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 710, Tue Oct 17 08:21:09 2000 UTC revision 1619, Wed Sep 8 18:19:45 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: 2004/09/08 13:20:00 CDT
18    Tag: blume-20040908-heap-magic
19    Description:
20    
21    Bumped the heap macig number to 0x09082004 to account for the changed
22    layout of the ML frame under MacOS X.
23    
24    ----------------------------------------------------------------------
25    Name: Allen Leung (leunga (at) reservoir (dot) com)
26    Date: 2004/09/03 11:26:00 EST
27    Tag: leunga-20040903-cygwin-install
28    Description:
29    
30       Added a patch to _arch-n-opsys to enable the Cygwin runtime.
31    The Cygwin runtime is turned on by setting the environment
32    variable SMLNJ_CYGWIN_RUNTIME to 1.
33    
34    ----------------------------------------------------------------------
35    Name: Matthias Blume (blume (at) tti - c (dot) org)
36    Date: 2004/08/31 17:15:00 CDT
37    Tag: blume-20040831-core
38    Description:
39    
40    Added some exports to src/compiler/core.cm upon request by J. Joemann.
41    
42    ----------------------------------------------------------------------
43    Name: Matthias Blume (blume (at) tti - c (dot) org)
44    Date: 2004/08/30 17:55:00 CDT
45    Tag: blume-20040830-installer
46    Description:
47    
48    Upon request by Johannes Joemann:
49    
50    - improved ML code of installer to fall back to coping when renaming
51      fails (i.e., when source and target are on different file systems);
52      the code compiles but has yet to be tested in anger
53    - removed mlrisc from list of default targets (config/targets)
54    
55    ----------------------------------------------------------------------
56    Name: Matthias Blume (blume (at) tti - c (dot) org)
57    Date: 2004/08/27 17:20:00 CDT
58    Tag: blume-20040827-ptreql
59    Description:
60    
61    Added ptreql primop to structure InlineT (upon request from Larry
62    Paulson).
63    
64    ----------------------------------------------------------------------
65    Name: Allen Leung (leunga (at) reservoir (dot) com)
66    Date: 2004/08/15 21:21:00 EST
67    Tag: leunga-110_48-udgraph
68    Description:
69    
70    Another bug fix from Carl Hauser:
71    
72    diff /net/niflab/smlnj48/src/MLRISC/graphs/udgraph.sml udgraph.sml
73    > 48c48
74    > <              | rmv((e as (k,_))::es,L) = rmv(es,if k = i then es else
75    > e::L)
76    > ---
77    > >              | rmv((e as (k,_))::es,L) = rmv(es,if k = i then L else e::L)
78    > Without this, any deletion of an edge in an undirected graph does severe
79    > violence to the graph.
80    
81    ----------------------------------------------------------------------
82    Name: Allen Leung (leunga (at) reservoir (dot) com)
83    Date: 2004/08/10 23:35:00 EST
84    Tag: leunga-110_48-ppc
85    Description:
86    
87       The IBM/MacOS syntax switch on PPC was incorrectly swapped.
88    
89    ----------------------------------------------------------------------
90    Name: Matthias Blume (blume (at) tti - c (dot) org)
91    Date: 2004/08/10 12:00:00 CDT
92    Tag: Release_110_48
93    Description:
94    
95    New working version (110.48).  NEW BOOTFILES!
96    
97    ----------------------------------------------------------------------
98    Name: Allen Leung (leunga (at) reservoir (dot) com)
99    Date: 2004/08/09 12:21:00 EST
100    Tag: leunga-110_47-dijsktra
101    Description:
102    
103    Bug fix from Carl Hauser:
104    
105    single_source_shortest_paths in dijkstra.sml was observed to get wrong
106    answers (by comparing to single_source_shortest_paths in bellman-ford.sml).
107    
108    The problem is that following the expression A.update(dist,s,Num.zero)
109    it is necessary to update the priority queue using Q.decreaseWeight(Q,s).
110    
111    ----------------------------------------------------------------------
112    Name: Matthias Blume (blume (at) tti - c (dot) org)
113    Date: 2004/08/06 18:10:00 CDT
114    Tag: blume-20040806-cmdline
115    Description:
116    
117    Fiddled with handling of command-line options:
118    
119      * sml now quits after processing the command line
120        if -H, -S, -h<n>, or -s<n> appears as the last
121        command-line argument
122      * a new option -q terminates the session when encountered on
123        the command line; subsequent arguments will be ignored
124      * bug fixes: short (erroneous) arguments are no longer ignored
125        completely
126    
127    ----------------------------------------------------------------------
128    Name: Allen Leung (leunga (at) reservoir (dot) com)
129    Date: 2004/08/04 18:17:00 EST
130    Tag: leunga-110_47-ppc-ibm-asm
131    Description:
132    
133      - Added minimal IBM assembly syntax support for PowerPC.
134    
135      - Cygwin: manually changed the file cygwin.def.  Some exported symbols have
136        been altered in the runtime.  We need an automatic way to keep the file
137        in sync.
138    
139    ----------------------------------------------------------------------
140    Name: Matthias Blume (blume (at) tti - c (dot) org)
141    Date: 2004/08/04 14:00:00 CDT
142    Tag: Release_110_47
143    Description:
144    
145    New working version (110.47).  NEW BOOTFILES!
146    
147    ----------------------------------------------------------------------
148    Name: Matthias Blume (blume (at) tti - c (dot) org)
149    Date: 2004/08/03 14:25:00 CDT
150    Tag: blume-20040803-callingconv
151    Description:
152    
153    Added low-level support for choosing C calling conventions by
154    twiddling the type of rawccall.  (See
155    src/compiler/Semant/types/cproto.sml for details.)
156    
157    ----------------------------------------------------------------------
158    Name: Matthias Blume (blume (at) tti - c (dot) org)
159    Date: 2004/08/02 15:55:00 CDT
160    Tag: blume-20040802-backout
161    Description:
162    
163    Backed out of change to win32-filesys.c.  The earlier patch to
164    get_file_time caused CM to produce files with the wrong time stamp.
165    
166    ----------------------------------------------------------------------
167    Name: Matthias Blume (blume (at) tti - c (dot) org)
168    Date: 2004/08/02 14:45:00 CDT
169    Tag: blume-20040802-nlffi-win32
170    Description:
171    
172    Added NLFFI support for Win32, adapted from a patch provided by David
173    Hansel.  This is currently completely untested.  Also, the issue
174    concerning stdcall vs. ccall is still unresolved.
175    
176    ----------------------------------------------------------------------
177    Name: Matthias Blume (blume (at) tti - c (dot) org)
178    Date: 2004/07/30 17:55:00 CDT
179    Tag: blume-20040730-various
180    Description:
181    
182    Gearing up towards 110.47...
183    
184    - various minor bugfixes to ml-nlffigen
185    - a beginning of a manual for nlffi
186    
187    - eliminated 'export name=value' in config/install.sh as this does
188      not work with certain versions of /bin/sh
189      (Thanks to David King at Motorola for catching this.)
190    
191    - several bugfixes provided or suggested by David Hansel at Reactive Systems:
192      - added a test for tm==NULL to gmtime.c and localtime.c
193      - applied patch for incorrect GetFileTime under win32
194      - toSeconds -> toMilliseconds in Win32/win32-process.sml
195    
196    ----------------------------------------------------------------------
197    Name: Matthias Blume (blume (at) tti - c (dot) org)
198    Date: 2004/07/21 18:20:00 CDT
199    Tag: blume-20040721-nlffigen
200    Description:
201    
202    - Fixed minor issue in ml-nlffigen: Now generate structure T_foo
203      for a typedef to an incomplete type, but leave out the "typ" member.
204      (This is just for consistency.)
205    - Started to produce what is supposed to become better (i.e., comprehensive)
206      documentation of what ml-nlffigen does and produces.
207    
208    ----------------------------------------------------------------------
209    Name: Matthias Blume (blume (at) tti - c (dot) org)
210    Date: 2004/07/14 16:25:00 CDT
211    Tag: blume-20040714-union
212    Description:
213    
214    Added C_UNION to c-calls/c-types.sml and updated the machinery
215    (ml-nlffigen, cproto.sml) that conveys C function interface
216    information to the code generator.
217    
218    However, the actual architecture-specific implementation of function
219    arguments and results that are C unions is still not implemented.
220    
221    ----------------------------------------------------------------------
222    Name: Allen Leung (leunga (at) reservoir (dot) com)
223    Date: 2004/07/14 14:38:00 EST
224    Tag: leunga-110_46_1-ppc-lwzu
225    Description:
226    
227        Added these instructions to the PowerPC architecture:
228        LBZU(X), LHZU(X), LWZU(X),
229        STWU(X), STFDU, STFSU
230    
231        etc...
232    
233        Note: I haven't added their instruction encoding into the description.
234    
235    ----------------------------------------------------------------------
236    Name: Allen Leung (leunga (at) reservoir (dot) com)
237    Date: 2004/07/13 15:04:00 EST
238    Tag: leunga-110_46_1-ppc-lwarx
239    Description:
240    
241       Added the two instructions LWARX and STWCX to the PowerPC
242    instruction set.
243    
244       A (untested) rewrite of loop-structure.sml.  The old version
245    is completely broken.
246    
247    ----------------------------------------------------------------------
248    Name: Matthias Blume (blume (at) tti - c (dot) org)
249    Date: 2004/07/13 13:50:00 CDT
250    Tag: blume-20040713-nlffi
251    Description:
252    
253    - use paramAlloc to report c-calls with too many arguments
254      (for PPC version where parameter area is pre-allocated)
255    - added ccall_maxargspace to machspec (to implement the above)
256    - made "make" commend in CM's "make" tool configurable
257    - added option (default: on) for passing the name of the SML/NJ's "bin"
258      directory to "make"; the call looks like this:
259    
260         make <options> SMLNJ_BINDIR=<dir> <target>
261    
262      This can be used by the Makefile to, e.g., pick the "right" version
263      of ml-nlffigen.
264    - minor code tweaks
265    
266    ----------------------------------------------------------------------
267    Name: Matthias Blume (blume (at) tti - c (dot) org)
268    Date: 2004/07/12 22:50:00 CDT
269    Tag: blume-110_46_1-macosx-nlffi
270    Description:
271    
272    NLFFI under Mac OS X now working (sort of).  This is largely untested,
273    though.
274    
275    Note:  1. You have to make a new, clean build of the runtime system.
276           2. There are new BOOTFILES, you have to use them!
277              (Doing the bootstrap process yourself would be *very* painful!
278               If you absolutely have to do it, build the system under
279               a different architecture and then cross-compile.)
280    
281    Version bumped to 110.46.1 to account for runtime data format changes.
282    
283    ----------------------------------------------------------------------
284    Name: Matthias Blume (blume (at) tti - c (dot) org)
285    Date: 2004/06/18 14:30:00 CDT
286    Tag: blume-20040618-unix
287    Description:
288    
289    Changed the implementation of structure Unix so that the same stream
290    is returned every time one of the {text,bin}{In,Out}streamOf functions
291    is invoked on the same proc.  This is not what the spec currently
292    says -- although IMO it arguably should.  (See discussion below.)
293    
294    ----------------------------------------------------------------------
295    Name: Matthias Blume (blume (at) tti - c (dot) org)
296    Date: 2004/06/17 18:15:00 CDT
297    Tag: Release_110_46
298    Description:
299    
300    New working version (110.46).  NEW BOOTFILES!
301    
302    ----------------------------------------------------------------------
303    Name: Matthias Blume (blume (at) tti - c (dot) org)
304    Date: 2004/06/17 17:20:00 CDT
305    Tag: blume-20040617-timer-unix
306    Description:
307    
308    Changed the interface of structures Timer and Unix to match the most
309    recent Basis spec.
310    
311    In the case of Unix there still seems to be an open/weird issue:
312    
313       The {text,bin}{In,Out}streamOf functions are supposed to create
314       fresh streams whenever they are called -- as opposed to have them
315       return the same stream every time.  This design is supposed to
316       prevent space leaks caused by proc values hanging on to streams.
317    
318       The reap function, on the other hand, is supposed to close the
319       streams.  This cannot be done without having a handle on the
320       stream in proc after all...
321    
322    I took the liberty to implement the following stopgap solution:
323    
324      The proc value hangs on to the most recently created stream(s).
325      Reap closes those.  If either or both of the two streams hadn't
326      been created at all yet, then reap will close the corresponding
327      file descriptors directly.
328    
329    PS: I don't understand the original space leak argument anymore.  If
330    a proc hangs on to the imperative stream, then I/O operations on those
331    will advance the state of the cached stream and avoid the space leak.
332    
333    ----------------------------------------------------------------------
334    Name: Matthias Blume (blume (at) tti - c (dot) org)
335    Date: 2004/05/28 16:45:00 CDT
336    Tag: blume-20040528-basis
337    Description:
338    
339    Added signature PACK_REAL and exported functor PrimIO.
340    
341    ----------------------------------------------------------------------
342    Name: Matthias Blume (blume (at) tti - c (dot) org)
343    Date: 2004/05/25 16:00:00 CDT
344    Tag: blume-20040525-group-owner
345    Description:
346    
347    CM now ignores (but still accepts) the "owner" information in group
348    descriptions.  The owner of a group is its next enclosing
349    library. Each group must have a unique owner.  (There is a virtual
350    "toplevel" library that owns groups which are not nested within a real
351    library.)  Previously, each group had to explicitly declare its owner,
352    and CM would check that such a declaration is correct.  The new scheme
353    is to have CM check that for each group there is precisely one owning
354    library.
355    
356    The advantage of the new scheme is that the programmer no longer needs
357    to maintain the somewhat annoying owner information.  The downside is
358    that CM cannot enforce the ownership rule across multiple runs of
359    CM.make.  Fortunately, enclosing the same group in two different
360    libraries A and B which are not part of the same program does not
361    cause real problems.
362    
363    ----------------------------------------------------------------------
364    Name: Matthias Blume (blume (at) tti - c (dot) org)
365    Date: 2004/05/20 16:00:00 CDT
366    Tag: blume-20040520-win32
367    Description:
368    
369    Made the win32 version work again.  (Strangely, a misplaced comma had
370    slipped into win32-process.c which prevented the runtime from being
371    compiled correctly.)
372    
373    Also, included a minor addition to ml-build.bat analogous to what was
374    done in blume-20040519-ml-build.
375    
376    ----------------------------------------------------------------------
377    Name: Matthias Blume (blume (at) tti - c (dot) org)
378    Date: 2004/05/19 22:10:00 CDT
379    Tag: blume-20040519-ml-build
380    Description:
381    
382    Arranged for ml-build to clean up after itself a little bit better.
383    The script generates a temporary SML source file and compiles it using
384    CM, so CM generates metadata (GUID, SKEL, objectfile) for it.  It now
385    gets rid of those at the end, so they don't accumulate under .cm.
386    
387    This required a minor change to install.sh because the name of the
388    metadata directory (default: .cm) is actually configurable at
389    installation time.
390    
391    ----------------------------------------------------------------------
392    Name: Matthias Blume (blume (at) tti - c (dot) org)
393    Date: 2004/05/18 15:50:00 CDT
394    Tag: blume-20040518-mkreader
395    Description:
396    
397    Added Posix.IO.mk{Bin,Text}{Reader,Writer} by lifting their respective
398    implementations from internal modules PosixBinPrimIO and PosixTextPrimIO.
