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 680, Mon Jul 3 06:35:55 2000 UTC revision 1633, Tue Sep 28 03:08:33 2004 UTC
# Line 8  Line 8 
8  The form of an entry should be:  The form of an entry should be:
9    
10  Name:  Name:
11  Date:  Date: yyyy/mm/dd
12  Tag: <post-commit CVS tag>  Tag: <post-commit CVS tag>
13  Description:  Description:
14    
15    ----------------------------------------------------------------------
16    Name: Matthias Blume (blume (at) tti - c (dot) org)
17    Date: 2004/09/27 22:08:00 CDT
18    Tag: blume-20040927-controls
19    Description:
20    
21    Added two pieces of functionality to the Controls interface:
22    
23    1.   val save'restore: 'a control -> unit -> unit
24    
25       grabs the current value of the control in stage 1 and restores it
26       in stage 2.
27    
28    2.   val set' : 'a control * 'a -> unit -> unit
29    
30       stores the given value into the control in stage 2 (i.e., delayed)
31       but does all error checking in stage 1.
32       (This is for string controls that need to do parse their argument --
33        something that might fail.  In some cases, notably in CM, one
34        already knows the intended argument but wants to delay the actual
35        assignment until a time when error recovery would be more difficult.)
36    
37    Changed the handling of controls in tool arguments to classes "sml" and
38    "lazysml":
39       - use Controls.save'restore as a more robust way of restoring the
40         old value (in particular: without having to re-parse the string)
41       - use controls to handle the "overload" keyword in the init group
42         (I believe this change actually fixes a long-standing obscure bug.)
43    
44    ----------------------------------------------------------------------
45    Name: Matthias Blume (blume (at) tti - c (dot) org)
46    Date: 2004/09/27 17:00:00 CDT
47    Tag: blume-20040927-lazysml
48    Description:
49    
50    Added a new tool class called "lazysml" to CM's tool chest.  The only
51    difference to "sml" is that compilation is done with Control.lazysml
52    set to true.  A source of class "lazysml" is automatically recognized
53    by a file name suffix of ".lml".
54    
55    In addition to the above feature, the original class "sml" now also
56    supports a tool argument "lazy" which has the same effect.  As a
57    result, the following three lines are equivalent:
58    
59        foo.sml : lazysml
60        foo.sml : sml (lazy)
61        foo.sml (lazy)
62    
63    The setting goes into effect both during parsing and during
64    compilation.  The original setting is restored right after parsing and
65    after compilation, respectively.
66    
67    In addition to all the above, there is also a general mechanism to set
68    ANY of the "controls" that are available at the command line via
69    "-C..." on a per-sml-file basis.  The same rules that apply for "lazy"
70    apply as well. (In fact, "lazy" is implemented as a special case of
71    the general mechanism.)
72    
73    The .cm file syntax uses a new keyword tool argument called "with".
74    There are several ways of indicating the desired settings:
75    
76        foo.sml (with:parser.quotations=true)
77        foo.sml (with:(name:parser.quotations value:true))
78        foo.sml (with:(name:name1 value:value1 name:name2 value:value2 ...))
79        foo.sml (with:(name1=value1 name2=value2 ...))
80        foo.sml (with:(name1=value1 name:name2 value:value2 name3=value3 ...))
81    
82        etc.
83    
84    Another possible abbreviation is to leave out the =v or value:v part
85    if the name refers to a boolean control (in which case the value is
86    taken to be true).  Thus, one could get lazy sml also by saying:
87    
88        foo.sml (with:parser.lazy-keyword=true)
89        foo.sml (with:parser.lazy-keyword)
90        foo.sml (with:(name:parser.lazy-keyword value:true))
91        foo.sml (with:(name:parser.lazy-keyword))
92    
93    ----------------------------------------------------------------------
94    Name: Matthias Blume (blume (at) tti - c (dot) org)
95    Date: 2004/09/24 16:22:00 CDT
96    Tag: blume-20040924-ppc-long-branch
97    Description:
98    
99    Turned message about "emiting long form of branch" off by default.
100    Added a control flag to turn it back on when desired.
101    
102    ----------------------------------------------------------------------
103    Name: Matthias Blume (blume (at) tti - c (dot) org)
104    Date: 2004/09/24 16:05:00 CDT
105    Tag: blume-20040924-rounding
106    Description:
107    
108    Applied patch for setting rounding modes under Mac OS X.  Thanks to
109    Melissa O'Neill for providing the code!
110    
111    ----------------------------------------------------------------------
112    Name: Matthias Blume (blume (at) tti - c (dot) org)
113    Date: 2004/09/23 17:30:00 CDT
114    Tag: blume-20040923-envvars
115    Description:
116    
117    1. Changed definition of type ControlRegistry.registry_tree to
118       include control_info (i.e., the name of the controlling environment
119       variable).
120    
121    2. Added command-line flags -e and -E to print the names of environment
122       variables that can be used to control internal settings.
123       (This uses the new API mentioned in 1.)
124    
125    ----------------------------------------------------------------------
126    Name: Matthias Blume (blume (at) tti - c (dot) org)
127    Date: 2004/09/13 16:50:00 CDT
128    Tag: Release_110_49
129    Description:
130    
131    New working version (110.49).  NEW BOOTFILES!
132    
133    ----------------------------------------------------------------------
134    Name: Matthias Blume
135    Date: 2004/09/13 16:20:00 CDT
136    Tag: blume-20040913-config-mlrisc
137    Description:
138    
139    Put target "mlrisc" back into the default list.
140    (There is no harm in having it, and some users have expressed their
141    wish to have "mlrisc" included by default.)
142    
143    ----------------------------------------------------------------------
144    Name: John Reppy
145    Date: 2004/09/13
146    Tag: jhr-20040913-signals
147    Description:
148    
149    Fixed the signal masking code to properly nest mask/unmask operations
150    on a per-signal basis.
151    
152    ----------------------------------------------------------------------
153    Name: Matthias Blume (blume (at) tti - c (dot) org)
154    Date: 2004/09/08 13:20:00 CDT
155    Tag: blume-20040908-heap-magic
156    Description:
157    
158    Bumped the heap macig number to 0x09082004 to account for the changed
159    layout of the ML frame under MacOS X.
160    
161    ----------------------------------------------------------------------
162    Name: Allen Leung (leunga (at) reservoir (dot) com)
163    Date: 2004/09/03 11:26:00 EST
164    Tag: leunga-20040903-cygwin-install
165    Description:
166    
167       Added a patch to _arch-n-opsys to enable the Cygwin runtime.
168    The Cygwin runtime is turned on by setting the environment
169    variable SMLNJ_CYGWIN_RUNTIME to 1.
170    
171    ----------------------------------------------------------------------
172    Name: Matthias Blume (blume (at) tti - c (dot) org)
173    Date: 2004/08/31 17:15:00 CDT
174    Tag: blume-20040831-core
175    Description:
176    
177    Added some exports to src/compiler/core.cm upon request by J. Joemann.
178    
179    ----------------------------------------------------------------------
180    Name: Matthias Blume (blume (at) tti - c (dot) org)
181    Date: 2004/08/30 17:55:00 CDT
182    Tag: blume-20040830-installer
183    Description:
184    
185    Upon request by Johannes Joemann:
186    
187    - improved ML code of installer to fall back to coping when renaming
188      fails (i.e., when source and target are on different file systems);
189      the code compiles but has yet to be tested in anger
190    - removed mlrisc from list of default targets (config/targets)
191    
192    ----------------------------------------------------------------------
193    Name: Matthias Blume (blume (at) tti - c (dot) org)
194    Date: 2004/08/27 17:20:00 CDT
195    Tag: blume-20040827-ptreql
196    Description:
197    
198    Added ptreql primop to structure InlineT (upon request from Larry
199    Paulson).
200    
201    ----------------------------------------------------------------------
202    Name: Allen Leung (leunga (at) reservoir (dot) com)
203    Date: 2004/08/15 21:21:00 EST
204    Tag: leunga-110_48-udgraph
205    Description:
206    
207    Another bug fix from Carl Hauser:
208    
209    diff /net/niflab/smlnj48/src/MLRISC/graphs/udgraph.sml udgraph.sml
210    > 48c48
211    > <              | rmv((e as (k,_))::es,L) = rmv(es,if k = i then es else
212    > e::L)
213    > ---
214    > >              | rmv((e as (k,_))::es,L) = rmv(es,if k = i then L else e::L)
215    > Without this, any deletion of an edge in an undirected graph does severe
216    > violence to the graph.
217    
218    ----------------------------------------------------------------------
219    Name: Allen Leung (leunga (at) reservoir (dot) com)
220    Date: 2004/08/10 23:35:00 EST
221    Tag: leunga-110_48-ppc
222    Description:
223    
224       The IBM/MacOS syntax switch on PPC was incorrectly swapped.
225    
226    ----------------------------------------------------------------------
227    Name: Matthias Blume (blume (at) tti - c (dot) org)
228    Date: 2004/08/10 12:00:00 CDT
229    Tag: Release_110_48
230    Description:
231    
232    New working version (110.48).  NEW BOOTFILES!
233    
234    ----------------------------------------------------------------------
235    Name: Allen Leung (leunga (at) reservoir (dot) com)
236    Date: 2004/08/09 12:21:00 EST
237    Tag: leunga-110_47-dijsktra
238    Description:
239    
240    Bug fix from Carl Hauser:
241    
242    single_source_shortest_paths in dijkstra.sml was observed to get wrong
243    answers (by comparing to single_source_shortest_paths in bellman-ford.sml).
244    
245    The problem is that following the expression A.update(dist,s,Num.zero)
246    it is necessary to update the priority queue using Q.decreaseWeight(Q,s).
247    
248    ----------------------------------------------------------------------
249    Name: Matthias Blume (blume (at) tti - c (dot) org)
250    Date: 2004/08/06 18:10:00 CDT
251    Tag: blume-20040806-cmdline
252    Description:
253    
254    Fiddled with handling of command-line options:
255    
256      * sml now quits after processing the command line
257        if -H, -S, -h<n>, or -s<n> appears as the last
258        command-line argument
259      * a new option -q terminates the session when encountered on
260        the command line; subsequent arguments will be ignored
261      * bug fixes: short (erroneous) arguments are no longer ignored
262        completely
263    
264    ----------------------------------------------------------------------
265    Name: Allen Leung (leunga (at) reservoir (dot) com)
266    Date: 2004/08/04 18:17:00 EST
267    Tag: leunga-110_47-ppc-ibm-asm
268    Description:
269    
270      - Added minimal IBM assembly syntax support for PowerPC.
271    
272      - Cygwin: manually changed the file cygwin.def.  Some exported symbols have
273        been altered in the runtime.  We need an automatic way to keep the file
274        in sync.
275    
276    ----------------------------------------------------------------------
277    Name: Matthias Blume (blume (at) tti - c (dot) org)
278    Date: 2004/08/04 14:00:00 CDT
279    Tag: Release_110_47
280    Description:
281    
282    New working version (110.47).  NEW BOOTFILES!
283    
284    ----------------------------------------------------------------------
285    Name: Matthias Blume (blume (at) tti - c (dot) org)
286    Date: 2004/08/03 14:25:00 CDT
287    Tag: blume-20040803-callingconv
288    Description:
289    
290    Added low-level support for choosing C calling conventions by
291    twiddling the type of rawccall.  (See
292    src/compiler/Semant/types/cproto.sml for details.)
293    
294    ----------------------------------------------------------------------
295    Name: Matthias Blume (blume (at) tti - c (dot) org)
296    Date: 2004/08/02 15:55:00 CDT
297    Tag: blume-20040802-backout
298    Description:
299    
300    Backed out of change to win32-filesys.c.  The earlier patch to
301    get_file_time caused CM to produce files with the wrong time stamp.
302    
303    ----------------------------------------------------------------------
304    Name: Matthias Blume (blume (at) tti - c (dot) org)
305    Date: 2004/08/02 14:45:00 CDT
306    Tag: blume-20040802-nlffi-win32
307    Description:
308    
309    Added NLFFI support for Win32, adapted from a patch provided by David
310    Hansel.  This is currently completely untested.  Also, the issue
311    concerning stdcall vs. ccall is still unresolved.
312    
313    ----------------------------------------------------------------------
314    Name: Matthias Blume (blume (at) tti - c (dot) org)
315    Date: 2004/07/30 17:55:00 CDT
316    Tag: blume-20040730-various
317    Description:
318    
319    Gearing up towards 110.47...
320    
321    - various minor bugfixes to ml-nlffigen
322    - a beginning of a manual for nlffi
323    
324    - eliminated 'export name=value' in config/install.sh as this does
325      not work with certain versions of /bin/sh
326      (Thanks to David King at Motorola for catching this.)
327    
328    - several bugfixes provided or suggested by David Hansel at Reactive Systems:
329      - added a test for tm==NULL to gmtime.c and localtime.c
330      - applied patch for incorrect GetFileTime under win32
331      - toSeconds -> toMilliseconds in Win32/win32-process.sml
332    
333    ----------------------------------------------------------------------
334    Name: Matthias Blume (blume (at) tti - c (dot) org)
335    Date: 2004/07/21 18:20:00 CDT
336    Tag: blume-20040721-nlffigen
337    Description:
338    
339    - Fixed minor issue in ml-nlffigen: Now generate structure T_foo
340      for a typedef to an incomplete type, but leave out the "typ" member.
341      (This is just for consistency.)
342    - Started to produce what is supposed to become better (i.e., comprehensive)
343      documentation of what ml-nlffigen does and produces.
344    
345    ----------------------------------------------------------------------
346    Name: Matthias Blume (blume (at) tti - c (dot) org)
347    Date: 2004/07/14 16:25:00 CDT
348    Tag: blume-20040714-union
349    Description:
350    
351    Added C_UNION to c-calls/c-types.sml and updated the machinery
352    (ml-nlffigen, cproto.sml) that conveys C function interface
353    information to the code generator.
354    
355    However, the actual architecture-specific implementation of function
356    arguments and results that are C unions is still not implemented.
357    
358    ----------------------------------------------------------------------
359    Name: Allen Leung (leunga (at) reservoir (dot) com)
360    Date: 2004/07/14 14:38:00 EST
361    Tag: leunga-110_46_1-ppc-lwzu
362    Description:
363    
364        Added these instructions to the PowerPC architecture:
365        LBZU(X), LHZU(X), LWZU(X),
366        STWU(X), STFDU, STFSU
367    
368        etc...
369    
370        Note: I haven't added their instruction encoding into the description.
371    
372    ----------------------------------------------------------------------
373    Name: Allen Leung (leunga (at) reservoir (dot) com)
374    Date: 2004/07/13 15:04:00 EST
375    Tag: leunga-110_46_1-ppc-lwarx
376    Description:
377    
378       Added the two instructions LWARX and STWCX to the PowerPC
379    instruction set.
380    
381       A (untested) rewrite of loop-structure.sml.  The old version
382    is completely broken.
383    
384    ----------------------------------------------------------------------
385    Name: Matthias Blume (blume (at) tti - c (dot) org)
386    Date: 2004/07/13 13:50:00 CDT
387    Tag: blume-20040713-nlffi
388    Description:
389    
390    - use paramAlloc to report c-calls with too many arguments
391      (for PPC version where parameter area is pre-allocated)
392    - added ccall_maxargspace to machspec (to implement the above)
393    - made "make" commend in CM's "make" tool configurable
394    - added option (default: on) for passing the name of the SML/NJ's "bin"
395      directory to "make"; the call looks like this:
396    
397         make <options> SMLNJ_BINDIR=<dir> <target>
398    
399      This can be used by the Makefile to, e.g., pick the "right" version
400      of ml-nlffigen.
401    - minor code tweaks
402    
403    ----------------------------------------------------------------------
404    Name: Matthias Blume (blume (at) tti - c (dot) org)
405    Date: 2004/07/12 22:50:00 CDT
406    Tag: blume-110_46_1-macosx-nlffi
407    Description:
408    
409    NLFFI under Mac OS X now working (sort of).  This is largely untested,
410    though.
411    
412    Note:  1. You have to make a new, clean build of the runtime system.
413           2. There are new BOOTFILES, you have to use them!
414              (Doing the bootstrap process yourself would be *very* painful!
415               If you absolutely have to do it, build the system under
416               a different architecture and then cross-compile.)
417    
418    Version bumped to 110.46.1 to account for runtime data format changes.
419    
420    ----------------------------------------------------------------------
421    Name: Matthias Blume (blume (at) tti - c (dot) org)
422    Date: 2004/06/18 14:30:00 CDT
423    Tag: blume-20040618-unix
424    Description:
425    
426    Changed the implementation of structure Unix so that the same stream
427    is returned every time one of the {text,bin}{In,Out}streamOf functions
428    is invoked on the same proc.  This is not what the spec currently
429    says -- although IMO it arguably should.  (See discussion below.)
430    
431    ----------------------------------------------------------------------
432    Name: Matthias Blume (blume (at) tti - c (dot) org)
433    Date: 2004/06/17 18:15:00 CDT
434    Tag: Release_110_46
435    Description:
436    
437    New working version (110.46).  NEW BOOTFILES!
438    
439    ----------------------------------------------------------------------
440    Name: Matthias Blume (blume (at) tti - c (dot) org)
441    Date: 2004/06/17 17:20:00 CDT
442    Tag: blume-20040617-timer-unix
443    Description:
444    
445    Changed the interface of structures Timer and Unix to match the most
446    recent Basis spec.
447    
448    In the case of Unix there still seems to be an open/weird issue:
449    
450       The {text,bin}{In,Out}streamOf functions are supposed to create
451       fresh streams whenever they are called -- as opposed to have them
452       return the same stream every time.  This design is supposed to
453       prevent space leaks caused by proc values hanging on to streams.
454    
455       The reap function, on the other hand, is supposed to close the
456       streams.  This cannot be done without having a handle on the
457       stream in proc after all...
