00:00 CDT
1012    Tag: blume-20030828-intinf
1013    Description:
1014    
1015    This is a major update which comes with a version number bump
1016    (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
1017    and an implementation of IntInf in the Basis.
1018    
1019    There are a fairly large number of related changes and updates throughout
1020    the system:
1021    
1022    Basis:
1023            - Implemented IntInf.
1024            - Made LargeInt a projection of IntInf (by filtering through INTEGER).
1025            - Added some missing Real64 operations, most notably Real.toLargeInt.
1026            - Added FixedInt as a synonym for Int32.
1027    
1028    compiler:
1029            * Added support for a built-in intinf type.
1030              - literals
1031              - pattern matching
1032              - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
1033              - overloading on literals and operations
1034    
1035            This required adding a primitive type intinf, some additional
1036            primops, and implementations for several non-trivial intinf
1037            operations in Core.  (The intinf type is completely abstract
1038            to the compiler; all operations get delegated back to the Core.)
1039    
1040            * Intinf equality is handled by polyequal.  However, the compiler
1041            does not print its usual warning in this case (since polyequal
1042            is the right thing to do there).
1043    
1044            * Improved the organization of structure InlineT.
1045    
1046            * A word about conversion primops:
1047            If conversions involving intinf do not cancel out during
1048            CPS contract, then the compiler must insert calls to Core functions.
1049            Since all core access must be resolved already during the FLINT
1050            translate phase, it would be too late a the time of CPS contract
1051            to add new Core calls.  For this reason, conversion primops
1052            for intinf carry two arguments:  1. the numeric argument that
1053            they are supposed to convert, and 2. the Core function that
1054            can help with this conversion if necessary.  If CPS contract
1055            eliminates a primop, then the associated Core function becomes
1056            dead and goes away.  Intinf conversion primops that do not get
1057            eliminated by CPS contract get rewritten into calls of their
1058            core functions by a separate, new phase.
1059    
1060    interactive system:
1061            - Control.Print.intinfDepth controls max length of intinf constants
1062            being printed. (Analogous to Control.Print.stringDepth.)
1063            - Cleanup in printutil and pputil: got rid of unused stuff and
1064            duplicates; replaced some of the code with code that makes better
1065            use of library functionality.
1066    
1067    CM:
1068            Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
1069            are no longer silent.
1070    
1071    CKIT:
1072            Fixed mismatched uses of Int32 and LargeInt.  I always decided
1073            in favor of LargeInt -- which is now the same as IntInf.
1074            CKIT-knowledgable people should check whether this is what's
1075            intended and otherwise change things back to using Int32 or
1076            FixedInt.
1077    
1078    Throughout the code:
1079            Started using IntInf.int literals and built-in operations
1080            (e.g., comparison with 0) where this seems appropriate.
1081    
1082    
1083    ----------------------------------------------------------------------
1084    Name: Dave MacQueen (dbm@cs.uchicago.edu)
1085    Date: 2003/08/13 11:36:00 CDT
1086    Tag: dbm-20030813-mcz-merge1
1087    Description:
1088    
1089    Merging changes from the mcz-branch development branch into trunk.
1090    These changes involve replacement of the emulated old prettyprinter
1091    interface with direct use of the SML/NJ Lib PP library, and fixing
1092    of a couple of bugs (895, 1186) relating to error messages.  A new
1093    prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
1094    has been added.
1095    
1096    ----------------------------------------------------------------------
1097    Name: Matthias Blume (blume (at) tti - c (dot) org)
1098    Date: 2003/08/11 15:45:00 CDT
1099    Tag: blume-20030811-windows
1100    Description:
1101    
1102    Version number bumped to 110.42.9.  NEW BOOTFILES!!!
1103    
1104        http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
1105    
1106    This patch restores SML/NJ's ability to run under win32.  There are a
1107    number of changes, including fixes for several bugs that had gone
1108    unnoticed until now:
1109    
1110      - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
1111        trying to run the win32 version from within cygwin.)
1112    
1113      - There are a number of simple .bat scripts that substitute for their
1114        corresponding Unix shell-scripts.  (See below.)
1115    
1116      - The internals of ml-build have been modified slightly.  The main
1117        difference is that instead of calling ".link-sml" (or link-sml.bat)
1118        using OS.Process.system, the ML process delegates this task back
1119        to the script.  Otherwise problems arise in mixed environments such
1120        as Cygwin where scripts look and work like Unix scripts, but
1121        where OS.Process.system cannot run them.
1122    
1123      - In CM, the srcpath pickler used native pathname syntax -- which
1124        is incorrect in the case of cross-compilation.  The new pickle format
1125        is independent of platform-specific naming conventions.
1126    
1127      - Path configuration files (such as lib/pathconfig) can now choose
1128        between native and standard syntax.  Placing a line of the form
1129    
1130            standard!
1131    
1132        into the file causes all subsequent paths to be interpreted using
1133        CM standard pathname syntax (= Unix conventions); a line
1134    
1135            native!
1136    
1137        switches back to native style.  This was needed so that
1138        path config files can be written portably, see src/system/pathconfig.
1139    
1140      - Runtime system:
1141    
1142          - win32-filesys.c:  get_file_time and set_file_time now
1143            access modification time, not creation time.
1144    
1145          - I/O code made aware of new array representation.
1146    
1147          - Bug fixes in X86.prim.masm.
1148    
1149       - src/system/makeml made aware of win32. (For use under cygwin
1150            and other Unix-environments for windows.)
1151    
1152       - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
1153         which caused BinIO.inputAll to fail consistently.
1154    
1155    .bat scripts:
1156    
1157       Windows .bat scripts assume that SMLNJ_HOME is defined.
1158    
1159       - sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
1160         applications (sml, ml-yacc, ml-lex).
1161       - ml-build.bat: analogous to ml-build.
1162       - config\install.bat:  Analogous to config/install.sh.  This requires
1163         that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
1164         (nmake etc. must be on the path, and vcvars32 must have been run.)
1165         Moreover, sources for ml-lex and ml-yacc need to exist under src,
1166         and the bootfile hierarchy must have been unpacked under
1167         sml.boot.x86-win32.
1168         The script is very primitive and does a poor job at error checking.
1169         It only installs the base system, ml-lex, and ml-yacc.  No other
1170         libraries are being installed (i.e., you get only those that
1171         are part of the compiler.)
1172       - link-sml.bat:  analogous to .link-sml, but not currently used
1173    
1174    Unrelated bug fixes:
1175    
1176      - ml-nlffigen now exports structures ST_* corresponding to incomplete
1177        types.
1178      - Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
1179        otherwise.)
1180    
1181    ----------------------------------------------------------------------
1182    Name: Dave MacQueen (macqueen@cs.uchicago.edu)
1183    Date: 2003/06/17
1184    Tag: macqueen-20030617-bug895
1185    Description:
1186    
1187    Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
1188    Tag will be used for new development branch (mcz-branch) for
1189    use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
1190    
1191    ----------------------------------------------------------------------
1192    Name: Matthias Blume (blume (at) tti - c (dot) org)
1193    Date: 2003/05/27 16:55:00 CDT
1194    Tag: blume-20030527-polyeq
1195    Description:
1196    
1197    Tried to eliminated most cases of polymorphic equality.
1198    
1199    ----------------------------------------------------------------------
1200    Name: Matthias Blume (blume (at) tti - c (dot) org)
1201    Date: 2003/05/21 17:45:00 CDT
1202    Tag: blume-20030517-complete
1203    Description:
1204    
1205    Two changes:
1206    
1207      1. Added a flag for controlling whether non-exhaustive bindings will
1208         be treated as errors (default is false).
1209      2. Cleaned up the *entire* source tree so that CMB.make goes through
1210         without a single non-exhaustive match- or bind warning.
1211    
1212    ----------------------------------------------------------------------
1213    Name: Matthias Blume (blume (at) tti - c (dot) org)
1214    Date: 2003/05/17 10:20:00 CDT
1215    Tag: blume-20030517-absyn
1216    Description:
1217    
1218    1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
1219    
1220       This mainly affects the quality of error messages.  However, some
1221       of the code is now more straightforward than before. (Treatment of
1222       the above four constructs in translate.sml is much simpler than
1223       the "macro-expansion" that was going on before.  Plus, the mach-
1224       compiler no longer gets invoked just to be able to compile an
1225       if-expression.)
1226    
1227    2. The ErrorMsg.Error exception is now caught and absorbed by the
1228       interactive loop.
1229    
1230    ----------------------------------------------------------------------
1231    Name: Allen Leung
1232    Date: 2003/05/16 13:05:00 CDT
1233    Tag: leunga-20030516-cygwin-runtime
1234    Description:
1235    
1236        Ported the runtime system to cygwin, which uses the unix
1237    x86-unix bin files.  Missing/buggy features:
1238    
1239       o getnetbyname, getnetbyaddr: these functions seem to be missing in
1240         the Cygwin library.
1241       o Ctrl-C handling may be flaky.
1242       o Windows system calls and Windows I/O are not supported.
1243    
1244       A new set of binfiles is located at:
1245    
1246         http://www.dorsai.org/~leunga/boot.x86-unix.tgz
1247    
1248       This is only needed for bootstrapping the cygwin version of smlnj.
1249       Other x86 versions can use the existing binfiles.
1250    
1251    ----------------------------------------------------------------------
1252    Name: Matthias Blume
1253    Date: 2003/04/08 15:42:00 CDT
1254    Tag: blume-20030408-listpair
1255    Description:
1256    
1257    1. Added a target 'mlrisc' to installer.
1258    
1259    2. Added missing elements to structure ListPair.
1260    
1261    ----------------------------------------------------------------------
1262    Name: Allen Leung
1263    Date: 2003/01/07 10:40:00 EST
1264    Tag: leunga-20030107-int-rem
1265    Description:
1266    
1267    Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
1268    The arguments to the SUBL instruction were swapped.
1269    
1270    ----------------------------------------------------------------------
1271    Name: Matthias Blume
1272    Date: 2002/12/12 16:25:00 EST
1273    Tag: blume-20021212-risc-ra
1274    Description:
1275    
1276    Fixed a serious bug in the rewrite code for FP spilling/reloading that
1277    sent the RA into an infinite loop when floating point registers get
1278    spilled.  (Because of this bug, e.g., nucleic stopped compiling between
1279    110.37 and 110.38.)
1280    There was another set of potential problems related to the handling of
1281    MLRISC annotations (but those did not yet cause real problems, apparently).
1282    
1283    ----------------------------------------------------------------------
1284    Name: Matthias Blume
1285    Date: 2002/12/06 22:40:00 EST
1286    Tag: blume-20021206-cm-fileid
1287    Description:
1288    
1289    Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
1290    This fixes the problem of CM getting confused by files that suddenly
1291    change their identity (e.g., by getting unlinked and recreated by some
1292    text editor such as vi).  There might be a better/cheaper/cleaner way
1293    of doing this, but for now this will have to do.
1294    
1295    ----------------------------------------------------------------------
1296    Name: Matthias Blume
1297    Date: 2002/10/28 09:50:00 EST
1298    Tag: blume-20021028-typecheck
1299    Description:
1300    
1301    Exported structure Typecheck from $smlnj/viscomp/core.cm.
1302    
1303    ----------------------------------------------------------------------
1304    Name: Matthias Blume
1305    Date: 2002/10/17 09:10:00 EDT
1306    Tag: Release_110_42
1307    Description:
1308    
1309    In good old tradition, there has been a slight hiccup so that we have
1310    to patch 110.42 after the fact.   The old release tag has been replaced
1311    (see below).
1312    
1313    The change solves a problem with two competing approaches the
1314    configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
1315    each other's way.
1316    
1317    This change only affects the runtime system code and the installer script.
1318    (No new bootfiles.)