399    
400    ----------------------------------------------------------------------
401    Name: Matthias Blume (blume (at) tti - c (dot) org)
402    Date: 2004/05/11 14:35:00 CDT
403    Tag: blume-20040511-win32sock
404    Description:
405    
406    Added previously missing support for many socket-related functions
407    under win32.  Thanks to David Hansel <hansel@reactive-systems.com>
408    for the voluminous patch!
409    
410    (I have not tested this patch under win32 yet.)
411    
412    Here is David's e-mail:
413    
414    Hi,
415    
416    Attached to this email you find a diff against sml/nj 110.45
417    that will enable socket support under Windows.
418    
419    To apply the patch (using unix or cygwin)
420    1) gunzip runtime.diff.gz
421    2) "cd" into "src/runtime" in the source tree of a fresh
422       110.45 installation.
423    3) patch -p 1 < [your/path/to]runtime.diff
424    
425    The code compiles fine but has NOT yet been extensively tested.
426    I only ran a few tests for basic socket client functionality
427    (which worked fine).  Especially the functions that use ioctl
428    are not tested at all and might not work (see below).
429    
430    I implemented this since we want to move to a newer version of sml/nj
431    but need socket support in order to use it.  This is the first time I
432    even had a look at the sml/nj source,  so please review my changes
433    before making this part of the distribution!  Here are a few issues
434    that I think might be better for someone to solve who is more
435    familiar with the sml/nj source (and socket programming):
436    
437    - getnetbyaddr.c and getnetbyname.c will raise a "not implemented"
438      exception since I could not figure out what the windows equivalent
439      of these functions is
440    
441    - In sockets-osdep.h there are a some #include statements that are
442      only used in a few files that include sockets-osdep.h
443    
444    - In smlnj-sock-lib.c, function init_fn() calls WSAStartup() but
445      does not process its return value since I don't know how to
446      report an error upwards.
447    
448    - It would probably be good to have a call to WSACleanup() when
449      the library is unloaded (if there is such a possibility).
450      Otherwise I think Windows will take care of this automatically
451      when the process finishes.
452    
453    - I used ioctlsocket() as a replacement for ioctl() but I have
454      no idea if that is actually the proper replacement on Windows.
455    
456    - All these issues are marked in the code by "FIXME" comments.
457    
458    We use sml/nj extensively in our products and are quite happy
459    with it.  I hope this contribution will help you.
460    
461    Keep up the good work!
462    
463    David
464    
465    ----------------------------------------------------------------------
466    Name: Matthias Blume (blume (at) tti - c (dot) org)
467    Date: 2004/05/11 14:20:00 CDT
468    Tag: blume-20040511-installml
469    Description:
470    
471    Fixed two bugs in installml script.  (Thanks to Vesa A. Norrman for
472    the patch.)
473    
474    ----------------------------------------------------------------------
475    Name: Matthias Blume (blume (at) tti - c (dot) org)
476    Date: 2004/05/11 14:05:00 CDT
477    Tag: blume-20040511-nlffi-netbsd
478    Description:
479    
480    Added support for nlffi under netbsd.  (Thanks to Vesa A. Norrman for
481    the patch.)
482    
483    ----------------------------------------------------------------------
484    Name: Matthias Blume (blume (at) tti - c (dot) org)
485    Date: 2004/05/11 12:05:00 CDT
486    Tag: blume-20040511-exports
487    Description:
488    
489    As per request by Adam Chlipala <adam@hcoop.net>, extended various
490    export lists in compiler-related .cm-files.
491    
492    ----------------------------------------------------------------------
493    Name: Matthias Blume (blume (at) tti - c (dot) org)
494    Date: 2004/05/11 11:35:00 CDT
495    Tag: blume-20040511-allsource
496    Description:
497    
498    The installer now honors the "src-smlnj" target again, although its meaning
499    has changed from "all sources required for the compiler" to "all sources
500    the installer knows about".  In other words, if you enable "src-smlnj"
501    in the "targets" file, then the installer will pull in sources for
502    everything.  (Notice that this refers to source code only.  Compiled
503    code is still only installed for modules that were requested explicitly
504    or which are required for other modules that were requested explicitly.)
505    
506    ----------------------------------------------------------------------
507    Name: Matthias Blume (blume (at) tti - c (dot) org)
508    Date: 2004/04/23 17:40:00 CDT
509    Tag: blume-20040423-ieee-scan
510    Description:
511    
512    Fixed IEEEReal.scan (and .fromString) so that if there is an overflow
513    in the exponent calculation we get INF or ZERO (depending on the mantissa
514    and the sign of the exponent).
515    
516    ----------------------------------------------------------------------
517    Name: Matthias Blume (blume (at) tti - c (dot) org)
518    Date: 2004/04/23 10:40:00 CDT
519    Tag: blume-20040423-ml-build
520    Description:
521    
522    The ml-build script now terminates with a non-0 status when something
523    goes wrong.
524    
525    ----------------------------------------------------------------------
526    Name: Matthias Blume (blume (at) tti - c (dot) org)
527    Date: 2004/04/22 16:35:00 CDT
528    Tag: blume-20040422-Option
529    Description:
530    
531    Made exception Option to be the same as exception Option.Option
532    (as it should be).
533    
534    ----------------------------------------------------------------------
535    Name: Allen Leung (leunga (at) reservoir (dot) com)
536    Date: 2004/03/19 14:40:00 EST
537    Tag: leunga-20040319-cygwin-nlffi
538    Description:
539    
540    Fixed the runtime so that ml-nlffi-lib runs on the cygwin version
541    of SML/NJ.  The problem is that
542    
543       lib = dlopen(NULL, ...)
544       f   = dlsym(lib, "malloc");
545    
546    does not work on Windows unless we explicitly export symbols
547    such as 'malloc' during linking.  We fixed this by explicitly
548    exporting the required symbols with the magic gcc incantation:
549    
550        -Wl,--export-all cygwin.def
551    
552    where cygwin.def is a file containing all the symbols that
553    we wish to export.
554    
555    I suspect this is a Windows problem and we'll have to
556    do the same (somehow with windows compilers) when
557    we build the native win32 version with the system
558    calls LoadLibrary/GetProcAddress.
559    
560    ----------------------------------------------------------------------
561    Name: Matthias Blume (blume (at) tti - c (dot) org)
562    Date: 2004/03/04 16:35:00 CST
563    Tag: blume-20040304-intinf-fmt
564    Description:
565    
566    Fixed problem with IntInf.fmt (sign would show up on the right instead
567    of on the left for BIN, OCT, and HEX).
568    
569    ----------------------------------------------------------------------
570    Name: Matthias Blume (blume (at) tti - c (dot) org)
571    Date: 2004/03/04 11:25:00 CST
572    Tag: blume-20040304-symlinks
573    Description:
574    
575    Fixed problem with installer script (unix only) where bin/ml-yacc and
576    friends pointed (via symlinks) to absolute locations instead of just
577    .run-sml.  This was reported by Vesa A Norrman.
578    
579    ----------------------------------------------------------------------
580    Name: Matthias Blume (blume (at) tti - c (dot) org)
581    Date: 2004/02/13 14:50:00 CST
582    Tag: Release_110_45
583    Description:
584    
585    New working version (110.45). New bootfiles.
586    
587    ----------------------------------------------------------------------
588    Name: Matthias Blume (blume (at) tti - c (dot) org)
589    Date: 2004/01/26 15:15:15 CST
590    Tag: blume-20040126-toplevel
591    Description:
592    
593    Improved handling of exceptions at the interactive toplevel.
594    
595    ----------------------------------------------------------------------
596    Name: Matthias Blume (blume (at) tti - c (dot) org)
597    Date: 2004/01/26 11:25:00
598    Tag: blume-20040126-app
599    Description:
600    
601    Type of top-level "app" corrected.
602    Added code for setting vp_limitPtrMask to Win32-specific runtime.
603    
604    ----------------------------------------------------------------------
605    Name: Matthias Blume (blume (at) tti - c (dot) org)
606    Date: 2003/11/18 17:10 CST
607    Tag: blume-20031118-basis-fiddle
608    Description:
609    
610    - changed Timer interface to what might become the spec
611    - POSIX_FLAGS -> BIT_FLAGS according to spec
612    - some other minor discrepancies wrt. spec eliminated
613    
614    ----------------------------------------------------------------------
615    Name: Matthias Blume (blume (at) tti - c (dot) org)
616    Date: 2003/11/06 12:00:00 CST
617    Tag: Release_110_44
618    Description:
619    
620    New working version (110.44). New bootfiles.
621    
622    ----------------------------------------------------------------------
623    Name: Matthias Blume (blume (at) tti - c (dot) org)
624    Date: 2003/11/04 11:50:00 CST
625    Tag: blume-20031104-move-libraries
626    Description:
627    
628    Eliminated the "dont_move_libraries" directive in config/targets.
629    (The mechanism was broken and could not be fixed easily.  Moreover,
630    there does not seem to be any reason not to move all libraries into
631    lib during installation.  I originally implemented this directive as a
632    backward-compatibility feature when I first introduced the new CM.
633    Now that things have been stable for a long time and going back to the
634    old CM is not an option, there is no reason to keep it around.)
635    
636    ----------------------------------------------------------------------
637    Name: Matthias Blume (blume (at) tti - c (dot) org)
638    Date: 2003/11/03 16:00:00 CST
639    Tag: blume-20031103-installdir
640    Description:
641    
642    Made installer honor INSTALLDIR variable again.  (Thanks to Chris
643    Richards for pointing out the problem and providing the solution.)
644    
645    ----------------------------------------------------------------------
646    Name: Matthias Blume (blume (at) tti - c (dot) org)
647    Date: 2003/10/01 17:05:00 CDT
648    Tag: blume-20031001-lal-mlrisc
649    Description:
650    
651    MLRISC bug fix from Lal.
652    
653    ----------------------------------------------------------------------
654    Name: Matthias Blume (blume (at) tti - c (dot) org)
655    Date: 2003/09/30 16:10:00 CDT
656    Tag: blume-20030930-primio-bat
657    Description:
658    
659    1. Added openVector, nullRd, and nullWr to PRIM_IO.
660    2. Improved .bat files (for Win32 port) to make things work under Win95.
661       (thanks to Aaron S. Hawley for this one)
662    
663    ----------------------------------------------------------------------
664    Name: Matthias Blume (blume (at) tti - c (dot) org)
665    Date: 2003/09/26 16:05:00 CDT
666    Tag: blume-20030926-wrappriv
667    Description:
668    
669    Added missing wrapper for privilege "primitive" in $smlnj/viscomp/core.cm.
670    
671    ----------------------------------------------------------------------
672    Name: Matthias Blume (blume (at) tti - c (dot) org)
673    Date: 2003/09/26 15:00:00 CDT
674    Tag: blume-20030926-110_43_3
675    Description:
676    
677    - additional cleanup
678    - version number bump, NEW BOOTFILES
679    
680    ----------------------------------------------------------------------
681    Name: Matthias Blume (blume (at) tti - c (dot) org)
682    Date: 2003/09/26 12:00:00 CDT
683    Tag: blume-20030926-ppautoload
684    Description:
685    
686    I modified the read-eval-print loop so that the autoloader gets
687    invoked whenever the prettyprinter tries to look up a symbol that
688    is not currently defined in the toplevel environment but which
689    appears in CM's autoload registry.  As a result, we see far fewer of
690    those ?.Foo.Bar.xxx names in the prettyprinter's output.
691    
692    In addition to this I tried to clean up some pieces of the Basis
693    implementation (e.g., Socket, Word8Array) in order to prevent other
694    instances of these ?.Foo.Bar.xxx names from being printed.
695    
696    The mechanism that picks names for types still needs some work, though.
697    (Right now it seems that if there is a type A.t which is defined to
698    be B.u, but B is unavailable at toplevel, then A.t gets printed as
699    "?.B.u" although the perhaps more sensible solution would be to use
700    "A.t" in this case.  In other words, the prettyprinter should follow
701    a chain of DEFtycs not farther than there are corresponding toplevel
702    names in the current environment.)
703    
704    ----------------------------------------------------------------------
705    Name: Matthias Blume (blume (at) tti - c (dot) org)
706    Date: 2003/09/24 16:31:00 CDT
707    Tag: blume-20030924-installer
708    Description:
709    
710    Another installer tweak:  All the ML code for the installer is now
711    compiled during CMB.make and put into a little library called
712    $smlnj/installer.cm.  The installation then simply invokes
713    
714       sml -m $smlnj/installer.cm
715    
716    and everything happens automagically.
717    
718    Win32:  ML code senses value of environment variable SMLNJ_HOME.
719    Unix:   ML code senses values of environment variables ROOT, CONFIGDIR,
720            and BINDIR.
721    
722    The new scheme guarantees that the ML code responsible for the installation
723    is in sync with the APIs of the main system.  Also, the installer is
724    somewhat faster because the installer script is precompiled.
725    
726    ----------------------------------------------------------------------
727    Name: Matthias Blume (blume (at) tti - c (dot) org)
728    Date: 2003/09/24 15:35:00 CDT
729    Tag: blume-20030924-synsock
730    Description:
731    
732    Added a signature SYNCHRONOUS_SOCKET to basis.cm.  This is like SOCKET
733    but excludes all non-blocking operations.  Defined SOCKET (in Basis)
734    and CML_SOCKET in terms of SYNCHRONOUS_SOCKET.  Removed superfluous
735    implementations of non-blocking operations from CML's Socket
736    structure.
737    
738    ----------------------------------------------------------------------
739    Name: Matthias Blume (blume (at) tti - c (dot) org)
740    Date: 2003/09/24 15:10:05 CDT
741    Tag: blume-20030924-sockets
742    Description:
743    
744    1. Fixed SOCKET API and implementation to match Basis spec.
745       This required changing the internal representation of sockets to one
746       that remembers (for each socket file descriptor) whether it is currently
747       blocking or non-blocking.  This state is maintained lazily (i.e., a system
748       call is made only if the state actually needs to change).
749    
750    2. OS-specific details of sockets were moved into separate files, thus
751       making it possible to unify the bulk of the socket implementations
752       between Unix and Win32.
753    
754    3. CML's socket API changed accordingly.
755       (Note that we need to remove non-blocking functions from this API
756       since they are redundant in the case of CML!)
757    
758    4. CML's socket implementation now makes use of non-blocking functions
759       provided by Basis, thus removing all OS-dependent code from this part
760       of CML.
761    
762    5. Changed Real64.precision from 52 to 53.  Minor cleanup in Real64 code.
763    
764    ----------------------------------------------------------------------
765    Name: Matthias Blume (blume (at) tti - c (dot) org)
766    Date: 2003/09/22 12:10:00 CDT
767    Tag: blume-20030922-110_43_2
768    Description:
769    
770    Made a new interim version and bootfiles for developer's bootstrapping
771    convenience.
772    
773       110.43.2 -- NEW BOOTFILES
774    
775    ----------------------------------------------------------------------
776    Name: Matthias Blume (blume (at) tti - c (dot) org)
777    Date: 2003/09/19 15:55:00 CDT
778    Tag: blume-20030919-cmdir
779    Description:
780    
781    1. new-install.sh -> install.sh
782    2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
783    3. tweaked installer so that another name instead of .cm can be chosen
784       at install time (by setting the CM_DIR_ARC environment variable
785       during installation); once installation is complete, the name is
786       fixed
787    
788    ----------------------------------------------------------------------
789    Name: Matthias Blume (blume (at) tti - c (dot) org)
790    Date: 2003/09/18 16:00:00 CDT
791    Tag: blume-20030918-110_43_1
792    Description:
793    
794    Made a new interim version and bootfiles for developer's bootstrapping
795    convenience.