458    
459    I took the liberty to implement the following stopgap solution:
460    
461      The proc value hangs on to the most recently created stream(s).
462      Reap closes those.  If either or both of the two streams hadn't
463      been created at all yet, then reap will close the corresponding
464      file descriptors directly.
465    
466    PS: I don't understand the original space leak argument anymore.  If
467    a proc hangs on to the imperative stream, then I/O operations on those
468    will advance the state of the cached stream and avoid the space leak.
469    
470    ----------------------------------------------------------------------
471    Name: Matthias Blume (blume (at) tti - c (dot) org)
472    Date: 2004/05/28 16:45:00 CDT
473    Tag: blume-20040528-basis
474    Description:
475    
476    Added signature PACK_REAL and exported functor PrimIO.
477    
478    ----------------------------------------------------------------------
479    Name: Matthias Blume (blume (at) tti - c (dot) org)
480    Date: 2004/05/25 16:00:00 CDT
481    Tag: blume-20040525-group-owner
482    Description:
483    
484    CM now ignores (but still accepts) the "owner" information in group
485    descriptions.  The owner of a group is its next enclosing
486    library. Each group must have a unique owner.  (There is a virtual
487    "toplevel" library that owns groups which are not nested within a real
488    library.)  Previously, each group had to explicitly declare its owner,
489    and CM would check that such a declaration is correct.  The new scheme
490    is to have CM check that for each group there is precisely one owning
491    library.
492    
493    The advantage of the new scheme is that the programmer no longer needs
494    to maintain the somewhat annoying owner information.  The downside is
495    that CM cannot enforce the ownership rule across multiple runs of
496    CM.make.  Fortunately, enclosing the same group in two different
497    libraries A and B which are not part of the same program does not
498    cause real problems.
499    
500    ----------------------------------------------------------------------
501    Name: Matthias Blume (blume (at) tti - c (dot) org)
502    Date: 2004/05/20 16:00:00 CDT
503    Tag: blume-20040520-win32
504    Description:
505    
506    Made the win32 version work again.  (Strangely, a misplaced comma had
507    slipped into win32-process.c which prevented the runtime from being
508    compiled correctly.)
509    
510    Also, included a minor addition to ml-build.bat analogous to what was
511    done in blume-20040519-ml-build.
512    
513    ----------------------------------------------------------------------
514    Name: Matthias Blume (blume (at) tti - c (dot) org)
515    Date: 2004/05/19 22:10:00 CDT
516    Tag: blume-20040519-ml-build
517    Description:
518    
519    Arranged for ml-build to clean up after itself a little bit better.
520    The script generates a temporary SML source file and compiles it using
521    CM, so CM generates metadata (GUID, SKEL, objectfile) for it.  It now
522    gets rid of those at the end, so they don't accumulate under .cm.
523    
524    This required a minor change to install.sh because the name of the
525    metadata directory (default: .cm) is actually configurable at
526    installation time.
527    
528    ----------------------------------------------------------------------
529    Name: Matthias Blume (blume (at) tti - c (dot) org)
530    Date: 2004/05/18 15:50:00 CDT
531    Tag: blume-20040518-mkreader
532    Description:
533    
534    Added Posix.IO.mk{Bin,Text}{Reader,Writer} by lifting their respective
535    implementations from internal modules PosixBinPrimIO and PosixTextPrimIO.
536    
537    ----------------------------------------------------------------------
538    Name: Matthias Blume (blume (at) tti - c (dot) org)
539    Date: 2004/05/11 14:35:00 CDT
540    Tag: blume-20040511-win32sock
541    Description:
542    
543    Added previously missing support for many socket-related functions
544    under win32.  Thanks to David Hansel <hansel@reactive-systems.com>
545    for the voluminous patch!
546    
547    (I have not tested this patch under win32 yet.)
548    
549    Here is David's e-mail:
550    
551    Hi,
552    
553    Attached to this email you find a diff against sml/nj 110.45
554    that will enable socket support under Windows.
555    
556    To apply the patch (using unix or cygwin)
557    1) gunzip runtime.diff.gz
558    2) "cd" into "src/runtime" in the source tree of a fresh
559       110.45 installation.
560    3) patch -p 1 < [your/path/to]runtime.diff
561    
562    The code compiles fine but has NOT yet been extensively tested.
563    I only ran a few tests for basic socket client functionality
564    (which worked fine).  Especially the functions that use ioctl
565    are not tested at all and might not work (see below).
566    
567    I implemented this since we want to move to a newer version of sml/nj
568    but need socket support in order to use it.  This is the first time I
569    even had a look at the sml/nj source,  so please review my changes
570    before making this part of the distribution!  Here are a few issues
571    that I think might be better for someone to solve who is more
572    familiar with the sml/nj source (and socket programming):
573    
574    - getnetbyaddr.c and getnetbyname.c will raise a "not implemented"
575      exception since I could not figure out what the windows equivalent
576      of these functions is
577    
578    - In sockets-osdep.h there are a some #include statements that are
579      only used in a few files that include sockets-osdep.h
580    
581    - In smlnj-sock-lib.c, function init_fn() calls WSAStartup() but
582      does not process its return value since I don't know how to
583      report an error upwards.
584    
585    - It would probably be good to have a call to WSACleanup() when
586      the library is unloaded (if there is such a possibility).
587      Otherwise I think Windows will take care of this automatically
588      when the process finishes.
589    
590    - I used ioctlsocket() as a replacement for ioctl() but I have
591      no idea if that is actually the proper replacement on Windows.
592    
593    - All these issues are marked in the code by "FIXME" comments.
594    
595    We use sml/nj extensively in our products and are quite happy
596    with it.  I hope this contribution will help you.
597    
598    Keep up the good work!
599    
600    David
601    
602    ----------------------------------------------------------------------
603    Name: Matthias Blume (blume (at) tti - c (dot) org)
604    Date: 2004/05/11 14:20:00 CDT
605    Tag: blume-20040511-installml
606    Description:
607    
608    Fixed two bugs in installml script.  (Thanks to Vesa A. Norrman for
609    the patch.)
610    
611    ----------------------------------------------------------------------
612    Name: Matthias Blume (blume (at) tti - c (dot) org)
613    Date: 2004/05/11 14:05:00 CDT
614    Tag: blume-20040511-nlffi-netbsd
615    Description:
616    
617    Added support for nlffi under netbsd.  (Thanks to Vesa A. Norrman for
618    the patch.)
619    
620    ----------------------------------------------------------------------
621    Name: Matthias Blume (blume (at) tti - c (dot) org)
622    Date: 2004/05/11 12:05:00 CDT
623    Tag: blume-20040511-exports
624    Description:
625    
626    As per request by Adam Chlipala <adam@hcoop.net>, extended various
627    export lists in compiler-related .cm-files.
628    
629    ----------------------------------------------------------------------
630    Name: Matthias Blume (blume (at) tti - c (dot) org)
631    Date: 2004/05/11 11:35:00 CDT
632    Tag: blume-20040511-allsource
633    Description:
634    
635    The installer now honors the "src-smlnj" target again, although its meaning
636    has changed from "all sources required for the compiler" to "all sources
637    the installer knows about".  In other words, if you enable "src-smlnj"
638    in the "targets" file, then the installer will pull in sources for
639    everything.  (Notice that this refers to source code only.  Compiled
640    code is still only installed for modules that were requested explicitly
641    or which are required for other modules that were requested explicitly.)
642    
643    ----------------------------------------------------------------------
644    Name: Matthias Blume (blume (at) tti - c (dot) org)
645    Date: 2004/04/23 17:40:00 CDT
646    Tag: blume-20040423-ieee-scan
647    Description:
648    
649    Fixed IEEEReal.scan (and .fromString) so that if there is an overflow
650    in the exponent calculation we get INF or ZERO (depending on the mantissa
651    and the sign of the exponent).
652    
653    ----------------------------------------------------------------------
654    Name: Matthias Blume (blume (at) tti - c (dot) org)
655    Date: 2004/04/23 10:40:00 CDT
656    Tag: blume-20040423-ml-build
657    Description:
658    
659    The ml-build script now terminates with a non-0 status when something
660    goes wrong.
661    
662    ----------------------------------------------------------------------
663    Name: Matthias Blume (blume (at) tti - c (dot) org)
664    Date: 2004/04/22 16:35:00 CDT
665    Tag: blume-20040422-Option
666    Description:
667    
668    Made exception Option to be the same as exception Option.Option
669    (as it should be).
670    
671    ----------------------------------------------------------------------
672    Name: Allen Leung (leunga (at) reservoir (dot) com)
673    Date: 2004/03/19 14:40:00 EST
674    Tag: leunga-20040319-cygwin-nlffi
675    Description:
676    
677    Fixed the runtime so that ml-nlffi-lib runs on the cygwin version
678    of SML/NJ.  The problem is that
679    
680       lib = dlopen(NULL, ...)
681       f   = dlsym(lib, "malloc");
682    
683    does not work on Windows unless we explicitly export symbols
684    such as 'malloc' during linking.  We fixed this by explicitly
685    exporting the required symbols with the magic gcc incantation:
686    
687        -Wl,--export-all cygwin.def
688    
689    where cygwin.def is a file containing all the symbols that
690    we wish to export.
691    
692    I suspect this is a Windows problem and we'll have to
693    do the same (somehow with windows compilers) when
694    we build the native win32 version with the system
695    calls LoadLibrary/GetProcAddress.
696    
697    ----------------------------------------------------------------------
698    Name: Matthias Blume (blume (at) tti - c (dot) org)
699    Date: 2004/03/04 16:35:00 CST
700    Tag: blume-20040304-intinf-fmt
701    Description:
702    
703    Fixed problem with IntInf.fmt (sign would show up on the right instead
704    of on the left for BIN, OCT, and HEX).
705    
706    ----------------------------------------------------------------------
707    Name: Matthias Blume (blume (at) tti - c (dot) org)
708    Date: 2004/03/04 11:25:00 CST
709    Tag: blume-20040304-symlinks
710    Description:
711    
712    Fixed problem with installer script (unix only) where bin/ml-yacc and
713    friends pointed (via symlinks) to absolute locations instead of just
714    .run-sml.  This was reported by Vesa A Norrman.
715    
716    ----------------------------------------------------------------------
717    Name: Matthias Blume (blume (at) tti - c (dot) org)
718    Date: 2004/02/13 14:50:00 CST
719    Tag: Release_110_45
720    Description:
721    
722    New working version (110.45). New bootfiles.
723    
724    ----------------------------------------------------------------------
725    Name: Matthias Blume (blume (at) tti - c (dot) org)
726    Date: 2004/01/26 15:15:15 CST
727    Tag: blume-20040126-toplevel
728    Description:
729    
730    Improved handling of exceptions at the interactive toplevel.
731    
732    ----------------------------------------------------------------------
733    Name: Matthias Blume (blume (at) tti - c (dot) org)
734    Date: 2004/01/26 11:25:00
735    Tag: blume-20040126-app
736    Description:
737    
738    Type of top-level "app" corrected.
739    Added code for setting vp_limitPtrMask to Win32-specific runtime.
740    
741    ----------------------------------------------------------------------
742    Name: Matthias Blume (blume (at) tti - c (dot) org)
743    Date: 2003/11/18 17:10 CST
744    Tag: blume-20031118-basis-fiddle
745    Description:
746    
747    - changed Timer interface to what might become the spec
748    - POSIX_FLAGS -> BIT_FLAGS according to spec
749    - some other minor discrepancies wrt. spec eliminated
750    
751    ----------------------------------------------------------------------
752    Name: Matthias Blume (blume (at) tti - c (dot) org)
753    Date: 2003/11/06 12:00:00 CST
754    Tag: Release_110_44
755    Description:
756    
757    New working version (110.44). New bootfiles.
758    
759    ----------------------------------------------------------------------
760    Name: Matthias Blume (blume (at) tti - c (dot) org)
761    Date: 2003/11/04 11:50:00 CST
762    Tag: blume-20031104-move-libraries
763    Description:
764    
765    Eliminated the "dont_move_libraries" directive in config/targets.
766    (The mechanism was broken and could not be fixed easily.  Moreover,
767    there does not seem to be any reason not to move all libraries into
768    lib during installation.  I originally implemented this directive as a
769    backward-compatibility feature when I first introduced the new CM.
770    Now that things have been stable for a long time and going back to the
771    old CM is not an option, there is no reason to keep it around.)
772    
773    ----------------------------------------------------------------------
774    Name: Matthias Blume (blume (at) tti - c (dot) org)
775    Date: 2003/11/03 16:00:00 CST
776    Tag: blume-20031103-installdir
777    Description:
778    
779    Made installer honor INSTALLDIR variable again.  (Thanks to Chris
780    Richards for pointing out the problem and providing the solution.)
781    
782    ----------------------------------------------------------------------
783    Name: Matthias Blume (blume (at) tti - c (dot) org)
784    Date: 2003/10/01 17:05:00 CDT
785    Tag: blume-20031001-lal-mlrisc
786    Description:
787    
788    MLRISC bug fix from Lal.
789    
790    ----------------------------------------------------------------------
791    Name: Matthias Blume (blume (at) tti - c (dot) org)
792    Date: 2003/09/30 16:10:00 CDT
793    Tag: blume-20030930-primio-bat
794    Description:
795    
796    1. Added openVector, nullRd, and nullWr to PRIM_IO.
797    2. Improved .bat files (for Win32 port) to make things work under Win95.
798       (thanks to Aaron S. Hawley for this one)
799    
800    ----------------------------------------------------------------------
801    Name: Matthias Blume (blume (at) tti - c (dot) org)
802    Date: 2003/09/26 16:05:00 CDT
803    Tag: blume-20030926-wrappriv
804    Description:
805    
806    Added missing wrapper for privilege "primitive" in $smlnj/viscomp/core.cm.
807    
808    ----------------------------------------------------------------------
809    Name: Matthias Blume (blume (at) tti - c (dot) org)
810    Date: 2003/09/26 15:00:00 CDT
811    Tag: blume-20030926-110_43_3
812    Description:
813    
814    - additional cleanup
815    - version number bump, NEW BOOTFILES
816    
817    ----------------------------------------------------------------------
818    Name: Matthias Blume (blume (at) tti - c (dot) org)
819    Date: 2003/09/26 12:00:00 CDT
820    Tag: blume-20030926-ppautoload
821    Description:
822    
823    I modified the read-eval-print loop so that the autoloader gets
824    invoked whenever the prettyprinter tries to look up a symbol that
825    is not currently defined in the toplevel environment but which
826    appears in CM's autoload registry.  As a result, we see far fewer of
827    those ?.Foo.Bar.xxx names in the prettyprinter's output.
828    
829    In addition to this I tried to clean up some pieces of the Basis
830    implementation (e.g., Socket, Word8Array) in order to prevent other
831    instances of these ?.Foo.Bar.xxx names from being printed.
832    
833    The mechanism that picks names for types still needs some work, though.
834    (Right now it seems that if there is a type A.t which is defined to
835    be B.u, but B is unavailable at toplevel, then A.t gets printed as
836    "?.B.u" although the perhaps more sensible solution would be to use
837    "A.t" in this case.  In other words, the prettyprinter should follow
838    a chain of DEFtycs not farther than there are corresponding toplevel
839    names in the current environment.)
840    
841    ----------------------------------------------------------------------
842    Name: Matthias Blume (blume (at) tti - c (dot) org)
843    Date: 2003/09/24 16:31:00 CDT
844    Tag: blume-20030924-installer
845    Description:
846    
847    Another installer tweak:  All the ML code for the installer is now
848    compiled during CMB.make and put into a little library called
849    $smlnj/installer.cm.  The installation then simply invokes
850    
851       sml -m $smlnj/installer.cm
852    
853    and everything happens automagically.
854    
855    Win32:  ML code senses value of environment variable SMLNJ_HOME.
856    Unix:   ML code senses values of environment variables ROOT, CONFIGDIR,
857            and BINDIR.
858    
859    The new scheme guarantees that the ML code responsible for the installation
860    is in sync with the APIs of the main system.  Also, the installer is
861    somewhat faster because the installer script is precompiled.
862    
863    ----------------------------------------------------------------------
864    Name: Matthias Blume (blume (at) tti - c (dot) org)
865    Date: 2003/09/24 15:35:00 CDT
866    Tag: blume-20030924-synsock
867    Description:
868    
869    Added a signature SYNCHRONOUS_SOCKET to basis.cm.  This is like SOCKET
870    but excludes all non-blocking operations.  Defined SOCKET (in Basis)
871    and CML_SOCKET in terms of SYNCHRONOUS_SOCKET.  Removed superfluous
872    implementations of non-blocking operations from CML's Socket
873    structure.
874    
875    ----------------------------------------------------------------------
876    Name: Matthias Blume (blume (at) tti - c (dot) org)
877    Date: 2003/09/24 15:10:05 CDT
878    Tag: blume-20030924-sockets
879    Description:
880    
881    1. Fixed SOCKET API and implementation to match Basis spec.
882       This required changing the internal representation of sockets to one
883       that remembers (for each socket file descriptor) whether it is currently
884       blocking or non-blocking.  This state is maintained lazily (i.e., a system
885       call is made only if the state actually needs to change).
886    
887    2. OS-specific details of sockets were moved into separate files, thus
888       making it possible to unify the bulk of the socket implementations
889       between Unix and Win32.
890    
891    3. CML's socket API changed accordingly.
892       (Note that we need to remove non-blocking functions from this API
893       since they are redundant in the case of CML!)
894    
895    4. CML's socket implementation now makes use of non-blocking functions
896       provided by Basis, thus removing all OS-dependent code from this part
897       of CML.
898    
899    5. Changed Real64.precision from 52 to 53.  Minor cleanup in Real64 code.
900    
901    ----------------------------------------------------------------------
902    Name: Matthias Blume (blume (at) tti - c (dot) org)
903    Date: 2003/09/22 12:10:00 CDT
904    Tag: blume-20030922-110_43_2
905    Description:
906    
907    Made a new interim version and bootfiles for developer's bootstrapping
908    convenience.