1319    
1320    ----------------------------------------------------------------------
1321    Name: Matthias Blume
1322    Date: 2002/10/16 12:00:00 EDT
1323    Tag: Release_110_42_removed
1324    Description:
1325    
1326    New working release.  New bootfiles.
1327    
1328    ----------------------------------------------------------------------
1329    Name: Matthias Blume
1330    Date: 2002/10/10 13:10:00 EDT
1331    Tag: blume-20021010-ppc-divs
1332    Description:
1333    
1334    The mltree operator DIVS must be implemented with an overflow check on
1335    the PPC because the hardware indicates divide-by-zero using "overflow" as
1336    well.
1337    
1338    ----------------------------------------------------------------------
1339    Name: Matthias Blume
1340    Date: 2002/07/23 11:45:00
1341    Tag: blume-20020723-smlnj-home
1342    Description:
1343    
1344    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
1345    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
1346    CM_PATHCONFIG is also set), the path configuration file is assumed
1347    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
1348    the entire tree to some other place and everything will "just work".
1349    
1350    (Companion commands such as ml-build and ml-makedepend also sense this
1351    variable.)
1352    
1353    ----------------------------------------------------------------------
1354    Name: Matthias Blume
1355    Date: 2002/07/12 21:19:00 EDT
1356    Tag: blume-20020712-liveness
1357    Description:
1358    
1359    Exported two useful "step" functions from liveness module (MLRISC).
1360    
1361    ----------------------------------------------------------------------
1362    Name: Matthias Blume
1363    Date: 2002/07/05 16:00 EDT
1364    Tag: Release_110_41
1365    Description:
1366    
1367    New working release.  New bootfiles.
1368    
1369    ----------------------------------------------------------------------
1370    Name: Matthias Blume
1371    Date: 2002/07/05 10:25:00 EDT
1372    Tag: blume-20020705-btimp
1373    Description:
1374    
1375    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
1376    other clients can set up backtracing support.
1377    
1378    ----------------------------------------------------------------------
1379    Name: Matthias Blume
1380    Date: 2002/06/25 17:23:00 EDT
1381    Tag: blume-20020625-fpmax
1382    Description:
1383    
1384    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
1385    case.  (The sense of the isNaN test was reversed -- which made min and
1386    max always return their first argument.)
1387    
1388    ----------------------------------------------------------------------
1389    Name: Matthias Blume
1390    Date: 2002/06/11
1391    Tag: blume-20020611-unixpath
1392    Description:
1393    
1394    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
1395    
1396    ----------------------------------------------------------------------
1397    Name: Matthias Blume
1398    Date: 2002/06/10 16:35:00 EDT
1399    Tag: blume-20020610-ieeereal
1400    Description:
1401    
1402    I back-ported my implementation of IEEEReal.fromString from the
1403    idlbasis-devel branch so that we can test it.
1404    
1405    Another small change is that ppDec tries to give more information
1406    than just "<sig>" in the case of functors.  However, this code is
1407    broken in some mysterious way if the functor's body's signature
1408    has not been declared by ascription but gets inferred from the
1409    implementation.  This needs fixing...
1410    
1411    ----------------------------------------------------------------------
1412    Name: Matthias Blume
1413    Date: 2002/05/31
1414    Tag: blume-20020531-btrace-mode
1415    Description:
1416    
1417    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
1418    the wayside when I switched over to using Controls everywhere.)
1419    
1420    ----------------------------------------------------------------------
1421    Name: Lal George
1422    Date: 2002/05/23 12:21:40 EDT
1423    Tag: george-20020523-visual-labels
1424    Description:
1425    
1426    Labels are now displayed in the graphical output to make
1427    the fall-through and target blocks obvious.
1428    
1429    ----------------------------------------------------------------------
1430    Name: Matthias Blume
1431    Date: 2002/05/22 11:03:00 EDT
1432    Tag: blume-20020522-shrink
1433    Description:
1434    
1435    John tweaked yesterday's fix for 1131 to handle an out-of-memory
1436    situation that comes up when allocating huge arrays.
1437    
1438    ----------------------------------------------------------------------
1439    Name: Matthias Blume
1440    Date: 2002/05/21 16:00:00 EDT
1441    Tag: Release_110_40
1442    Description:
1443    
1444    New working release (110.40).  New bootfiles.
1445    
1446    [Also: John Reppy fixed GC bug 1131.]
1447    
1448    ----------------------------------------------------------------------
1449    Name: Matthias Blume
1450    Date: 2002/05/21 12:35:00 EDT
1451    Tag: blume-20020521-cmdoc
1452    Description:
1453    
1454    CM documentation update.
1455    
1456    ----------------------------------------------------------------------
1457    Name: Matthias Blume
1458    Date: 2002/05/21 10:55:00 EDT
1459    Tag: blume-20020521-misc
1460    Description:
1461    
1462    - John tweaked runtime to be silent on heap export (except when
1463      GC messages are on).
1464    - I added a few more things (cross-compiling versions of CMB) to
1465      config/preloads (as suggestions).
1466    
1467    ----------------------------------------------------------------------
1468    Name: Matthias Blume
1469    Date: 2002/05/20 22:25:00 EDT
1470    Tag: blume-20020520-controls
1471    Description:
1472    
1473    - Added ControlUtil structure to control-lib.cm.
1474    - Use it throughout.
1475    - Used Controls facility to define MLRISC controls (as opposed to
1476      registering MLRISC control ref cells with Controls after the fact)
1477    - Fixed messed-up controls priorities.
1478    
1479    * Removed again all the stuff from config/preloads that one wouldn't
1480      be able to preload at the time the initial heap image is built.
1481      (Many libraries, e.g., CML, do not exist yet at this time.  The
1482       only libraries that can be preloaded via config/preloads are those
1483       that come bundled with the bootfiles.)
1484    
1485    ----------------------------------------------------------------------
1486    Name: Matthias Blume
1487    Date: 2002/05/20 10:59:00 EDT
1488    Tag: blume-20020520-preloads
1489    Description:
1490    
1491    Added a lot of commented-out suggestions for things to be included
1492    in config/preloads.
1493    
1494    ----------------------------------------------------------------------
1495    Name: Allen Leung
1496    Date: 2002/05/18 14:20:00 EDT
1497    Tag: leunga-20020518-mdl
1498    Description:
1499    
1500    o Made the mdl tool stuff compile and run again.
1501    o I've disabled all the stuff that depends on RTL specifications; they
1502      are all badly broken anyway.
1503    
1504    ----------------------------------------------------------------------
1505    Name: Matthias Blume
1506    Date: 2002/05/17 16:49:00 EDT
1507    Tag: blume-20020517-controls
1508    Description:
1509    
1510    0. John Reppy made several modifications to the SML/NJ library.
1511       In particular, there is a shiny new controls-lib.cm.
1512    
1513    1. Pushed new controls interface through compiler so that everything
1514       compiles again.
1515    
1516    2. Added FormatComb and FORMAT_COMB to the CML version of the
1517       SML/NJ library (so that CML compiles again).
1518    
1519    3. Modified init scripts because XXX_DEFAULT environment variables
1520       are no longer with us.  (Boot-time initialization is now done
1521       using the same environment variables that are also used for
1522       startup-time initialization of controls.)
1523    
1524    ----------------------------------------------------------------------
1525    Name: Lal George
1526    Date: 2002/05/15 09:20:10 EDT
1527    Tag: george-20020515-pseudo-op-decls
1528    Description:
1529    
1530            All pseudo-ops emitted before the first segment declaration
1531    such as TEXT, DATA, and BSS directives are assumed to be global
1532    declarations and are emitted first in the assembly file. This is
1533    useful in a number of situations where one has pseudo-ops that are not
1534    specific to any segment, and also works around the constraint that one
1535    cannot have client pseudo-ops in the TEXT segment.
1536    
1537            Because no segment is associated with these declarations it is
1538    an error to allocate any space or objects before the first segment
1539    directive and an exception will be raised. However, we cannot make
1540    this check for client pseudo-ops.
1541    
1542            These top level declarations are a field in the CFG graph_info.
1543    In theory you can continue to add to this field after the CFG has been
1544    built -- provided you know what you are doing;-)
1545    
1546    ----------------------------------------------------------------------
1547    Name: Matthias Blume
1548    Date: 2002/05/13 16:40:00 EDT
1549    Tag: blume-20020513-pp-etc
1550    Description:
1551    
1552    A few minor bugfixes:
1553    
1554      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
1555        (Bogus printouts for redefined bindings still occur.  Compiler
1556        bug should no longer occur now.  We need to redo the prettyprinter
1557        from scratch.)
1558    
1559      - CM pathname printer now also adds escape sequences for ( and )
1560    
1561      - commend and docu fixes for ml-nlffi
1562    
1563    ----------------------------------------------------------------------
1564    Name: Matthias Blume
1565    Date: 2002/05/10 16:40:00 EDT
1566    Tag: blume-20020510-erg-textio
1567    Description:
1568    
1569    Applied the following bugfix provided by Emden Gansner:
1570    
1571        Output is corrupted when outputSubstr is used rather than output.
1572    
1573        The problem occurs when a substring
1574    
1575            ss = (s, dataStart, dataLen)
1576    
1577        where dataStart > 0, fills a stream buffer with avail bytes left.
1578        avail bytes of s, starting at index dataStart, are copied into the
1579        buffer, the buffer is flushed, and then the remaining dataLen-avail
1580        bytes of ss are copied into the beginning of the buffer. Instead of
1581        starting this copy at index dataStart+avail in s, the current code
1582        starts the copy at index avail.
1583    
1584        Fix:
1585        In text-io-fn.sml, change line 695 from
1586             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
1587        to
1588             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
1589    
1590    ----------------------------------------------------------------------
1591    Name: Matthias Blume
1592    Date: 2002/04/12 13:55:00 EDT
1593    Tag: blume-20020412-assyntax
1594    Description:
1595    
1596    1. Grabbed newer assyntax.h from the XFree86 project.
1597    2. Fiddled with how to compile X86.prim.asm without warnings.
1598    3. (Very) Minor cleanup in CM.
1599    
1600    ----------------------------------------------------------------------
1601    Name: Matthias Blume
1602    Date: 2002/04/01 (no joke!) 17:07:00 EST
1603    Tag: blume-20020401-x86div
1604    Description:
1605    
1606    Added full support for div/mod/rem/quot on the x86, using the machine
1607    instruction's two results (without clumsily recomputing the remainder)
1608    directly where appropriate.
1609    
1610    Some more extensive power-of-two support was added to the x86 instruction
1611    selector (avoiding expensive divs, mods, and muls where they can be
1612    replaced with cheaper shifts and masks).  However, this sort of thing
1613    ought to be done earlier, e.g., within the CPS optimizer so that
1614    all architectures benefit from it.
1615    
1616    The compiler compiles to a fixed point, but changes might be somewhat
1617    fragile nevertheless.  Please, report any strange things that you might
1618    see wrt. div/mod/quot/rem...
1619    
1620    ----------------------------------------------------------------------
1621    Name: Matthias Blume
1622    Date: 2002/03/29 17:22:00
1623    Tag: blume-20020329-div
1624    Description:
1625    
1626    Fixed my broken div/mod logic.  Unfortunately, this means that the
1627    inline code for div/mod now has one more comparison than before.
1628    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
1629    The problem was with quotient = 0, because that alone does not tell
1630    us which way the rounding went.  One then has to look at whether
1631    remainder and divisor have the same sign...  :(
1632    
1633    Anyway, I replaced the bootfiles with fresh ones...
1634    
1635    ----------------------------------------------------------------------
1636    Name: Matthias Blume
1637    Date: 2002/03/29 14:10:00 EST
1638    Tag: blume-20020329-inlprims
1639    Description:
1640    
1641    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
1642    
1643    Primops have changed. This means that the bin/boot-file formats have
1644    changed as well.
1645    
1646    To make sure that there is no confusion, I made a new version.
1647    
1648    
1649    CHANGES:
1650    
1651    * removed REMT from mltree (remainder should never overflow).