796    
797       110.43.1 -- NEW BOOTFILES
798    
799    ----------------------------------------------------------------------
800    Name: Matthias Blume (blume (at) tti - c (dot) org)
801    Date: 2003/09/18 15:20:00 CDT
802    Tag: blume-20030918-misc
803    Description:
804    
805    1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
806       this seems to be controversial at the moment) and used those in
807       Posix.ProcEnv.times.
808    
809    2. Added Time.{from,to}Nanoseconds to Time.
810    
811    3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
812       For example, fromLargeInt never needs to look at more than 3 "big
813       digits" to get its 53 bits of precision.
814    
815    ----------------------------------------------------------------------
816    Name: Matthias Blume (blume (at) tti - c (dot) org)
817    Date: 2003/09/17 16:30:00 CDT
818    Tag: blume-20030917-real32-slices
819    Description:
820    
821    Added an entry to the primitive environment
822    (compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
823    added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
824    
825    Removed some of the "magic" constants in real64.sml and replaced them
826    with code that generates these values from their corresponding
827    integer counterparts.
828    
829    Made all(?) the slice-related changes to the Basis and made everything
830    compile again...
831    
832    ----------------------------------------------------------------------
833    Name: Matthias Blume (blume (at) tti - c (dot) org)
834    Date: 2003/09/15 17:45:00 CDT
835    Tag: blume-20030915-rbase
836    Description:
837    
838    Fixed bug in Real.fromLargeInt.
839    
840    ----------------------------------------------------------------------
841    Name: Matthias Blume (blume (at) tti - c (dot) org)
842    Date: 2003/09/13 18:11:00 CDT
843    Tag: blume-20030913-libinstall
844    Description:
845    
846    Minor bugfix in config/libinstall (set anchor with path to
847    standalone tool after installing it, otherwise libraries that
848    need ml-lex or ml-yacc won't compile the first time the installer
849    runs).
850    
851    ----------------------------------------------------------------------
852    Name: Matthias Blume (blume (at) tti - c (dot) org)
853    Date: 2003/09/12 11:45:00 CDT
854    Tag: blume-20030912-various
855    Description:
856    
857     - fixed bug in Real.toLargeInt
858     - fixed bug in Posix.ProcEnv.times
859     - changed inputLine functions to return an option
860     - minor installer improvements / bugfixes
861     - changed default @SMLalloc parameter for x86/celeron to 64k
862    
863    ----------------------------------------------------------------------
864    Name: Matthias Blume (blume (at) tti - c (dot) org)
865    Date: 2003/09/09 22:00:00 CDT
866    Tag: Release_110_43
867    Description:
868    
869        New working release 110.43.  New bootfiles.
870    
871    ----------------------------------------------------------------------
872    Name: Matthias Blume (blume (at) tti - c (dot) org)
873    Date: 2003/09/09 19:20:00 CDT
874    Tag: blume-20030909-installer
875    Description:
876    
877    Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
878    Modified config/install.bat to take advantage of it.  Also modified
879    config/install.sh (and called it config/new-install.sh) to take advantage
880    of it on Unix systems.  (The SML code is (supposed to be) platform-
881    independent.)
882    
883    The installer can now install everything under Win32
884    as well as under *nix as long as it compiles.
885    
886    Other changes:
887    
888      - made CML compile again under Win32
889      - made eXene compile under Win32 (by providing a fake structure UnixSock
890        and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
891      - fixed a bug in nowhere: it assumed that type OS.Process.status is the
892        same as type int;  under Win32 it isn't
893      - fixed some slice-related problems in the win32-specific parts of CML
894      - added a functor argument "sameVol" to os-path-fn.sml in the Basis
895        (under Win32, the volume name is case-insensitive, and the
896        OS.Path code compares volume names for equality)
897    
898    ----------------------------------------------------------------------
899    Name: Matthias Blume (blume (at) tti - c (dot) org)
900    Date: 2003/09/08 11:55:00 CDT
901    Tag: blume-20030908-fullpath
902    Description:
903    
904    Made Win32 version of OS.FileSys.fullPath return current directory
905    when given an empty string.  This is what the spec says, and incidentally,
906    CM depends on it.  (CM otherwise goes into an infinite loop in certain
907    cases when presented with the name of a non-existing .cm file.)
908    
909    ----------------------------------------------------------------------
910    Name: Matthias Blume (blume (at) tti - c (dot) org)
911    Date: 2003/09/04 16:30:00 CDT
912    Tag: blume-20030905-slices-etc
913    Description:
914    
915      1. Changed interface to vectors and arrays in Basis to match
916         (draft) Basis spec.
917      2. Added signatures and implementations of slices according to
918         Basis spec.
919      3. Edited source code throughout the system to make it compile again
920         under 1. and 2.  (In some cases code had to be added to have it
921         match the new signatures.)
922      4. MLRISC should be backward-compatible: the copies of the originals
923         of files that needed to change under 3. were retained, the .cm files
924         check the compiler version number and use old versions when
925         appropriate.
926      5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
927         dirstream -> string option (in accordance with Basis spec).
928      6. When generating code that counts lines, ml-lex used function
929         CharVector.foldli, taking advantage of its old interface.
930         This has been replaced with the corresponding code from
931         CharVectorSlice.  (html-lex must be re-lexed!)
932      7. BitArray in smlnj-lib/Util has been extended/modified to match the
933         new MONO_ARRAY signature.  (Do we need BitArraySlice?)
934      8. Removed temporary additions (fromInternal, toInternal) from the
935         (now obsolete) IntInf in smlnj-lib/Util.
936      9. Cleaned up structure Byte.
937     10. Added localOffset, scan, and fromString to Date (according to spec).
938         Cleaned/corrected implementation of Date.
939         (Still need to check for correctness; implement better canonicalizeDate.)
940     11. Added "scan" to signature IEEE_REAL.
941     12. Some improvements to IntInf [in particular: efficiency-hack for
942         mod and rem when second operand is 2 (for parity checks).]
943     13. Changed representation of type Time.time, using a single IntInf.int
944         value counting microseconds.  This considerably simplified the
945         implementation of structure Time.  We now support negative time
946         values; scan and fromString handle signs.
947     14. Functor PrimIO now takes two additional arguments (VectorSlice and
948         ArraySlice).
949    
950    ----------------------------------------------------------------------
951    Name: Matthias Blume (blume (at) tti - c (dot) org)
952    Date: 2003/08/28 17:00:00 CDT
953    Tag: blume-20030828-intinf
954    Description:
955    
956    This is a major update which comes with a version number bump
957    (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
958    and an implementation of IntInf in the Basis.
959    
960    There are a fairly large number of related changes and updates throughout
961    the system:
962    
963    Basis:
964            - Implemented IntInf.
965            - Made LargeInt a projection of IntInf (by filtering through INTEGER).
966            - Added some missing Real64 operations, most notably Real.toLargeInt.
967            - Added FixedInt as a synonym for Int32.
968    
969    compiler:
970            * Added support for a built-in intinf type.
971              - literals
972              - pattern matching
973              - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
974              - overloading on literals and operations
975    
976            This required adding a primitive type intinf, some additional
977            primops, and implementations for several non-trivial intinf
978            operations in Core.  (The intinf type is completely abstract
979            to the compiler; all operations get delegated back to the Core.)
980    
981            * Intinf equality is handled by polyequal.  However, the compiler
982            does not print its usual warning in this case (since polyequal
983            is the right thing to do there).
984    
985            * Improved the organization of structure InlineT.
986    
987            * A word about conversion primops:
988            If conversions involving intinf do not cancel out during
989            CPS contract, then the compiler must insert calls to Core functions.
990            Since all core access must be resolved already during the FLINT
991            translate phase, it would be too late a the time of CPS contract
992            to add new Core calls.  For this reason, conversion primops
993            for intinf carry two arguments:  1. the numeric argument that
994            they are supposed to convert, and 2. the Core function that
995            can help with this conversion if necessary.  If CPS contract
996            eliminates a primop, then the associated Core function becomes
997            dead and goes away.  Intinf conversion primops that do not get
998            eliminated by CPS contract get rewritten into calls of their
999            core functions by a separate, new phase.
1000    
1001    interactive system:
1002            - Control.Print.intinfDepth controls max length of intinf constants
1003            being printed. (Analogous to Control.Print.stringDepth.)
1004            - Cleanup in printutil and pputil: got rid of unused stuff and
1005            duplicates; replaced some of the code with code that makes better
1006            use of library functionality.
1007    
1008    CM:
1009            Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
1010            are no longer silent.
1011    
1012    CKIT:
1013            Fixed mismatched uses of Int32 and LargeInt.  I always decided
1014            in favor of LargeInt -- which is now the same as IntInf.
1015            CKIT-knowledgable people should check whether this is what's
1016            intended and otherwise change things back to using Int32 or
1017            FixedInt.
1018    
1019    Throughout the code:
1020            Started using IntInf.int literals and built-in operations
1021            (e.g., comparison with 0) where this seems appropriate.
1022    
1023    
1024    ----------------------------------------------------------------------
1025    Name: Dave MacQueen (dbm@cs.uchicago.edu)
1026    Date: 2003/08/13 11:36:00 CDT
1027    Tag: dbm-20030813-mcz-merge1
1028    Description:
1029    
1030    Merging changes from the mcz-branch development branch into trunk.
1031    These changes involve replacement of the emulated old prettyprinter
1032    interface with direct use of the SML/NJ Lib PP library, and fixing
1033    of a couple of bugs (895, 1186) relating to error messages.  A new
1034    prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
1035    has been added.
1036    
1037    ----------------------------------------------------------------------
1038    Name: Matthias Blume (blume (at) tti - c (dot) org)
1039    Date: 2003/08/11 15:45:00 CDT
1040    Tag: blume-20030811-windows
1041    Description:
1042    
1043    Version number bumped to 110.42.9.  NEW BOOTFILES!!!
1044    
1045        http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
1046    
1047    This patch restores SML/NJ's ability to run under win32.  There are a
1048    number of changes, including fixes for several bugs that had gone
1049    unnoticed until now:
1050    
1051      - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
1052        trying to run the win32 version from within cygwin.)
1053    
1054      - There are a number of simple .bat scripts that substitute for their
1055        corresponding Unix shell-scripts.  (See below.)
1056    
1057      - The internals of ml-build have been modified slightly.  The main
1058        difference is that instead of calling ".link-sml" (or link-sml.bat)
1059        using OS.Process.system, the ML process delegates this task back
1060        to the script.  Otherwise problems arise in mixed environments such
1061        as Cygwin where scripts look and work like Unix scripts, but
1062        where OS.Process.system cannot run them.
1063    
1064      - In CM, the srcpath pickler used native pathname syntax -- which
1065        is incorrect in the case of cross-compilation.  The new pickle format
1066        is independent of platform-specific naming conventions.
1067    
1068      - Path configuration files (such as lib/pathconfig) can now choose
1069        between native and standard syntax.  Placing a line of the form
1070    
1071            standard!
1072    
1073        into the file causes all subsequent paths to be interpreted using
1074        CM standard pathname syntax (= Unix conventions); a line
1075    
1076            native!
1077    
1078        switches back to native style.  This was needed so that
1079        path config files can be written portably, see src/system/pathconfig.
1080    
1081      - Runtime system:
1082    
1083          - win32-filesys.c:  get_file_time and set_file_time now
1084            access modification time, not creation time.
1085    
1086          - I/O code made aware of new array representation.
1087    
1088          - Bug fixes in X86.prim.masm.
1089    
1090       - src/system/makeml made aware of win32. (For use under cygwin
1091            and other Unix-environments for windows.)
1092    
1093       - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
1094         which caused BinIO.inputAll to fail consistently.
1095    
1096    .bat scripts:
1097    
1098       Windows .bat scripts assume that SMLNJ_HOME is defined.
1099    
1100       - sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
1101         applications (sml, ml-yacc, ml-lex).
1102       - ml-build.bat: analogous to ml-build.
1103       - config\install.bat:  Analogous to config/install.sh.  This requires
1104         that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
1105         (nmake etc. must be on the path, and vcvars32 must have been run.)
1106         Moreover, sources for ml-lex and ml-yacc need to exist under src,
1107         and the bootfile hierarchy must have been unpacked under
1108         sml.boot.x86-win32.
1109         The script is very primitive and does a poor job at error checking.
1110         It only installs the base system, ml-lex, and ml-yacc.  No other
1111         libraries are being installed (i.e., you get only those that
1112         are part of the compiler.)
1113       - link-sml.bat:  analogous to .link-sml, but not currently used
1114    
1115    Unrelated bug fixes:
1116    
1117      - ml-nlffigen now exports structures ST_* corresponding to incomplete
1118        types.
1119      - Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
1120        otherwise.)
1121    
1122    ----------------------------------------------------------------------
1123    Name: Dave MacQueen (macqueen@cs.uchicago.edu)
1124    Date: 2003/06/17
1125    Tag: macqueen-20030617-bug895
1126    Description:
1127    
1128    Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
1129    Tag will be used for new development branch (mcz-branch) for
1130    use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
1131    
1132    ----------------------------------------------------------------------
1133    Name: Matthias Blume (blume (at) tti - c (dot) org)
1134    Date: 2003/05/27 16:55:00 CDT
1135    Tag: blume-20030527-polyeq
1136    Description:
1137    
1138    Tried to eliminated most cases of polymorphic equality.
1139    
1140    ----------------------------------------------------------------------
1141    Name: Matthias Blume (blume (at) tti - c (dot) org)
1142    Date: 2003/05/21 17:45:00 CDT
1143    Tag: blume-20030517-complete
1144    Description:
1145    
1146    Two changes:
1147    
1148      1. Added a flag for controlling whether non-exhaustive bindings will
1149         be treated as errors (default is false).
1150      2. Cleaned up the *entire* source tree so that CMB.make goes through
1151         without a single non-exhaustive match- or bind warning.
1152    
1153    ----------------------------------------------------------------------
1154    Name: Matthias Blume (blume (at) tti - c (dot) org)
1155    Date: 2003/05/17 10:20:00 CDT
1156    Tag: blume-20030517-absyn
1157    Description:
1158    
1159    1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
1160    
1161       This mainly affects the quality of error messages.  However, some
1162       of the code is now more straightforward than before. (Treatment of
1163       the above four constructs in translate.sml is much simpler than
1164       the "macro-expansion" that was going on before.  Plus, the mach-
1165       compiler no longer gets invoked just to be able to compile an
1166       if-expression.)
1167    
1168    2. The ErrorMsg.Error exception is now caught and absorbed by the
1169       interactive loop.
1170    
1171    ----------------------------------------------------------------------
1172    Name: Allen Leung
1173    Date: 2003/05/16 13:05:00 CDT
1174    Tag: leunga-20030516-cygwin-runtime
1175    Description:
1176    
1177        Ported the runtime system to cygwin, which uses the unix
1178    x86-unix bin files.  Missing/buggy features:
1179    
1180       o getnetbyname, getnetbyaddr: these functions seem to be missing in
1181         the Cygwin library.