909    
910       110.43.2 -- NEW BOOTFILES
911    
912    ----------------------------------------------------------------------
913    Name: Matthias Blume (blume (at) tti - c (dot) org)
914    Date: 2003/09/19 15:55:00 CDT
915    Tag: blume-20030919-cmdir
916    Description:
917    
918    1. new-install.sh -> install.sh
919    2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
920    3. tweaked installer so that another name instead of .cm can be chosen
921       at install time (by setting the CM_DIR_ARC environment variable
922       during installation); once installation is complete, the name is
923       fixed
924    
925    ----------------------------------------------------------------------
926    Name: Matthias Blume (blume (at) tti - c (dot) org)
927    Date: 2003/09/18 16:00:00 CDT
928    Tag: blume-20030918-110_43_1
929    Description:
930    
931    Made a new interim version and bootfiles for developer's bootstrapping
932    convenience.
933    
934       110.43.1 -- NEW BOOTFILES
935    
936    ----------------------------------------------------------------------
937    Name: Matthias Blume (blume (at) tti - c (dot) org)
938    Date: 2003/09/18 15:20:00 CDT
939    Tag: blume-20030918-misc
940    Description:
941    
942    1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
943       this seems to be controversial at the moment) and used those in
944       Posix.ProcEnv.times.
945    
946    2. Added Time.{from,to}Nanoseconds to Time.
947    
948    3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
949       For example, fromLargeInt never needs to look at more than 3 "big
950       digits" to get its 53 bits of precision.
951    
952    ----------------------------------------------------------------------
953    Name: Matthias Blume (blume (at) tti - c (dot) org)
954    Date: 2003/09/17 16:30:00 CDT
955    Tag: blume-20030917-real32-slices
956    Description:
957    
958    Added an entry to the primitive environment
959    (compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
960    added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
961    
962    Removed some of the "magic" constants in real64.sml and replaced them
963    with code that generates these values from their corresponding
964    integer counterparts.
965    
966    Made all(?) the slice-related changes to the Basis and made everything
967    compile again...
968    
969    ----------------------------------------------------------------------
970    Name: Matthias Blume (blume (at) tti - c (dot) org)
971    Date: 2003/09/15 17:45:00 CDT
972    Tag: blume-20030915-rbase
973    Description:
974    
975    Fixed bug in Real.fromLargeInt.
976    
977    ----------------------------------------------------------------------
978    Name: Matthias Blume (blume (at) tti - c (dot) org)
979    Date: 2003/09/13 18:11:00 CDT
980    Tag: blume-20030913-libinstall
981    Description:
982    
983    Minor bugfix in config/libinstall (set anchor with path to
984    standalone tool after installing it, otherwise libraries that
985    need ml-lex or ml-yacc won't compile the first time the installer
986    runs).
987    
988    ----------------------------------------------------------------------
989    Name: Matthias Blume (blume (at) tti - c (dot) org)
990    Date: 2003/09/12 11:45:00 CDT
991    Tag: blume-20030912-various
992    Description:
993    
994     - fixed bug in Real.toLargeInt
995     - fixed bug in Posix.ProcEnv.times
996     - changed inputLine functions to return an option
997     - minor installer improvements / bugfixes
998     - changed default @SMLalloc parameter for x86/celeron to 64k
999    
1000    ----------------------------------------------------------------------
1001    Name: Matthias Blume (blume (at) tti - c (dot) org)
1002    Date: 2003/09/09 22:00:00 CDT
1003    Tag: Release_110_43
1004    Description:
1005    
1006        New working release 110.43.  New bootfiles.
1007    
1008    ----------------------------------------------------------------------
1009    Name: Matthias Blume (blume (at) tti - c (dot) org)
1010    Date: 2003/09/09 19:20:00 CDT
1011    Tag: blume-20030909-installer
1012    Description:
1013    
1014    Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
1015    Modified config/install.bat to take advantage of it.  Also modified
1016    config/install.sh (and called it config/new-install.sh) to take advantage
1017    of it on Unix systems.  (The SML code is (supposed to be) platform-
1018    independent.)
1019    
1020    The installer can now install everything under Win32
1021    as well as under *nix as long as it compiles.
1022    
1023    Other changes:
1024    
1025      - made CML compile again under Win32
1026      - made eXene compile under Win32 (by providing a fake structure UnixSock
1027        and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
1028      - fixed a bug in nowhere: it assumed that type OS.Process.status is the
1029        same as type int;  under Win32 it isn't
1030      - fixed some slice-related problems in the win32-specific parts of CML
1031      - added a functor argument "sameVol" to os-path-fn.sml in the Basis
1032        (under Win32, the volume name is case-insensitive, and the
1033        OS.Path code compares volume names for equality)
1034    
1035    ----------------------------------------------------------------------
1036    Name: Matthias Blume (blume (at) tti - c (dot) org)
1037    Date: 2003/09/08 11:55:00 CDT
1038    Tag: blume-20030908-fullpath
1039    Description:
1040    
1041    Made Win32 version of OS.FileSys.fullPath return current directory
1042    when given an empty string.  This is what the spec says, and incidentally,
1043    CM depends on it.  (CM otherwise goes into an infinite loop in certain
1044    cases when presented with the name of a non-existing .cm file.)
1045    
1046    ----------------------------------------------------------------------
1047    Name: Matthias Blume (blume (at) tti - c (dot) org)
1048    Date: 2003/09/04 16:30:00 CDT
1049    Tag: blume-20030905-slices-etc
1050    Description:
1051    
1052      1. Changed interface to vectors and arrays in Basis to match
1053         (draft) Basis spec.
1054      2. Added signatures and implementations of slices according to
1055         Basis spec.
1056      3. Edited source code throughout the system to make it compile again
1057         under 1. and 2.  (In some cases code had to be added to have it
1058         match the new signatures.)
1059      4. MLRISC should be backward-compatible: the copies of the originals
1060         of files that needed to change under 3. were retained, the .cm files
1061         check the compiler version number and use old versions when
1062         appropriate.
1063      5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
1064         dirstream -> string option (in accordance with Basis spec).
1065      6. When generating code that counts lines, ml-lex used function
1066         CharVector.foldli, taking advantage of its old interface.
1067         This has been replaced with the corresponding code from
1068         CharVectorSlice.  (html-lex must be re-lexed!)
1069      7. BitArray in smlnj-lib/Util has been extended/modified to match the
1070         new MONO_ARRAY signature.  (Do we need BitArraySlice?)
1071      8. Removed temporary additions (fromInternal, toInternal) from the
1072         (now obsolete) IntInf in smlnj-lib/Util.
1073      9. Cleaned up structure Byte.
1074     10. Added localOffset, scan, and fromString to Date (according to spec).
1075         Cleaned/corrected implementation of Date.
1076         (Still need to check for correctness; implement better canonicalizeDate.)
1077     11. Added "scan" to signature IEEE_REAL.
1078     12. Some improvements to IntInf [in particular: efficiency-hack for
1079         mod and rem when second operand is 2 (for parity checks).]
1080     13. Changed representation of type Time.time, using a single IntInf.int
1081         value counting microseconds.  This considerably simplified the
1082         implementation of structure Time.  We now support negative time
1083         values; scan and fromString handle signs.
1084     14. Functor PrimIO now takes two additional arguments (VectorSlice and
1085         ArraySlice).
1086    
1087    ----------------------------------------------------------------------
1088    Name: Matthias Blume (blume (at) tti - c (dot) org)
1089    Date: 2003/08/28 17:00:00 CDT
1090    Tag: blume-20030828-intinf
1091    Description:
1092    
1093    This is a major update which comes with a version number bump
1094    (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
1095    and an implementation of IntInf in the Basis.
1096    
1097    There are a fairly large number of related changes and updates throughout
1098    the system:
1099    
1100    Basis:
1101            - Implemented IntInf.
1102            - Made LargeInt a projection of IntInf (by filtering through INTEGER).
1103            - Added some missing Real64 operations, most notably Real.toLargeInt.
1104            - Added FixedInt as a synonym for Int32.
1105    
1106    compiler:
1107            * Added support for a built-in intinf type.
1108              - literals
1109              - pattern matching
1110              - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
1111              - overloading on literals and operations
1112    
1113            This required adding a primitive type intinf, some additional
1114            primops, and implementations for several non-trivial intinf
1115            operations in Core.  (The intinf type is completely abstract
1116            to the compiler; all operations get delegated back to the Core.)
1117    
1118            * Intinf equality is handled by polyequal.  However, the compiler
1119            does not print its usual warning in this case (since polyequal
1120            is the right thing to do there).
1121    
1122            * Improved the organization of structure InlineT.
1123    
1124            * A word about conversion primops:
1125            If conversions involving intinf do not cancel out during
1126            CPS contract, then the compiler must insert calls to Core functions.
1127            Since all core access must be resolved already during the FLINT
1128            translate phase, it would be too late a the time of CPS contract
1129            to add new Core calls.  For this reason, conversion primops
1130            for intinf carry two arguments:  1. the numeric argument that
1131            they are supposed to convert, and 2. the Core function that
1132            can help with this conversion if necessary.  If CPS contract
1133            eliminates a primop, then the associated Core function becomes
1134            dead and goes away.  Intinf conversion primops that do not get
1135            eliminated by CPS contract get rewritten into calls of their
1136            core functions by a separate, new phase.
1137    
1138    interactive system:
1139            - Control.Print.intinfDepth controls max length of intinf constants
1140            being printed. (Analogous to Control.Print.stringDepth.)
1141            - Cleanup in printutil and pputil: got rid of unused stuff and
1142            duplicates; replaced some of the code with code that makes better
1143            use of library functionality.
1144    
1145    CM:
1146            Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
1147            are no longer silent.
1148    
1149    CKIT:
1150            Fixed mismatched uses of Int32 and LargeInt.  I always decided
1151            in favor of LargeInt -- which is now the same as IntInf.
1152            CKIT-knowledgable people should check whether this is what's
1153            intended and otherwise change things back to using Int32 or
1154            FixedInt.
1155    
1156    Throughout the code:
1157            Started using IntInf.int literals and built-in operations
1158            (e.g., comparison with 0) where this seems appropriate.
1159    
1160    
1161    ----------------------------------------------------------------------
1162    Name: Dave MacQueen (dbm@cs.uchicago.edu)
1163    Date: 2003/08/13 11:36:00 CDT
1164    Tag: dbm-20030813-mcz-merge1
1165    Description:
1166    
1167    Merging changes from the mcz-branch development branch into trunk.
1168    These changes involve replacement of the emulated old prettyprinter
1169    interface with direct use of the SML/NJ Lib PP library, and fixing
1170    of a couple of bugs (895, 1186) relating to error messages.  A new
1171    prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
1172    has been added.
1173    
1174    ----------------------------------------------------------------------
1175    Name: Matthias Blume (blume (at) tti - c (dot) org)
1176    Date: 2003/08/11 15:45:00 CDT
1177    Tag: blume-20030811-windows
1178    Description:
1179    
1180    Version number bumped to 110.42.9.  NEW BOOTFILES!!!
1181    
1182        http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
1183    
1184    This patch restores SML/NJ's ability to run under win32.  There are a
1185    number of changes, including fixes for several bugs that had gone
1186    unnoticed until now:
1187    
1188      - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
1189        trying to run the win32 version from within cygwin.)
1190    
1191      - There are a number of simple .bat scripts that substitute for their
1192        corresponding Unix shell-scripts.  (See below.)
1193    
1194      - The internals of ml-build have been modified slightly.  The main
1195        difference is that instead of calling ".link-sml" (or link-sml.bat)
1196        using OS.Process.system, the ML process delegates this task back
1197        to the script.  Otherwise problems arise in mixed environments such
1198        as Cygwin where scripts look and work like Unix scripts, but
1199        where OS.Process.system cannot run them.
1200    
1201      - In CM, the srcpath pickler used native pathname syntax -- which
1202        is incorrect in the case of cross-compilation.  The new pickle format
1203        is independent of platform-specific naming conventions.
1204    
1205      - Path configuration files (such as lib/pathconfig) can now choose
1206        between native and standard syntax.  Placing a line of the form
1207    
1208            standard!
1209    
1210        into the file causes all subsequent paths to be interpreted using
1211        CM standard pathname syntax (= Unix conventions); a line
1212    
1213            native!
1214    
1215        switches back to native style.  This was needed so that
1216        path config files can be written portably, see src/system/pathconfig.
1217    
1218      - Runtime system:
1219    
1220          - win32-filesys.c:  get_file_time and set_file_time now
1221            access modification time, not creation time.
1222    
1223          - I/O code made aware of new array representation.
1224    
1225          - Bug fixes in X86.prim.masm.
1226    
1227       - src/system/makeml made aware of win32. (For use under cygwin
1228            and other Unix-environments for windows.)
1229    
1230       - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
1231         which caused BinIO.inputAll to fail consistently.
1232    
1233    .bat scripts:
1234    
1235       Windows .bat scripts assume that SMLNJ_HOME is defined.
1236    
1237       - sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
1238         applications (sml, ml-yacc, ml-lex).
1239       - ml-build.bat: analogous to ml-build.
1240       - config\install.bat:  Analogous to config/install.sh.  This requires
1241         that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
1242         (nmake etc. must be on the path, and vcvars32 must have been run.)
1243         Moreover, sources for ml-lex and ml-yacc need to exist under src,
1244         and the bootfile hierarchy must have been unpacked under
1245         sml.boot.x86-win32.
1246         The script is very primitive and does a poor job at error checking.
1247         It only installs the base system, ml-lex, and ml-yacc.  No other
1248         libraries are being installed (i.e., you get only those that
1249         are part of the compiler.)
1250       - link-sml.bat:  analogous to .link-sml, but not currently used
1251    
1252    Unrelated bug fixes:
1253    
1254      - ml-nlffigen now exports structures ST_* corresponding to incomplete
1255        types.
1256      - Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
1257        otherwise.)
1258    
1259    ----------------------------------------------------------------------
1260    Name: Dave MacQueen (macqueen@cs.uchicago.edu)
1261    Date: 2003/06/17
1262    Tag: macqueen-20030617-bug895
1263    Description:
1264    
1265    Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
1266    Tag will be used for new development branch (mcz-branch) for
1267    use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
1268    
1269    ----------------------------------------------------------------------
1270    Name: Matthias Blume (blume (at) tti - c (dot) org)
1271    Date: 2003/05/27 16:55:00 CDT
1272    Tag: blume-20030527-polyeq
1273    Description:
1274    
1275    Tried to eliminated most cases of polymorphic equality.
1276    
1277    ----------------------------------------------------------------------
1278    Name: Matthias Blume (blume (at) tti - c (dot) org)
1279    Date: 2003/05/21 17:45:00 CDT
1280    Tag: blume-20030517-complete
1281    Description:
1282    
1283    Two changes:
1284    
1285      1. Added a flag for controlling whether non-exhaustive bindings will
1286         be treated as errors (default is false).
1287      2. Cleaned up the *entire* source tree so that CMB.make goes through
1288         without a single non-exhaustive match- or bind warning.
1289    
1290    ----------------------------------------------------------------------
1291    Name: Matthias Blume (blume (at) tti - c (dot) org)
1292    Date: 2003/05/17 10:20:00 CDT
1293    Tag: blume-20030517-absyn
1294    Description:
1295    
1296    1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
1297    
1298       This mainly affects the quality of error messages.  However, some
1299       of the code is now more straightforward than before. (Treatment of
1300       the above four constructs in translate.sml is much simpler than
1301       the "macro-expansion" that was going on before.  Plus, the mach-
1302       compiler no longer gets invoked just to be able to compile an
1303       if-expression.)
1304    
1305    2. The ErrorMsg.Error exception is now caught and absorbed by the
1306       interactive loop.
1307    
1308    ----------------------------------------------------------------------
1309    Name: Allen Leung
1310    Date: 2003/05/16 13:05:00 CDT
1311    Tag: leunga-20030516-cygwin-runtime
1312    Description:
1313    
1314        Ported the runtime system to cygwin, which uses the unix
1315    x86-unix bin files.  Missing/buggy features:
1316    
1317       o getnetbyname, getnetbyaddr: these functions seem to be missing in
1318         the Cygwin library.
1319       o Ctrl-C handling may be flaky.
1320       o Windows system calls and Windows I/O are not supported.
1321    
1322       A new set of binfiles is located at:
1323    
1324         http://www.dorsai.org/~leunga/boot.x86-unix.tgz
1325    
1326       This is only needed for bootstrapping the cygwin version of smlnj.
1327       Other x86 versions can use the existing binfiles.
1328    
1329    ----------------------------------------------------------------------
1330    Name: Matthias Blume
1331    Date: 2003/04/08 15:42:00 CDT
1332    Tag: blume-20030408-listpair
1333    Description:
1334    
1335    1. Added a target 'mlrisc' to installer.
1336    
1337    2. Added missing elements to structure ListPair.
1338    
1339    ----------------------------------------------------------------------
1340    Name: Allen Leung
1341    Date: 2003/01/07 10:40:00 EST
1342    Tag: leunga-20030107-int-rem
1343    Description:
1344    
1345    Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
1346    The arguments to the SUBL instruction were swapped.
1347    
1348    ----------------------------------------------------------------------
1349    Name: Matthias Blume
1350    Date: 2002/12/12 16:25:00 EST
1351    Tag: blume-20021212-risc-ra
1352    Description:
1353    
1354    Fixed a serious bug in the rewrite code for FP spilling/reloading that
1355    sent the RA into an infinite loop when floating point registers get
1356    spilled.  (Because of this bug, e.g., nucleic stopped compiling between
1357    110.37 and 110.38.)
1358    There was another set of potential problems related to the handling of
1359    MLRISC annotations (but those did not yet cause real problems, apparently).
1360    
1361    ----------------------------------------------------------------------
1362    Name: Matthias Blume
1363    Date: 2002/12/06 22:40:00 EST
1364    Tag: blume-20021206-cm-fileid
1365    Description:
1366    
1367    Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
1368    This fixes the problem of CM getting confused by files that suddenly
1369    change their identity (e.g., by getting unlinked and recreated by some
1370    text editor such as vi).  There might be a better/cheaper/cleaner way
1371    of doing this, but for now this will have to do.