1652    
1653    * added primops to deal with divisions of all flavors to the frontend
1654    
1655    * handled these primops all the way through so they map to their respective
1656      MLRISC support
1657    
1658    * used these primops in the implementation of Int, Int32, Word, Word32
1659    
1660    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
1661    
1662    * parameterized INLMIN, INLMAX, and INLABS by a numkind
1663    
1664    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
1665      floating point
1666    
1667    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
1668      and Real (but Real.abs maps to a separate floating-point-only primop)
1669    
1670    
1671    TODO items:
1672    
1673    * Hacked Alpha32 instruction selection, disabling the selection of REMx
1674      instructions because the machine instruction encoder cannot handle
1675      them.  (Hppa, PPC, and Sparc instruction selection did not handle
1676      REM in the first place, and REM is supported by the x86 machine coder.)
1677    
1678    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
1679      selection phase.  (The two can be streamlined because the hardware
1680      delivers both quotient and remainder at the same time anyway.)
1681    
1682    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
1683      (Currently the behavior is inconsistent both across architectures and
1684      wrt. the draft Basis spec.)
1685    
1686    * Word8 should eventually be handled natively, too.
1687    
1688    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
1689      as if there currently is no execution path that could trigger it in
1690      SML/NJ.  (The assumptions underlying functions arith and promotable do not
1691      hold for things like multiplication and division.)
1692    
1693    ----------------------------------------------------------------------
1694    Name: Matthias Blume
1695    Date: 2002/03/27 16:27:00 EST
1696    Tag: blume-20020327-mlrisc-divisions
1697    Description:
1698    
1699    Added support for all four division operations (ML's div, mod, quot,
1700    and rem) to MLRISC.  In the course of doing so, I also rationalized
1701    the naming (no more annoying switch-around of DIV and QUOT), by
1702    parameterizing the operation by div_rounding_mode (which can be either
1703    DIV_TO_ZERO or DIV_TO_NEGINF).
1704    
1705    The generic MLTreeGen functor takes care of compiling all four
1706    operations down to only round-to-zero div.
1707    
1708    Missing pieces:
1709    
1710      * Doing something smarter than relying on MLTreeGen on architectures
1711        like, e.g., the x86 where hardware division delivers both quotient and
1712        remainder at the same time.  With this, the implementation of the
1713        round-to-neginf operations could be further streamlined.
1714    
1715      * Remove inlining support for div/mod/rem from the frontend and replace it
1716        with primops that get carried through to the backend.  Do this for all
1717        int and word types.
1718    
1719    ----------------------------------------------------------------------
1720    Name: Matthias Blume
1721    Date: 2002/03/25 17:25:00 EST
1722    Tag: blume-20020325-divmod
1723    Description:
1724    
1725    I improved (hopefully without breaking them) the implementation of Int.div,
1726    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
1727    advantage of the following observations:
1728    
1729      Let  q = x quot y      r = x rem y
1730           d = x div  y      m = x mod y
1731    
1732    where "quot" is the round-to-zero version of integer division that
1733    hardware usually provides.  Then we have:
1734    
1735         r = x - q * y        where neither the * nor the - will overflow
1736         d = if q >= 0 orelse x = q * y then q else q - 1
1737                              where neither the * nor the - will overflow
1738         m = if q >= 0 orelse r = 0 then r else r + y
1739                              where the + will not overflow
1740    
1741    This results in substantial simplification of the generated code.
1742    The following table shows the number of CFG nodes and edges generated
1743    for
1744            fun f (x, y) = x OPER y
1745            (* with OPER \in div, mod, quot, rem *)
1746    
1747    
1748        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
1749        --------------------------------------------------------
1750         div |         24 |         39 |         12 |         16
1751         mod |         41 |         71 |         12 |         16
1752        quot |          8 |         10 |          8 |         10
1753         rem |         10 |         14 |          8 |         10
1754    
1755    
1756    ----------------------------------------------------------------------
1757    Name: Matthias Blume
1758    Date: 2002/03/25 22:06:00 EST
1759    Tag: blume-20020325-cprotobug
1760    Description:
1761    
1762    Fixed a bug in cproto (c prototype decoder).
1763    
1764    ----------------------------------------------------------------------
1765    Name: Matthias Blume
1766    Date: 2002/03/25 16:00:00 EST
1767    Tag: blume-20020325-raw-primops
1768    Description:
1769    
1770    I did some cleanup to Allen's new primop code and
1771    replaced yesterday's bootfiles with new ones.
1772    (But they are stored in the same place.)
1773    
1774    ----------------------------------------------------------------------
1775    Name: Matthias Blume
1776    Date: 2002/03/24 22:40:00 EST
1777    Tag: blume-20020324-bootfiles
1778    Description:
1779    
1780    Made the bootfiles that Allen asked for.
1781    
1782    ----------------------------------------------------------------------
1783    Name: Allen Leung
1784    Date: 2002/03/23 15:50:00 EST
1785    Tag: leunga-20020323-flint-cps-rcc-primops
1786    Description:
1787    
1788      1. Changes to FLINT primops:
1789    
1790        (* make a call to a C-function;
1791         * The primop carries C function prototype information and specifies
1792         * which of its (ML-) arguments are floating point. C prototype
1793         * information is for use by the backend, ML information is for
1794         * use by the CPS converter. *)
1795      | RAW_CCALL of { c_proto: CTypes.c_proto,
1796                       ml_args: ccall_type list,
1797                       ml_res_opt: ccall_type option,
1798                       reentrant : bool
1799                     } option
1800       (* Allocate uninitialized storage on the heap.
1801        * The record is meant to hold short-lived C objects, i.e., they
1802        * are not ML pointers.  With the tag, the representation is
1803        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
1804        *)
1805      | RAW_RECORD of {tag:bool,sz:int}
1806      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
1807    
1808      2.  These CPS primops are now overloaded:
1809    
1810           rawload of {kind:numkind}
1811           rawstore of {kind:numkind}
1812    
1813          The one argument form is:
1814    
1815             rawload {kind} address
1816    
1817          The two argument form is:
1818    
1819             rawload {kind} [ml object, byte-offset]
1820    
1821      3. RAW_CCALL/RCC now takes two extra arguments:
1822    
1823         a. The first is whether the C call is reentrant, i.e., whether
1824            ML state should be saved and restored.
1825         b. The second argument is a string argument specifying the name of
1826            library and the C function.
1827    
1828         These things are currently not handled in the code generator, yet.
1829    
1830      4. In CProto,
1831    
1832         An encoding type of "bool" means "ml object" and is mapped into
1833         C prototype of PTR.  Note that "bool" is different than "string",
1834         even though "string" is also mapped into PTR, because "bool"
1835         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
1836    
1837      5. Pickler/unpicker
1838    
1839         Changed to handle RAW_RECORD and newest RAW_CCALL
1840    
1841      6. MLRiscGen,
1842    
1843         1. Changed to handle the new rawload/rawstore/rawrecord operators.
1844         2. Code for handling C Calls has been moved to a new module CPSCCalls,
1845            in the file CodeGen/cpscompile/cps-c-calls.sml
1846    
1847      7. Added the conditional move operator
1848    
1849             condmove of branch
1850    
1851         to cps.  Generation of this is still buggy so it is currently
1852         disabled.
1853    
1854    ----------------------------------------------------------------------
1855    Name: Lal George
1856    Date: 2002/03/22 14:18:25 EST
1857    Tag: george-20020322-cps-branch-prob
1858    Description:
1859    
1860    Implemented the Ball-Larus branch prediction-heuristics, and
1861    incorporated graphical viewers for control flow graphs.
1862    
1863    Ball-Larus Heuristics:
1864    ---------------------
1865    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
1866    
1867    By design it uses the Dempster-Shafer theory for combining
1868    probabilities.  For example, in the function:
1869    
1870        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
1871    
1872    the ball-larus heuristics predicts that the n=0 is unlikely
1873    (OH-heuristic), and the 'then' branch is unlikely because of the
1874    RH-heuristic -- giving the 'then' branch an even lower combined
1875    probability using the Dempster-Shafer theory.
1876    
1877    Finally, John Reppy's loop analysis in MLRISC, further lowers the
1878    probability of the 'then' branch because of the loop in the else
1879    branch.
1880    
1881    
1882    Graphical Viewing:
1883    ------------------
1884    I merely plugged in Allen's graphical viewers into the compiler. The
1885    additional code is not much. At the top level, saying:
1886    
1887            Control.MLRISC.getFlag "cfg-graphical-view" := true;
1888    
1889    will display the graphical view of the control flow graph just before
1890    back-patching.  daVinci must be in your path for this to work. If
1891    daVinci is not available, then the default viewer can be changed
1892    using:
1893    
1894            Control.MLRISC.getString "viewer"
1895    
1896    which can be set to "dot" or "vcg" for the corresponding viewers. Of
1897    course, these viewers must be in your path.
1898    
1899    The above will display the compilation unit at the level of clusters,
1900    many of which are small, boring, and un-interesting. Also setting:
1901    
1902            Control.MLRISC.getInt "cfg-graphical-view_size"
1903    
1904    will display clusters that are larger than the value set by the above.
1905    
1906    
1907    ----------------------------------------------------------------------
1908    Name: Matthias Blume
1909    Date: 2002/03/21 22:20:00 EST
1910    Tag: blume-20020321-kmp-bugfix
1911    Description:
1912    
1913    Changed the interface to the KMP routine in PreString and fixed
1914    a minor bug in one place where it was used.
1915    
1916    ----------------------------------------------------------------------
1917    Name: Allen Leung
1918    Date: 2002/03/21 20:30:00 EST
1919    Tag: leunga-20020321-cfg
1920    Description:
1921    
1922      Fixed a potential problem in cfg edge splitting.
1923    
1924    ----------------------------------------------------------------------
1925    Name: Allen Leung
1926    Date: 2002/03/21 17:15:00 EST
1927    Tag: leunga-20020321-x86-fp-cfg
1928    Description:
1929    
1930      1. Recoded the buggy parts of x86-fp.
1931    
1932         a. All the block reordering code has been removed.
1933            We now depend on the block placement phases to do this work.
1934    
1935         b. Critical edge splitting code has been simplified and moved into the
1936            CFG modules, as where they belong.
1937    
1938         Both of these were quite buggy and complex.  The code is now much, much
1939         simpler.
1940    
1941      2. X86 backend.
1942    
1943         a. Added instructions for 64-bit support.  Instruction selection for
1944            64-bit has not been committed, however, since that
1945            requires changes to MLTREE which haven't been approved by
1946            Lal and John.
1947    
1948         b. Added support for FUCOMI and FUCOMIP when generating code for
1949            PentiumPro and above.  We only generate these instructions in
1950            the fast-fp mode.
1951    
1952         c. Added cases for JP and JNP in X86FreqProps.
1953    
1954      3. CFG
1955    
1956         CFG now has a bunch of methods for edge splitting and merging.
1957    
1958      4. Machine description.
1959    
1960         John's simplification of MLTREE_BASIS.fcond broke a few machine
1961         description things:
1962    
1963         rtl-build.{sig,sml} and hppa.mdl fixed.
1964    
1965         NOTE: the machine description stuff in the repository is still broken.
1966               Again, I can't put my fixes in because that involves
1967               changes to MLTREE.
1968    
1969    ----------------------------------------------------------------------
1970    Name: Matthias Blume
1971    Date: 2002/03/20 15:55:00 EST
1972    Tag: blume-20020320-kmp
1973    Description:
1974    
1975    Implemented Knuth-Morris-Pratt string matching in PreString and used
1976    it for String.isSubstring, Substring.isSubstring, and
1977    Substring.position.
1978    
1979    (Might need some stress-testing.  Simple examples worked fine.)
1980    
1981    ----------------------------------------------------------------------
1982    Name: Matthias Blume
1983    Date: 2002/03/19 16:37:00 EST
1984    Tag: blume-20020319-witnesses
1985    Description:
1986    
1987    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
1988    
1989    This implements a generic mechanism for changing constness qualifiers
1990    anywhere within big C types without resorting to outright "casts".
1991    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
1992    modify the constness at the outermost level.)