1182       o Ctrl-C handling may be flaky.
1183       o Windows system calls and Windows I/O are not supported.
1184    
1185       A new set of binfiles is located at:
1186    
1187         http://www.dorsai.org/~leunga/boot.x86-unix.tgz
1188    
1189       This is only needed for bootstrapping the cygwin version of smlnj.
1190       Other x86 versions can use the existing binfiles.
1191    
1192    ----------------------------------------------------------------------
1193    Name: Matthias Blume
1194    Date: 2003/04/08 15:42:00 CDT
1195    Tag: blume-20030408-listpair
1196    Description:
1197    
1198    1. Added a target 'mlrisc' to installer.
1199    
1200    2. Added missing elements to structure ListPair.
1201    
1202    ----------------------------------------------------------------------
1203    Name: Allen Leung
1204    Date: 2003/01/07 10:40:00 EST
1205    Tag: leunga-20030107-int-rem
1206    Description:
1207    
1208    Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
1209    The arguments to the SUBL instruction were swapped.
1210    
1211    ----------------------------------------------------------------------
1212    Name: Matthias Blume
1213    Date: 2002/12/12 16:25:00 EST
1214    Tag: blume-20021212-risc-ra
1215    Description:
1216    
1217    Fixed a serious bug in the rewrite code for FP spilling/reloading that
1218    sent the RA into an infinite loop when floating point registers get
1219    spilled.  (Because of this bug, e.g., nucleic stopped compiling between
1220    110.37 and 110.38.)
1221    There was another set of potential problems related to the handling of
1222    MLRISC annotations (but those did not yet cause real problems, apparently).
1223    
1224    ----------------------------------------------------------------------
1225    Name: Matthias Blume
1226    Date: 2002/12/06 22:40:00 EST
1227    Tag: blume-20021206-cm-fileid
1228    Description:
1229    
1230    Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
1231    This fixes the problem of CM getting confused by files that suddenly
1232    change their identity (e.g., by getting unlinked and recreated by some
1233    text editor such as vi).  There might be a better/cheaper/cleaner way
1234    of doing this, but for now this will have to do.
1235    
1236    ----------------------------------------------------------------------
1237    Name: Matthias Blume
1238    Date: 2002/10/28 09:50:00 EST
1239    Tag: blume-20021028-typecheck
1240    Description:
1241    
1242    Exported structure Typecheck from $smlnj/viscomp/core.cm.
1243    
1244    ----------------------------------------------------------------------
1245    Name: Matthias Blume
1246    Date: 2002/10/17 09:10:00 EDT
1247    Tag: Release_110_42
1248    Description:
1249    
1250    In good old tradition, there has been a slight hiccup so that we have
1251    to patch 110.42 after the fact.   The old release tag has been replaced
1252    (see below).
1253    
1254    The change solves a problem with two competing approaches the
1255    configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
1256    each other's way.
1257    
1258    This change only affects the runtime system code and the installer script.
1259    (No new bootfiles.)
1260    
1261    ----------------------------------------------------------------------
1262    Name: Matthias Blume
1263    Date: 2002/10/16 12:00:00 EDT
1264    Tag: Release_110_42_removed
1265    Description:
1266    
1267    New working release.  New bootfiles.
1268    
1269    ----------------------------------------------------------------------
1270    Name: Matthias Blume
1271    Date: 2002/10/10 13:10:00 EDT
1272    Tag: blume-20021010-ppc-divs
1273    Description:
1274    
1275    The mltree operator DIVS must be implemented with an overflow check on
1276    the PPC because the hardware indicates divide-by-zero using "overflow" as
1277    well.
1278    
1279    ----------------------------------------------------------------------
1280    Name: Matthias Blume
1281    Date: 2002/07/23 11:45:00
1282    Tag: blume-20020723-smlnj-home
1283    Description:
1284    
1285    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
1286    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
1287    CM_PATHCONFIG is also set), the path configuration file is assumed
1288    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
1289    the entire tree to some other place and everything will "just work".
1290    
1291    (Companion commands such as ml-build and ml-makedepend also sense this
1292    variable.)
1293    
1294    ----------------------------------------------------------------------
1295    Name: Matthias Blume
1296    Date: 2002/07/12 21:19:00 EDT
1297    Tag: blume-20020712-liveness
1298    Description:
1299    
1300    Exported two useful "step" functions from liveness module (MLRISC).
1301    
1302    ----------------------------------------------------------------------
1303    Name: Matthias Blume
1304    Date: 2002/07/05 16:00 EDT
1305    Tag: Release_110_41
1306    Description:
1307    
1308    New working release.  New bootfiles.
1309    
1310    ----------------------------------------------------------------------
1311    Name: Matthias Blume
1312    Date: 2002/07/05 10:25:00 EDT
1313    Tag: blume-20020705-btimp
1314    Description:
1315    
1316    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
1317    other clients can set up backtracing support.
1318    
1319    ----------------------------------------------------------------------
1320    Name: Matthias Blume
1321    Date: 2002/06/25 17:23:00 EDT
1322    Tag: blume-20020625-fpmax
1323    Description:
1324    
1325    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
1326    case.  (The sense of the isNaN test was reversed -- which made min and
1327    max always return their first argument.)
1328    
1329    ----------------------------------------------------------------------
1330    Name: Matthias Blume
1331    Date: 2002/06/11
1332    Tag: blume-20020611-unixpath
1333    Description:
1334    
1335    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
1336    
1337    ----------------------------------------------------------------------
1338    Name: Matthias Blume
1339    Date: 2002/06/10 16:35:00 EDT
1340    Tag: blume-20020610-ieeereal
1341    Description:
1342    
1343    I back-ported my implementation of IEEEReal.fromString from the
1344    idlbasis-devel branch so that we can test it.
1345    
1346    Another small change is that ppDec tries to give more information
1347    than just "<sig>" in the case of functors.  However, this code is
1348    broken in some mysterious way if the functor's body's signature
1349    has not been declared by ascription but gets inferred from the
1350    implementation.  This needs fixing...
1351    
1352    ----------------------------------------------------------------------
1353    Name: Matthias Blume
1354    Date: 2002/05/31
1355    Tag: blume-20020531-btrace-mode
1356    Description:
1357    
1358    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
1359    the wayside when I switched over to using Controls everywhere.)
1360    
1361    ----------------------------------------------------------------------
1362    Name: Lal George
1363    Date: 2002/05/23 12:21:40 EDT
1364    Tag: george-20020523-visual-labels
1365    Description:
1366    
1367    Labels are now displayed in the graphical output to make
1368    the fall-through and target blocks obvious.
1369    
1370    ----------------------------------------------------------------------
1371    Name: Matthias Blume
1372    Date: 2002/05/22 11:03:00 EDT
1373    Tag: blume-20020522-shrink
1374    Description:
1375    
1376    John tweaked yesterday's fix for 1131 to handle an out-of-memory
1377    situation that comes up when allocating huge arrays.
1378    
1379    ----------------------------------------------------------------------
1380    Name: Matthias Blume
1381    Date: 2002/05/21 16:00:00 EDT
1382    Tag: Release_110_40
1383    Description:
1384    
1385    New working release (110.40).  New bootfiles.
1386    
1387    [Also: John Reppy fixed GC bug 1131.]
1388    
1389    ----------------------------------------------------------------------
1390    Name: Matthias Blume
1391    Date: 2002/05/21 12:35:00 EDT
1392    Tag: blume-20020521-cmdoc
1393    Description:
1394    
1395    CM documentation update.
1396    
1397    ----------------------------------------------------------------------
1398    Name: Matthias Blume
1399    Date: 2002/05/21 10:55:00 EDT
1400    Tag: blume-20020521-misc
1401    Description:
1402    
1403    - John tweaked runtime to be silent on heap export (except when
1404      GC messages are on).
1405    - I added a few more things (cross-compiling versions of CMB) to
1406      config/preloads (as suggestions).
1407    
1408    ----------------------------------------------------------------------
1409    Name: Matthias Blume
1410    Date: 2002/05/20 22:25:00 EDT
1411    Tag: blume-20020520-controls
1412    Description:
1413    
1414    - Added ControlUtil structure to control-lib.cm.
1415    - Use it throughout.
1416    - Used Controls facility to define MLRISC controls (as opposed to
1417      registering MLRISC control ref cells with Controls after the fact)
1418    - Fixed messed-up controls priorities.
1419    
1420    * Removed again all the stuff from config/preloads that one wouldn't
1421      be able to preload at the time the initial heap image is built.
1422      (Many libraries, e.g., CML, do not exist yet at this time.  The
1423       only libraries that can be preloaded via config/preloads are those
1424       that come bundled with the bootfiles.)
1425    
1426    ----------------------------------------------------------------------
1427    Name: Matthias Blume
1428    Date: 2002/05/20 10:59:00 EDT
1429    Tag: blume-20020520-preloads
1430    Description:
1431    
1432    Added a lot of commented-out suggestions for things to be included
1433    in config/preloads.
1434    
1435    ----------------------------------------------------------------------
1436    Name: Allen Leung
1437    Date: 2002/05/18 14:20:00 EDT
1438    Tag: leunga-20020518-mdl
1439    Description:
1440    
1441    o Made the mdl tool stuff compile and run again.
1442    o I've disabled all the stuff that depends on RTL specifications; they
1443      are all badly broken anyway.
1444    
1445    ----------------------------------------------------------------------
1446    Name: Matthias Blume
1447    Date: 2002/05/17 16:49:00 EDT
1448    Tag: blume-20020517-controls
1449    Description:
1450    
1451    0. John Reppy made several modifications to the SML/NJ library.
1452       In particular, there is a shiny new controls-lib.cm.
1453    
1454    1. Pushed new controls interface through compiler so that everything
1455       compiles again.
1456    
1457    2. Added FormatComb and FORMAT_COMB to the CML version of the
1458       SML/NJ library (so that CML compiles again).
1459    
1460    3. Modified init scripts because XXX_DEFAULT environment variables
1461       are no longer with us.  (Boot-time initialization is now done
1462       using the same environment variables that are also used for
1463       startup-time initialization of controls.)
1464    
1465    ----------------------------------------------------------------------
1466    Name: Lal George
1467    Date: 2002/05/15 09:20:10 EDT
1468    Tag: george-20020515-pseudo-op-decls
1469    Description:
1470    
1471            All pseudo-ops emitted before the first segment declaration
1472    such as TEXT, DATA, and BSS directives are assumed to be global
1473    declarations and are emitted first in the assembly file. This is
1474    useful in a number of situations where one has pseudo-ops that are not
1475    specific to any segment, and also works around the constraint that one
1476    cannot have client pseudo-ops in the TEXT segment.
1477    
1478            Because no segment is associated with these declarations it is
1479    an error to allocate any space or objects before the first segment
1480    directive and an exception will be raised. However, we cannot make
1481    this check for client pseudo-ops.
1482    
1483            These top level declarations are a field in the CFG graph_info.
1484    In theory you can continue to add to this field after the CFG has been
1485    built -- provided you know what you are doing;-)
1486    
1487    ----------------------------------------------------------------------
1488    Name: Matthias Blume
1489    Date: 2002/05/13 16:40:00 EDT
1490    Tag: blume-20020513-pp-etc
1491    Description:
1492    
1493    A few minor bugfixes:
1494    
1495      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
1496        (Bogus printouts for redefined bindings still occur.  Compiler
1497        bug should no longer occur now.  We need to redo the prettyprinter
1498        from scratch.)
1499    
1500      - CM pathname printer now also adds escape sequences for ( and )
1501    
1502      - commend and docu fixes for ml-nlffi
1503    
1504    ----------------------------------------------------------------------
1505    Name: Matthias Blume
1506    Date: 2002/05/10 16:40:00 EDT
1507    Tag: blume-20020510-erg-textio
1508    Description:
1509    
1510    Applied the following bugfix provided by Emden Gansner:
1511    
1512        Output is corrupted when outputSubstr is used rather than output.
1513    
1514        The problem occurs when a substring
1515    
1516            ss = (s, dataStart, dataLen)
1517    
1518        where dataStart > 0, fills a stream buffer with avail bytes left.
1519        avail bytes of s, starting at index dataStart, are copied into the
1520        buffer, the buffer is flushed, and then the remaining dataLen-avail
1521        bytes of ss are copied into the beginning of the buffer. Instead of
1522        starting this copy at index dataStart+avail in s, the current code
1523        starts the copy at index avail.
1524    
1525        Fix:
1526        In text-io-fn.sml, change line 695 from
1527             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
1528        to
1529             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
1530    
1531    ----------------------------------------------------------------------
1532    Name: Matthias Blume
1533    Date: 2002/04/12 13:55:00 EDT
1534    Tag: blume-20020412-assyntax
1535    Description:
1536    
1537    1. Grabbed newer assyntax.h from the XFree86 project.
1538    2. Fiddled with how to compile X86.prim.asm without warnings.
1539    3. (Very) Minor cleanup in CM.
1540    
1541    ----------------------------------------------------------------------
1542    Name: Matthias Blume
1543    Date: 2002/04/01 (no joke!) 17:07:00 EST
1544    Tag: blume-20020401-x86div
1545    Description:
1546    
1547    Added full support for div/mod/rem/quot on the x86, using the machine
1548    instruction's two results (without clumsily recomputing the remainder)
1549    directly where appropriate.
1550    
1551    Some more extensive power-of-two support was added to the x86 instruction
1552    selector (avoiding expensive divs, mods, and muls where they can be
1553    replaced with cheaper shifts and masks).  However, this sort of thing
1554    ought to be done earlier, e.g., within the CPS optimizer so that
1555    all architectures benefit from it.
1556    
1557    The compiler compiles to a fixed point, but changes might be somewhat
1558    fragile nevertheless.  Please, report any strange things that you might
1559    see wrt. div/mod/quot/rem...
1560    
1561    ----------------------------------------------------------------------
1562    Name: Matthias Blume
1563    Date: 2002/03/29 17:22:00
1564    Tag: blume-20020329-div
1565    Description:
1566    
1567    Fixed my broken div/mod logic.  Unfortunately, this means that the
1568    inline code for div/mod now has one more comparison than before.
1569    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
1570    The problem was with quotient = 0, because that alone does not tell
1571    us which way the rounding went.  One then has to look at whether
1572    remainder and divisor have the same sign...  :(
1573    
1574    Anyway, I replaced the bootfiles with fresh ones...
1575    
1576    ----------------------------------------------------------------------
1577    Name: Matthias Blume
1578    Date: 2002/03/29 14:10:00 EST
1579    Tag: blume-20020329-inlprims
1580    Description:
1581    
1582    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
1583    
1584    Primops have changed. This means that the bin/boot-file formats have
1585    changed as well.
1586    
1587    To make sure that there is no confusion, I made a new version.
1588    
1589    
1590    CHANGES:
1591    
1592    * removed REMT from mltree (remainder should never overflow).
1593    
1594    * added primops to deal with divisions of all flavors to the frontend
1595    
1596    * handled these primops all the way through so they map to their respective
1597      MLRISC support
1598    
1599    * used these primops in the implementation of Int, Int32, Word, Word32
1600    
1601    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
1602    
1603    * parameterized INLMIN, INLMAX, and INLABS by a numkind
1604    
1605    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
1606      floating point
1607    
1608    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
1609      and Real (but Real.abs maps to a separate floating-point-only primop)
1610    
1611    
1612    TODO items:
1613    
1614    * Hacked Alpha32 instruction selection, disabling the selection of REMx
1615      instructions because the machine instruction encoder cannot handle
1616      them.  (Hppa, PPC, and Sparc instruction selection did not handle
1617      REM in the first place, and REM is supported by the x86 machine coder.)
1618    
1619    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
1620      selection phase.  (The two can be streamlined because the hardware
1621      delivers both quotient and remainder at the same time anyway.)