1372    
1373    ----------------------------------------------------------------------
1374    Name: Matthias Blume
1375    Date: 2002/10/28 09:50:00 EST
1376    Tag: blume-20021028-typecheck
1377    Description:
1378    
1379    Exported structure Typecheck from $smlnj/viscomp/core.cm.
1380    
1381    ----------------------------------------------------------------------
1382    Name: Matthias Blume
1383    Date: 2002/10/17 09:10:00 EDT
1384    Tag: Release_110_42
1385    Description:
1386    
1387    In good old tradition, there has been a slight hiccup so that we have
1388    to patch 110.42 after the fact.   The old release tag has been replaced
1389    (see below).
1390    
1391    The change solves a problem with two competing approaches the
1392    configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
1393    each other's way.
1394    
1395    This change only affects the runtime system code and the installer script.
1396    (No new bootfiles.)
1397    
1398    ----------------------------------------------------------------------
1399    Name: Matthias Blume
1400    Date: 2002/10/16 12:00:00 EDT
1401    Tag: Release_110_42_removed
1402    Description:
1403    
1404    New working release.  New bootfiles.
1405    
1406    ----------------------------------------------------------------------
1407    Name: Matthias Blume
1408    Date: 2002/10/10 13:10:00 EDT
1409    Tag: blume-20021010-ppc-divs
1410    Description:
1411    
1412    The mltree operator DIVS must be implemented with an overflow check on
1413    the PPC because the hardware indicates divide-by-zero using "overflow" as
1414    well.
1415    
1416    ----------------------------------------------------------------------
1417    Name: Matthias Blume
1418    Date: 2002/07/23 11:45:00
1419    Tag: blume-20020723-smlnj-home
1420    Description:
1421    
1422    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
1423    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
1424    CM_PATHCONFIG is also set), the path configuration file is assumed
1425    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
1426    the entire tree to some other place and everything will "just work".
1427    
1428    (Companion commands such as ml-build and ml-makedepend also sense this
1429    variable.)
1430    
1431    ----------------------------------------------------------------------
1432    Name: Matthias Blume
1433    Date: 2002/07/12 21:19:00 EDT
1434    Tag: blume-20020712-liveness
1435    Description:
1436    
1437    Exported two useful "step" functions from liveness module (MLRISC).
1438    
1439    ----------------------------------------------------------------------
1440    Name: Matthias Blume
1441    Date: 2002/07/05 16:00 EDT
1442    Tag: Release_110_41
1443    Description:
1444    
1445    New working release.  New bootfiles.
1446    
1447    ----------------------------------------------------------------------
1448    Name: Matthias Blume
1449    Date: 2002/07/05 10:25:00 EDT
1450    Tag: blume-20020705-btimp
1451    Description:
1452    
1453    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
1454    other clients can set up backtracing support.
1455    
1456    ----------------------------------------------------------------------
1457    Name: Matthias Blume
1458    Date: 2002/06/25 17:23:00 EDT
1459    Tag: blume-20020625-fpmax
1460    Description:
1461    
1462    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
1463    case.  (The sense of the isNaN test was reversed -- which made min and
1464    max always return their first argument.)
1465    
1466    ----------------------------------------------------------------------
1467    Name: Matthias Blume
1468    Date: 2002/06/11
1469    Tag: blume-20020611-unixpath
1470    Description:
1471    
1472    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
1473    
1474    ----------------------------------------------------------------------
1475    Name: Matthias Blume
1476    Date: 2002/06/10 16:35:00 EDT
1477    Tag: blume-20020610-ieeereal
1478    Description:
1479    
1480    I back-ported my implementation of IEEEReal.fromString from the
1481    idlbasis-devel branch so that we can test it.
1482    
1483    Another small change is that ppDec tries to give more information
1484    than just "<sig>" in the case of functors.  However, this code is
1485    broken in some mysterious way if the functor's body's signature
1486    has not been declared by ascription but gets inferred from the
1487    implementation.  This needs fixing...
1488    
1489    ----------------------------------------------------------------------
1490    Name: Matthias Blume
1491    Date: 2002/05/31
1492    Tag: blume-20020531-btrace-mode
1493    Description:
1494    
1495    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
1496    the wayside when I switched over to using Controls everywhere.)
1497    
1498    ----------------------------------------------------------------------
1499    Name: Lal George
1500    Date: 2002/05/23 12:21:40 EDT
1501    Tag: george-20020523-visual-labels
1502    Description:
1503    
1504    Labels are now displayed in the graphical output to make
1505    the fall-through and target blocks obvious.
1506    
1507    ----------------------------------------------------------------------
1508    Name: Matthias Blume
1509    Date: 2002/05/22 11:03:00 EDT
1510    Tag: blume-20020522-shrink
1511    Description:
1512    
1513    John tweaked yesterday's fix for 1131 to handle an out-of-memory
1514    situation that comes up when allocating huge arrays.
1515    
1516    ----------------------------------------------------------------------
1517    Name: Matthias Blume
1518    Date: 2002/05/21 16:00:00 EDT
1519    Tag: Release_110_40
1520    Description:
1521    
1522    New working release (110.40).  New bootfiles.
1523    
1524    [Also: John Reppy fixed GC bug 1131.]
1525    
1526    ----------------------------------------------------------------------
1527    Name: Matthias Blume
1528    Date: 2002/05/21 12:35:00 EDT
1529    Tag: blume-20020521-cmdoc
1530    Description:
1531    
1532    CM documentation update.
1533    
1534    ----------------------------------------------------------------------
1535    Name: Matthias Blume
1536    Date: 2002/05/21 10:55:00 EDT
1537    Tag: blume-20020521-misc
1538    Description:
1539    
1540    - John tweaked runtime to be silent on heap export (except when
1541      GC messages are on).
1542    - I added a few more things (cross-compiling versions of CMB) to
1543      config/preloads (as suggestions).
1544    
1545    ----------------------------------------------------------------------
1546    Name: Matthias Blume
1547    Date: 2002/05/20 22:25:00 EDT
1548    Tag: blume-20020520-controls
1549    Description:
1550    
1551    - Added ControlUtil structure to control-lib.cm.
1552    - Use it throughout.
1553    - Used Controls facility to define MLRISC controls (as opposed to
1554      registering MLRISC control ref cells with Controls after the fact)
1555    - Fixed messed-up controls priorities.
1556    
1557    * Removed again all the stuff from config/preloads that one wouldn't
1558      be able to preload at the time the initial heap image is built.
1559      (Many libraries, e.g., CML, do not exist yet at this time.  The
1560       only libraries that can be preloaded via config/preloads are those
1561       that come bundled with the bootfiles.)
1562    
1563    ----------------------------------------------------------------------
1564    Name: Matthias Blume
1565    Date: 2002/05/20 10:59:00 EDT
1566    Tag: blume-20020520-preloads
1567    Description:
1568    
1569    Added a lot of commented-out suggestions for things to be included
1570    in config/preloads.
1571    
1572    ----------------------------------------------------------------------
1573    Name: Allen Leung
1574    Date: 2002/05/18 14:20:00 EDT
1575    Tag: leunga-20020518-mdl
1576    Description:
1577    
1578    o Made the mdl tool stuff compile and run again.
1579    o I've disabled all the stuff that depends on RTL specifications; they
1580      are all badly broken anyway.
1581    
1582    ----------------------------------------------------------------------
1583    Name: Matthias Blume
1584    Date: 2002/05/17 16:49:00 EDT
1585    Tag: blume-20020517-controls
1586    Description:
1587    
1588    0. John Reppy made several modifications to the SML/NJ library.
1589       In particular, there is a shiny new controls-lib.cm.
1590    
1591    1. Pushed new controls interface through compiler so that everything
1592       compiles again.
1593    
1594    2. Added FormatComb and FORMAT_COMB to the CML version of the
1595       SML/NJ library (so that CML compiles again).
1596    
1597    3. Modified init scripts because XXX_DEFAULT environment variables
1598       are no longer with us.  (Boot-time initialization is now done
1599       using the same environment variables that are also used for
1600       startup-time initialization of controls.)
1601    
1602    ----------------------------------------------------------------------
1603    Name: Lal George
1604    Date: 2002/05/15 09:20:10 EDT
1605    Tag: george-20020515-pseudo-op-decls
1606    Description:
1607    
1608            All pseudo-ops emitted before the first segment declaration
1609    such as TEXT, DATA, and BSS directives are assumed to be global
1610    declarations and are emitted first in the assembly file. This is
1611    useful in a number of situations where one has pseudo-ops that are not
1612    specific to any segment, and also works around the constraint that one
1613    cannot have client pseudo-ops in the TEXT segment.
1614    
1615            Because no segment is associated with these declarations it is
1616    an error to allocate any space or objects before the first segment
1617    directive and an exception will be raised. However, we cannot make
1618    this check for client pseudo-ops.
1619    
1620            These top level declarations are a field in the CFG graph_info.
1621    In theory you can continue to add to this field after the CFG has been
1622    built -- provided you know what you are doing;-)
1623    
1624    ----------------------------------------------------------------------
1625    Name: Matthias Blume
1626    Date: 2002/05/13 16:40:00 EDT
1627    Tag: blume-20020513-pp-etc
1628    Description:
1629    
1630    A few minor bugfixes:
1631    
1632      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
1633        (Bogus printouts for redefined bindings still occur.  Compiler
1634        bug should no longer occur now.  We need to redo the prettyprinter
1635        from scratch.)
1636    
1637      - CM pathname printer now also adds escape sequences for ( and )
1638    
1639      - commend and docu fixes for ml-nlffi
1640    
1641    ----------------------------------------------------------------------
1642    Name: Matthias Blume
1643    Date: 2002/05/10 16:40:00 EDT
1644    Tag: blume-20020510-erg-textio
1645    Description:
1646    
1647    Applied the following bugfix provided by Emden Gansner:
1648    
1649        Output is corrupted when outputSubstr is used rather than output.
1650    
1651        The problem occurs when a substring
1652    
1653            ss = (s, dataStart, dataLen)
1654    
1655        where dataStart > 0, fills a stream buffer with avail bytes left.
1656        avail bytes of s, starting at index dataStart, are copied into the
1657        buffer, the buffer is flushed, and then the remaining dataLen-avail
1658        bytes of ss are copied into the beginning of the buffer. Instead of
1659        starting this copy at index dataStart+avail in s, the current code
1660        starts the copy at index avail.
1661    
1662        Fix:
1663        In text-io-fn.sml, change line 695 from
1664             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
1665        to
1666             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
1667    
1668    ----------------------------------------------------------------------
1669    Name: Matthias Blume
1670    Date: 2002/04/12 13:55:00 EDT
1671    Tag: blume-20020412-assyntax
1672    Description:
1673    
1674    1. Grabbed newer assyntax.h from the XFree86 project.
1675    2. Fiddled with how to compile X86.prim.asm without warnings.
1676    3. (Very) Minor cleanup in CM.
1677    
1678    ----------------------------------------------------------------------
1679    Name: Matthias Blume
1680    Date: 2002/04/01 (no joke!) 17:07:00 EST
1681    Tag: blume-20020401-x86div
1682    Description:
1683    
1684    Added full support for div/mod/rem/quot on the x86, using the machine
1685    instruction's two results (without clumsily recomputing the remainder)
1686    directly where appropriate.
1687    
1688    Some more extensive power-of-two support was added to the x86 instruction
1689    selector (avoiding expensive divs, mods, and muls where they can be
1690    replaced with cheaper shifts and masks).  However, this sort of thing
1691    ought to be done earlier, e.g., within the CPS optimizer so that
1692    all architectures benefit from it.
1693    
1694    The compiler compiles to a fixed point, but changes might be somewhat
1695    fragile nevertheless.  Please, report any strange things that you might
1696    see wrt. div/mod/quot/rem...
1697    
1698    ----------------------------------------------------------------------
1699    Name: Matthias Blume
1700    Date: 2002/03/29 17:22:00
1701    Tag: blume-20020329-div
1702    Description:
1703    
1704    Fixed my broken div/mod logic.  Unfortunately, this means that the
1705    inline code for div/mod now has one more comparison than before.
1706    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
1707    The problem was with quotient = 0, because that alone does not tell
1708    us which way the rounding went.  One then has to look at whether
1709    remainder and divisor have the same sign...  :(
1710    
1711    Anyway, I replaced the bootfiles with fresh ones...
1712    
1713    ----------------------------------------------------------------------
1714    Name: Matthias Blume
1715    Date: 2002/03/29 14:10:00 EST
1716    Tag: blume-20020329-inlprims
1717    Description:
1718    
1719    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
1720    
1721    Primops have changed. This means that the bin/boot-file formats have
1722    changed as well.
1723    
1724    To make sure that there is no confusion, I made a new version.
1725    
1726    
1727    CHANGES:
1728    
1729    * removed REMT from mltree (remainder should never overflow).
1730    
1731    * added primops to deal with divisions of all flavors to the frontend
1732    
1733    * handled these primops all the way through so they map to their respective
1734      MLRISC support
1735    
1736    * used these primops in the implementation of Int, Int32, Word, Word32
1737    
1738    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
1739    
1740    * parameterized INLMIN, INLMAX, and INLABS by a numkind
1741    
1742    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
1743      floating point
1744    
1745    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
1746      and Real (but Real.abs maps to a separate floating-point-only primop)
1747    
1748    
1749    TODO items:
1750    
1751    * Hacked Alpha32 instruction selection, disabling the selection of REMx
1752      instructions because the machine instruction encoder cannot handle
1753      them.  (Hppa, PPC, and Sparc instruction selection did not handle
1754      REM in the first place, and REM is supported by the x86 machine coder.)
1755    
1756    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
1757      selection phase.  (The two can be streamlined because the hardware
1758      delivers both quotient and remainder at the same time anyway.)
1759    
1760    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
1761      (Currently the behavior is inconsistent both across architectures and
1762      wrt. the draft Basis spec.)
1763    
1764    * Word8 should eventually be handled natively, too.
1765    
1766    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
1767      as if there currently is no execution path that could trigger it in
1768      SML/NJ.  (The assumptions underlying functions arith and promotable do not
1769      hold for things like multiplication and division.)
1770    
1771    ----------------------------------------------------------------------
1772    Name: Matthias Blume
1773    Date: 2002/03/27 16:27:00 EST
1774    Tag: blume-20020327-mlrisc-divisions
1775    Description:
1776    
1777    Added support for all four division operations (ML's div, mod, quot,
1778    and rem) to MLRISC.  In the course of doing so, I also rationalized
1779    the naming (no more annoying switch-around of DIV and QUOT), by
1780    parameterizing the operation by div_rounding_mode (which can be either
1781    DIV_TO_ZERO or DIV_TO_NEGINF).
1782    
1783    The generic MLTreeGen functor takes care of compiling all four
1784    operations down to only round-to-zero div.
1785    
1786    Missing pieces:
1787    
1788      * Doing something smarter than relying on MLTreeGen on architectures
1789        like, e.g., the x86 where hardware division delivers both quotient and
1790        remainder at the same time.  With this, the implementation of the
1791        round-to-neginf operations could be further streamlined.
1792    
1793      * Remove inlining support for div/mod/rem from the frontend and replace it
1794        with primops that get carried through to the backend.  Do this for all
1795        int and word types.
1796    
1797    ----------------------------------------------------------------------
1798    Name: Matthias Blume
1799    Date: 2002/03/25 17:25:00 EST
1800    Tag: blume-20020325-divmod
1801    Description:
1802    
1803    I improved (hopefully without breaking them) the implementation of Int.div,
1804    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
1805    advantage of the following observations:
1806    
1807      Let  q = x quot y      r = x rem y
1808           d = x div  y      m = x mod y
1809    
1810    where "quot" is the round-to-zero version of integer division that
1811    hardware usually provides.  Then we have:
1812    
1813         r = x - q * y        where neither the * nor the - will overflow
1814         d = if q >= 0 orelse x = q * y then q else q - 1
1815                              where neither the * nor the - will overflow
1816         m = if q >= 0 orelse r = 0 then r else r + y
1817                              where the + will not overflow
1818    
1819    This results in substantial simplification of the generated code.
1820    The following table shows the number of CFG nodes and edges generated
1821    for
1822            fun f (x, y) = x OPER y
1823            (* with OPER \in div, mod, quot, rem *)
1824    
1825    
1826        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
1827        --------------------------------------------------------
1828         div |         24 |         39 |         12 |         16
1829         mod |         41 |         71 |         12 |         16
1830        quot |          8 |         10 |          8 |         10
1831         rem |         10 |         14 |          8 |         10
1832    
1833    
1834    ----------------------------------------------------------------------
1835    Name: Matthias Blume
1836    Date: 2002/03/25 22:06:00 EST
1837    Tag: blume-20020325-cprotobug
1838    Description:
1839    
1840    Fixed a bug in cproto (c prototype decoder).
1841    
1842    ----------------------------------------------------------------------
1843    Name: Matthias Blume
1844    Date: 2002/03/25 16:00:00 EST
1845    Tag: blume-20020325-raw-primops
1846    Description:
1847    
1848    I did some cleanup to Allen's new primop code and
1849    replaced yesterday's bootfiles with new ones.
1850    (But they are stored in the same place.)
1851    
1852    ----------------------------------------------------------------------
1853    Name: Matthias Blume
1854    Date: 2002/03/24 22:40:00 EST
1855    Tag: blume-20020324-bootfiles
1856    Description:
1857    
1858    Made the bootfiles that Allen asked for.
1859    
1860    ----------------------------------------------------------------------
1861    Name: Allen Leung
1862    Date: 2002/03/23 15:50:00 EST
1863    Tag: leunga-20020323-flint-cps-rcc-primops
1864    Description:
1865    
1866      1. Changes to FLINT primops:
1867    
1868        (* make a call to a C-function;
1869         * The primop carries C function prototype information and specifies
1870         * which of its (ML-) arguments are floating point. C prototype
1871         * information is for use by the backend, ML information is for
1872         * use by the CPS converter. *)
1873      | RAW_CCALL of { c_proto: CTypes.c_proto,
1874                       ml_args: ccall_type list,
1875                       ml_res_opt: ccall_type option,
1876                       reentrant : bool
1877                     } option
1878       (* Allocate uninitialized storage on the heap.