1993    The implementation of "convert" is based on the idea of "witness"
1994    values -- values that are not used by the operation but whose types
1995    "testify" to their applicability.  On the implementation side, "convert"
1996    is simply a projection (returning its second curried argument).  With
1997    cross-module inlining, it should not result in any machine code being
1998    generated.
1999    
2000    ----------------------------------------------------------------------
2001    Name: Matthias Blume
2002    Date: 2002/03/15 16:40:00 EST
2003    Tag: blume-20020315-basis
2004    Description:
2005    
2006    Provided (preliminary?) implementations for
2007    
2008      {String,Substring}.{concatWith,isSuffix,isSubstring}
2009    
2010    and
2011    
2012      Substring.full
2013    
2014    Those are in the Basis spec but they were missing in SML/NJ.
2015    
2016    ----------------------------------------------------------------------
2017    Name: Matthias Blume
2018    Date: 2002/03/14 21:30:00 EST
2019    Tag: blume-20020314-controls
2020    Description:
2021    
2022    Controls:
2023    ---------
2024    
2025    1. Factored out the recently-added Controls : CONTROLS stuff and put
2026       it into its own library $/controls-lib.cm.  The source tree for
2027       this is under src/smlnj-lib/Controls.
2028    
2029    2. Changed the names of types and functions in this interface, so they
2030       make a bit more "sense":
2031    
2032          module -> registry
2033          'a registry -> 'a group
2034    
2035    3. The interface now deals in ref cells only.  The getter/setter interface
2036       is (mostly) gone.
2037    
2038    4. Added a function that lets one register an already-existing ref cell.
2039    
2040    5. Made the corresponding modifications to the rest of the code so that
2041       everything compiles again.
2042    
2043    6. Changed the implementation of Controls.MLRISC back to something closer
2044       to the original.  In particular, this module (and therefore MLRISC)
2045       does not depend on Controls.  There now is some link-time code in
2046       int-sys.sml that registers the MLRISC controls with the Controls
2047       module.
2048    
2049    CM:
2050    ---
2051    
2052      * One can now specify the lambda-split aggressiveness in init.cmi.
2053    
2054    ----------------------------------------------------------------------
2055    Name: Allen Leung
2056    Date: 2002/03/13 17:30:00 EST
2057    Tag: leunga-20020313-x86-fp-unary
2058    Description:
2059    
2060    Bug fix for:
2061    
2062    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
2063    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
2064    > - fun f(x,(y,z)) = Real.~ y;
2065    > [autoloading]
2066    > [autoloading done]
2067    >       fchsl   (%eax), 184(%esp)
2068    > Error: MLRisc bug: X86MCEmitter.emitInstr
2069    >
2070    > uncaught exception Error
2071    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
2072    
2073    The problem was that the code generator did not generate any fp registers
2074    in this case, and the ra didn't know that it needed to run the X86FP phase to
2075    translate the pseudo fp instruction.   This only happened with unary fp
2076    operators in certain situations.
2077    
2078    ----------------------------------------------------------------------
2079    Name: Matthias Blume
2080    Date: 2002/03/13 14:00:00 EST
2081    Tag: blume-20020313-overload-etc
2082    Description:
2083    
2084    1. Added _overload as a synonym for overload for backward compatibility.
2085       (Control.overloadKW must be true for either version to be accepted.)
2086    
2087    2. Fixed bug in install script that caused more things to be installed
2088       than what was requested in config/targets.
2089    
2090    3. Made CM aware of the (_)overload construct so that autoloading
2091       works.
2092    
2093    ----------------------------------------------------------------------
2094    Name: Matthias Blume
2095    Date: 2002/03/12 22:03:00 EST
2096    Tag: blume-20020312-url
2097    Description:
2098    
2099    Forgot to update BOOT and srcarchiveurl.
2100    
2101    ----------------------------------------------------------------------
2102    Name: Matthias Blume
2103    Date: 2002/03/12 17:30:00 EST
2104    Tag: blume-20020312-version110392
2105    Description:
2106    
2107    Yet another version number bump (because of small changes to the
2108    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
2109    
2110    Changes:
2111    
2112      The new pid generation scheme described a few weeks ago was overly
2113      complicated.  I implemented a new mechanism that is simpler and
2114      provides a bit more "stability":  Once CM has seen a compilation
2115      unit, it keeps its identity constant (as long as you do not delete
2116      those crucial CM/GUID/* files).  This means that when you change
2117      an interface, compile, then go back to the old interface, and
2118      compile again, you arrive at the original pid.
2119    
2120      There now also is a mechanism that instructs CM to use the plain
2121      environment hash as a module's pid (effectively making its GUID
2122      the empty string).  For this, "noguid" must be specified as an
2123      option to the .sml file in question within its .cm file.
2124      This is most useful for code that is being generated by tools such
2125      as ml-nlffigen (because during development programmers tend to
2126      erase the tool's entire output directory tree including CM's cached
2127      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
2128      revert to the old, broken behavior of SML/NJ, but in specific cases
2129      where there is no danger of interface confusion, its use is ok
2130      (I think).
2131    
2132      ml-nlffigen by default generates "noguid" annotations.  They can be
2133      turned off by specifying -guid in its command line.
2134    
2135    ----------------------------------------------------------------------
2136    Name: Lal George
2137    Date: 2002/03/12 12 14:42:36 EST
2138    Tag: george-20020312-frequency-computation
2139    Description:
2140    
2141    Integrated jump chaining and static block frequency into the
2142    compiler. More details and numbers later.
2143    
2144    ----------------------------------------------------------------------
2145    Name: Lal George
2146    Date: 2002/03/11 11 22:38:53 EST
2147    Tag: george-20020311-jump-chain-elim
2148    Description:
2149    
2150    Tested the jump chain elimination on all architectures (except the
2151    hppa).  This is on by default right now and is profitable for the
2152    alpha and x86, however, it may not be profitable for the sparc and ppc
2153    when compiling the compiler.
2154    
2155    The gc test will typically jump to a label at the end of the cluster,
2156    where there is another jump to an external cluster containing the actual
2157    code to invoke gc. This is to allow factoring of common gc invocation
2158    sequences. That is to say, we generate:
2159    
2160            f:
2161               testgc
2162               ja   L1      % jump if above to L1
2163    
2164            L1:
2165               jmp L2
2166    
2167    
2168    After jump chain elimination the 'ja L1' instructions is converted to
2169    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
2170    up being implemented in their long form (if L2 is far away) using:
2171    
2172            jbe     L3      % jump if below or equal to L3
2173            jmp     L2
2174         L3:
2175            ...
2176    
2177    
2178    For large compilation units L2  may be far away.
2179    
2180    
2181    ----------------------------------------------------------------------
2182    Name: Matthias Blume
2183    Date: 2002/03/11 13:30:00 EST
2184    Tag: blume-20020311-mltreeeval
2185    Description:
2186    
2187    A functor parameter was missing.
2188    
2189    ----------------------------------------------------------------------
2190    Name: Allen Leung
2191    Date: 2002/03/11 10:30:00 EST
2192    Tag: leunga-20020311-runtime-string0
2193    Description:
2194    
2195       The representation of the empty string now points to a
2196    legal null terminated C string instead of unit.  It is now possible
2197    to convert an ML string into C string with InlineT.CharVector.getData.
2198    This compiles into one single machine instruction.
2199    
2200    ----------------------------------------------------------------------
2201    Name: Allen Leung
2202    Date: 2002/03/10 23:55:00 EST
2203    Tag: leunga-20020310-x86-call
2204    Description:
2205    
2206       Added machine generation for CALL instruction (relative displacement mode)
2207    
2208    ----------------------------------------------------------------------
2209    Name: Matthias Blume
2210    Date: 2002/03/08 16:05:00
2211    Tag: blume-20020308-entrypoints
2212    Description:
2213    
2214    Version number bumped to 110.39.1.  NEW BOOTFILES!
2215    
2216    Entrypoints: non-zero offset into a code object where execution should begin.
2217    
2218    - Added the notion of an entrypoint to CodeObj.
2219    - Added reading/writing of entrypoint info to Binfile.
2220    - Made runtime system bootloader aware of entrypoints.
2221    - Use the address of the label of the first function given to mlriscGen
2222      as the entrypoint.  This address is currently always 0, but it will
2223      not be 0 once we turn on block placement.
2224    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
2225      with entry points) from mlriscGen.
2226    
2227    ----------------------------------------------------------------------
2228    Name: Allen Leung
2229    Date: 2002/03/07 20:45:00 EST
2230    Tag: leunga-20020307-x86-cmov
2231    Description:
2232    
2233       Bug fixes for CMOVcc on x86.
2234    
2235       1. Added machine code generation for CMOVcc
2236       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
2237       3. CMOVcc cannot have an immediate operand as argument.
2238    
2239    ----------------------------------------------------------------------
2240    Name: Matthias Blume
2241    Date: 2002/03/07 16:15:00 EST
2242    Tag: blume-20020307-controls
2243    Description:
2244    
2245    This is a very large but mostly boring patch which makes (almost)
2246    every tuneable compiler knob (i.e., pretty much everything under
2247    Control.* plus a few other things) configurable via both the command
2248    line and environment variables in the style CM did its configuration
2249    until now.
2250    
2251    Try starting sml with '-h' (or, if you are brave, '-H')
2252    
2253    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
2254    implements the underlying generic mechanism.
2255    
2256    The interface to some of the existing such facilities has changed somewhat.
2257    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
2258    (The getFoo interface is still there for backward-compatibility, but its
2259    use is deprecated.)
2260    
2261    The ml-build script passes -Cxxx=yyy command-line arguments through so
2262    that one can now twiddle the compiler settings when using this "batch"
2263    compiler.
2264    
2265    TODO items:
2266    
2267    We should go through and throw out all controls that are no longer
2268    connected to anything.  Moreover, we should go through and provide
2269    meaningful (and correct!) documentation strings for those controls
2270    that still are connected.
2271    
2272    Currently, multiple calls to Controls.new are accepted (only the first
2273    has any effect).  Eventually we should make sure that every control
2274    is being made (via Controls.new) exactly once.  Future access can then
2275    be done using Controls.acc.
2276    
2277    Finally, it would probably be a good idea to use the getter-setter
2278    interface to controls rather than ref cells.  For the time being, both
2279    styles are provided by the Controls module, but getter-setter pairs are
2280    better if thread-safety is of any concern because they can be wrapped.
2281    
2282    *****************************************
2283    
2284    One bug fix: The function blockPlacement in three of the MLRISC
2285    backpatch files used to be hard-wired to one of two possibilities at
2286    link time (according to the value of the placementFlag).  But (I
2287    think) it should rather sense the flag every time.
2288    
2289    *****************************************
2290    
2291    Other assorted changes (by other people who did not supply a HISTORY entry):
2292    
2293    1. the cross-module inliner now works much better (Monnier)
2294    2. representation of weights, frequencies, and probabilities in MLRISC
2295       changed in preparation of using those for weighted block placement
2296       (Reppy, George)
2297    
2298    ----------------------------------------------------------------------
2299    Name: Lal George
2300    Date: 2002/03/07 14:44:24 EST 2002
2301    Tag: george-20020307-weighted-block-placement
2302    
2303    Tested the weighted block placement optimization on all architectures
2304    (except the hppa) using AMPL to generate the block and edge frequencies.
2305    Changes were required in the machine properties to correctly
2306    categorize trap instructions. There is an MLRISC flag
2307    "weighted-block-placement" that can be used to enable weighted block
2308    placement, but this will be ineffective without block/edge
2309    frequencies (coming soon).
2310    
2311    
2312    ----------------------------------------------------------------------
2313    Name: Lal George
2314    Date: 2002/03/05 17:24:48 EST
2315    Tag: george-20020305-linkage-cluster
2316    
2317    In order to support the block placement optimization, a new cluster
2318    is generated as the very first cluster (called the linkage cluster).
2319    It contains a single jump to the 'real' entry point for the compilation
2320    unit. Block placement has no effect on the linkage cluster itself, but
2321    all the other clusters  have full freedom in the manner in which they
2322    reorder blocks or functions.