1622    
1623    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
1624      (Currently the behavior is inconsistent both across architectures and
1625      wrt. the draft Basis spec.)
1626    
1627    * Word8 should eventually be handled natively, too.
1628    
1629    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
1630      as if there currently is no execution path that could trigger it in
1631      SML/NJ.  (The assumptions underlying functions arith and promotable do not
1632      hold for things like multiplication and division.)
1633    
1634    ----------------------------------------------------------------------
1635    Name: Matthias Blume
1636    Date: 2002/03/27 16:27:00 EST
1637    Tag: blume-20020327-mlrisc-divisions
1638    Description:
1639    
1640    Added support for all four division operations (ML's div, mod, quot,
1641    and rem) to MLRISC.  In the course of doing so, I also rationalized
1642    the naming (no more annoying switch-around of DIV and QUOT), by
1643    parameterizing the operation by div_rounding_mode (which can be either
1644    DIV_TO_ZERO or DIV_TO_NEGINF).
1645    
1646    The generic MLTreeGen functor takes care of compiling all four
1647    operations down to only round-to-zero div.
1648    
1649    Missing pieces:
1650    
1651      * Doing something smarter than relying on MLTreeGen on architectures
1652        like, e.g., the x86 where hardware division delivers both quotient and
1653        remainder at the same time.  With this, the implementation of the
1654        round-to-neginf operations could be further streamlined.
1655    
1656      * Remove inlining support for div/mod/rem from the frontend and replace it
1657        with primops that get carried through to the backend.  Do this for all
1658        int and word types.
1659    
1660    ----------------------------------------------------------------------
1661    Name: Matthias Blume
1662    Date: 2002/03/25 17:25:00 EST
1663    Tag: blume-20020325-divmod
1664    Description:
1665    
1666    I improved (hopefully without breaking them) the implementation of Int.div,
1667    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
1668    advantage of the following observations:
1669    
1670      Let  q = x quot y      r = x rem y
1671           d = x div  y      m = x mod y
1672    
1673    where "quot" is the round-to-zero version of integer division that
1674    hardware usually provides.  Then we have:
1675    
1676         r = x - q * y        where neither the * nor the - will overflow
1677         d = if q >= 0 orelse x = q * y then q else q - 1
1678                              where neither the * nor the - will overflow
1679         m = if q >= 0 orelse r = 0 then r else r + y
1680                              where the + will not overflow
1681    
1682    This results in substantial simplification of the generated code.
1683    The following table shows the number of CFG nodes and edges generated
1684    for
1685            fun f (x, y) = x OPER y
1686            (* with OPER \in div, mod, quot, rem *)
1687    
1688    
1689        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
1690        --------------------------------------------------------
1691         div |         24 |         39 |         12 |         16
1692         mod |         41 |         71 |         12 |         16
1693        quot |          8 |         10 |          8 |         10
1694         rem |         10 |         14 |          8 |         10
1695    
1696    
1697    ----------------------------------------------------------------------
1698    Name: Matthias Blume
1699    Date: 2002/03/25 22:06:00 EST
1700    Tag: blume-20020325-cprotobug
1701    Description:
1702    
1703    Fixed a bug in cproto (c prototype decoder).
1704    
1705    ----------------------------------------------------------------------
1706    Name: Matthias Blume
1707    Date: 2002/03/25 16:00:00 EST
1708    Tag: blume-20020325-raw-primops
1709    Description:
1710    
1711    I did some cleanup to Allen's new primop code and
1712    replaced yesterday's bootfiles with new ones.
1713    (But they are stored in the same place.)
1714    
1715    ----------------------------------------------------------------------
1716    Name: Matthias Blume
1717    Date: 2002/03/24 22:40:00 EST
1718    Tag: blume-20020324-bootfiles
1719    Description:
1720    
1721    Made the bootfiles that Allen asked for.
1722    
1723    ----------------------------------------------------------------------
1724    Name: Allen Leung
1725    Date: 2002/03/23 15:50:00 EST
1726    Tag: leunga-20020323-flint-cps-rcc-primops
1727    Description:
1728    
1729      1. Changes to FLINT primops:
1730    
1731        (* make a call to a C-function;
1732         * The primop carries C function prototype information and specifies
1733         * which of its (ML-) arguments are floating point. C prototype
1734         * information is for use by the backend, ML information is for
1735         * use by the CPS converter. *)
1736      | RAW_CCALL of { c_proto: CTypes.c_proto,
1737                       ml_args: ccall_type list,
1738                       ml_res_opt: ccall_type option,
1739                       reentrant : bool
1740                     } option
1741       (* Allocate uninitialized storage on the heap.
1742        * The record is meant to hold short-lived C objects, i.e., they
1743        * are not ML pointers.  With the tag, the representation is
1744        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
1745        *)
1746      | RAW_RECORD of {tag:bool,sz:int}
1747      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
1748    
1749      2.  These CPS primops are now overloaded:
1750    
1751           rawload of {kind:numkind}
1752           rawstore of {kind:numkind}
1753    
1754          The one argument form is:
1755    
1756             rawload {kind} address
1757    
1758          The two argument form is:
1759    
1760             rawload {kind} [ml object, byte-offset]
1761    
1762      3. RAW_CCALL/RCC now takes two extra arguments:
1763    
1764         a. The first is whether the C call is reentrant, i.e., whether
1765            ML state should be saved and restored.
1766         b. The second argument is a string argument specifying the name of
1767            library and the C function.
1768    
1769         These things are currently not handled in the code generator, yet.
1770    
1771      4. In CProto,
1772    
1773         An encoding type of "bool" means "ml object" and is mapped into
1774         C prototype of PTR.  Note that "bool" is different than "string",
1775         even though "string" is also mapped into PTR, because "bool"
1776         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
1777    
1778      5. Pickler/unpicker
1779    
1780         Changed to handle RAW_RECORD and newest RAW_CCALL
1781    
1782      6. MLRiscGen,
1783    
1784         1. Changed to handle the new rawload/rawstore/rawrecord operators.
1785         2. Code for handling C Calls has been moved to a new module CPSCCalls,
1786            in the file CodeGen/cpscompile/cps-c-calls.sml
1787    
1788      7. Added the conditional move operator
1789    
1790             condmove of branch
1791    
1792         to cps.  Generation of this is still buggy so it is currently
1793         disabled.
1794    
1795    ----------------------------------------------------------------------
1796    Name: Lal George
1797    Date: 2002/03/22 14:18:25 EST
1798    Tag: george-20020322-cps-branch-prob
1799    Description:
1800    
1801    Implemented the Ball-Larus branch prediction-heuristics, and
1802    incorporated graphical viewers for control flow graphs.
1803    
1804    Ball-Larus Heuristics:
1805    ---------------------
1806    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
1807    
1808    By design it uses the Dempster-Shafer theory for combining
1809    probabilities.  For example, in the function:
1810    
1811        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
1812    
1813    the ball-larus heuristics predicts that the n=0 is unlikely
1814    (OH-heuristic), and the 'then' branch is unlikely because of the
1815    RH-heuristic -- giving the 'then' branch an even lower combined
1816    probability using the Dempster-Shafer theory.
1817    
1818    Finally, John Reppy's loop analysis in MLRISC, further lowers the
1819    probability of the 'then' branch because of the loop in the else
1820    branch.
1821    
1822    
1823    Graphical Viewing:
1824    ------------------
1825    I merely plugged in Allen's graphical viewers into the compiler. The
1826    additional code is not much. At the top level, saying:
1827    
1828            Control.MLRISC.getFlag "cfg-graphical-view" := true;
1829    
1830    will display the graphical view of the control flow graph just before
1831    back-patching.  daVinci must be in your path for this to work. If
1832    daVinci is not available, then the default viewer can be changed
1833    using:
1834    
1835            Control.MLRISC.getString "viewer"
1836    
1837    which can be set to "dot" or "vcg" for the corresponding viewers. Of
1838    course, these viewers must be in your path.
1839    
1840    The above will display the compilation unit at the level of clusters,
1841    many of which are small, boring, and un-interesting. Also setting:
1842    
1843            Control.MLRISC.getInt "cfg-graphical-view_size"
1844    
1845    will display clusters that are larger than the value set by the above.
1846    
1847    
1848    ----------------------------------------------------------------------
1849    Name: Matthias Blume
1850    Date: 2002/03/21 22:20:00 EST
1851    Tag: blume-20020321-kmp-bugfix
1852    Description:
1853    
1854    Changed the interface to the KMP routine in PreString and fixed
1855    a minor bug in one place where it was used.
1856    
1857    ----------------------------------------------------------------------
1858    Name: Allen Leung
1859    Date: 2002/03/21 20:30:00 EST
1860    Tag: leunga-20020321-cfg
1861    Description:
1862    
1863      Fixed a potential problem in cfg edge splitting.
1864    
1865    ----------------------------------------------------------------------
1866    Name: Allen Leung
1867    Date: 2002/03/21 17:15:00 EST
1868    Tag: leunga-20020321-x86-fp-cfg
1869    Description:
1870    
1871      1. Recoded the buggy parts of x86-fp.
1872    
1873         a. All the block reordering code has been removed.
1874            We now depend on the block placement phases to do this work.
1875    
1876         b. Critical edge splitting code has been simplified and moved into the
1877            CFG modules, as where they belong.
1878    
1879         Both of these were quite buggy and complex.  The code is now much, much
1880         simpler.
1881    
1882      2. X86 backend.
1883    
1884         a. Added instructions for 64-bit support.  Instruction selection for
1885            64-bit has not been committed, however, since that
1886            requires changes to MLTREE which haven't been approved by
1887            Lal and John.
1888    
1889         b. Added support for FUCOMI and FUCOMIP when generating code for
1890            PentiumPro and above.  We only generate these instructions in
1891            the fast-fp mode.
1892    
1893         c. Added cases for JP and JNP in X86FreqProps.
1894    
1895      3. CFG
1896    
1897         CFG now has a bunch of methods for edge splitting and merging.
1898    
1899      4. Machine description.
1900    
1901         John's simplification of MLTREE_BASIS.fcond broke a few machine
1902         description things:
1903    
1904         rtl-build.{sig,sml} and hppa.mdl fixed.
1905    
1906         NOTE: the machine description stuff in the repository is still broken.
1907               Again, I can't put my fixes in because that involves
1908               changes to MLTREE.
1909    
1910    ----------------------------------------------------------------------
1911    Name: Matthias Blume
1912    Date: 2002/03/20 15:55:00 EST
1913    Tag: blume-20020320-kmp
1914    Description:
1915    
1916    Implemented Knuth-Morris-Pratt string matching in PreString and used
1917    it for String.isSubstring, Substring.isSubstring, and
1918    Substring.position.
1919    
1920    (Might need some stress-testing.  Simple examples worked fine.)
1921    
1922    ----------------------------------------------------------------------
1923    Name: Matthias Blume
1924    Date: 2002/03/19 16:37:00 EST
1925    Tag: blume-20020319-witnesses
1926    Description:
1927    
1928    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
1929    
1930    This implements a generic mechanism for changing constness qualifiers
1931    anywhere within big C types without resorting to outright "casts".
1932    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
1933    modify the constness at the outermost level.)
1934    The implementation of "convert" is based on the idea of "witness"
1935    values -- values that are not used by the operation but whose types
1936    "testify" to their applicability.  On the implementation side, "convert"
1937    is simply a projection (returning its second curried argument).  With
1938    cross-module inlining, it should not result in any machine code being
1939    generated.
1940    
1941    ----------------------------------------------------------------------
1942    Name: Matthias Blume
1943    Date: 2002/03/15 16:40:00 EST
1944    Tag: blume-20020315-basis
1945    Description:
1946    
1947    Provided (preliminary?) implementations for
1948    
1949      {String,Substring}.{concatWith,isSuffix,isSubstring}
1950    
1951    and
1952    
1953      Substring.full
1954    
1955    Those are in the Basis spec but they were missing in SML/NJ.
1956    
1957    ----------------------------------------------------------------------
1958    Name: Matthias Blume
1959    Date: 2002/03/14 21:30:00 EST
1960    Tag: blume-20020314-controls
1961    Description:
1962    
1963    Controls:
1964    ---------
1965    
1966    1. Factored out the recently-added Controls : CONTROLS stuff and put
1967       it into its own library $/controls-lib.cm.  The source tree for
1968       this is under src/smlnj-lib/Controls.
1969    
1970    2. Changed the names of types and functions in this interface, so they
1971       make a bit more "sense":
1972    
1973          module -> registry
1974          'a registry -> 'a group
1975    
1976    3. The interface now deals in ref cells only.  The getter/setter interface
1977       is (mostly) gone.
1978    
1979    4. Added a function that lets one register an already-existing ref cell.
1980    
1981    5. Made the corresponding modifications to the rest of the code so that
1982       everything compiles again.
1983    
1984    6. Changed the implementation of Controls.MLRISC back to something closer
1985       to the original.  In particular, this module (and therefore MLRISC)
1986       does not depend on Controls.  There now is some link-time code in
1987       int-sys.sml that registers the MLRISC controls with the Controls
1988       module.
1989    
1990    CM:
1991    ---
1992    
1993      * One can now specify the lambda-split aggressiveness in init.cmi.
1994    
1995    ----------------------------------------------------------------------
1996    Name: Allen Leung
1997    Date: 2002/03/13 17:30:00 EST
1998    Tag: leunga-20020313-x86-fp-unary
1999    Description:
2000    
2001    Bug fix for:
2002    
2003    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
2004    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
2005    > - fun f(x,(y,z)) = Real.~ y;
2006    > [autoloading]
2007    > [autoloading done]
2008    >       fchsl   (%eax), 184(%esp)
2009    > Error: MLRisc bug: X86MCEmitter.emitInstr
2010    >
2011    > uncaught exception Error
2012    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
2013    
2014    The problem was that the code generator did not generate any fp registers
2015    in this case, and the ra didn't know that it needed to run the X86FP phase to
2016    translate the pseudo fp instruction.   This only happened with unary fp
2017    operators in certain situations.
2018    
2019    ----------------------------------------------------------------------
2020    Name: Matthias Blume
2021    Date: 2002/03/13 14:00:00 EST
2022    Tag: blume-20020313-overload-etc
2023    Description:
2024    
2025    1. Added _overload as a synonym for overload for backward compatibility.
2026       (Control.overloadKW must be true for either version to be accepted.)
2027    
2028    2. Fixed bug in install script that caused more things to be installed
2029       than what was requested in config/targets.
2030    
2031    3. Made CM aware of the (_)overload construct so that autoloading
2032       works.