1879        * The record is meant to hold short-lived C objects, i.e., they
1880        * are not ML pointers.  With the tag, the representation is
1881        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
1882        *)
1883      | RAW_RECORD of {tag:bool,sz:int}
1884      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
1885    
1886      2.  These CPS primops are now overloaded:
1887    
1888           rawload of {kind:numkind}
1889           rawstore of {kind:numkind}
1890    
1891          The one argument form is:
1892    
1893             rawload {kind} address
1894    
1895          The two argument form is:
1896    
1897             rawload {kind} [ml object, byte-offset]
1898    
1899      3. RAW_CCALL/RCC now takes two extra arguments:
1900    
1901         a. The first is whether the C call is reentrant, i.e., whether
1902            ML state should be saved and restored.
1903         b. The second argument is a string argument specifying the name of
1904            library and the C function.
1905    
1906         These things are currently not handled in the code generator, yet.
1907    
1908      4. In CProto,
1909    
1910         An encoding type of "bool" means "ml object" and is mapped into
1911         C prototype of PTR.  Note that "bool" is different than "string",
1912         even though "string" is also mapped into PTR, because "bool"
1913         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
1914    
1915      5. Pickler/unpicker
1916    
1917         Changed to handle RAW_RECORD and newest RAW_CCALL
1918    
1919      6. MLRiscGen,
1920    
1921         1. Changed to handle the new rawload/rawstore/rawrecord operators.
1922         2. Code for handling C Calls has been moved to a new module CPSCCalls,
1923            in the file CodeGen/cpscompile/cps-c-calls.sml
1924    
1925      7. Added the conditional move operator
1926    
1927             condmove of branch
1928    
1929         to cps.  Generation of this is still buggy so it is currently
1930         disabled.
1931    
1932    ----------------------------------------------------------------------
1933    Name: Lal George
1934    Date: 2002/03/22 14:18:25 EST
1935    Tag: george-20020322-cps-branch-prob
1936    Description:
1937    
1938    Implemented the Ball-Larus branch prediction-heuristics, and
1939    incorporated graphical viewers for control flow graphs.
1940    
1941    Ball-Larus Heuristics:
1942    ---------------------
1943    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
1944    
1945    By design it uses the Dempster-Shafer theory for combining
1946    probabilities.  For example, in the function:
1947    
1948        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
1949    
1950    the ball-larus heuristics predicts that the n=0 is unlikely
1951    (OH-heuristic), and the 'then' branch is unlikely because of the
1952    RH-heuristic -- giving the 'then' branch an even lower combined
1953    probability using the Dempster-Shafer theory.
1954    
1955    Finally, John Reppy's loop analysis in MLRISC, further lowers the
1956    probability of the 'then' branch because of the loop in the else
1957    branch.
1958    
1959    
1960    Graphical Viewing:
1961    ------------------
1962    I merely plugged in Allen's graphical viewers into the compiler. The
1963    additional code is not much. At the top level, saying:
1964    
1965            Control.MLRISC.getFlag "cfg-graphical-view" := true;
1966    
1967    will display the graphical view of the control flow graph just before
1968    back-patching.  daVinci must be in your path for this to work. If
1969    daVinci is not available, then the default viewer can be changed
1970    using:
1971    
1972            Control.MLRISC.getString "viewer"
1973    
1974    which can be set to "dot" or "vcg" for the corresponding viewers. Of
1975    course, these viewers must be in your path.
1976    
1977    The above will display the compilation unit at the level of clusters,
1978    many of which are small, boring, and un-interesting. Also setting:
1979    
1980            Control.MLRISC.getInt "cfg-graphical-view_size"
1981    
1982    will display clusters that are larger than the value set by the above.
1983    
1984    
1985    ----------------------------------------------------------------------
1986    Name: Matthias Blume
1987    Date: 2002/03/21 22:20:00 EST
1988    Tag: blume-20020321-kmp-bugfix
1989    Description:
1990    
1991    Changed the interface to the KMP routine in PreString and fixed
1992    a minor bug in one place where it was used.
1993    
1994    ----------------------------------------------------------------------
1995    Name: Allen Leung
1996    Date: 2002/03/21 20:30:00 EST
1997    Tag: leunga-20020321-cfg
1998    Description:
1999    
2000      Fixed a potential problem in cfg edge splitting.
2001    
2002    ----------------------------------------------------------------------
2003    Name: Allen Leung
2004    Date: 2002/03/21 17:15:00 EST
2005    Tag: leunga-20020321-x86-fp-cfg
2006    Description:
2007    
2008      1. Recoded the buggy parts of x86-fp.
2009    
2010         a. All the block reordering code has been removed.
2011            We now depend on the block placement phases to do this work.
2012    
2013         b. Critical edge splitting code has been simplified and moved into the
2014            CFG modules, as where they belong.
2015    
2016         Both of these were quite buggy and complex.  The code is now much, much
2017         simpler.
2018    
2019      2. X86 backend.
2020    
2021         a. Added instructions for 64-bit support.  Instruction selection for
2022            64-bit has not been committed, however, since that
2023            requires changes to MLTREE which haven't been approved by
2024            Lal and John.
2025    
2026         b. Added support for FUCOMI and FUCOMIP when generating code for
2027            PentiumPro and above.  We only generate these instructions in
2028            the fast-fp mode.
2029    
2030         c. Added cases for JP and JNP in X86FreqProps.
2031    
2032      3. CFG
2033    
2034         CFG now has a bunch of methods for edge splitting and merging.
2035    
2036      4. Machine description.
2037    
2038         John's simplification of MLTREE_BASIS.fcond broke a few machine
2039         description things:
2040    
2041         rtl-build.{sig,sml} and hppa.mdl fixed.
2042    
2043         NOTE: the machine description stuff in the repository is still broken.
2044               Again, I can't put my fixes in because that involves
2045               changes to MLTREE.
2046    
2047    ----------------------------------------------------------------------
2048    Name: Matthias Blume
2049    Date: 2002/03/20 15:55:00 EST
2050    Tag: blume-20020320-kmp
2051    Description:
2052    
2053    Implemented Knuth-Morris-Pratt string matching in PreString and used
2054    it for String.isSubstring, Substring.isSubstring, and
2055    Substring.position.
2056    
2057    (Might need some stress-testing.  Simple examples worked fine.)
2058    
2059    ----------------------------------------------------------------------
2060    Name: Matthias Blume
2061    Date: 2002/03/19 16:37:00 EST
2062    Tag: blume-20020319-witnesses
2063    Description:
2064    
2065    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
2066    
2067    This implements a generic mechanism for changing constness qualifiers
2068    anywhere within big C types without resorting to outright "casts".
2069    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
2070    modify the constness at the outermost level.)
2071    The implementation of "convert" is based on the idea of "witness"
2072    values -- values that are not used by the operation but whose types
2073    "testify" to their applicability.  On the implementation side, "convert"
2074    is simply a projection (returning its second curried argument).  With
2075    cross-module inlining, it should not result in any machine code being
2076    generated.
2077    
2078    ----------------------------------------------------------------------
2079    Name: Matthias Blume
2080    Date: 2002/03/15 16:40:00 EST
2081    Tag: blume-20020315-basis
2082    Description:
2083    
2084    Provided (preliminary?) implementations for
2085    
2086      {String,Substring}.{concatWith,isSuffix,isSubstring}
2087    
2088    and
2089    
2090      Substring.full
2091    
2092    Those are in the Basis spec but they were missing in SML/NJ.
2093    
2094    ----------------------------------------------------------------------
2095    Name: Matthias Blume
2096    Date: 2002/03/14 21:30:00 EST
2097    Tag: blume-20020314-controls
2098    Description:
2099    
2100    Controls:
2101    ---------
2102    
2103    1. Factored out the recently-added Controls : CONTROLS stuff and put
2104       it into its own library $/controls-lib.cm.  The source tree for
2105       this is under src/smlnj-lib/Controls.
2106    
2107    2. Changed the names of types and functions in this interface, so they
2108       make a bit more "sense":
2109    
2110          module -> registry
2111          'a registry -> 'a group
2112    
2113    3. The interface now deals in ref cells only.  The getter/setter interface
2114       is (mostly) gone.
2115    
2116    4. Added a function that lets one register an already-existing ref cell.
2117    
2118    5. Made the corresponding modifications to the rest of the code so that
2119       everything compiles again.
2120    
2121    6. Changed the implementation of Controls.MLRISC back to something closer
2122       to the original.  In particular, this module (and therefore MLRISC)
2123       does not depend on Controls.  There now is some link-time code in
2124       int-sys.sml that registers the MLRISC controls with the Controls
2125       module.
2126    
2127    CM:
2128    ---
2129    
2130      * One can now specify the lambda-split aggressiveness in init.cmi.
2131    
2132    ----------------------------------------------------------------------
2133    Name: Allen Leung
2134    Date: 2002/03/13 17:30:00 EST
2135    Tag: leunga-20020313-x86-fp-unary
2136    Description:
2137    
2138    Bug fix for:
2139    
2140    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
2141    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
2142    > - fun f(x,(y,z)) = Real.~ y;
2143    > [autoloading]
2144    > [autoloading done]
2145    >       fchsl   (%eax), 184(%esp)
2146    > Error: MLRisc bug: X86MCEmitter.emitInstr
2147    >
2148    > uncaught exception Error
2149    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
2150    
2151    The problem was that the code generator did not generate any fp registers
2152    in this case, and the ra didn't know that it needed to run the X86FP phase to
2153    translate the pseudo fp instruction.   This only happened with unary fp
2154    operators in certain situations.
2155    
2156    ----------------------------------------------------------------------
2157    Name: Matthias Blume
2158    Date: 2002/03/13 14:00:00 EST
2159    Tag: blume-20020313-overload-etc
2160    Description:
2161    
2162    1. Added _overload as a synonym for overload for backward compatibility.
2163       (Control.overloadKW must be true for either version to be accepted.)
2164    
2165    2. Fixed bug in install script that caused more things to be installed
2166       than what was requested in config/targets.
2167    
2168    3. Made CM aware of the (_)overload construct so that autoloading
2169       works.
2170    
2171    ----------------------------------------------------------------------
2172    Name: Matthias Blume
2173    Date: 2002/03/12 22:03:00 EST
2174    Tag: blume-20020312-url
2175    Description:
2176    
2177    Forgot to update BOOT and srcarchiveurl.
2178    
2179    ----------------------------------------------------------------------
2180    Name: Matthias Blume
2181    Date: 2002/03/12 17:30:00 EST
2182    Tag: blume-20020312-version110392
2183    Description:
2184    
2185    Yet another version number bump (because of small changes to the
2186    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
2187    
2188    Changes:
2189    
2190      The new pid generation scheme described a few weeks ago was overly
2191      complicated.  I implemented a new mechanism that is simpler and
2192      provides a bit more "stability":  Once CM has seen a compilation
2193      unit, it keeps its identity constant (as long as you do not delete
2194      those crucial CM/GUID/* files).  This means that when you change
2195      an interface, compile, then go back to the old interface, and
2196      compile again, you arrive at the original pid.
2197    
2198      There now also is a mechanism that instructs CM to use the plain
2199      environment hash as a module's pid (effectively making its GUID
2200      the empty string).  For this, "noguid" must be specified as an
2201      option to the .sml file in question within its .cm file.
2202      This is most useful for code that is being generated by tools such
2203      as ml-nlffigen (because during development programmers tend to
2204      erase the tool's entire output directory tree including CM's cached
2205      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
2206      revert to the old, broken behavior of SML/NJ, but in specific cases
2207      where there is no danger of interface confusion, its use is ok
2208      (I think).
2209    
2210      ml-nlffigen by default generates "noguid" annotations.  They can be
2211      turned off by specifying -guid in its command line.
2212    
2213    ----------------------------------------------------------------------
2214    Name: Lal George
2215    Date: 2002/03/12 12 14:42:36 EST
2216    Tag: george-20020312-frequency-computation
2217    Description:
2218    
2219    Integrated jump chaining and static block frequency into the
2220    compiler. More details and numbers later.
2221    
2222    ----------------------------------------------------------------------
2223    Name: Lal George
2224    Date: 2002/03/11 11 22:38:53 EST
2225    Tag: george-20020311-jump-chain-elim
2226    Description:
2227    
2228    Tested the jump chain elimination on all architectures (except the
2229    hppa).  This is on by default right now and is profitable for the
2230    alpha and x86, however, it may not be profitable for the sparc and ppc
2231    when compiling the compiler.
2232    
2233    The gc test will typically jump to a label at the end of the cluster,
2234    where there is another jump to an external cluster containing the actual
2235    code to invoke gc. This is to allow factoring of common gc invocation
2236    sequences. That is to say, we generate:
2237    
2238            f:
2239               testgc
2240               ja   L1      % jump if above to L1
2241    
2242            L1:
2243               jmp L2
2244    
2245    
2246    After jump chain elimination the 'ja L1' instructions is converted to
2247    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
2248    up being implemented in their long form (if L2 is far away) using:
2249    
2250            jbe     L3      % jump if below or equal to L3
2251            jmp     L2
2252         L3:
2253            ...
2254    
2255    
2256    For large compilation units L2  may be far away.
2257    
2258    
2259    ----------------------------------------------------------------------
2260    Name: Matthias Blume
2261    Date: 2002/03/11 13:30:00 EST
2262    Tag: blume-20020311-mltreeeval
2263    Description:
2264    
2265    A functor parameter was missing.
2266    
2267    ----------------------------------------------------------------------
2268    Name: Allen Leung
2269    Date: 2002/03/11 10:30:00 EST
2270    Tag: leunga-20020311-runtime-string0
2271    Description:
2272    
2273       The representation of the empty string now points to a
2274    legal null terminated C string instead of unit.  It is now possible
2275    to convert an ML string into C string with InlineT.CharVector.getData.
2276    This compiles into one single machine instruction.
2277    
2278    ----------------------------------------------------------------------
2279    Name: Allen Leung
2280    Date: 2002/03/10 23:55:00 EST
2281    Tag: leunga-20020310-x86-call
2282    Description:
2283    
2284       Added machine generation for CALL instruction (relative displacement mode)
2285    
2286    ----------------------------------------------------------------------
2287    Name: Matthias Blume
2288    Date: 2002/03/08 16:05:00
2289    Tag: blume-20020308-entrypoints
2290    Description:
2291    
2292    Version number bumped to 110.39.1.  NEW BOOTFILES!
2293    
2294    Entrypoints: non-zero offset into a code object where execution should begin.
2295    
2296    - Added the notion of an entrypoint to CodeObj.
2297    - Added reading/writing of entrypoint info to Binfile.
2298    - Made runtime system bootloader aware of entrypoints.
2299    - Use the address of the label of the first function given to mlriscGen
2300      as the entrypoint.  This address is currently always 0, but it will
2301      not be 0 once we turn on block placement.
2302    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
2303      with entry points) from mlriscGen.
2304    
2305    ----------------------------------------------------------------------
2306    Name: Allen Leung
2307    Date: 2002/03/07 20:45:00 EST
2308    Tag: leunga-20020307-x86-cmov
2309    Description:
2310    
2311       Bug fixes for CMOVcc on x86.
2312    
2313       1. Added machine code generation for CMOVcc
2314       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
2315       3. CMOVcc cannot have an immediate operand as argument.
2316    
2317    ----------------------------------------------------------------------
2318    Name: Matthias Blume
2319    Date: 2002/03/07 16:15:00 EST
2320    Tag: blume-20020307-controls
2321    Description:
2322    
2323    This is a very large but mostly boring patch which makes (almost)
2324    every tuneable compiler knob (i.e., pretty much everything under
2325    Control.* plus a few other things) configurable via both the command
2326    line and environment variables in the style CM did its configuration
2327    until now.
2328    
2329    Try starting sml with '-h' (or, if you are brave, '-H')
2330    
2331    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
2332    implements the underlying generic mechanism.
2333    
2334    The interface to some of the existing such facilities has changed somewhat.
2335    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
2336    (The getFoo interface is still there for backward-compatibility, but its
2337    use is deprecated.)
2338    
2339    The ml-build script passes -Cxxx=yyy command-line arguments through so
2340    that one can now twiddle the compiler settings when using this "batch"
2341    compiler.
2342    
2343    TODO items:
2344    
2345    We should go through and throw out all controls that are no longer
2346    connected to anything.  Moreover, we should go through and provide
2347    meaningful (and correct!) documentation strings for those controls
2348    that still are connected.
2349    
2350    Currently, multiple calls to Controls.new are accepted (only the first
2351    has any effect).  Eventually we should make sure that every control
2352    is being made (via Controls.new) exactly once.  Future access can then
2353    be done using Controls.acc.
2354    
2355    Finally, it would probably be a good idea to use the getter-setter
2356    interface to controls rather than ref cells.  For the time being, both
2357    styles are provided by the Controls module, but getter-setter pairs are
2358    better if thread-safety is of any concern because they can be wrapped.
2359    
2360    *****************************************
2361    
2362    One bug fix: The function blockPlacement in three of the MLRISC
2363    backpatch files used to be hard-wired to one of two possibilities at
2364    link time (according to the value of the placementFlag).  But (I
2365    think) it should rather sense the flag every time.
2366    
2367    *****************************************
2368    
2369    Other assorted changes (by other people who did not supply a HISTORY entry):
2370    
2371    1. the cross-module inliner now works much better (Monnier)
2372    2. representation of weights, frequencies, and probabilities in MLRISC
2373       changed in preparation of using those for weighted block placement
2374       (Reppy, George)
2375    
2376    ----------------------------------------------------------------------
2377    Name: Lal George
2378    Date: 2002/03/07 14:44:24 EST 2002
2379    Tag: george-20020307-weighted-block-placement
2380    
2381    Tested the weighted block placement optimization on all architectures
2382    (except the hppa) using AMPL to generate the block and edge frequencies.