2323    
2324    On the x86 the typical linkage code that is generated is:
2325       ----------------------
2326            .align 2
2327       L0:
2328            addl    $L1-L0, 72(%esp)
2329            jmp     L1
2330    
2331    
2332            .align  2
2333       L1:
2334       ----------------------
2335    
2336    72(%esp) is the memory location for the stdlink register. This
2337    must contain the address of the CPS function being called. In the
2338    above example, it contains the address of  L0; before
2339    calling L1 (the real entry point for the compilation unit), it
2340    must contain the address for L1, and hence
2341    
2342            addl $L1-L0, 72(%esp)
2343    
2344    I have tested this on all architectures except the hppa.The increase
2345    in code size is of course negligible
2346    
2347    ----------------------------------------------------------------------
2348    Name: Allen Leung
2349    Date: 2002/03/03 13:20:00 EST
2350    Tag: leunga-20020303-mlrisc-tools
2351    
2352      Added #[ ... ] expressions to mlrisc tools
2353    
2354    ----------------------------------------------------------------------
2355    Name: Matthias Blume
2356    Date: 2002/02/27 12:29:00 EST
2357    Tag: blume-20020227-cdebug
2358    Description:
2359    
2360    - made types in structure C and C_Debug to be equal
2361    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
2362    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
2363    
2364    ----------------------------------------------------------------------
2365    Name: Matthias Blume
2366    Date: 2002/02/26 12:00:00 EST
2367    Tag: blume-20020226-ffi
2368    Description:
2369    
2370    1. Fixed a minor bug in CM's "noweb" tool:
2371       If numbering is turned off, then truly don't number (i.e., do not
2372       supply the -L option to noweb).  The previous behavior was to supply
2373       -L'' -- which caused noweb to use the "default" line numbering scheme.
2374       Thanks to Chris Richards for pointing this out (and supplying the fix).
2375    
2376    2. Once again, I reworked some aspects of the FFI:
2377    
2378       A. The incomplete/complete type business:
2379    
2380       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
2381         gone!
2382       - ML types representing an incomplete type are now *equal* to
2383         ML types representing their corresponding complete types (just like
2384         in C).  This is still safe because ml-nlffigen will not generate
2385         RTTI for incomplete types, nor will it generate functions that
2386         require access to such RTTI.   But when ML code generated from both
2387         incomplete and complete versions of the C type meet, the ML types
2388         are trivially interoperable.
2389    
2390         NOTE:  These changes restore the full generality of the translation
2391         (which was previously lost when I eliminated functorization)!
2392    
2393       B. Enum types:
2394    
2395       - Structure C now has a type constructor "enum" that is similar to
2396         how the "su" constructor works.  However, "enum" is not a phantom
2397         type because each "T enum" has values (and is isomorphic to
2398         MLRep.Signed.int).
2399       - There are generic access operations for enum objects (using
2400         MLRep.Signed.int).
2401       - ml-nlffigen will generate a structure E_foo for each "enum foo".
2402         * The structure contains the definition of type "mlrep" (the ML-side
2403         representation type of the enum).  Normally, mlrep is the same
2404         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
2405         then mlrep will be defined as a datatype -- thus facilitating
2406         pattern matching on mlrep values.
2407         ("-ec" will be suppressed if there are duplicate values in an
2408          enumeration.)
2409         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
2410         will be generated for each C enum constant xxx.
2411         * Conversion functions m2i and i2m convert between mlrep and
2412         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
2413         * Coversion functions c and ml convert between mlrep and "tag enum".
2414         * Access functions (get/set) fetch and store mlrep values.
2415       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
2416         enumerations are merged into one single enumeration represented by
2417         structure E_'.
2418    
2419    ----------------------------------------------------------------------
2420    Name: Allen Leung
2421    Date: 2002/02/25 04:45:00 EST
2422    Tag: leunga-20020225-cps-spill
2423    
2424    This is a new implementation of the CPS spill phase.
2425    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
2426    In case of problems, replace it with the old file spill.sml
2427    
2428    The current compiler runs into some serious performance problems when
2429    constructing a large record.  This can happen when we try to compile a
2430    structure with many items.  Even a very simple structure like the following
2431    makes the compiler slow down.
2432    
2433        structure Foo = struct
2434           val x_1 = 0w1 : Word32.int
2435           val x_2 = 0w2 : Word32.int
2436           val x_3 = 0w3 : Word32.int
2437           ...
2438           val x_N = 0wN : Word32.int
2439        end
2440    
2441    The following table shows the compile time, from N=1000 to N=4000,
2442    with the old compiler:
2443    
2444    N
2445    1000   CPS 100 spill                           0.04u  0.00s  0.00g
2446           MLRISC ra                               0.06u  0.00s  0.05g
2447              (spills = 0 reloads = 0)
2448           TOTAL                                   0.63u  0.07s  0.21g
2449    
2450    1100   CPS 100 spill                           8.25u  0.32s  0.64g
2451           MLRISC ra                               5.68u  0.59s  3.93g
2452              (spills = 0 reloads = 0)
2453           TOTAL                                   14.71u  0.99s  4.81g
2454    
2455    1500   CPS 100 spill                           58.55u  2.34s  1.74g
2456           MLRISC ra                               5.54u  0.65s  3.91g
2457              (spills = 543 reloads = 1082)
2458           TOTAL                                   65.40u  3.13s  6.00g
2459    
2460    2000   CPS 100 spill                           126.69u  4.84s  3.08g
2461           MLRISC ra                               0.80u  0.10s  0.55g
2462              (spills = 42 reloads = 84)
2463           TOTAL                                   129.42u  5.10s  4.13g
2464    
2465    3000   CPS 100 spill                           675.59u  19.03s  11.64g
2466           MLRISC ra                               2.69u  0.27s  1.38g
2467              (spills = 62 reloads = 124)
2468           TOTAL                                   682.48u  19.61s  13.99g
2469    
2470    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
2471           MLRISC ra                               4.96u  0.27s  2.72g
2472              (spills = 85 reloads = 170)
2473           TOTAL                                   2375.26u  57.21s  48.00g
2474    
2475    As you can see the old cps spill module suffers from some serious
2476    performance problem.  But since I cannot decipher the old code fully,
2477    instead of patching the problems up, I'm reimplementing it
2478    with a different algorithm.  The new code is more modular,
2479    smaller when compiled, and substantially faster
2480    (O(n log n) time and O(n) space).  Timing of the new spill module:
2481    
2482    4000  CPS 100 spill                           0.02u  0.00s  0.00g
2483          MLRISC ra                               0.25u  0.02s  0.15g
2484             (spills=1 reloads=3)
2485          TOTAL                                   7.74u  0.34s  1.62g
2486    
2487    Implementation details:
2488    
2489    As far as I can tell, the purpose of the CPS spill module is to make sure the
2490    number of live variables at any program point (the bandwidth)
2491    does not exceed a certain limit, which is determined by the
2492    size of the spill area.
2493    
2494    When the bandwidth is too large, we decrease the register pressure by
2495    packing live variables into spill records.  How we achieve this is
2496    completely different than what we did in the old code.
2497    
2498    First, there is something about the MLRiscGen code generator
2499    that we should be aware of:
2500    
2501    o MLRiscGen performs code motion!
2502    
2503       In particular, it will move floating point computations and
2504       address computations involving only the heap pointer to
2505       their use sites (if there is only a single use).
2506       What this means is that if we have a CPS record construction
2507       statement
2508    
2509           RECORD(k,vl,w,e)
2510    
2511       we should never count the new record address w as live if w
2512       has only one use (which is often the case).
2513    
2514       We should do something similar to floating point, but the transformation
2515       there is much more complex, so I won't deal with that.
2516    
2517    Secondly, there are now two new cps primops at our disposal:
2518    
2519     1. rawrecord of record_kind option
2520        This pure operator allocates some uninitialized storage from the heap.
2521        There are two forms:
2522    
2523         rawrecord NONE [INT n]  allocates a tagless record of length n
2524         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
2525                                     and initializes the tag.
2526    
2527     2. rawupdate of cty
2528          rawupdate cty (v,i,x)
2529          Assigns to x to the ith component of record v.
2530          The storelist is not updated.
2531    
2532    We use these new primops for both spilling and increment record construction.
2533    
2534     1. Spilling.
2535    
2536        This is implemented with a linear scan algorithm (but generalized
2537        to trees).  The algorithm will create a single spill record at the
2538        beginning of the cps function and use rawupdate to spill to it,
2539        and SELECT or SELp to reload from it.  So both spills and reloads
2540        are fine-grain operations.  In contrast, in the old algorithm
2541        "spills" have to be bundled together in records.
2542    
2543        Ideally, we should sink the spill record construction to where
2544        it is needed.  We can even split the spill record into multiple ones
2545        at the places where they are needed.  But CPS is not a good
2546        representation for global code motion, so I'll keep it simple and
2547        am not attempting this.
2548    
2549     2. Incremental record construction (aka record splitting).
2550    
2551        Long records with many component values which are simulatenously live
2552        (recall that single use record addresses are not considered to
2553         be live) are constructed with rawrecord and rawupdate.
2554        We allocate space on the heap with rawrecord first, then gradually
2555        fill it in with rawupdate.  This is the technique suggested to me
2556        by Matthias.
2557    
2558        Some restrictions on when this is applicable:
2559        1. It is not a VECTOR record.  The code generator currently does not handle
2560           this case. VECTOR record uses double indirection like arrays.
2561        2. All the record component values are defined in the same "basic block"
2562           as the record constructor.  This is to prevent speculative
2563           record construction.
2564    
2565    ----------------------------------------------------------------------
2566    Name: Allen Leung
2567    Date: 2002/02/22 01:02:00 EST
2568    Tag: leunga-20020222-mlrisc-tools
2569    
2570    Minor bug fixes in the parser and rewriter
2571    
2572    ----------------------------------------------------------------------
2573    Name: Allen Leung
2574    Date: 2002/02/21 20:20:00 EST
2575    Tag: leunga-20020221-peephole
2576    
2577    Regenerated the peephole files.  Some contained typos in the specification
2578    and some didn't compile because of pretty printing bugs in the old version
2579    of 'nowhere'.
2580    
2581    ----------------------------------------------------------------------
2582    Name: Allen Leung
2583    Date: 2002/02/19 20:20:00 EST
2584    Tag: leunga-20020219-mlrisc-tools
2585    Description:
2586    
2587       Minor bug fixes to the mlrisc-tools library:
2588    
2589       1.  Fixed up parsing colon suffixed keywords
2590       2.  Added the ability to shut the error messages up
2591       3.  Reimplemented the pretty printer and fixed up/improved
2592           the pretty printing of handle and -> types.
2593       4.  Fixed up generation of literal symbols in the nowhere tool.
2594       5.  Added some SML keywords to to sml.sty
2595    
2596    ----------------------------------------------------------------------
2597    Name: Matthias Blume
2598    Date: 2002/02/19 16:20:00 EST
2599    Tag: blume-20020219-cmffi
2600    Description:
2601    
2602    A wild mix of changes, some minor, some major:
2603    
2604    * All C FFI-related libraries are now anchored under $c:
2605        $/c.cm      --> $c/c.cm
2606        $/c-int.cm  --> $c/internals/c-int.cm
2607        $/memory.cm --> $c/memory/memory.cm
2608    
2609    * "make" tool (in CM) now treats its argument pathname slightly
2610      differently:
2611        1. If the native expansion is an absolute name, then before invoking
2612           the "make" command on it, CM will apply OS.Path.mkRelative
2613           (with relativeTo = OS.FileSys.getDir()) to it.
2614        2. The argument will be passed through to subsequent phases of CM
2615           processing without "going native".  In particular, if the argument
2616           was an anchored path, then "make" will not lose track of that anchor.
2617    
2618    * Compiler backends now "know" their respective C calling conventions
2619      instead of having to be told about it by ml-nlffigen.  This relieves
2620      ml-nlffigen from one of its burdens.
