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