2383    Changes were required in the machine properties to correctly
2384    categorize trap instructions. There is an MLRISC flag
2385    "weighted-block-placement" that can be used to enable weighted block
2386    placement, but this will be ineffective without block/edge
2387    frequencies (coming soon).
2388    
2389    
2390    ----------------------------------------------------------------------
2391    Name: Lal George
2392    Date: 2002/03/05 17:24:48 EST
2393    Tag: george-20020305-linkage-cluster
2394    
2395    In order to support the block placement optimization, a new cluster
2396    is generated as the very first cluster (called the linkage cluster).
2397    It contains a single jump to the 'real' entry point for the compilation
2398    unit. Block placement has no effect on the linkage cluster itself, but
2399    all the other clusters  have full freedom in the manner in which they
2400    reorder blocks or functions.
2401    
2402    On the x86 the typical linkage code that is generated is:
2403       ----------------------
2404            .align 2
2405       L0:
2406            addl    $L1-L0, 72(%esp)
2407            jmp     L1
2408    
2409    
2410            .align  2
2411       L1:
2412       ----------------------
2413    
2414    72(%esp) is the memory location for the stdlink register. This
2415    must contain the address of the CPS function being called. In the
2416    above example, it contains the address of  L0; before
2417    calling L1 (the real entry point for the compilation unit), it
2418    must contain the address for L1, and hence
2419    
2420            addl $L1-L0, 72(%esp)
2421    
2422    I have tested this on all architectures except the hppa.The increase
2423    in code size is of course negligible
2424    
2425    ----------------------------------------------------------------------
2426    Name: Allen Leung
2427    Date: 2002/03/03 13:20:00 EST
2428    Tag: leunga-20020303-mlrisc-tools
2429    
2430      Added #[ ... ] expressions to mlrisc tools
2431    
2432    ----------------------------------------------------------------------
2433    Name: Matthias Blume
2434    Date: 2002/02/27 12:29:00 EST
2435    Tag: blume-20020227-cdebug
2436    Description:
2437    
2438    - made types in structure C and C_Debug to be equal
2439    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
2440    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
2441    
2442    ----------------------------------------------------------------------
2443    Name: Matthias Blume
2444    Date: 2002/02/26 12:00:00 EST
2445    Tag: blume-20020226-ffi
2446    Description:
2447    
2448    1. Fixed a minor bug in CM's "noweb" tool:
2449       If numbering is turned off, then truly don't number (i.e., do not
2450       supply the -L option to noweb).  The previous behavior was to supply
2451       -L'' -- which caused noweb to use the "default" line numbering scheme.
2452       Thanks to Chris Richards for pointing this out (and supplying the fix).
2453    
2454    2. Once again, I reworked some aspects of the FFI:
2455    
2456       A. The incomplete/complete type business:
2457    
2458       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
2459         gone!
2460       - ML types representing an incomplete type are now *equal* to
2461         ML types representing their corresponding complete types (just like
2462         in C).  This is still safe because ml-nlffigen will not generate
2463         RTTI for incomplete types, nor will it generate functions that
2464         require access to such RTTI.   But when ML code generated from both
2465         incomplete and complete versions of the C type meet, the ML types
2466         are trivially interoperable.
2467    
2468         NOTE:  These changes restore the full generality of the translation
2469         (which was previously lost when I eliminated functorization)!
2470    
2471       B. Enum types:
2472    
2473       - Structure C now has a type constructor "enum" that is similar to
2474         how the "su" constructor works.  However, "enum" is not a phantom
2475         type because each "T enum" has values (and is isomorphic to
2476         MLRep.Signed.int).
2477       - There are generic access operations for enum objects (using
2478         MLRep.Signed.int).
2479       - ml-nlffigen will generate a structure E_foo for each "enum foo".
2480         * The structure contains the definition of type "mlrep" (the ML-side
2481         representation type of the enum).  Normally, mlrep is the same
2482         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
2483         then mlrep will be defined as a datatype -- thus facilitating
2484         pattern matching on mlrep values.
2485         ("-ec" will be suppressed if there are duplicate values in an
2486          enumeration.)
2487         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
2488         will be generated for each C enum constant xxx.
2489         * Conversion functions m2i and i2m convert between mlrep and
2490         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
2491         * Coversion functions c and ml convert between mlrep and "tag enum".
2492         * Access functions (get/set) fetch and store mlrep values.
2493       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
2494         enumerations are merged into one single enumeration represented by
2495         structure E_'.
2496    
2497    ----------------------------------------------------------------------
2498    Name: Allen Leung
2499    Date: 2002/02/25 04:45:00 EST
2500    Tag: leunga-20020225-cps-spill
2501    
2502    This is a new implementation of the CPS spill phase.
2503    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
2504    In case of problems, replace it with the old file spill.sml
2505    
2506    The current compiler runs into some serious performance problems when
2507    constructing a large record.  This can happen when we try to compile a
2508    structure with many items.  Even a very simple structure like the following
2509    makes the compiler slow down.
2510    
2511        structure Foo = struct
2512           val x_1 = 0w1 : Word32.int
2513           val x_2 = 0w2 : Word32.int
2514           val x_3 = 0w3 : Word32.int
2515           ...
2516           val x_N = 0wN : Word32.int
2517        end
2518    
2519    The following table shows the compile time, from N=1000 to N=4000,
2520    with the old compiler:
2521    
2522    N
2523    1000   CPS 100 spill                           0.04u  0.00s  0.00g
2524           MLRISC ra                               0.06u  0.00s  0.05g
2525              (spills = 0 reloads = 0)
2526           TOTAL                                   0.63u  0.07s  0.21g
2527    
2528    1100   CPS 100 spill                           8.25u  0.32s  0.64g
2529           MLRISC ra                               5.68u  0.59s  3.93g
2530              (spills = 0 reloads = 0)
2531           TOTAL                                   14.71u  0.99s  4.81g
2532    
2533    1500   CPS 100 spill                           58.55u  2.34s  1.74g
2534           MLRISC ra                               5.54u  0.65s  3.91g
2535              (spills = 543 reloads = 1082)
2536           TOTAL                                   65.40u  3.13s  6.00g
2537    
2538    2000   CPS 100 spill                           126.69u  4.84s  3.08g
2539           MLRISC ra                               0.80u  0.10s  0.55g
2540              (spills = 42 reloads = 84)
2541           TOTAL                                   129.42u  5.10s  4.13g
2542    
2543    3000   CPS 100 spill                           675.59u  19.03s  11.64g
2544           MLRISC ra                               2.69u  0.27s  1.38g
2545              (spills = 62 reloads = 124)
2546           TOTAL                                   682.48u  19.61s  13.99g
2547    
2548    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
2549           MLRISC ra                               4.96u  0.27s  2.72g
2550              (spills = 85 reloads = 170)
2551           TOTAL                                   2375.26u  57.21s  48.00g
2552    
2553    As you can see the old cps spill module suffers from some serious
2554    performance problem.  But since I cannot decipher the old code fully,
2555    instead of patching the problems up, I'm reimplementing it
2556    with a different algorithm.  The new code is more modular,
2557    smaller when compiled, and substantially faster
2558    (O(n log n) time and O(n) space).  Timing of the new spill module:
2559    
2560    4000  CPS 100 spill                           0.02u  0.00s  0.00g
2561          MLRISC ra                               0.25u  0.02s  0.15g
2562             (spills=1 reloads=3)
2563          TOTAL                                   7.74u  0.34s  1.62g
2564    
2565    Implementation details:
2566    
2567    As far as I can tell, the purpose of the CPS spill module is to make sure the
2568    number of live variables at any program point (the bandwidth)
2569    does not exceed a certain limit, which is determined by the
2570    size of the spill area.
2571    
2572    When the bandwidth is too large, we decrease the register pressure by
2573    packing live variables into spill records.  How we achieve this is
2574    completely different than what we did in the old code.
2575    
2576    First, there is something about the MLRiscGen code generator
2577    that we should be aware of:
2578    
2579    o MLRiscGen performs code motion!
2580    
2581       In particular, it will move floating point computations and
2582       address computations involving only the heap pointer to
2583       their use sites (if there is only a single use).
2584       What this means is that if we have a CPS record construction
2585       statement
2586    
2587           RECORD(k,vl,w,e)
2588    
2589       we should never count the new record address w as live if w
2590       has only one use (which is often the case).
2591    
2592       We should do something similar to floating point, but the transformation
2593       there is much more complex, so I won't deal with that.
2594    
2595    Secondly, there are now two new cps primops at our disposal:
2596    
2597     1. rawrecord of record_kind option
2598        This pure operator allocates some uninitialized storage from the heap.
2599        There are two forms:
2600    
2601         rawrecord NONE [INT n]  allocates a tagless record of length n
2602         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
2603                                     and initializes the tag.
2604    
2605     2. rawupdate of cty
2606          rawupdate cty (v,i,x)
2607          Assigns to x to the ith component of record v.
2608          The storelist is not updated.
2609    
2610    We use these new primops for both spilling and increment record construction.
2611    
2612     1. Spilling.
2613    
2614        This is implemented with a linear scan algorithm (but generalized
2615        to trees).  The algorithm will create a single spill record at the
2616        beginning of the cps function and use rawupdate to spill to it,
2617        and SELECT or SELp to reload from it.  So both spills and reloads
2618        are fine-grain operations.  In contrast, in the old algorithm
2619        "spills" have to be bundled together in records.
2620    
2621        Ideally, we should sink the spill record construction to where
2622        it is needed.  We can even split the spill record into multiple ones
2623        at the places where they are needed.  But CPS is not a good
2624        representation for global code motion, so I'll keep it simple and
2625        am not attempting this.
2626    
2627     2. Incremental record construction (aka record splitting).
2628    
2629        Long records with many component values which are simulatenously live
2630        (recall that single use record addresses are not considered to
2631         be live) are constructed with rawrecord and rawupdate.
2632        We allocate space on the heap with rawrecord first, then gradually
2633        fill it in with rawupdate.  This is the technique suggested to me
2634        by Matthias.
2635    
2636        Some restrictions on when this is applicable:
2637        1. It is not a VECTOR record.  The code generator currently does not handle
2638           this case. VECTOR record uses double indirection like arrays.
2639        2. All the record component values are defined in the same "basic block"
2640           as the record constructor.  This is to prevent speculative
2641           record construction.
2642    
2643    ----------------------------------------------------------------------
2644    Name: Allen Leung
2645    Date: 2002/02/22 01:02:00 EST
2646    Tag: leunga-20020222-mlrisc-tools
2647    
2648    Minor bug fixes in the parser and rewriter
2649    
2650    ----------------------------------------------------------------------
2651    Name: Allen Leung
2652    Date: 2002/02/21 20:20:00 EST
2653    Tag: leunga-20020221-peephole
2654    
2655    Regenerated the peephole files.  Some contained typos in the specification
2656    and some didn't compile because of pretty printing bugs in the old version
2657    of 'nowhere'.
2658    
2659    ----------------------------------------------------------------------
2660    Name: Allen Leung
2661    Date: 2002/02/19 20:20:00 EST
2662    Tag: leunga-20020219-mlrisc-tools
2663    Description:
2664    
2665       Minor bug fixes to the mlrisc-tools library:
2666    
2667       1.  Fixed up parsing colon suffixed keywords
2668       2.  Added the ability to shut the error messages up
2669       3.  Reimplemented the pretty printer and fixed up/improved
2670           the pretty printing of handle and -> types.
2671       4.  Fixed up generation of literal symbols in the nowhere tool.
2672       5.  Added some SML keywords to to sml.sty
2673    
2674    ----------------------------------------------------------------------
2675    Name: Matthias Blume
2676    Date: 2002/02/19 16:20:00 EST
2677    Tag: blume-20020219-cmffi
2678    Description:
2679    
2680    A wild mix of changes, some minor, some major:
2681    
2682    * All C FFI-related libraries are now anchored under $c:
2683        $/c.cm      --> $c/c.cm
2684        $/c-int.cm  --> $c/internals/c-int.cm
2685        $/memory.cm --> $c/memory/memory.cm
2686    
2687    * "make" tool (in CM) now treats its argument pathname slightly
2688      differently:
2689        1. If the native expansion is an absolute name, then before invoking
2690           the "make" command on it, CM will apply OS.Path.mkRelative
2691           (with relativeTo = OS.FileSys.getDir()) to it.
2692        2. The argument will be passed through to subsequent phases of CM
2693           processing without "going native".  In particular, if the argument
2694           was an anchored path, then "make" will not lose track of that anchor.
2695    
2696    * Compiler backends now "know" their respective C calling conventions
2697      instead of having to be told about it by ml-nlffigen.  This relieves
2698      ml-nlffigen from one of its burdens.
2699    
2700    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
2701    
2702    * Export C_DEBUG and C_Debug from $c/c.cm.
2703    
2704    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
2705      subtyping relationship between incomplete pointers and their complete
2706      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
2707      with the convention of instantiating 'o with ('t, 'c) obj whenever
2708      the pointer target type is complete.  In the incomplete case, 'o
2709      will be instantiated with some "'c iobj" -- a type obtained by
2710      using one of the functors PointerToIncompleteType or PointerToCompleteType.
2711    
2712      Operations that work on both incomplete and complete pointer types are
2713      typed as taking an 'o ptr while operations that require the target to
2714      be known are typed as taking some ('t, 'c) obj ptr.
2715    
2716      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
2717      where void is an eqtype without any values.  This makes it possible
2718      to work on voidptr values using functions meant to operate on light
2719      incomplete pointers.
2720    
2721    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
2722      been vastly simplified.
2723    
2724    ----------------------------------------------------------------------
2725    Name: Matthias Blume
2726    Date: 2002/02/19 10:48:00 EST
2727    Tag: blume-20020219-pqfix
2728    Description:
2729    
2730    Applied Chris Okasaki's bug fix for priority queues.
2731    
2732    ----------------------------------------------------------------------
2733    Name: Matthias Blume
2734    Date: 2002/02/15 17:05:00
2735    Tag: Release_110_39
2736    Description:
2737    
2738    Last-minute retagging is becoming a tradition... :-(
2739    
2740    This is the working release 110.39.
2741    
2742    ----------------------------------------------------------------------
2743    Name: Matthias Blume
2744    Date: 2002/02/15 16:00:00 EST
2745    Tag: Release_110_39-orig
2746    Description:
2747    
2748    Working release 110.39.  New bootfiles.
2749    
2750    (Update: There was a small bug in the installer so it wouldn't work
2751    with all shells.  So I retagged. -Matthias)
2752    
2753    ----------------------------------------------------------------------
2754    Name: Matthias Blume
2755    Date: 2002/02/15 14:17:00 EST
2756    Tag: blume-20020215-showbindings
2757    Description:
2758    
2759    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
2760    the latter can be useful for exploring what bindings are available at
2761    the interactive prompt.  (The first function returns only the list
2762    of symbols that are really bound, the second prints those but also the
2763    ones that CM's autoloading mechanism knows about.)
2764    
2765    ----------------------------------------------------------------------
2766    Name: Matthias Blume
2767    Date: 2002/02/15 12:08:00 EST
2768    Tag: blume-20020215-iptrs
2769    Description:
2770    
2771    Two improvements to ml-nlffigen:
2772    
2773      1. Write files only if they do not exist or if their current contents
2774         do not coincide with what's being written.  (That is, avoid messing
2775         with the time stamps unless absolutely necessary.)
2776    
2777      2. Implement a "repository" mechanism for generated files related
2778         to "incomplete pointer types".   See the README file for details.
2779    
2780    ----------------------------------------------------------------------
2781    Name: Matthias Blume
2782    Date: 2002/02/14 11:50:00 EST
2783    Tag: blume-20020214-quote
2784    Description:
2785    
2786    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
2787    because of the new and improved tag generation scheme.  (Thanks to Allen
2788    Leung for pointing it out.)
2789    
2790    ----------------------------------------------------------------------
2791    Name: Lal George
2792    Date: 2002/02/14 09:55:27 EST 2002
2793    Tag: george-20020214-isabelle-bug
2794    Description:
2795    
2796    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
2797    of Isabelle on the x86.
2798    
2799    From Allen:
2800    -----------
2801     I've found the problem:
2802    
2803         in ra-core.sml, I use the counter "blocked" to keep track of the
2804         true number of elements in the freeze queue.  When the counter goes
2805         to zero, I skip examining the queue.  But I've messed up the
2806         bookkeeping in combine():
2807    
2808             else ();
2809             case !ucol of
2810               PSEUDO => (if !cntv > 0 then
2811                     (if !cntu > 0 then blocked := !blocked - 1 else ();
2812                                        ^^^^^^^^^^^^^^^^^^^^^^^
2813                      moveu := mergeMoveList(!movev, !moveu)
2814                     )
2815                  else ();
2816    
2817         combine() is called to coalesce two nodes u and v.
2818         I think I was thinking that if the move counts of u and v are both
2819         greater than zero then after they are coalesced then one node is
2820         removed from the freeze queue.  Apparently I was thinking that
2821         both u and v are of low degree, but that's clearly not necessarily true.
2822    
2823    
2824    02/12/2002:
2825        Here's the patch.  HOL now compiles.
2826    
2827        I don't know how this impact on performance (compile
2828        time or runtime).  This bug caused the RA (especially on the x86)
2829        to go thru the potential spill phase when there are still nodes on the
2830        freeze queue.
2831    
2832    
2833    
2834    
2835    ----------------------------------------------------------------------
2836    Name: Matthias Blume
2837    Date: 2002/02/13 22:40:00 EST
2838    Tag: blume-20020213-fptr-rtti
2839    Description:
2840    
2841    Fixed a bug in ml-nlffigen that was introduced with one of the previous
2842    updates.
2843    
2844    ----------------------------------------------------------------------
2845    Name: Matthias Blume
2846    Date: 2002/02/13 16:41:00 EST
2847    Tag: blume-20020213-cmlpq
2848    Description:
2849    
2850    Added new priority queue export symbols (which have just been added to
2851    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
2852    not compile and the installer would choke.)