2621    
2622    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
2623    
2624    * Export C_DEBUG and C_Debug from $c/c.cm.
2625    
2626    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
2627      subtyping relationship between incomplete pointers and their complete
2628      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
2629      with the convention of instantiating 'o with ('t, 'c) obj whenever
2630      the pointer target type is complete.  In the incomplete case, 'o
2631      will be instantiated with some "'c iobj" -- a type obtained by
2632      using one of the functors PointerToIncompleteType or PointerToCompleteType.
2633    
2634      Operations that work on both incomplete and complete pointer types are
2635      typed as taking an 'o ptr while operations that require the target to
2636      be known are typed as taking some ('t, 'c) obj ptr.
2637    
2638      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
2639      where void is an eqtype without any values.  This makes it possible
2640      to work on voidptr values using functions meant to operate on light
2641      incomplete pointers.
2642    
2643    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
2644      been vastly simplified.
2645    
2646    ----------------------------------------------------------------------
2647    Name: Matthias Blume
2648    Date: 2002/02/19 10:48:00 EST
2649    Tag: blume-20020219-pqfix
2650    Description:
2651    
2652    Applied Chris Okasaki's bug fix for priority queues.
2653    
2654    ----------------------------------------------------------------------
2655    Name: Matthias Blume
2656    Date: 2002/02/15 17:05:00
2657    Tag: Release_110_39
2658    Description:
2659    
2660    Last-minute retagging is becoming a tradition... :-(
2661    
2662    This is the working release 110.39.
2663    
2664    ----------------------------------------------------------------------
2665    Name: Matthias Blume
2666    Date: 2002/02/15 16:00:00 EST
2667    Tag: Release_110_39-orig
2668    Description:
2669    
2670    Working release 110.39.  New bootfiles.
2671    
2672    (Update: There was a small bug in the installer so it wouldn't work
2673    with all shells.  So I retagged. -Matthias)
2674    
2675    ----------------------------------------------------------------------
2676    Name: Matthias Blume
2677    Date: 2002/02/15 14:17:00 EST
2678    Tag: blume-20020215-showbindings
2679    Description:
2680    
2681    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
2682    the latter can be useful for exploring what bindings are available at
2683    the interactive prompt.  (The first function returns only the list
2684    of symbols that are really bound, the second prints those but also the
2685    ones that CM's autoloading mechanism knows about.)
2686    
2687    ----------------------------------------------------------------------
2688    Name: Matthias Blume
2689    Date: 2002/02/15 12:08:00 EST
2690    Tag: blume-20020215-iptrs
2691    Description:
2692    
2693    Two improvements to ml-nlffigen:
2694    
2695      1. Write files only if they do not exist or if their current contents
2696         do not coincide with what's being written.  (That is, avoid messing
2697         with the time stamps unless absolutely necessary.)
2698    
2699      2. Implement a "repository" mechanism for generated files related
2700         to "incomplete pointer types".   See the README file for details.
2701    
2702    ----------------------------------------------------------------------
2703    Name: Matthias Blume
2704    Date: 2002/02/14 11:50:00 EST
2705    Tag: blume-20020214-quote
2706    Description:
2707    
2708    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
2709    because of the new and improved tag generation scheme.  (Thanks to Allen
2710    Leung for pointing it out.)
2711    
2712    ----------------------------------------------------------------------
2713    Name: Lal George
2714    Date: 2002/02/14 09:55:27 EST 2002
2715    Tag: george-20020214-isabelle-bug
2716    Description:
2717    
2718    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
2719    of Isabelle on the x86.
2720    
2721    From Allen:
2722    -----------
2723     I've found the problem:
2724    
2725         in ra-core.sml, I use the counter "blocked" to keep track of the
2726         true number of elements in the freeze queue.  When the counter goes
2727         to zero, I skip examining the queue.  But I've messed up the
2728         bookkeeping in combine():
2729    
2730             else ();
2731             case !ucol of
2732               PSEUDO => (if !cntv > 0 then
2733                     (if !cntu > 0 then blocked := !blocked - 1 else ();
2734                                        ^^^^^^^^^^^^^^^^^^^^^^^
2735                      moveu := mergeMoveList(!movev, !moveu)
2736                     )
2737                  else ();
2738    
2739         combine() is called to coalesce two nodes u and v.
2740         I think I was thinking that if the move counts of u and v are both
2741         greater than zero then after they are coalesced then one node is
2742         removed from the freeze queue.  Apparently I was thinking that
2743         both u and v are of low degree, but that's clearly not necessarily true.
2744    
2745    
2746    02/12/2002:
2747        Here's the patch.  HOL now compiles.
2748    
2749        I don't know how this impact on performance (compile
2750        time or runtime).  This bug caused the RA (especially on the x86)
2751        to go thru the potential spill phase when there are still nodes on the
2752        freeze queue.
2753    
2754    
2755    
2756    
2757    ----------------------------------------------------------------------
2758    Name: Matthias Blume
2759    Date: 2002/02/13 22:40:00 EST
2760    Tag: blume-20020213-fptr-rtti
2761    Description:
2762    
2763    Fixed a bug in ml-nlffigen that was introduced with one of the previous
2764    updates.
2765    
2766    ----------------------------------------------------------------------
2767    Name: Matthias Blume
2768    Date: 2002/02/13 16:41:00 EST
2769    Tag: blume-20020213-cmlpq
2770    Description:
2771    
2772    Added new priority queue export symbols (which have just been added to
2773    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
2774    not compile and the installer would choke.)
2775    
2776    ----------------------------------------------------------------------
2777    Name: Matthias Blume
2778    Date: 2002/02/13 16:15:00 EST
2779    Tag: blume-20020213-various
2780    Description:
2781    
2782    1. More tweaks to ml-nlffigen:
2783    
2784       - better internal datastructures (resulting in slight speedup)
2785       - "-match" option requires exact match
2786       - "localized" gensym counters (untagged structs/unions nested within
2787         other structs/unions or within typedefs get a fresh counter; their
2788         tag will be prefixed by a concatenation of their parents' tags)
2789       - bug fixes (related to calculation of transitive closure of types
2790         to be included in the output)
2791    
2792    2. Minor Basis updates:
2793    
2794       - added implementations for List.collate and Option.app
2795    
2796    ----------------------------------------------------------------------
2797    Name: Matthias Blume
2798    Date: 2002/02/11 15:55:00 EST
2799    Tag: blume-20020211-gensym
2800    Description:
2801    
2802    Added a "-gensym" option to command line of ml-nlffigen.  This can be
2803    used to specify a "stem" -- a string that is inserted in all "gensym'd"
2804    names (ML structure names that correspond to unnamed C structs, unions,
2805    and enums), so that separate runs of ml-nlffigen do not clash.
2806    
2807    ----------------------------------------------------------------------
2808    Name: Matthias Blume
2809    Date: 2002/02/11 12:05:00 EST
2810    Tag: blume-20020211-gensml
2811    Description:
2812    
2813    A quick fix for a problem with GenSML (in the pgraph-util library):
2814    Make generation of toplevel "local" optional.  (Strictly speaking,
2815    signature definitions within "local" are not legal SML.)
2816    
2817    Other than that: updates to INSTALL and cm/TODO.
2818    
2819    ----------------------------------------------------------------------
2820    Name: Matthias Blume
2821    Date: 2002/02/08 15:00:00 EST
2822    Tag: blume-20020208-uniquepid
2823    Description:
2824    
2825    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
2826    
2827    1. The installer (config/install.sh) has gotten smarter:
2828    
2829         - Configuration options are a bit easier to specify now
2830           (in config/targets).
2831         - Bug in recognizing .tar.bz2 files fixed.
2832         - Installer automatically resolves dependencies between
2833           configuration options (e.g., if you ask for eXene, you will
2834           also get cml -- regardless whether you asked for it or not).
2835         - Installer can run in "quieter mode" by setting the environment
2836           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
2837           "completely silent", though.
2838         - Build HashCons library as part of smlnj-lib.
2839    
2840    2. A new scheme for assigning persistent identifiers to compilation
2841       units (and, by extension, to types etc.) has been put into place.
2842       This fixes a long-standing bug where types and even dynamic values
2843       can get internally confused, thereby compromising type safety
2844       (abstraction) and dynamic correctness.  See
2845    
2846         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
2847    
2848       for an example of how things could go wrong until now.
2849    
2850       The downside of the new scheme is that pids are not quite as
2851       persistent as they used to be: CM will generate a fresh pid
2852       for every compilation unit that it thinks it sees for the first
2853       time.  That means that if you compile starting from a clean, fresh
2854       source tree at two different times, you end up with different
2855       binaries.
2856    
2857       Cutoff recompilation, however, has not been compromised because
2858       CM keeps pid information in special caches between runs.
2859    
2860    ----------------------------------------------------------------------
2861    Name: Lal George
2862    Date: 2002/02/07 15:34:13 EST 2002
2863    Tag: <none>
2864    Description:
2865    
2866    Compilers that generate assembly code may produce  global labels
2867    whose value is resolved at link time. The various peephole optimization
2868    modules did not take this in account.
2869    
2870    TODO. The Labels.addrOf function should really return an option
2871    type so that clients are forced to deal with this issue, rather
2872    than an exception being raised.
2873    
2874    ----------------------------------------------------------------------
2875    Name: Lal George
2876    Date: 2002/02/06 13:55:02 EST
2877    Tag: george-20020206-ra-breakup
2878    Description:
2879    
2880    1. A bug fix from Allen.
2881    
2882        A typo causes extra fstp %st(0)'s to be generated at compensation
2883        edges, which might cause stack underflow traps at runtime.  This
2884        occurs in fft where there are extraneous fstps right before the 'into'
2885        trap instruction (in this case they are harmless since none of the
2886        integers overflow.)
2887    
2888    2. Pulled out various utility modules that were embedded in the modules
2889       of the register allocator. I need these modules for other purposes, but
2890       they are not complete enough to put into a library (just yet).
2891    ----------------------------------------------------------------------
2892    Name: Matthias Blume
2893    Date: 2002/01/31 16:05:00 EST
2894    Tag: blume-20020131-sparc-ccalls
2895    Description:
2896    
2897    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
2898       of extra stack space by mistake.  Fixed.
2899    
2900    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
2901    
2902    ----------------------------------------------------------------------
2903    Name: Allen Leung
2904    Date: 2002/01/30
2905    Tag: leunga-20020130-nowhere-bug-fix
2906    Description:
2907    
2908       MLRISC bug fixes:
2909       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
2910       2. MachineInt.fromString was negating its value.
2911    
2912    ----------------------------------------------------------------------
2913    Name: Matthias Blume
2914    Date: 2002/01/29
2915    Tag: blume-20020129-INSTALL
2916    Description:
2917    
2918    - Added somewhat detailed installation instructions (file INSTALL).
2919    - Fixed curl-detection bug in config/install.sh.
2920    - It is now possible to select the URL getter using the URLGETTER
2921      environment variable:
2922    
2923          not set / "unknown"      --> automatic detection (script tries wget,
2924                                       curl, and lynx)
2925          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
2926                                       how to properly invoke them)
2927          other                    --> use $URLGETTER directly, it must take
2928                                       precisely two command-line arguments
2929                                       (source URL and destination file name)
2930    
2931    ----------------------------------------------------------------------
2932    Name: Matthias Blume
2933    Date: 2002/01/28
2934    Tag: blume-20020128-sparc-ccalls
2935    Description:
2936    
2937    - Fixed problem with calculation of "used" registers in sparc-c-calls.
2938    - Make use of the allocParam argument in sparc-c-calls.
2939    
2940    ----------------------------------------------------------------------
2941    Name: Matthias Blume
2942    Date: 2002/01/28
2943    Tag: blume-20020128-allocParam
2944    Description:
2945    
2946    John Reppy:  Changes c-calls API to accept client-callback for
2947    allocating extra stack space.
2948    me: Corresponding changes to mlriscGen (using a dummy argument that
2949        does not change the current behavior).