2033    
2034    ----------------------------------------------------------------------
2035    Name: Matthias Blume
2036    Date: 2002/03/12 22:03:00 EST
2037    Tag: blume-20020312-url
2038    Description:
2039    
2040    Forgot to update BOOT and srcarchiveurl.
2041    
2042    ----------------------------------------------------------------------
2043    Name: Matthias Blume
2044    Date: 2002/03/12 17:30:00 EST
2045    Tag: blume-20020312-version110392
2046    Description:
2047    
2048    Yet another version number bump (because of small changes to the
2049    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
2050    
2051    Changes:
2052    
2053      The new pid generation scheme described a few weeks ago was overly
2054      complicated.  I implemented a new mechanism that is simpler and
2055      provides a bit more "stability":  Once CM has seen a compilation
2056      unit, it keeps its identity constant (as long as you do not delete
2057      those crucial CM/GUID/* files).  This means that when you change
2058      an interface, compile, then go back to the old interface, and
2059      compile again, you arrive at the original pid.
2060    
2061      There now also is a mechanism that instructs CM to use the plain
2062      environment hash as a module's pid (effectively making its GUID
2063      the empty string).  For this, "noguid" must be specified as an
2064      option to the .sml file in question within its .cm file.
2065      This is most useful for code that is being generated by tools such
2066      as ml-nlffigen (because during development programmers tend to
2067      erase the tool's entire output directory tree including CM's cached
2068      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
2069      revert to the old, broken behavior of SML/NJ, but in specific cases
2070      where there is no danger of interface confusion, its use is ok
2071      (I think).
2072    
2073      ml-nlffigen by default generates "noguid" annotations.  They can be
2074      turned off by specifying -guid in its command line.
2075    
2076    ----------------------------------------------------------------------
2077    Name: Lal George
2078    Date: 2002/03/12 12 14:42:36 EST
2079    Tag: george-20020312-frequency-computation
2080    Description:
2081    
2082    Integrated jump chaining and static block frequency into the
2083    compiler. More details and numbers later.
2084    
2085    ----------------------------------------------------------------------
2086    Name: Lal George
2087    Date: 2002/03/11 11 22:38:53 EST
2088    Tag: george-20020311-jump-chain-elim
2089    Description:
2090    
2091    Tested the jump chain elimination on all architectures (except the
2092    hppa).  This is on by default right now and is profitable for the
2093    alpha and x86, however, it may not be profitable for the sparc and ppc
2094    when compiling the compiler.
2095    
2096    The gc test will typically jump to a label at the end of the cluster,
2097    where there is another jump to an external cluster containing the actual
2098    code to invoke gc. This is to allow factoring of common gc invocation
2099    sequences. That is to say, we generate:
2100    
2101            f:
2102               testgc
2103               ja   L1      % jump if above to L1
2104    
2105            L1:
2106               jmp L2
2107    
2108    
2109    After jump chain elimination the 'ja L1' instructions is converted to
2110    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
2111    up being implemented in their long form (if L2 is far away) using:
2112    
2113            jbe     L3      % jump if below or equal to L3
2114            jmp     L2
2115         L3:
2116            ...
2117    
2118    
2119    For large compilation units L2  may be far away.
2120    
2121    
2122    ----------------------------------------------------------------------
2123    Name: Matthias Blume
2124    Date: 2002/03/11 13:30:00 EST
2125    Tag: blume-20020311-mltreeeval
2126    Description:
2127    
2128    A functor parameter was missing.
2129    
2130    ----------------------------------------------------------------------
2131    Name: Allen Leung
2132    Date: 2002/03/11 10:30:00 EST
2133    Tag: leunga-20020311-runtime-string0
2134    Description:
2135    
2136       The representation of the empty string now points to a
2137    legal null terminated C string instead of unit.  It is now possible
2138    to convert an ML string into C string with InlineT.CharVector.getData.
2139    This compiles into one single machine instruction.
2140    
2141    ----------------------------------------------------------------------
2142    Name: Allen Leung
2143    Date: 2002/03/10 23:55:00 EST
2144    Tag: leunga-20020310-x86-call
2145    Description:
2146    
2147       Added machine generation for CALL instruction (relative displacement mode)
2148    
2149    ----------------------------------------------------------------------
2150    Name: Matthias Blume
2151    Date: 2002/03/08 16:05:00
2152    Tag: blume-20020308-entrypoints
2153    Description:
2154    
2155    Version number bumped to 110.39.1.  NEW BOOTFILES!
2156    
2157    Entrypoints: non-zero offset into a code object where execution should begin.
2158    
2159    - Added the notion of an entrypoint to CodeObj.
2160    - Added reading/writing of entrypoint info to Binfile.
2161    - Made runtime system bootloader aware of entrypoints.
2162    - Use the address of the label of the first function given to mlriscGen
2163      as the entrypoint.  This address is currently always 0, but it will
2164      not be 0 once we turn on block placement.
2165    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
2166      with entry points) from mlriscGen.
2167    
2168    ----------------------------------------------------------------------
2169    Name: Allen Leung
2170    Date: 2002/03/07 20:45:00 EST
2171    Tag: leunga-20020307-x86-cmov
2172    Description:
2173    
2174       Bug fixes for CMOVcc on x86.
2175    
2176       1. Added machine code generation for CMOVcc
2177       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
2178       3. CMOVcc cannot have an immediate operand as argument.
2179    
2180    ----------------------------------------------------------------------
2181    Name: Matthias Blume
2182    Date: 2002/03/07 16:15:00 EST
2183    Tag: blume-20020307-controls
2184    Description:
2185    
2186    This is a very large but mostly boring patch which makes (almost)
2187    every tuneable compiler knob (i.e., pretty much everything under
2188    Control.* plus a few other things) configurable via both the command
2189    line and environment variables in the style CM did its configuration
2190    until now.
2191    
2192    Try starting sml with '-h' (or, if you are brave, '-H')
2193    
2194    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
2195    implements the underlying generic mechanism.
2196    
2197    The interface to some of the existing such facilities has changed somewhat.
2198    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
2199    (The getFoo interface is still there for backward-compatibility, but its
2200    use is deprecated.)
2201    
2202    The ml-build script passes -Cxxx=yyy command-line arguments through so
2203    that one can now twiddle the compiler settings when using this "batch"
2204    compiler.
2205    
2206    TODO items:
2207    
2208    We should go through and throw out all controls that are no longer
2209    connected to anything.  Moreover, we should go through and provide
2210    meaningful (and correct!) documentation strings for those controls
2211    that still are connected.
2212    
2213    Currently, multiple calls to Controls.new are accepted (only the first
2214    has any effect).  Eventually we should make sure that every control
2215    is being made (via Controls.new) exactly once.  Future access can then
2216    be done using Controls.acc.
2217    
2218    Finally, it would probably be a good idea to use the getter-setter
2219    interface to controls rather than ref cells.  For the time being, both
2220    styles are provided by the Controls module, but getter-setter pairs are
2221    better if thread-safety is of any concern because they can be wrapped.
2222    
2223    *****************************************
2224    
2225    One bug fix: The function blockPlacement in three of the MLRISC
2226    backpatch files used to be hard-wired to one of two possibilities at
2227    link time (according to the value of the placementFlag).  But (I
2228    think) it should rather sense the flag every time.
2229    
2230    *****************************************
2231    
2232    Other assorted changes (by other people who did not supply a HISTORY entry):
2233    
2234    1. the cross-module inliner now works much better (Monnier)
2235    2. representation of weights, frequencies, and probabilities in MLRISC
2236       changed in preparation of using those for weighted block placement
2237       (Reppy, George)
2238    
2239    ----------------------------------------------------------------------
2240    Name: Lal George
2241    Date: 2002/03/07 14:44:24 EST 2002
2242    Tag: george-20020307-weighted-block-placement
2243    
2244    Tested the weighted block placement optimization on all architectures
2245    (except the hppa) using AMPL to generate the block and edge frequencies.
2246    Changes were required in the machine properties to correctly
2247    categorize trap instructions. There is an MLRISC flag
2248    "weighted-block-placement" that can be used to enable weighted block
2249    placement, but this will be ineffective without block/edge
2250    frequencies (coming soon).
2251    
2252    
2253    ----------------------------------------------------------------------
2254    Name: Lal George
2255    Date: 2002/03/05 17:24:48 EST
2256    Tag: george-20020305-linkage-cluster
2257    
2258    In order to support the block placement optimization, a new cluster
2259    is generated as the very first cluster (called the linkage cluster).
2260    It contains a single jump to the 'real' entry point for the compilation
2261    unit. Block placement has no effect on the linkage cluster itself, but
2262    all the other clusters  have full freedom in the manner in which they
2263    reorder blocks or functions.
2264    
2265    On the x86 the typical linkage code that is generated is:
2266       ----------------------
2267            .align 2
2268       L0:
2269            addl    $L1-L0, 72(%esp)
2270            jmp     L1
2271    
2272    
2273            .align  2
2274       L1:
2275       ----------------------
2276    
2277    72(%esp) is the memory location for the stdlink register. This
2278    must contain the address of the CPS function being called. In the
2279    above example, it contains the address of  L0; before
2280    calling L1 (the real entry point for the compilation unit), it
2281    must contain the address for L1, and hence
2282    
2283            addl $L1-L0, 72(%esp)
2284    
2285    I have tested this on all architectures except the hppa.The increase
2286    in code size is of course negligible
2287    
2288    ----------------------------------------------------------------------
2289    Name: Allen Leung
2290    Date: 2002/03/03 13:20:00 EST
2291    Tag: leunga-20020303-mlrisc-tools
2292    
2293      Added #[ ... ] expressions to mlrisc tools
2294    
2295    ----------------------------------------------------------------------
2296    Name: Matthias Blume
2297    Date: 2002/02/27 12:29:00 EST
2298    Tag: blume-20020227-cdebug
2299    Description:
2300    
2301    - made types in structure C and C_Debug to be equal
2302    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
2303    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
2304    
2305    ----------------------------------------------------------------------
2306    Name: Matthias Blume
2307    Date: 2002/02/26 12:00:00 EST
2308    Tag: blume-20020226-ffi
2309    Description:
2310    
2311    1. Fixed a minor bug in CM's "noweb" tool:
2312       If numbering is turned off, then truly don't number (i.e., do not
2313       supply the -L option to noweb).  The previous behavior was to supply
2314       -L'' -- which caused noweb to use the "default" line numbering scheme.
2315       Thanks to Chris Richards for pointing this out (and supplying the fix).
2316    
2317    2. Once again, I reworked some aspects of the FFI:
2318    
2319       A. The incomplete/complete type business:
2320    
2321       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
2322         gone!
2323       - ML types representing an incomplete type are now *equal* to
2324         ML types representing their corresponding complete types (just like
2325         in C).  This is still safe because ml-nlffigen will not generate
2326         RTTI for incomplete types, nor will it generate functions that
2327         require access to such RTTI.   But when ML code generated from both
2328         incomplete and complete versions of the C type meet, the ML types
2329         are trivially interoperable.
2330    
2331         NOTE:  These changes restore the full generality of the translation
2332         (which was previously lost when I eliminated functorization)!
2333    
2334       B. Enum types:
2335    
2336       - Structure C now has a type constructor "enum" that is similar to
2337         how the "su" constructor works.  However, "enum" is not a phantom
2338         type because each "T enum" has values (and is isomorphic to
2339         MLRep.Signed.int).
2340       - There are generic access operations for enum objects (using
2341         MLRep.Signed.int).
2342       - ml-nlffigen will generate a structure E_foo for each "enum foo".
2343         * The structure contains the definition of type "mlrep" (the ML-side
2344         representation type of the enum).  Normally, mlrep is the same
2345         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
2346         then mlrep will be defined as a datatype -- thus facilitating
2347         pattern matching on mlrep values.
2348         ("-ec" will be suppressed if there are duplicate values in an
2349          enumeration.)
2350         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
2351         will be generated for each C enum constant xxx.
2352         * Conversion functions m2i and i2m convert between mlrep and
2353         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
2354         * Coversion functions c and ml convert between mlrep and "tag enum".
2355         * Access functions (get/set) fetch and store mlrep values.
2356       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
2357         enumerations are merged into one single enumeration represented by
2358         structure E_'.
2359    
2360    ----------------------------------------------------------------------
2361    Name: Allen Leung
2362    Date: 2002/02/25 04:45:00 EST
2363    Tag: leunga-20020225-cps-spill
2364    
2365    This is a new implementation of the CPS spill phase.
2366    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
2367    In case of problems, replace it with the old file spill.sml
2368    
2369    The current compiler runs into some serious performance problems when
2370    constructing a large record.  This can happen when we try to compile a
2371    structure with many items.  Even a very simple structure like the following
2372    makes the compiler slow down.
2373    
2374        structure Foo = struct
2375           val x_1 = 0w1 : Word32.int
2376           val x_2 = 0w2 : Word32.int
2377           val x_3 = 0w3 : Word32.int
2378           ...
2379           val x_N = 0wN : Word32.int
2380        end
2381    
2382    The following table shows the compile time, from N=1000 to N=4000,
2383    with the old compiler:
2384    
2385    N
2386    1000   CPS 100 spill                           0.04u  0.00s  0.00g
2387           MLRISC ra                               0.06u  0.00s  0.05g
2388              (spills = 0 reloads = 0)
2389           TOTAL                                   0.63u  0.07s  0.21g
2390    
2391    1100   CPS 100 spill                           8.25u  0.32s  0.64g
2392           MLRISC ra                               5.68u  0.59s  3.93g
2393              (spills = 0 reloads = 0)
2394           TOTAL                                   14.71u  0.99s  4.81g
2395    
2396    1500   CPS 100 spill                           58.55u  2.34s  1.74g
2397           MLRISC ra                               5.54u  0.65s  3.91g
2398              (spills = 543 reloads = 1082)
2399           TOTAL                                   65.40u  3.13s  6.00g
2400    
2401    2000   CPS 100 spill                           126.69u  4.84s  3.08g
2402           MLRISC ra                               0.80u  0.10s  0.55g
2403              (spills = 42 reloads = 84)
2404           TOTAL                                   129.42u  5.10s  4.13g
2405    
2406    3000   CPS 100 spill                           675.59u  19.03s  11.64g
2407           MLRISC ra                               2.69u  0.27s  1.38g
2408              (spills = 62 reloads = 124)
2409           TOTAL                                   682.48u  19.61s  13.99g
2410    
2411    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
2412           MLRISC ra                               4.96u  0.27s  2.72g
2413              (spills = 85 reloads = 170)
2414           TOTAL                                   2375.26u  57.21s  48.00g
2415    
2416    As you can see the old cps spill module suffers from some serious
2417    performance problem.  But since I cannot decipher the old code fully,
2418    instead of patching the problems up, I'm reimplementing it
2419    with a different algorithm.  The new code is more modular,
2420    smaller when compiled, and substantially faster
2421    (O(n log n) time and O(n) space).  Timing of the new spill module:
2422    
2423    4000  CPS 100 spill                           0.02u  0.00s  0.00g
2424          MLRISC ra                               0.25u  0.02s  0.15g
2425             (spills=1 reloads=3)
2426          TOTAL                                   7.74u  0.34s  1.62g
2427    
2428    Implementation details:
2429    
2430    As far as I can tell, the purpose of the CPS spill module is to make sure the
2431    number of live variables at any program point (the bandwidth)
2432    does not exceed a certain limit, which is determined by the
2433    size of the spill area.
2434    
2435    When the bandwidth is too large, we decrease the register pressure by
2436    packing live variables into spill records.  How we achieve this is
2437    completely different than what we did in the old code.
2438    
2439    First, there is something about the MLRiscGen code generator
2440    that we should be aware of:
2441    
2442    o MLRiscGen performs code motion!