2853    
2854    ----------------------------------------------------------------------
2855    Name: Matthias Blume
2856    Date: 2002/02/13 16:15:00 EST
2857    Tag: blume-20020213-various
2858    Description:
2859    
2860    1. More tweaks to ml-nlffigen:
2861    
2862       - better internal datastructures (resulting in slight speedup)
2863       - "-match" option requires exact match
2864       - "localized" gensym counters (untagged structs/unions nested within
2865         other structs/unions or within typedefs get a fresh counter; their
2866         tag will be prefixed by a concatenation of their parents' tags)
2867       - bug fixes (related to calculation of transitive closure of types
2868         to be included in the output)
2869    
2870    2. Minor Basis updates:
2871    
2872       - added implementations for List.collate and Option.app
2873    
2874    ----------------------------------------------------------------------
2875    Name: Matthias Blume
2876    Date: 2002/02/11 15:55:00 EST
2877    Tag: blume-20020211-gensym
2878    Description:
2879    
2880    Added a "-gensym" option to command line of ml-nlffigen.  This can be
2881    used to specify a "stem" -- a string that is inserted in all "gensym'd"
2882    names (ML structure names that correspond to unnamed C structs, unions,
2883    and enums), so that separate runs of ml-nlffigen do not clash.
2884    
2885    ----------------------------------------------------------------------
2886    Name: Matthias Blume
2887    Date: 2002/02/11 12:05:00 EST
2888    Tag: blume-20020211-gensml
2889    Description:
2890    
2891    A quick fix for a problem with GenSML (in the pgraph-util library):
2892    Make generation of toplevel "local" optional.  (Strictly speaking,
2893    signature definitions within "local" are not legal SML.)
2894    
2895    Other than that: updates to INSTALL and cm/TODO.
2896    
2897    ----------------------------------------------------------------------
2898    Name: Matthias Blume
2899    Date: 2002/02/08 15:00:00 EST
2900    Tag: blume-20020208-uniquepid
2901    Description:
2902    
2903    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
2904    
2905    1. The installer (config/install.sh) has gotten smarter:
2906    
2907         - Configuration options are a bit easier to specify now
2908           (in config/targets).
2909         - Bug in recognizing .tar.bz2 files fixed.
2910         - Installer automatically resolves dependencies between
2911           configuration options (e.g., if you ask for eXene, you will
2912           also get cml -- regardless whether you asked for it or not).
2913         - Installer can run in "quieter mode" by setting the environment
2914           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
2915           "completely silent", though.
2916         - Build HashCons library as part of smlnj-lib.
2917    
2918    2. A new scheme for assigning persistent identifiers to compilation
2919       units (and, by extension, to types etc.) has been put into place.
2920       This fixes a long-standing bug where types and even dynamic values
2921       can get internally confused, thereby compromising type safety
2922       (abstraction) and dynamic correctness.  See
2923    
2924         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
2925    
2926       for an example of how things could go wrong until now.
2927    
2928       The downside of the new scheme is that pids are not quite as
2929       persistent as they used to be: CM will generate a fresh pid
2930       for every compilation unit that it thinks it sees for the first
2931       time.  That means that if you compile starting from a clean, fresh
2932       source tree at two different times, you end up with different
2933       binaries.
2934    
2935       Cutoff recompilation, however, has not been compromised because
2936       CM keeps pid information in special caches between runs.
2937    
2938    ----------------------------------------------------------------------
2939    Name: Lal George
2940    Date: 2002/02/07 15:34:13 EST 2002
2941    Tag: <none>
2942    Description:
2943    
2944    Compilers that generate assembly code may produce  global labels
2945    whose value is resolved at link time. The various peephole optimization
2946    modules did not take this in account.
2947    
2948    TODO. The Labels.addrOf function should really return an option
2949    type so that clients are forced to deal with this issue, rather
2950    than an exception being raised.
2951    
2952    ----------------------------------------------------------------------
2953    Name: Lal George
2954    Date: 2002/02/06 13:55:02 EST
2955    Tag: george-20020206-ra-breakup
2956    Description:
2957    
2958    1. A bug fix from Allen.
2959    
2960        A typo causes extra fstp %st(0)'s to be generated at compensation
2961        edges, which might cause stack underflow traps at runtime.  This
2962        occurs in fft where there are extraneous fstps right before the 'into'
2963        trap instruction (in this case they are harmless since none of the
2964        integers overflow.)
2965    
2966    2. Pulled out various utility modules that were embedded in the modules
2967       of the register allocator. I need these modules for other purposes, but
2968       they are not complete enough to put into a library (just yet).
2969    ----------------------------------------------------------------------
2970    Name: Matthias Blume
2971    Date: 2002/01/31 16:05:00 EST
2972    Tag: blume-20020131-sparc-ccalls
2973    Description:
2974    
2975    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
2976       of extra stack space by mistake.  Fixed.
2977    
2978    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
2979    
2980    ----------------------------------------------------------------------
2981    Name: Allen Leung
2982    Date: 2002/01/30
2983    Tag: leunga-20020130-nowhere-bug-fix
2984    Description:
2985    
2986       MLRISC bug fixes:
2987       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
2988       2. MachineInt.fromString was negating its value.
2989    
2990    ----------------------------------------------------------------------
2991    Name: Matthias Blume
2992    Date: 2002/01/29
2993    Tag: blume-20020129-INSTALL
2994    Description:
2995    
2996    - Added somewhat detailed installation instructions (file INSTALL).
2997    - Fixed curl-detection bug in config/install.sh.
2998    - It is now possible to select the URL getter using the URLGETTER
2999      environment variable:
3000    
3001          not set / "unknown"      --> automatic detection (script tries wget,
3002                                       curl, and lynx)
3003          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
3004                                       how to properly invoke them)
3005          other                    --> use $URLGETTER directly, it must take
3006                                       precisely two command-line arguments
3007                                       (source URL and destination file name)
3008    
3009    ----------------------------------------------------------------------
3010    Name: Matthias Blume
3011    Date: 2002/01/28
3012    Tag: blume-20020128-sparc-ccalls
3013    Description:
3014    
3015    - Fixed problem with calculation of "used" registers in sparc-c-calls.
3016    - Make use of the allocParam argument in sparc-c-calls.
3017    
3018    ----------------------------------------------------------------------
3019    Name: Matthias Blume
3020    Date: 2002/01/28
3021    Tag: blume-20020128-allocParam
3022    Description:
3023    
3024    John Reppy:  Changes c-calls API to accept client-callback for
3025    allocating extra stack space.
3026    me: Corresponding changes to mlriscGen (using a dummy argument that
3027        does not change the current behavior).
3028    
3029    ----------------------------------------------------------------------
3030    Name: Matthias Blume
3031    Date: 2002/01/28 12:00:00
3032    Tag: Release_110_38
3033    Description:
3034    
3035    This time for real!!!
3036    
3037    ----------------------------------------------------------------------
3038    Name: Matthias Blume
3039    Date: 2002/01/28 10:56:00 EST
3040    Tag: blume-20020128-retraction
3041    Description:
3042    
3043    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
3044       with blume-Release_110_38-retracted.)
3045    
3046    1. Fixed a problem with incorrect rounding modes in real64.sml.
3047       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
3048    
3049    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
3050       and enums fixed.  The naming of corresponding ML identifiers should
3051       now be consistent again.
3052    
3053    ----------------------------------------------------------------------
3054    Name: Allen Leung
3055    Date: 2002/01/27
3056    Tag: leunga-20020127-nowhere
3057    Description:
3058    
3059       Added a target called nowhere in the configuration scripts.
3060       Enabling this will build the MLRISC 'nowhere' tool (for translating
3061       programs with where-clauses into legal SML code) during installation.
3062    
3063    ----------------------------------------------------------------------
3064    Name: Matthias Blume
3065    Date: 2002/01/25 21:27:00 EST
3066    Tag: blume-Release_110_38-retracted
3067    Description:
3068    
3069    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
3070    
3071    README will be added later.
3072    
3073    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
3074    (Release_110_38) removed.  Reason: Last-minute bug fixes.
3075    
3076    ----------------------------------------------------------------------
3077    Name: Matthias Blume
3078    Date: 2002/01/25
3079    Tag: blume-20020125-ffi
3080    Description:
3081    
3082    A large number of tweaks and improvements to ml-nlffi-lib and
3083    ml-nlffigen:
3084    
3085       - ML represenation types have been streamlined
3086       - getter and setter functions work with concrete values, not abstract
3087         ones where possible
3088       - ml-nlffigen command line more flexible (see README file there)
3089       - some bugs have been fixed (hopefully)
3090    
3091    ----------------------------------------------------------------------
3092    Name: Lal George
3093    Date: 2002/01/24
3094    Tag: george-20020124-risc-ra-interface
3095    Description:
3096    
3097       There is a dramatic simplification in the interface to the
3098       register allocator for RISC architectures as a result of making
3099       parallel copy instructions explicit.
3100    
3101    ----------------------------------------------------------------------
3102    Name: Matthias Blume
3103    Date: 2002/01/22
3104    Tag: blume-20020122-x86-ccalls
3105    Description:
3106    
3107    Bug fix for c-calls on x86 (having to do with how char- and
3108    short-arguments are being handled).
3109    
3110    ----------------------------------------------------------------------
3111    Name: Matthias Blume
3112    Date: 2002/01/21
3113    Tag: blume-20020121-ff
3114    Description:
3115    
3116    Another day of fiddling with the FFI...
3117    
3118    1. Bug fix/workaround:  CKIT does not complain about negative array
3119       dimensions, so ml-nlffigen has to guard itself against this possibility.
3120       (Otherwise a negative dimension would send it into an infinite loop.)
3121    
3122    2. Some of the abstract types (light objects, light pointers, most "base"
3123       types) in structure C are now eqtypes.
3124    
3125    3. Added constructors and test functions for NULL function pointers.
3126    
3127    ----------------------------------------------------------------------
3128    Name: Matthias Blume
3129    Date: 2002/01/18
3130    Tag: blume-20020118-ready-for-new-release
3131    Description:
3132    
3133    Made config/srcarchiveurl point to a new place.  (Will provide boot
3134    files shortly.)
3135    
3136    Maybe we christen this to be 110.38?
3137    
3138    ----------------------------------------------------------------------
3139    Name: Matthias Blume
3140    Date: 2002/01/18
3141    Tag: blume-20020118-more-ffifiddle
3142    Description:
3143    
3144    Today's FFI fiddling:
3145    
3146      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
3147        C.Set.* that use concrete (MLRep.*) arguments and results instead
3148        of abstract ones.
3149    
3150      - Provided word-style bit operations etc. for "int" representation
3151        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
3152        and Long.
3153    
3154    ----------------------------------------------------------------------
3155    Name: Matthias Blume
3156    Date: 2002/01/18
3157    Tag: blume-20020118-use-x86-fp
3158    Description:
3159    
3160    Now that x86-fast-fp seems to be working, I turned it back on again
3161    by default.  (Seems to work fine now, even with the FFI.)
3162    
3163    Other than that, I added some documentation about the FFI to
3164    src/ml-nlffigen/README and updated the FFI test examples in
3165    src/ml-nlffi-lib/Tests/*.
3166    
3167    ----------------------------------------------------------------------
3168    Name: Allen Leung
3169    Date: 2002/01/17
3170    Tag: leunga-20020117-x86-fast-fp-call
3171    Description:
3172    
3173       1. Fixed a problem with handling return fp values when x86's fast fp
3174          mode is turned on.
3175    
3176       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
3177          of %f32.
3178    
3179       3. Added a constructor INT32lit to the ast of MLRISC tools.
3180    
3181    ----------------------------------------------------------------------
3182    Name: Matthias Blume
3183    Date: 2002/01/16
3184    Tag: blume-20020116-ffifiddle
3185    Description:
3186    
3187    More fiddling with the FFI interface:
3188    
3189     - Make constness 'c instead of rw wherever possible.  This eliminates
3190       the need for certain explicit coercions.  (However, due to ML's
3191       value polymorphism, there will still be many cases where explicit
3192       coercions are necessary.  Phantom types are not the whole answer
3193       to modeling a subtyping relationship in ML.)
3194    
3195     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
3196    
3197     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
3198       clumsy workaround for varargs problem.)
3199    
3200    ----------------------------------------------------------------------
3201    Name: Lal George
3202    Date: 2002/01/15
3203    Tag: <none>
3204    Description:
3205    
3206    1. Since COPY instructions are no longer native to the architecture,
3207       a generic functor can be used to implement the expandCopies function.
3208    
3209    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
3210       TEXT segment.
3211    
3212    ----------------------------------------------------------------------
3213    Name: Matthias Blume
3214    Date: 2002/01/15
3215    Tag: blume-20020115-ffiupdates
3216    Description:
3217    
3218    1. Fix for bug resulting in single-precision float values being returned
3219       incorrectly from FFI calls.
3220    
3221    2. Small modifications to C FFI API:
3222    
3223        - memory-allocation routines return straight objects (no options)
3224          and raise an exception in out-of-memory situations
3225        - unsafe extensions to cast between function pointers and pointers
3226          from/to ints
3227        - added structure C_Debug as an alternative to structure C where
3228          pointer-dereferencing (|*| and |*!) always check for null-pointers
3229        - added open_lib' to DynLinkage;  open_lib' works like open_lib
3230          but also takes a (possibly empty) list of existing library handles
3231          that the current library depends on
3232    
3233    ----------------------------------------------------------------------
3234    Name: Matthias Blume
3235    Date: 2002/01/10
3236    Tag: blume-20020110-newffigen
3237    Description:
3238    
3239    1. Updates to portable graph code.
3240    
3241    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
3242       more scalable now so that even huge interfaces such as the one
3243       for GTK compile in finite time and space. :-)
3244       See src/ml-nlffigen/README for details on what's new.
3245    
3246    ----------------------------------------------------------------------
3247    Name: Lal George
3248    Date: 2001/01/09 14:31:35 EST 2002
3249    Tag: george-20011206-rm-native-copy
3250    Description:
3251    
3252            Removed the native COPY and FCOPY instructions
3253            from all the architectures and replaced it with the
3254            explicit COPY instruction from the previous commit.
3255    
3256            It is now possible to simplify many of the optimizations
3257            modules that manipulate copies. This has not been
3258            done in this change.
3259    
3260    ----------------------------------------------------------------------
3261    Name: Lal George
3262    Date: 2001/12/06 16:50:13 EST 2001
3263    Tag: george-20011206-mlrisc-instruction
3264    Description:
3265    
3266    Changed the representation of instructions from being fully abstract
3267    to being partially concrete. That is to say:
3268    
3269      from
3270            type instruction
3271    
3272      to
3273            type instr                              (* machine instruction *)
3274    
3275            datatype instruction =
3276                LIVE of {regs: C.cellset, spilled: C.cellset}
3277              | KILL of {regs: C.cellset, spilled: C.cellset}
3278              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
3279              | ANNOTATION of {i: instruction, a: Annotations.annotation}
3280              | INSTR of instr
3281    
3282    This makes the handling of certain special instructions that appear on
3283    all architectures easier and uniform.
3284    
3285    LIVE and KILL say that a list of registers are live or killed at the
3286    program point where they appear. No spill code is generated when an
3287    element of the 'regs' field is spilled, but the register is moved to
3288    the 'spilled' (which is present, more for debugging than anything else).
3289    
3290    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
3291    We used to generate:
3292    
3293            DEFFREG f1
3294            f1 := f2 + f3
3295            trapb
3296    
3297    but now generate:
3298    
3299            f1 := f2 + f3
3300            trapb
3301            LIVE {regs=[f1,f2,f3], spilled=[]}
3302    
3303    Furthermore, the DEFFREG (hack) required that all floating point instruction
3304    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
3305    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
3306    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
3307    this architecture).
3308    
3309    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
3310    all the architectures. This will result in further simplification of the
3311    register allocator that must be aware of them for coalescing purposes, and
3312    will also simplify certain aspects of the machine description that provides
3313    callbacks related to parallel copies.
3314    
3315    ANNOTATION should be obvious, and now INSTR represents the honest to God
3316    machine instruction set!
3317    
3318    The <arch>/instructions/<arch>Instr.sml files define certain utility
3319    functions for making porting easier -- essentially converting upper case
3320    to lower case. All machine instructions (of type instr) are in upper case,
3321    and the lower case form generates an MLRISC instruction. For example on
3322    the alpha we have:
3323    
3324      datatype instr =
3325         LDA of {r:cell, b:cell, d:operand}
3326       | ...
3327    
3328      val lda : {r:cell, b:cell, d:operand} -> instruction
3329        ...
3330    
3331    where lda is just (INSTR o LDA), etc.
3332    
3333    ----------------------------------------------------------------------
3334    Name: Matthias Blume
3335    Date: 2001/11/22 21:40:00 EST
3336    Tag: Release_110_37
3337    Description:
3338    
3339    Release 110.37.  This time for real.
3340    
3341    ----------------------------------------------------------------------
3342    Name: Matthias Blume
3343    Date: 2001/11/21 16:35:00 EST
3344    Tag: blume-20011121-foot-in-mouth
3345    Description:
3346    
3347    Removed the "Release_110_37" tag because of a serious bug.
3348    This will be re-tagged once the bug is fixed.
3349    
3350    ----------------------------------------------------------------------
3351    Name: Matthias Blume
3352    Date: 2001/11/21 16:14:00 EST
3353    Tag: blume-20011121-forgottenfile
3354    Description:
3355    
3356    Forgot to add a file.  (Just a .tex-file -- part of
3357    the CM manual source.)
3358    
3359    ----------------------------------------------------------------------
3360    Name: Matthias Blume
3361    Date: 2001/11/21 16:10:00 EST
3362    Tag: blume-20011121-invalid_110_37
3363    Description:
3364    
3365    Note: I removed the original tag "Release_110_37" from this commit
3366          because we found a serious bug in all non-x86 backends.