2950    
2951    ----------------------------------------------------------------------
2952    Name: Matthias Blume
2953    Date: 2002/01/28 12:00:00
2954    Tag: Release_110_38
2955    Description:
2956    
2957    This time for real!!!
2958    
2959    ----------------------------------------------------------------------
2960    Name: Matthias Blume
2961    Date: 2002/01/28 10:56:00 EST
2962    Tag: blume-20020128-retraction
2963    Description:
2964    
2965    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
2966       with blume-Release_110_38-retracted.)
2967    
2968    1. Fixed a problem with incorrect rounding modes in real64.sml.
2969       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
2970    
2971    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
2972       and enums fixed.  The naming of corresponding ML identifiers should
2973       now be consistent again.
2974    
2975    ----------------------------------------------------------------------
2976    Name: Allen Leung
2977    Date: 2002/01/27
2978    Tag: leunga-20020127-nowhere
2979    Description:
2980    
2981       Added a target called nowhere in the configuration scripts.
2982       Enabling this will build the MLRISC 'nowhere' tool (for translating
2983       programs with where-clauses into legal SML code) during installation.
2984    
2985    ----------------------------------------------------------------------
2986    Name: Matthias Blume
2987    Date: 2002/01/25 21:27:00 EST
2988    Tag: blume-Release_110_38-retracted
2989    Description:
2990    
2991    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
2992    
2993    README will be added later.
2994    
2995    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
2996    (Release_110_38) removed.  Reason: Last-minute bug fixes.
2997    
2998    ----------------------------------------------------------------------
2999    Name: Matthias Blume
3000    Date: 2002/01/25
3001    Tag: blume-20020125-ffi
3002    Description:
3003    
3004    A large number of tweaks and improvements to ml-nlffi-lib and
3005    ml-nlffigen:
3006    
3007       - ML represenation types have been streamlined
3008       - getter and setter functions work with concrete values, not abstract
3009         ones where possible
3010       - ml-nlffigen command line more flexible (see README file there)
3011       - some bugs have been fixed (hopefully)
3012    
3013    ----------------------------------------------------------------------
3014    Name: Lal George
3015    Date: 2002/01/24
3016    Tag: george-20020124-risc-ra-interface
3017    Description:
3018    
3019       There is a dramatic simplification in the interface to the
3020       register allocator for RISC architectures as a result of making
3021       parallel copy instructions explicit.
3022    
3023    ----------------------------------------------------------------------
3024    Name: Matthias Blume
3025    Date: 2002/01/22
3026    Tag: blume-20020122-x86-ccalls
3027    Description:
3028    
3029    Bug fix for c-calls on x86 (having to do with how char- and
3030    short-arguments are being handled).
3031    
3032    ----------------------------------------------------------------------
3033    Name: Matthias Blume
3034    Date: 2002/01/21
3035    Tag: blume-20020121-ff
3036    Description:
3037    
3038    Another day of fiddling with the FFI...
3039    
3040    1. Bug fix/workaround:  CKIT does not complain about negative array
3041       dimensions, so ml-nlffigen has to guard itself against this possibility.
3042       (Otherwise a negative dimension would send it into an infinite loop.)
3043    
3044    2. Some of the abstract types (light objects, light pointers, most "base"
3045       types) in structure C are now eqtypes.
3046    
3047    3. Added constructors and test functions for NULL function pointers.
3048    
3049    ----------------------------------------------------------------------
3050    Name: Matthias Blume
3051    Date: 2002/01/18
3052    Tag: blume-20020118-ready-for-new-release
3053    Description:
3054    
3055    Made config/srcarchiveurl point to a new place.  (Will provide boot
3056    files shortly.)
3057    
3058    Maybe we christen this to be 110.38?
3059    
3060    ----------------------------------------------------------------------
3061    Name: Matthias Blume
3062    Date: 2002/01/18
3063    Tag: blume-20020118-more-ffifiddle
3064    Description:
3065    
3066    Today's FFI fiddling:
3067    
3068      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
3069        C.Set.* that use concrete (MLRep.*) arguments and results instead
3070        of abstract ones.
3071    
3072      - Provided word-style bit operations etc. for "int" representation
3073        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
3074        and Long.
3075    
3076    ----------------------------------------------------------------------
3077    Name: Matthias Blume
3078    Date: 2002/01/18
3079    Tag: blume-20020118-use-x86-fp
3080    Description:
3081    
3082    Now that x86-fast-fp seems to be working, I turned it back on again
3083    by default.  (Seems to work fine now, even with the FFI.)
3084    
3085    Other than that, I added some documentation about the FFI to
3086    src/ml-nlffigen/README and updated the FFI test examples in
3087    src/ml-nlffi-lib/Tests/*.
3088    
3089    ----------------------------------------------------------------------
3090    Name: Allen Leung
3091    Date: 2002/01/17
3092    Tag: leunga-20020117-x86-fast-fp-call
3093    Description:
3094    
3095       1. Fixed a problem with handling return fp values when x86's fast fp
3096          mode is turned on.
3097    
3098       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
3099          of %f32.
3100    
3101       3. Added a constructor INT32lit to the ast of MLRISC tools.
3102    
3103    ----------------------------------------------------------------------
3104    Name: Matthias Blume
3105    Date: 2002/01/16
3106    Tag: blume-20020116-ffifiddle
3107    Description:
3108    
3109    More fiddling with the FFI interface:
3110    
3111     - Make constness 'c instead of rw wherever possible.  This eliminates
3112       the need for certain explicit coercions.  (However, due to ML's
3113       value polymorphism, there will still be many cases where explicit
3114       coercions are necessary.  Phantom types are not the whole answer
3115       to modeling a subtyping relationship in ML.)
3116    
3117     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
3118    
3119     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
3120       clumsy workaround for varargs problem.)
3121    
3122    ----------------------------------------------------------------------
3123    Name: Lal George
3124    Date: 2002/01/15
3125    Tag: <none>
3126    Description:
3127    
3128    1. Since COPY instructions are no longer native to the architecture,
3129       a generic functor can be used to implement the expandCopies function.
3130    
3131    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
3132       TEXT segment.
3133    
3134    ----------------------------------------------------------------------
3135    Name: Matthias Blume
3136    Date: 2002/01/15
3137    Tag: blume-20020115-ffiupdates
3138    Description:
3139    
3140    1. Fix for bug resulting in single-precision float values being returned
3141       incorrectly from FFI calls.
3142    
3143    2. Small modifications to C FFI API:
3144    
3145        - memory-allocation routines return straight objects (no options)
3146          and raise an exception in out-of-memory situations
3147        - unsafe extensions to cast between function pointers and pointers
3148          from/to ints
3149        - added structure C_Debug as an alternative to structure C where
3150          pointer-dereferencing (|*| and |*!) always check for null-pointers
3151        - added open_lib' to DynLinkage;  open_lib' works like open_lib
3152          but also takes a (possibly empty) list of existing library handles
3153          that the current library depends on
3154    
3155    ----------------------------------------------------------------------
3156    Name: Matthias Blume
3157    Date: 2002/01/10
3158    Tag: blume-20020110-newffigen
3159    Description:
3160    
3161    1. Updates to portable graph code.
3162    
3163    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
3164       more scalable now so that even huge interfaces such as the one
3165       for GTK compile in finite time and space. :-)
3166       See src/ml-nlffigen/README for details on what's new.
3167    
3168    ----------------------------------------------------------------------
3169    Name: Lal George
3170    Date: 2001/01/09 14:31:35 EST 2002
3171    Tag: george-20011206-rm-native-copy
3172    Description:
3173    
3174            Removed the native COPY and FCOPY instructions
3175            from all the architectures and replaced it with the
3176            explicit COPY instruction from the previous commit.
3177    
3178            It is now possible to simplify many of the optimizations
3179            modules that manipulate copies. This has not been
3180            done in this change.
3181    
3182    ----------------------------------------------------------------------
3183    Name: Lal George
3184    Date: 2001/12/06 16:50:13 EST 2001
3185    Tag: george-20011206-mlrisc-instruction
3186    Description:
3187    
3188    Changed the representation of instructions from being fully abstract
3189    to being partially concrete. That is to say:
3190    
3191      from
3192            type instruction
3193    
3194      to
3195            type instr                              (* machine instruction *)
3196    
3197            datatype instruction =
3198                LIVE of {regs: C.cellset, spilled: C.cellset}
3199              | KILL of {regs: C.cellset, spilled: C.cellset}
3200              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
3201              | ANNOTATION of {i: instruction, a: Annotations.annotation}
3202              | INSTR of instr
3203    
3204    This makes the handling of certain special instructions that appear on
3205    all architectures easier and uniform.
3206    
3207    LIVE and KILL say that a list of registers are live or killed at the
3208    program point where they appear. No spill code is generated when an
3209    element of the 'regs' field is spilled, but the register is moved to
3210    the 'spilled' (which is present, more for debugging than anything else).
3211    
3212    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
3213    We used to generate:
3214    
3215            DEFFREG f1
3216            f1 := f2 + f3
3217            trapb
3218    
3219    but now generate:
3220    
3221            f1 := f2 + f3
3222            trapb
3223            LIVE {regs=[f1,f2,f3], spilled=[]}
3224    
3225    Furthermore, the DEFFREG (hack) required that all floating point instruction
3226    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
3227    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
3228    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
3229    this architecture).
3230    
3231    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
3232    all the architectures. This will result in further simplification of the
3233    register allocator that must be aware of them for coalescing purposes, and
3234    will also simplify certain aspects of the machine description that provides
3235    callbacks related to parallel copies.
3236    
3237    ANNOTATION should be obvious, and now INSTR represents the honest to God
3238    machine instruction set!
3239    
3240    The <arch>/instructions/<arch>Instr.sml files define certain utility
3241    functions for making porting easier -- essentially converting upper case
3242    to lower case. All machine instructions (of type instr) are in upper case,
3243    and the lower case form generates an MLRISC instruction. For example on
3244    the alpha we have:
3245    
3246      datatype instr =
3247         LDA of {r:cell, b:cell, d:operand}
3248       | ...
3249    
3250      val lda : {r:cell, b:cell, d:operand} -> instruction
3251        ...
3252    
3253    where lda is just (INSTR o LDA), etc.
3254    
3255    ----------------------------------------------------------------------
3256    Name: Matthias Blume
3257    Date: 2001/11/22 21:40:00 EST
3258    Tag: Release_110_37
3259    Description:
3260    
3261    Release 110.37.  This time for real.
3262    
3263    ----------------------------------------------------------------------
3264    Name: Matthias Blume
3265    Date: 2001/11/21 16:35:00 EST
3266    Tag: blume-20011121-foot-in-mouth
3267    Description:
3268    
3269    Removed the "Release_110_37" tag because of a serious bug.
3270    This will be re-tagged once the bug is fixed.
3271    
3272    ----------------------------------------------------------------------
3273    Name: Matthias Blume
3274    Date: 2001/11/21 16:14:00 EST
3275    Tag: blume-20011121-forgottenfile
3276    Description:
3277    
3278    Forgot to add a file.  (Just a .tex-file -- part of
3279    the CM manual source.)
3280    
3281    ----------------------------------------------------------------------
3282    Name: Matthias Blume
3283    Date: 2001/11/21 16:10:00 EST
3284    Tag: blume-20011121-invalid_110_37
3285    Description:
3286    
3287    Note: I removed the original tag "Release_110_37" from this commit
3288          because we found a serious bug in all non-x86 backends.
3289          - Matthias
3290    
3291    1. Modifications to the SML/NJ code generator and to the runtime system
3292       so that code object name strings are directly inserted into code
3293       objects at code generation time.  The only business the runtime system
3294       has with this is now to read the name strings on occasions.
3295       (The encoding of the name string has also changed somewhat.)
3296    
3297    2. CM now implements a simple "set calculus" for specifying export lists.
3298       In particular, it is now possible to refer to the export lists of
3299       other libraries/groups/sources and form unions as well as differences.