2443    
2444       In particular, it will move floating point computations and
2445       address computations involving only the heap pointer to
2446       their use sites (if there is only a single use).
2447       What this means is that if we have a CPS record construction
2448       statement
2449    
2450           RECORD(k,vl,w,e)
2451    
2452       we should never count the new record address w as live if w
2453       has only one use (which is often the case).
2454    
2455       We should do something similar to floating point, but the transformation
2456       there is much more complex, so I won't deal with that.
2457    
2458    Secondly, there are now two new cps primops at our disposal:
2459    
2460     1. rawrecord of record_kind option
2461        This pure operator allocates some uninitialized storage from the heap.
2462        There are two forms:
2463    
2464         rawrecord NONE [INT n]  allocates a tagless record of length n
2465         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
2466                                     and initializes the tag.
2467    
2468     2. rawupdate of cty
2469          rawupdate cty (v,i,x)
2470          Assigns to x to the ith component of record v.
2471          The storelist is not updated.
2472    
2473    We use these new primops for both spilling and increment record construction.
2474    
2475     1. Spilling.
2476    
2477        This is implemented with a linear scan algorithm (but generalized
2478        to trees).  The algorithm will create a single spill record at the
2479        beginning of the cps function and use rawupdate to spill to it,
2480        and SELECT or SELp to reload from it.  So both spills and reloads
2481        are fine-grain operations.  In contrast, in the old algorithm
2482        "spills" have to be bundled together in records.
2483    
2484        Ideally, we should sink the spill record construction to where
2485        it is needed.  We can even split the spill record into multiple ones
2486        at the places where they are needed.  But CPS is not a good
2487        representation for global code motion, so I'll keep it simple and
2488        am not attempting this.
2489    
2490     2. Incremental record construction (aka record splitting).
2491    
2492        Long records with many component values which are simulatenously live
2493        (recall that single use record addresses are not considered to
2494         be live) are constructed with rawrecord and rawupdate.
2495        We allocate space on the heap with rawrecord first, then gradually
2496        fill it in with rawupdate.  This is the technique suggested to me
2497        by Matthias.
2498    
2499        Some restrictions on when this is applicable:
2500        1. It is not a VECTOR record.  The code generator currently does not handle
2501           this case. VECTOR record uses double indirection like arrays.
2502        2. All the record component values are defined in the same "basic block"
2503           as the record constructor.  This is to prevent speculative
2504           record construction.
2505    
2506    ----------------------------------------------------------------------
2507    Name: Allen Leung
2508    Date: 2002/02/22 01:02:00 EST
2509    Tag: leunga-20020222-mlrisc-tools
2510    
2511    Minor bug fixes in the parser and rewriter
2512    
2513    ----------------------------------------------------------------------
2514    Name: Allen Leung
2515    Date: 2002/02/21 20:20:00 EST
2516    Tag: leunga-20020221-peephole
2517    
2518    Regenerated the peephole files.  Some contained typos in the specification
2519    and some didn't compile because of pretty printing bugs in the old version
2520    of 'nowhere'.
2521    
2522    ----------------------------------------------------------------------
2523    Name: Allen Leung
2524    Date: 2002/02/19 20:20:00 EST
2525    Tag: leunga-20020219-mlrisc-tools
2526    Description:
2527    
2528       Minor bug fixes to the mlrisc-tools library:
2529    
2530       1.  Fixed up parsing colon suffixed keywords
2531       2.  Added the ability to shut the error messages up
2532       3.  Reimplemented the pretty printer and fixed up/improved
2533           the pretty printing of handle and -> types.
2534       4.  Fixed up generation of literal symbols in the nowhere tool.
2535       5.  Added some SML keywords to to sml.sty
2536    
2537    ----------------------------------------------------------------------
2538    Name: Matthias Blume
2539    Date: 2002/02/19 16:20:00 EST
2540    Tag: blume-20020219-cmffi
2541    Description:
2542    
2543    A wild mix of changes, some minor, some major:
2544    
2545    * All C FFI-related libraries are now anchored under $c:
2546        $/c.cm      --> $c/c.cm
2547        $/c-int.cm  --> $c/internals/c-int.cm
2548        $/memory.cm --> $c/memory/memory.cm
2549    
2550    * "make" tool (in CM) now treats its argument pathname slightly
2551      differently:
2552        1. If the native expansion is an absolute name, then before invoking
2553           the "make" command on it, CM will apply OS.Path.mkRelative
2554           (with relativeTo = OS.FileSys.getDir()) to it.
2555        2. The argument will be passed through to subsequent phases of CM
2556           processing without "going native".  In particular, if the argument
2557           was an anchored path, then "make" will not lose track of that anchor.
2558    
2559    * Compiler backends now "know" their respective C calling conventions
2560      instead of having to be told about it by ml-nlffigen.  This relieves
2561      ml-nlffigen from one of its burdens.
2562    
2563    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
2564    
2565    * Export C_DEBUG and C_Debug from $c/c.cm.
2566    
2567    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
2568      subtyping relationship between incomplete pointers and their complete
2569      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
2570      with the convention of instantiating 'o with ('t, 'c) obj whenever
2571      the pointer target type is complete.  In the incomplete case, 'o
2572      will be instantiated with some "'c iobj" -- a type obtained by
2573      using one of the functors PointerToIncompleteType or PointerToCompleteType.
2574    
2575      Operations that work on both incomplete and complete pointer types are
2576      typed as taking an 'o ptr while operations that require the target to
2577      be known are typed as taking some ('t, 'c) obj ptr.
2578    
2579      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
2580      where void is an eqtype without any values.  This makes it possible
2581      to work on voidptr values using functions meant to operate on light
2582      incomplete pointers.
2583    
2584    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
2585      been vastly simplified.
2586    
2587    ----------------------------------------------------------------------
2588    Name: Matthias Blume
2589    Date: 2002/02/19 10:48:00 EST
2590    Tag: blume-20020219-pqfix
2591    Description:
2592    
2593    Applied Chris Okasaki's bug fix for priority queues.
2594    
2595    ----------------------------------------------------------------------
2596    Name: Matthias Blume
2597    Date: 2002/02/15 17:05:00
2598    Tag: Release_110_39
2599    Description:
2600    
2601    Last-minute retagging is becoming a tradition... :-(
2602    
2603    This is the working release 110.39.
2604    
2605    ----------------------------------------------------------------------
2606    Name: Matthias Blume
2607    Date: 2002/02/15 16:00:00 EST
2608    Tag: Release_110_39-orig
2609    Description:
2610    
2611    Working release 110.39.  New bootfiles.
2612    
2613    (Update: There was a small bug in the installer so it wouldn't work
2614    with all shells.  So I retagged. -Matthias)
2615    
2616    ----------------------------------------------------------------------
2617    Name: Matthias Blume
2618    Date: 2002/02/15 14:17:00 EST
2619    Tag: blume-20020215-showbindings
2620    Description:
2621    
2622    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
2623    the latter can be useful for exploring what bindings are available at
2624    the interactive prompt.  (The first function returns only the list
2625    of symbols that are really bound, the second prints those but also the
2626    ones that CM's autoloading mechanism knows about.)
2627    
2628    ----------------------------------------------------------------------
2629    Name: Matthias Blume
2630    Date: 2002/02/15 12:08:00 EST
2631    Tag: blume-20020215-iptrs
2632    Description:
2633    
2634    Two improvements to ml-nlffigen:
2635    
2636      1. Write files only if they do not exist or if their current contents
2637         do not coincide with what's being written.  (That is, avoid messing
2638         with the time stamps unless absolutely necessary.)
2639    
2640      2. Implement a "repository" mechanism for generated files related
2641         to "incomplete pointer types".   See the README file for details.
2642    
2643    ----------------------------------------------------------------------
2644    Name: Matthias Blume
2645    Date: 2002/02/14 11:50:00 EST
2646    Tag: blume-20020214-quote
2647    Description:
2648    
2649    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
2650    because of the new and improved tag generation scheme.  (Thanks to Allen
2651    Leung for pointing it out.)
2652    
2653    ----------------------------------------------------------------------
2654    Name: Lal George
2655    Date: 2002/02/14 09:55:27 EST 2002
2656    Tag: george-20020214-isabelle-bug
2657    Description:
2658    
2659    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
2660    of Isabelle on the x86.
2661    
2662    From Allen:
2663    -----------
2664     I've found the problem:
2665    
2666         in ra-core.sml, I use the counter "blocked" to keep track of the
2667         true number of elements in the freeze queue.  When the counter goes
2668         to zero, I skip examining the queue.  But I've messed up the
2669         bookkeeping in combine():
2670    
2671             else ();
2672             case !ucol of
2673               PSEUDO => (if !cntv > 0 then
2674                     (if !cntu > 0 then blocked := !blocked - 1 else ();
2675                                        ^^^^^^^^^^^^^^^^^^^^^^^
2676                      moveu := mergeMoveList(!movev, !moveu)
2677                     )
2678                  else ();
2679    
2680         combine() is called to coalesce two nodes u and v.
2681         I think I was thinking that if the move counts of u and v are both
2682         greater than zero then after they are coalesced then one node is
2683         removed from the freeze queue.  Apparently I was thinking that
2684         both u and v are of low degree, but that's clearly not necessarily true.
2685    
2686    
2687    02/12/2002:
2688        Here's the patch.  HOL now compiles.
2689    
2690        I don't know how this impact on performance (compile
2691        time or runtime).  This bug caused the RA (especially on the x86)
2692        to go thru the potential spill phase when there are still nodes on the
2693        freeze queue.
2694    
2695    
2696    
2697    
2698    ----------------------------------------------------------------------
2699    Name: Matthias Blume
2700    Date: 2002/02/13 22:40:00 EST
2701    Tag: blume-20020213-fptr-rtti
2702    Description:
2703    
2704    Fixed a bug in ml-nlffigen that was introduced with one of the previous
2705    updates.
2706    
2707    ----------------------------------------------------------------------
2708    Name: Matthias Blume
2709    Date: 2002/02/13 16:41:00 EST
2710    Tag: blume-20020213-cmlpq
2711    Description:
2712    
2713    Added new priority queue export symbols (which have just been added to
2714    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
2715    not compile and the installer would choke.)
2716    
2717    ----------------------------------------------------------------------
2718    Name: Matthias Blume
2719    Date: 2002/02/13 16:15:00 EST
2720    Tag: blume-20020213-various
2721    Description:
2722    
2723    1. More tweaks to ml-nlffigen:
2724    
2725       - better internal datastructures (resulting in slight speedup)
2726       - "-match" option requires exact match
2727       - "localized" gensym counters (untagged structs/unions nested within
2728         other structs/unions or within typedefs get a fresh counter; their
2729         tag will be prefixed by a concatenation of their parents' tags)
2730       - bug fixes (related to calculation of transitive closure of types
2731         to be included in the output)
2732    
2733    2. Minor Basis updates:
2734    
2735       - added implementations for List.collate and Option.app
2736    
2737    ----------------------------------------------------------------------
2738    Name: Matthias Blume
2739    Date: 2002/02/11 15:55:00 EST
2740    Tag: blume-20020211-gensym
2741    Description:
2742    
2743    Added a "-gensym" option to command line of ml-nlffigen.  This can be
2744    used to specify a "stem" -- a string that is inserted in all "gensym'd"
2745    names (ML structure names that correspond to unnamed C structs, unions,
2746    and enums), so that separate runs of ml-nlffigen do not clash.
2747    
2748    ----------------------------------------------------------------------
2749    Name: Matthias Blume
2750    Date: 2002/02/11 12:05:00 EST
2751    Tag: blume-20020211-gensml
2752    Description:
2753    
2754    A quick fix for a problem with GenSML (in the pgraph-util library):
2755    Make generation of toplevel "local" optional.  (Strictly speaking,
2756    signature definitions within "local" are not legal SML.)
2757    
2758    Other than that: updates to INSTALL and cm/TODO.
2759    
2760    ----------------------------------------------------------------------
2761    Name: Matthias Blume
2762    Date: 2002/02/08 15:00:00 EST
2763    Tag: blume-20020208-uniquepid
2764    Description:
2765    
2766    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
2767    
2768    1. The installer (config/install.sh) has gotten smarter:
2769    
2770         - Configuration options are a bit easier to specify now
2771           (in config/targets).
2772         - Bug in recognizing .tar.bz2 files fixed.
2773         - Installer automatically resolves dependencies between
2774           configuration options (e.g., if you ask for eXene, you will
2775           also get cml -- regardless whether you asked for it or not).
2776         - Installer can run in "quieter mode" by setting the environment
2777           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
2778           "completely silent", though.
2779         - Build HashCons library as part of smlnj-lib.
2780    
2781    2. A new scheme for assigning persistent identifiers to compilation
2782       units (and, by extension, to types etc.) has been put into place.
2783       This fixes a long-standing bug where types and even dynamic values
2784       can get internally confused, thereby compromising type safety
2785       (abstraction) and dynamic correctness.  See
2786    
2787         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
2788    
2789       for an example of how things could go wrong until now.
2790    
2791       The downside of the new scheme is that pids are not quite as
2792       persistent as they used to be: CM will generate a fresh pid
2793       for every compilation unit that it thinks it sees for the first
2794       time.  That means that if you compile starting from a clean, fresh
2795       source tree at two different times, you end up with different
2796       binaries.
2797    
2798       Cutoff recompilation, however, has not been compromised because
2799       CM keeps pid information in special caches between runs.
2800    
2801    ----------------------------------------------------------------------
2802    Name: Lal George
2803    Date: 2002/02/07 15:34:13 EST 2002
2804    Tag: <none>
2805    Description:
2806    
2807    Compilers that generate assembly code may produce  global labels
2808    whose value is resolved at link time. The various peephole optimization
2809    modules did not take this in account.
2810    
2811    TODO. The Labels.addrOf function should really return an option
2812    type so that clients are forced to deal with this issue, rather
2813    than an exception being raised.
2814    
2815    ----------------------------------------------------------------------
2816    Name: Lal George
2817    Date: 2002/02/06 13:55:02 EST
2818    Tag: george-20020206-ra-breakup
2819    Description:
2820    
2821    1. A bug fix from Allen.
2822    
2823        A typo causes extra fstp %st(0)'s to be generated at compensation
2824        edges, which might cause stack underflow traps at runtime.  This
2825        occurs in fft where there are extraneous fstps right before the 'into'
2826        trap instruction (in this case they are harmless since none of the
2827        integers overflow.)
2828    
2829    2. Pulled out various utility modules that were embedded in the modules
2830       of the register allocator. I need these modules for other purposes, but
2831       they are not complete enough to put into a library (just yet).
2832    ----------------------------------------------------------------------
2833    Name: Matthias Blume
2834    Date: 2002/01/31 16:05:00 EST
2835    Tag: blume-20020131-sparc-ccalls
2836    Description:
2837    
2838    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
2839       of extra stack space by mistake.  Fixed.
2840    
2841    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
2842    
2843    ----------------------------------------------------------------------
2844    Name: Allen Leung
2845    Date: 2002/01/30
2846    Tag: leunga-20020130-nowhere-bug-fix
2847    Description:
2848    
2849       MLRISC bug fixes:
2850       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
2851       2. MachineInt.fromString was negating its value.
2852    
2853    ----------------------------------------------------------------------
2854    Name: Matthias Blume
2855    Date: 2002/01/29
2856    Tag: blume-20020129-INSTALL
2857    Description:
2858    
2859    - Added somewhat detailed installation instructions (file INSTALL).