3367          - Matthias
3368    
3369    1. Modifications to the SML/NJ code generator and to the runtime system
3370       so that code object name strings are directly inserted into code
3371       objects at code generation time.  The only business the runtime system
3372       has with this is now to read the name strings on occasions.
3373       (The encoding of the name string has also changed somewhat.)
3374    
3375    2. CM now implements a simple "set calculus" for specifying export lists.
3376       In particular, it is now possible to refer to the export lists of
3377       other libraries/groups/sources and form unions as well as differences.
3378       See the latest CM manual for details.
3379    
3380    3. An separate notion of "proxy" libraries has again be eliminated from
3381       CM's model.  (Proxy libraries are now simply a special case of using
3382       the export list calculus.)
3383    
3384    4. Some of the existing libraries now take advantage of the new set
3385       calculus.
3386       (Notice that not all libraries have been converted because some
3387       of the existing .cm-files are supposed to be backward compatible
3388       with 110.0.x.)
3389    
3390    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
3391       "General.exnMessage"!  The former relies on a certain hook to be
3392       initialized, and that often does not happen in the stand-alone case.)
3393    
3394    ----------------------------------------------------------------------
3395    Name: Lal George
3396    Date: 2001/11/21  13:56:18 EST
3397    Tag: george-2001121-pseudo-ops
3398    Description:
3399    
3400      Implemented a complete redesign of MLRISC pseudo-ops. Now there
3401      ought to never be any question of incompatabilities with
3402      pseudo-op syntax expected by host assemblers.
3403    
3404      For now, only modules supporting GAS syntax are implemented
3405      but more should follow, such as MASM, and vendor assembler
3406      syntax, e.g. IBM as, Sun as, etc.
3407    
3408    ----------------------------------------------------------------------
3409    Name: Matthias Blume
3410    Date: 2001/11/14 11:52:00 EST
3411    Tag: blume-20011114-srcname
3412    Description:
3413    
3414    1. Routed the name of the current source file to mlriscgen where it
3415       should be directly emitted into the code object.  (This last part
3416       is yet to be done.)
3417    
3418    2. Some cleanup of the pgraph code to make it match the proposal that
3419       I put out the other day.  (The proposal notwithstanding, things are
3420       still in flux here.)
3421    
3422    ----------------------------------------------------------------------
3423    Name: Lal George
3424    Date: 2001/11/14 09:44:04 EST
3425    Tag:
3426    Description:
3427    
3428      Fix for a backpatching bug reported by Allen.
3429    
3430      Because the boundary between short and long span-dependent
3431      instructions is +/- 128, there are an astounding number of
3432      span-dependent instructions whose size is over estimated.
3433    
3434      Allen came up with the idea of letting the size of span
3435      dependent instructions be non-monotonic, for a maxIter
3436      number of times, after which the size must be monotonically
3437      increasing.
3438    
3439      This table shows the number of span-dependent instructions
3440      whose size was over-estimated as a function of maxIter, for the
3441      file Parse/parse/ml.grm.sml:
3442    
3443         maxIter            # of instructions:
3444            10                      687
3445            20                      438
3446            30                      198
3447            40                        0
3448    
3449      In compiling the compiler, there is no significant difference in
3450      compilation speed between maxIter=10 and maxIter=40. Actually,
3451      my measurements showed that maxIter=40 was a tad faster than
3452      maxIter=10! Also 96% of the  files in the compiler reach a fix
3453      point within 13 iterations, so fixing maxIter at 40, while high,
3454      is okay.
3455    
3456    ----------------------------------------------------------------------
3457    Name: Matthias Blume
3458    Date: 2001/10/31 15:25:00 EST
3459    Tag: blume-20011031-pgraph
3460    Description:
3461    
3462    CKIT:
3463    * Changed the "Function" constructor of type Ast.ctype to carry optional
3464      argument identifiers.
3465    * Changed the return type of TypeUtil.getFunction accordingly.
3466    * Type equality ignores the argument names.
3467    * TypeUtil.composite tries to preserve argument names but gives up quickly
3468      if there is a mismatch.
3469    
3470    installation script:
3471    * attempts to use "curl" if available (unless "wget" is available as well)
3472    
3473    CM:
3474    * has an experimental implementation of "portable graphs" which I will
3475      soon propose as an implementation-independent library format
3476    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
3477    
3478    NLFFI-LIB:
3479    * some cleanup (all cosmetic)
3480    
3481    NLFFIGEN:
3482    * temporarily disabled the mechanism that suppresses ML output for
3483      C definitions whose identifiers start with an underscore character
3484    * generate val bindings for enum constants
3485    * user can request that only one style (light or heavy) is being used;
3486      default is to use both (command-line arguments: -heavy and -light)
3487    * fixed bug in handling of function types involving incomplete pointers
3488    * generate ML entry points that take record arguments (i.e., using
3489      named arguments) for C functions that have a prototype with named
3490      arguments
3491      (see changes to CKIT)
3492    
3493    ----------------------------------------------------------------------
3494    Name: Allen Leung
3495    Date: 2001/10/27 20:34:00 EDT
3496    Tag: leunga-20011027-x86-fast-fp-call
3497    Description:
3498    
3499       Fixed the bug described in blume-20010920-slowfp.
3500    
3501       The fix involves
3502          1. generating FCOPYs in FSTP in ia32-svid
3503          2. marking a CALL with the appropriate annotation
3504    
3505    ----------------------------------------------------------------------
3506    Name: Matthias Blume
3507    Date: 2001/10/16 11:32:00 EDT
3508    Tag: blume-20011016-netbsd
3509    Description:
3510    
3511    Underscore patch from Chris Richards (fixing problem with compiling
3512    runtime system under recent NetBSD).
3513    
3514    ----------------------------------------------------------------------
3515    Name: Allen Leung
3516    Date: 2001/10/12 17:18:32 EDT 2001
3517    Tag: leung-20011012-x86-printflowgraph
3518    Description:
3519    
3520    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
3521    
3522    ----------------------------------------------------------------------
3523    Name: Lal George
3524    Date: 2001/10/11 23:51:34 EDT
3525    Tag: george-20011011-too-many-instrs
3526    Description:
3527    
3528    The representation of a program point never expected to see more
3529    than 65536 instructions in a basic block!
3530    
3531    ----------------------------------------------------------------------
3532    Name: Lal George
3533    Date: 2001/10/09 09:41:37 EDT
3534    Tag: george-20011008-mlrisc-labels
3535    Description:
3536    
3537    Changed the machine description files to support printing of
3538    local and global labels in assembly code, based on host assembler
3539    conventions.
3540    
3541    ----------------------------------------------------------------------
3542    Name: Matthias Blume
3543    Date: 2001/09/25 15:25:00 EDT
3544    Tag: blume-20010925-exninfo
3545    Description:
3546    
3547    I provided a non-hook implementation of exnName (at the toplevel) and
3548    made the "dummy" implementation of exnMessage (at the toplevel) more
3549    useful: if nothing gets "hooked in", then at least you are going to
3550    see the exception name and a message indicating why you don't see more.
3551    
3552    [For the time being, programs that need exnMessage and want to use
3553    ml-build should either use General.exnMessage (strongly recommended) or
3554    refer to structure General at some other point so that CM sees a
3555    static dependency.]
3556    
3557    [Similar remarks go for "print" and "use":  If you want to use their
3558    functionality in stand-alone programs generated by ml-build, then use
3559    TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
3560    
3561    ----------------------------------------------------------------------
3562    Name: Matthias Blume
3563    Date: 2001/09/20 17:28:00 EDT
3564    Tag: blume-20010920-slowfp
3565    Description:
3566    
3567    Allen says that x86-fast-fp is not safe yet, so I turned it off again...
3568    
3569    ----------------------------------------------------------------------
3570    Name: Matthias Blume
3571    Date: 2001/09/20 17:20:00 EDT
3572    Tag: blume-20010920-canonicalpaths
3573    Description:
3574    
3575    0. Updated the BOOT file (something that I forgot to do earlier).
3576    
3577    1. Small internal change to CM so that it avoids "/../" in filenames
3578       as much as possible (but only where it is safe).
3579    
3580    2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
3581       that arguments that contain delimiters are passed through correctly.
3582       This change also means that all "special" arguments of the form
3583       @SMLxxx... must come first.
3584    
3585    3. Changed install script to put relative anchor names for tool commands
3586       into pathconfig.
3587    
3588    ----------------------------------------------------------------------
3589    Name: Matthias Blume
3590    Date: 2001/09/18 15:35:00 EDT
3591    Tag: blume-20010918-readme11036
3592    Description:
3593    
3594    Added README files.
3595    
3596    ----------------------------------------------------------------------
3597    Name: Matthias Blume
3598    Date: 2001/09/18 11:45:00 EDT
3599    Tag: Release_110_36 (retag)
3600    Description:
3601    
3602    Fixed mistake in config/preloads. Retagged as 110.36.
3603    
3604    ----------------------------------------------------------------------
3605    Name: Matthias Blume
3606    Date: 2001/09/18 09:40:00 EDT
3607    Tag: Release_110_36_orig (tag changed)
3608    Description:
3609    
3610    New version (110.36).  New bootfiles.
3611    
3612    ----------------------------------------------------------------------
3613    Name: Matthias Blume
3614    Date: 2001/09/14 16:15:00 EDT
3615    Tag: blume-20010914-x86fastfp
3616    Description:
3617    
3618    John committed some changes that Allen made, in particular a (hopefully)
3619    correctly working version of the x86-fp module.
3620    
3621    I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
3622    flag to "true".  Everything seems to compile to a fixpoint ok, and
3623    "mandelbrot" speeds up by about 15%.
3624    
3625    ----------------------------------------------------------------------
3626    Name: Matthias Blume
3627    Date: 2001/09/13 11:20:00 EDT
3628    Tag: blume-20010913-minimal
3629    Description:
3630    
3631    1. Stefan Monnier's patch to fix a miscompilation problem that
3632       was brought to light by John Reppy's work on Moby.
3633    
3634    2. Implemented a minimal "structure Compiler" that contains just
3635       "version" and "architecture".  The minimal version will be
3636       available when the full version is not.  This is for backward-
3637       compatibility with code that wants to test Compiler.version.
3638    
3639    ----------------------------------------------------------------------
3640    Name: Matthias Blume
3641    Date: 2001/08/28 14:03:00 EDT
3642    Tag: blume-20010828-ml-lex
3643    Description:
3644    
3645    Fix for bug 1581, received from Neophytos Michael.
3646    
3647    ----------------------------------------------------------------------
3648    Name: Matthias Blume
3649    Date: 2001/08/27 11:20:00 EDT
3650    Tag: blume-20010827-readme11035
3651    Description:
3652    
3653    Fleshed out the README file for 110.35.
3654    
3655    ----------------------------------------------------------------------
3656    Name: Matthias Blume
3657    Date: 2001/08/24 17:10:00 EDT
3658    Tag: Release_110_35
3659    Description:
3660    
3661    New version number (110.35).  New bootfiles.
3662    
3663    ----------------------------------------------------------------------
3664    Name: Lal George
3665    Date: 2001/08/24 13:47:18 EDT 2001
3666    Tag: george-20010824-MLRISC-graphs
3667    Description:
3668    
3669     removed clusters from MLRISC completely and replaced with graphs.
3670    
3671    ----------------------------------------------------------------------
3672    Name: Matthias Blume
3673    Date: 2001/08/23 17:50:00 EDT
3674    Tag: blume-20010823-toplevel
3675    Description:
3676    
3677    - some reorganization of the code that implements various kinds of
3678      environments in the compiler (static, dynamic, symbolic, combined)
3679    - re-implemented the EnvRef module so that evalStream works properly
3680      (if the stream contains references to "use", "CM.make", etc.)
3681    - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
3682    
3683    ----------------------------------------------------------------------
3684    Name: Matthias Blume
3685    Date: 2001/08/20 15:50 EDT
3686    Tag: blume20010820-slipup
3687    Description:
3688    
3689    I forgot to commit a few files.  Here they are...
3690    
3691    ----------------------------------------------------------------------
3692    Name: Matthias Blume
3693    Date: 2001/08/20 15:35:00 EDT
3694    Tag: blume-20010820-debugprof
3695    Description:
3696    
3697    !!!! NEW BOOTFILES !!!!
3698    
3699    This is another round of reorganizing the compiler sources.  This
3700    time the main goal was to factor out all the "instrumentation"
3701    passes (for profiling and backtracing) into their own library.
3702    The difficulty was to do it in such a way that it does not depend
3703    on elaborate.cm but only on elabdata.cm.
3704    
3705    Therefore there have been further changes to both elaborate.cm and
3706    elabdata.cm -- more "generic" things have been moved from the former
3707    to the latter.  As a result, I was forced to split the assignment
3708    of numbers indicating "primtyc"s into two portions: SML-generic and
3709    SML/NJ-specific.  Since it would have been awkward to maintain,
3710    I bit the bullet and actually _changed_ the mapping between these
3711    numbers and primtycs.  The bottom line of this is that you need
3712    a new set of bin- and bootfiles.
3713    
3714    I have built new bootfiles for all architectures, so doing a fresh
3715    checkout and config/install.sh should be all you need.
3716    
3717    The newly created library's name is
3718    
3719        $smlnj/viscomp/debugprof.cm
3720    
3721    and its sources live under
3722    
3723        src/compiler/DebugProf
3724    
3725    ----------------------------------------------------------------------
3726    Name: Matthias Blume
3727    Date: 2001/08/15 17:15:00 EDT
3728    Tag: blume-20010815-compreorg
3729    Description:
3730    
3731    This is a first cut at reorganizing the CM libraries that make up the
3732    core of the compiler.  The idea is to separate out pieces that could
3733    be used independently by tools, e.g., the parser, the typechecker, etc.
3734    
3735    The current status is a step in this direction, but it is not quite
3736    satisfactory yet.  Expect more changes in the future.
3737    
3738    Here is the current (new) organization...
3739    
3740        What used to be $smlnj/viscomp/core.cm is now divided into
3741        six CM libraries:
3742    
3743             $smlnj/viscomp/basics.cm
3744                           /parser.cm
3745                           /elabdata.cm
3746                           /elaborate.cm
3747                           /execute.cm
3748                           /core.cm
3749    
3750        The CM files for these libraries live under src/system/smlnj/viscomp.
3751        All these libraries are proxy libraries that contain precisely
3752        one CM library component.  Here are the locations of the components
3753        (all within the src/compiler tree):
3754    
3755             Basics/basics.cm
3756             Parse/parser.cm
3757             ElabData/elabdata.cm
3758             Elaborator/elaborate.cm
3759             Execution/execute.cm
3760             core.cm
3761    
3762         [This organization is the same that has been used already
3763         for a while for the architecture-specific parts of the visible
3764         compiler and for the old version of core.cm.]
3765    
3766         As you will notice, many source files have been moved from their
3767         respective original locations to a new home in one of the above
3768         subtrees.
3769    
3770         The division of labor between the new libraries is the following:
3771    
3772             basics.cm:
3773                - Simple, basic definitions that pertain to many (or all) of
3774                  the other libraries.
3775             parser.cm:
3776                - The SML parser, producing output of type Ast.dec.
3777                - The type family for Ast is also defined and exported here.
3778             elabdata.cm:
3779                - The datatypes that describe input and output of the elaborator.
3780                  This includes types, absyn, and static environments.
3781             elaborator.cm:
3782                - The SML/NJ type checker and elaborator.
3783                  This maps an Ast.dec (with a given static environment) to
3784                  an Absyn.dec (with a new static environment).
3785                - This libraries implements certain modules that used to be
3786                  structures as functors (to remove dependencies on FLINT).
3787             execute.cm:
3788                - Everything having to do with executing binary code objects.
3789                - Dynamic environments.
3790             core.cm:
3791                - SML/NJ-specific instantiations of the elaborator and MLRISC.
3792                - Top-level modules.
3793                - FLINT (this should eventually become its own library)
3794    
3795    Notes:
3796    
3797    I am not 100% happy with the way I separated the elaborator (and its
3798    data structures) from FLINT.  Two instances of the same problem:
3799    
3800        1. Data structures contain certain fields that carry FLINT-specific
3801           information.  I hacked around this using exn and the property list
3802           module from smlnj-lib.  But the fact that there are middle-end
3803           specific fields around at all is a bit annoying.
3804    
3805        2. The elaborator calculates certain FLINT-related information.  I tried
3806           to make this as abstract as I could using functorization, but, again,
3807           the fact that the elaborator has to perform calculations on behalf
3808           of the middle-end at all is not nice.
3809    
3810        3. Having to used exn and property lists is unfortunate because it
3811           weakens type checking.  The other alternative (parameterizing
3812           nearly *everything*) is not appealing, though.
3813    
3814    I removed the "rebinding =" warning hack because due to the new organization
3815    it was awkward to maintain it.  As a result, the compiler now issues some of
3816    these warnings when compiling init.cmi during bootstrap compilation. On
3817    the plus side, you also get a warning when you do, for example:
3818       val op = = Int32.+
3819    which was not the case up to now.
3820    
3821    I placed "assign" and "deref" into the _Core structure so that the
3822    code that deals with the "lazy" keyword can find them there.  This
3823    removes the need for having access to the primitive environment
3824    during elaboration.
3825    
3826    ----------------------------------------------------------------------
3827    Name: Matthias Blume
3828    Date: 2001/08/13
3829    Tag: blume-20010813-closures
3830    Description:
3831    
3832    This fix was sent to us by Zhong Shao.  It is supposed to improve the
3833    performance of certain loops by avoiding needless closure allocation.
3834    
3835    ----------------------------------------------------------------------
3836    Name: Lal George
3837    Date: 2001/07/31 10:03:23 EDT 2001
3838    Tag: george-20010731-x86-fmalloc
3839    Description: Fixed bug in x86 calls
3840    
3841        There was a bug where call instructions would mysteriously
3842        vanish. The call instruction had to be one that returned
3843        a floating point value.
3844    
3845    ----------------------------------------------------------------------
3846    N