3300       See the latest CM manual for details.
3301    
3302    3. An separate notion of "proxy" libraries has again be eliminated from
3303       CM's model.  (Proxy libraries are now simply a special case of using
3304       the export list calculus.)
3305    
3306    4. Some of the existing libraries now take advantage of the new set
3307       calculus.
3308       (Notice that not all libraries have been converted because some
3309       of the existing .cm-files are supposed to be backward compatible
3310       with 110.0.x.)
3311    
3312    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
3313       "General.exnMessage"!  The former relies on a certain hook to be
3314       initialized, and that often does not happen in the stand-alone case.)
3315    
3316    ----------------------------------------------------------------------
3317    Name: Lal George
3318    Date: 2001/11/21  13:56:18 EST
3319    Tag: george-2001121-pseudo-ops
3320    Description:
3321    
3322      Implemented a complete redesign of MLRISC pseudo-ops. Now there
3323      ought to never be any question of incompatabilities with
3324      pseudo-op syntax expected by host assemblers.
3325    
3326      For now, only modules supporting GAS syntax are implemented
3327      but more should follow, such as MASM, and vendor assembler
3328      syntax, e.g. IBM as, Sun as, etc.
3329    
3330    ----------------------------------------------------------------------
3331    Name: Matthias Blume
3332    Date: 2001/11/14 11:52:00 EST
3333    Tag: blume-20011114-srcname
3334    Description:
3335    
3336    1. Routed the name of the current source file to mlriscgen where it
3337       should be directly emitted into the code object.  (This last part
3338       is yet to be done.)
3339    
3340    2. Some cleanup of the pgraph code to make it match the proposal that
3341       I put out the other day.  (The proposal notwithstanding, things are
3342       still in flux here.)
3343    
3344    ----------------------------------------------------------------------
3345    Name: Lal George
3346    Date: 2001/11/14 09:44:04 EST
3347    Tag:
3348    Description:
3349    
3350      Fix for a backpatching bug reported by Allen.
3351    
3352      Because the boundary between short and long span-dependent
3353      instructions is +/- 128, there are an astounding number of
3354      span-dependent instructions whose size is over estimated.
3355    
3356      Allen came up with the idea of letting the size of span
3357      dependent instructions be non-monotonic, for a maxIter
3358      number of times, after which the size must be monotonically
3359      increasing.
3360    
3361      This table shows the number of span-dependent instructions
3362      whose size was over-estimated as a function of maxIter, for the
3363      file Parse/parse/ml.grm.sml:
3364    
3365         maxIter            # of instructions:
3366            10                      687
3367            20                      438
3368            30                      198
3369            40                        0
3370    
3371      In compiling the compiler, there is no significant difference in
3372      compilation speed between maxIter=10 and maxIter=40. Actually,
3373      my measurements showed that maxIter=40 was a tad faster than
3374      maxIter=10! Also 96% of the  files in the compiler reach a fix
3375      point within 13 iterations, so fixing maxIter at 40, while high,
3376      is okay.
3377    
3378    ----------------------------------------------------------------------
3379    Name: Matthias Blume
3380    Date: 2001/10/31 15:25:00 EST
3381    Tag: blume-20011031-pgraph
3382    Description:
3383    
3384    CKIT:
3385    * Changed the "Function" constructor of type Ast.ctype to carry optional
3386      argument identifiers.
3387    * Changed the return type of TypeUtil.getFunction accordingly.
3388    * Type equality ignores the argument names.
3389    * TypeUtil.composite tries to preserve argument names but gives up quickly
3390      if there is a mismatch.
3391    
3392    installation script:
3393    * attempts to use "curl" if available (unless "wget" is available as well)
3394    
3395    CM:
3396    * has an experimental implementation of "portable graphs" which I will
3397      soon propose as an implementation-independent library format
3398    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
3399    
3400    NLFFI-LIB:
3401    * some cleanup (all cosmetic)
3402    
3403    NLFFIGEN:
3404    * temporarily disabled the mechanism that suppresses ML output for
3405      C definitions whose identifiers start with an underscore character
3406    * generate val bindings for enum constants
3407    * user can request that only one style (light or heavy) is being used;
3408      default is to use both (command-line arguments: -heavy and -light)
3409    * fixed bug in handling of function types involving incomplete pointers
3410    * generate ML entry points that take record arguments (i.e., using
3411      named arguments) for C functions that have a prototype with named
3412      arguments
3413      (see changes to CKIT)
3414    
3415    ----------------------------------------------------------------------
3416    Name: Allen Leung
3417    Date: 2001/10/27 20:34:00 EDT
3418    Tag: leunga-20011027-x86-fast-fp-call
3419    Description:
3420    
3421       Fixed the bug described in blume-20010920-slowfp.
3422    
3423       The fix involves
3424          1. generating FCOPYs in FSTP in ia32-svid
3425          2. marking a CALL with the appropriate annotation
3426    
3427    ----------------------------------------------------------------------
3428    Name: Matthias Blume
3429    Date: 2001/10/16 11:32:00 EDT
3430    Tag: blume-20011016-netbsd
3431    Description:
3432    
3433    Underscore patch from Chris Richards (fixing problem with compiling
3434    runtime system under recent NetBSD).
3435    
3436    ----------------------------------------------------------------------
3437    Name: Allen Leung
3438    Date: 2001/10/12 17:18:32 EDT 2001
3439    Tag: leung-20011012-x86-printflowgraph
3440    Description:
3441    
3442    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
3443    
3444    ----------------------------------------------------------------------
3445    Name: Lal George
3446    Date: 2001/10/11 23:51:34 EDT
3447    Tag: george-20011011-too-many-instrs
3448    Description:
3449    
3450    The representation of a program point never expected to see more
3451    than 65536 instructions in a basic block!
3452    
3453    ----------------------------------------------------------------------
3454    Name: Lal George
3455    Date: 2001/10/09 09:41:37 EDT
3456    Tag: george-20011008-mlrisc-labels
3457    Description:
3458    
3459    Changed the machine description files to support printing of
3460    local and global labels in assembly code, based on host assembler
3461    conventions.
3462    
3463    ----------------------------------------------------------------------
3464    Name: Matthias Blume
3465    Date: 2001/09/25 15:25:00 EDT
3466    Tag: blume-20010925-exninfo
3467    Description:
3468    
3469    I provided a non-hook implementation of exnName (at the toplevel) and
3470    made the "dummy" implementation of exnMessage (at the toplevel) more
3471    useful: if nothing gets "hooked in", then at least you are going to
3472    see the exception name and a message indicating why you don't see more.
3473    
3474    [For the time being, programs that need exnMessage and want to use
3475    ml-build should either use General.exnMessage (strongly recommended) or
3476    refer to structure General at some other point so that CM sees a
3477    static dependency.]
3478    
3479    [Similar remarks go for "print" and "use":  If you want to use their
3480    functionality in stand-alone programs generated by ml-build, then use
3481    TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
3482    
3483    ----------------------------------------------------------------------
3484    Name: Matthias Blume
3485    Date: 2001/09/20 17:28:00 EDT
3486    Tag: blume-20010920-slowfp
3487    Description:
3488    
3489    Allen says that x86-fast-fp is not safe yet, so I turned it off again...
3490    
3491    ----------------------------------------------------------------------
3492    Name: Matthias Blume
3493    Date: 2001/09/20 17:20:00 EDT
3494    Tag: blume-20010920-canonicalpaths
3495    Description:
3496    
3497    0. Updated the BOOT file (something that I forgot to do earlier).
3498    
3499    1. Small internal change to CM so that it avoids "/../" in filenames
3500       as much as possible (but only where it is safe).
3501    
3502    2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
3503       that arguments that contain delimiters are passed through correctly.
3504       This change also means that all "special" arguments of the form
3505       @SMLxxx... must come first.
3506    
3507    3. Changed install script to put relative anchor names for tool commands
3508       into pathconfig.
3509    
3510    ----------------------------------------------------------------------
3511    Name: Matthias Blume
3512    Date: 2001/09/18 15:35:00 EDT
3513    Tag: blume-20010918-readme11036
3514    Description:
3515    
3516    Added README files.
3517    
3518    ----------------------------------------------------------------------
3519    Name: Matthias Blume
3520    Date: 2001/09/18 11:45:00 EDT
3521    Tag: Release_110_36 (retag)
3522    Description:
3523    
3524    Fixed mistake in config/preloads. Retagged as 110.36.
3525    
3526    ----------------------------------------------------------------------
3527    Name: Matthias Blume
3528    Date: 2001/09/18 09:40:00 EDT
3529    Tag: Release_110_36_orig (tag changed)
3530    Description:
3531    
3532    New version (110.36).  New bootfiles.
3533    
3534    ----------------------------------------------------------------------
3535    Name: Matthias Blume
3536    Date: 2001/09/14 16:15:00 EDT
3537    Tag: blume-20010914-x86fastfp
3538    Description:
3539    
3540    John committed some changes that Allen made, in particular a (hopefully)
3541    correctly working version of the x86-fp module.
3542    
3543    I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
3544    flag to "true".  Everything seems to compile to a fixpoint ok, and
3545    "mandelbrot" speeds up by about 15%.
3546    
3547    ----------------------------------------------------------------------
3548    Name: Matthias Blume
3549    Date: 2001/09/13 11:20:00 EDT
3550    Tag: blume-20010913-minimal
3551    Description:
3552    
3553    1. Stefan Monnier's patch to fix a miscompilation problem that
3554       was brought to light by John Reppy's work on Moby.
3555    
3556    2. Implemented a minimal "structure Compiler" that contains just
3557       "version" and "architecture".  The minimal version will be
3558       available when the full version is not.  This is for backward-
3559       compatibility with code that wants to test Compiler.version.
3560    
3561    ----------------------------------------------------------------------
3562    Name: Matthias Blume
3563    Date: 2001/08/28 14:03:00 EDT
3564    Tag: blume-20010828-ml-lex
3565    Description:
3566    
3567    Fix for bug 1581, received from Neophytos Michael.
3568    
3569    ----------------------------------------------------------------------
3570    Name: Matthias Blume
3571    Date: 2001/08/27 11:20:00 EDT
3572    Tag: blume-20010827-readme11035
3573    Description:
3574    
3575    Fleshed out the README file for 110.35.
3576    
3577    ----------------------------------------------------------------------
3578    Name: Matthias Blume
3579    Date: 2001/08/24 17:10:00 EDT
3580    Tag: Release_110_35
3581    Description:
3582    
3583    New version number (110.35).  New bootfiles.
3584    
3585    ----------------------------------------------------------------------
3586    Name: Lal George
3587    Date: 2001/08/24 13:47:18 EDT 2001
3588    Tag: george-20010824-MLRISC-graphs
3589    Description:
3590    
3591     removed clusters from MLRISC completely and replaced with graphs.
3592    
3593    ----------------------------------------------------------------------
3594    Name: Matthias Blume
3595    Date: 2001/08/23 17:50:00 EDT
3596    Tag: blume-20010823-toplevel
3597    Description:
3598    
3599    - some reorganization of the code that implements various kinds of
3600      environments in the compiler (static, dynamic, symbolic, combined)
3601    - re-implemented the EnvRef module so that evalStream works properly
3602      (if the stream contains references to "use", "CM.make", etc.)
3603    - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
3604    
3605    ----------------------------------------------------------------------
3606    Name: Matthias Blume
3607    Date: 2001/08/20 15:50 EDT
3608    Tag: blume20010820-slipup
3609    Description:
3610    
3611    I forgot to commit a few files.  Here they are...
3612    
3613    ----------------------------------------------------------------------
3614    Name: Matthias Blume
3615    Date: 2001/08/20 15:35:00 EDT
3616    Tag: blume-20010820-debugprof
3617    Description:
3618    
3619    !!!! NEW BOOTFILES !!!!
3620    
3621    This is another round of reorganizing the compiler sources.  This
3622    time the main goal was to factor out all the "instrumentation"
3623    passes (for profiling and backtracing) into their own library.
3624    The difficulty was to do it in such a way that it does not depend