2860    - Fixed curl-detection bug in config/install.sh.
2861    - It is now possible to select the URL getter using the URLGETTER
2862      environment variable:
2863    
2864          not set / "unknown"      --> automatic detection (script tries wget,
2865                                       curl, and lynx)
2866          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
2867                                       how to properly invoke them)
2868          other                    --> use $URLGETTER directly, it must take
2869                                       precisely two command-line arguments
2870                                       (source URL and destination file name)
2871    
2872    ----------------------------------------------------------------------
2873    Name: Matthias Blume
2874    Date: 2002/01/28
2875    Tag: blume-20020128-sparc-ccalls
2876    Description:
2877    
2878    - Fixed problem with calculation of "used" registers in sparc-c-calls.
2879    - Make use of the allocParam argument in sparc-c-calls.
2880    
2881    ----------------------------------------------------------------------
2882    Name: Matthias Blume
2883    Date: 2002/01/28
2884    Tag: blume-20020128-allocParam
2885    Description:
2886    
2887    John Reppy:  Changes c-calls API to accept client-callback for
2888    allocating extra stack space.
2889    me: Corresponding changes to mlriscGen (using a dummy argument that
2890        does not change the current behavior).
2891    
2892    ----------------------------------------------------------------------
2893    Name: Matthias Blume
2894    Date: 2002/01/28 12:00:00
2895    Tag: Release_110_38
2896    Description:
2897    
2898    This time for real!!!
2899    
2900    ----------------------------------------------------------------------
2901    Name: Matthias Blume
2902    Date: 2002/01/28 10:56:00 EST
2903    Tag: blume-20020128-retraction
2904    Description:
2905    
2906    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
2907       with blume-Release_110_38-retracted.)
2908    
2909    1. Fixed a problem with incorrect rounding modes in real64.sml.
2910       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
2911    
2912    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
2913       and enums fixed.  The naming of corresponding ML identifiers should
2914       now be consistent again.
2915    
2916    ----------------------------------------------------------------------
2917    Name: Allen Leung
2918    Date: 2002/01/27
2919    Tag: leunga-20020127-nowhere
2920    Description:
2921    
2922       Added a target called nowhere in the configuration scripts.
2923       Enabling this will build the MLRISC 'nowhere' tool (for translating
2924       programs with where-clauses into legal SML code) during installation.
2925    
2926    ----------------------------------------------------------------------
2927    Name: Matthias Blume
2928    Date: 2002/01/25 21:27:00 EST
2929    Tag: blume-Release_110_38-retracted
2930    Description:
2931    
2932    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
2933    
2934    README will be added later.
2935    
2936    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
2937    (Release_110_38) removed.  Reason: Last-minute bug fixes.
2938    
2939    ----------------------------------------------------------------------
2940    Name: Matthias Blume
2941    Date: 2002/01/25
2942    Tag: blume-20020125-ffi
2943    Description:
2944    
2945    A large number of tweaks and improvements to ml-nlffi-lib and
2946    ml-nlffigen:
2947    
2948       - ML represenation types have been streamlined
2949       - getter and setter functions work with concrete values, not abstract
2950         ones where possible
2951       - ml-nlffigen command line more flexible (see README file there)
2952       - some bugs have been fixed (hopefully)
2953    
2954    ----------------------------------------------------------------------
2955    Name: Lal George
2956    Date: 2002/01/24
2957    Tag: george-20020124-risc-ra-interface
2958    Description:
2959    
2960       There is a dramatic simplification in the interface to the
2961       register allocator for RISC architectures as a result of making
2962       parallel copy instructions explicit.
2963    
2964    ----------------------------------------------------------------------
2965    Name: Matthias Blume
2966    Date: 2002/01/22
2967    Tag: blume-20020122-x86-ccalls
2968    Description:
2969    
2970    Bug fix for c-calls on x86 (having to do with how char- and
2971    short-arguments are being handled).
2972    
2973    ----------------------------------------------------------------------
2974    Name: Matthias Blume
2975    Date: 2002/01/21
2976    Tag: blume-20020121-ff
2977    Description:
2978    
2979    Another day of fiddling with the FFI...
2980    
2981    1. Bug fix/workaround:  CKIT does not complain about negative array
2982       dimensions, so ml-nlffigen has to guard itself against this possibility.
2983       (Otherwise a negative dimension would send it into an infinite loop.)
2984    
2985    2. Some of the abstract types (light objects, light pointers, most "base"
2986       types) in structure C are now eqtypes.
2987    
2988    3. Added constructors and test functions for NULL function pointers.
2989    
2990    ----------------------------------------------------------------------
2991    Name: Matthias Blume
2992    Date: 2002/01/18
2993    Tag: blume-20020118-ready-for-new-release
2994    Description:
2995    
2996    Made config/srcarchiveurl point to a new place.  (Will provide boot
2997    files shortly.)
2998    
2999    Maybe we christen this to be 110.38?
3000    
3001    ----------------------------------------------------------------------
3002    Name: Matthias Blume
3003    Date: 2002/01/18
3004    Tag: blume-20020118-more-ffifiddle
3005    Description:
3006    
3007    Today's FFI fiddling:
3008    
3009      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
3010        C.Set.* that use concrete (MLRep.*) arguments and results instead
3011        of abstract ones.
3012    
3013      - Provided word-style bit operations etc. for "int" representation
3014        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
3015        and Long.
3016    
3017    ----------------------------------------------------------------------
3018    Name: Matthias Blume
3019    Date: 2002/01/18
3020    Tag: blume-20020118-use-x86-fp
3021    Description:
3022    
3023    Now that x86-fast-fp seems to be working, I turned it back on again
3024    by default.  (Seems to work fine now, even with the FFI.)
3025    
3026    Other than that, I added some documentation about the FFI to
3027    src/ml-nlffigen/README and updated the FFI test examples in
3028    src/ml-nlffi-lib/Tests/*.
3029    
3030    ----------------------------------------------------------------------
3031    Name: Allen Leung
3032    Date: 2002/01/17
3033    Tag: leunga-20020117-x86-fast-fp-call
3034    Description:
3035    
3036       1. Fixed a problem with handling return fp values when x86's fast fp
3037          mode is turned on.
3038    
3039       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
3040          of %f32.
3041    
3042       3. Added a constructor INT32lit to the ast of MLRISC tools.
3043    
3044    ----------------------------------------------------------------------
3045    Name: Matthias Blume
3046    Date: 2002/01/16
3047    Tag: blume-20020116-ffifiddle
3048    Description:
3049    
3050    More fiddling with the FFI interface:
3051    
3052     - Make constness 'c instead of rw wherever possible.  This eliminates
3053       the need for certain explicit coercions.  (However, due to ML's
3054       value polymorphism, there will still be many cases where explicit
3055       coercions are necessary.  Phantom types are not the whole answer
3056       to modeling a subtyping relationship in ML.)
3057    
3058     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
3059    
3060     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
3061       clumsy workaround for varargs problem.)
3062    
3063    ----------------------------------------------------------------------
3064    Name: Lal George
3065    Date: 2002/01/15
3066    Tag: <none>
3067    Description:
3068    
3069    1. Since COPY instructions are no longer native to the architecture,
3070       a generic functor can be used to implement the expandCopies function.
3071    
3072    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
3073       TEXT segment.
3074    
3075    ----------------------------------------------------------------------
3076    Name: Matthias Blume
3077    Date: 2002/01/15
3078    Tag: blume-20020115-ffiupdates
3079    Description:
3080    
3081    1. Fix for bug resulting in single-precision float values being returned
3082       incorrectly from FFI calls.
3083    
3084    2. Small modifications to C FFI API:
3085    
3086        - memory-allocation routines return straight objects (no options)
3087          and raise an exception in out-of-memory situations
3088        - unsafe extensions to cast between function pointers and pointers
3089          from/to ints
3090        - added structure C_Debug as an alternative to structure C where
3091          pointer-dereferencing (|*| and |*!) always check for null-pointers
3092        - added open_lib' to DynLinkage;  open_lib' works like open_lib
3093          but also takes a (possibly empty) list of existing library handles
3094          that the current library depends on
3095    
3096    ----------------------------------------------------------------------
3097    Name: Matthias Blume
3098    Date: 2002/01/10
3099    Tag: blume-20020110-newffigen
3100    Description:
3101    
3102    1. Updates to portable graph code.
3103    
3104    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
3105       more scalable now so that even huge interfaces such as the one
3106       for GTK compile in finite time and space. :-)
3107       See src/ml-nlffigen/README for details on what's new.
3108    
3109    ----------------------------------------------------------------------
3110    Name: Lal George
3111    Date: 2001/01/09 14:31:35 EST 2002
3112    Tag: george-20011206-rm-native-copy
3113    Description:
3114    
3115            Removed the native COPY and FCOPY instructions
3116            from all the architectures and replaced it with the
3117            explicit COPY instruction from the previous commit.
3118    
3119            It is now possible to simplify many of the optimizations
3120            modules that manipulate copies. This has not been
3121            done in this change.
3122    
3123    ----------------------------------------------------------------------
3124    Name: Lal George
3125    Date: 2001/12/06 16:50:13 EST 2001
3126    Tag: george-20011206-mlrisc-instruction
3127    Description:
3128    
3129    Changed the representation of instructions from being fully abstract
3130    to being partially concrete. That is to say:
3131    
3132      from
3133            type instruction
3134    
3135      to
3136            type instr                              (* machine instruction *)
3137    
3138            datatype instruction =
3139                LIVE of {regs: C.cellset, spilled: C.cellset}
3140              | KILL of {regs: C.cellset, spilled: C.cellset}
3141              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
3142              | ANNOTATION of {i: instruction, a: Annotations.annotation}
3143              | INSTR of instr
3144    
3145    This makes the handling of certain special instructions that appear on
3146    all architectures easier and uniform.
3147    
3148    LIVE and KILL say that a list of registers are live or killed at the
3149    program point where they appear. No spill code is generated when an
3150    element of the 'regs' field is spilled, but the register is moved to
3151    the 'spilled' (which is present, more for debugging than anything else).
3152    
3153    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
3154    We used to generate:
3155    
3156            DEFFREG f1
3157            f1 := f2 + f3
3158            trapb
3159    
3160    but now generate:
3161    
3162            f1 := f2 + f3
3163            trapb
3164            LIVE {regs=[f1,f2,f3], spilled=[]}
3165    
3166    Furthermore, the DEFFREG (hack) required that all floating point instruction
3167    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
3168    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
3169    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
3170    this architecture).
3171    
3172    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
3173    all the architectures. This will result in further simplification of the
3174    register allocator that must be aware of them for coalescing purposes, and
3175    will also simplify certain aspects of the machine description that provides
3176    callbacks related to parallel copies.
3177    
3178    ANNOTATION should be obvious, and now INSTR represents the honest to God
3179    machine instruction set!
3180    
3181    The <arch>/instructions/<arch>Instr.sml files define certain utility
3182    functions for making porting easier -- essentially converting upper case
3183    to lower case. All machine instructions (of type instr) are in upper case,
3184    and the lower case form generates an MLRISC instruction. For example on
3185    the alpha we have:
3186    
3187      datatype instr =
3188         LDA of {r:cell, b:cell, d:operand}
3189       | ...
3190    
3191      val lda : {r:cell, b:cell, d:operand} -> instruction
3192        ...
3193    
3194    where lda is just (INSTR o LDA), etc.
3195    
3196    ----------------------------------------------------------------------
3197    Name: Matthias Blume
3198    Date: 2001/11/22 21:40:00 EST
3199    Tag: Release_110_37
3200    Description:
3201    
3202    Release 110.37.  This time for real.
3203    
3204    ----------------------------------------------------------------------
3205    Name: Matthias Blume
3206    Date: 2001/11/21 16:35:00 EST
3207    Tag: blume-20011121-foot-in-mouth
3208    Description:
3209    
3210    Removed the "Release_110_37" tag because of a serious bug.
3211    This will be re-tagged once the bug is fixed.
3212    
3213    ----------------------------------------------------------------------
3214    Name: Matthias Blume
3215    Date: 2001/11/21 16:14:00 EST
3216    Tag: blume-20011121-forgottenfile
3217    Description:
3218    
3219    Forgot to add a file.  (Just a .tex-file -- part of
3220    the CM manual source.)
3221    
3222    ----------------------------------------------------------------------
3223    Name: Matthias Blume
3224    Date: 2001/11/21 16:10:00 EST
3225    Tag: blume-20011121-invalid_110_37
3226    Description:
3227    
3228    Note: I removed the original tag "Release_110_37" from this commit
3229          because we found a serious bug in all non-x86 backends.
3230          - Matthias
3231    
3232    1. Modifications to the SML/NJ code generator and to the runtime system
3233       so that code object name strings are directly inserted into code
3234       objects at code generation time.  The only business the runtime system
3235       has with this is now to read the name strings on occasions.
3236       (The encoding of the name string has also changed somewhat.)
3237    
3238    2. CM now implements a simple "set calculus" for specifying export lists.
3239       In particular, it is now possible to refer to the export lists of
3240       other libraries/groups/sources and form unions as well as differences.
3241       See the latest CM manual for details.
3242    
3243    3. An separate notion of "proxy" libraries has again be eliminated from
3244       CM's model.  (Proxy libraries are now simply a special case of using
3245       the export list calculus.)
3246    
3247    4. Some of the existing libraries now take advantage of the new set
3248       calculus.
3249       (Notice that not all libraries have been converted because some
3250       of the existing .cm-files are supposed to be backward compatible
3251       with 110.0.x.)
3252    
3253    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
3254       "General.exnMessage"!  The former relies on a certain hook to be
3255       initialized, and that often does not happen in the stand-alone case.)
3256    
3257    ----------------------------------------------------------------------
3258    Name: Lal George
3259    Date: 2001/11/21  13:56:18 EST
3260    Tag: george-2001121-pseudo-ops
3261    Description:
3262    
3263      Implemented a complete redesign of MLRISC pseudo-ops. Now there
3264      ought to never be any question of incompatabilities with
3265      pseudo-op syntax expected by host assemblers.
3266    
3267      For now, only modules supporting GAS syntax are implemented
3268      but more should follow, such as MASM, and vendor assembler
3269      syntax, e.g. IBM as, Sun as, etc.
3270    
3271    ----------------------------------------------------------------------
3272    Name: Matthias Blume
3273    Date: 2001/11/14 11:52:00 EST
3274    Tag: blume-20011114-srcname
3275    Description:
3276    
3277    1. Routed the name of the current source file to mlriscgen where it
3278       should be directly emitted into the code object.  (This last part
3279       is yet to be done.)
3280    
3281    2. Some cleanup of the pgraph code to make it match the proposal that
3282       I put out the other day.  (The proposal notwithstanding, things are
3283       still in flux here.)
3284    
3285    ----------------------------------------------------------------------
3286    Name: Lal George
3287    Date: 2001/11/14 09:44:04 EST
3288    Tag:
3289    Description:
3290    
3291      Fix for a backpatching bug reported by Allen.
3292    
3293      Because the boundary between short and long span-dependent
3294      instructions is +/- 128, there are an astounding number of
3295      span-dependent instructions whose size is over estimated.
3296    
3297      Allen came up with the idea of letting the size of span
3298      dependent instructions be non-monotonic, for a maxIter
3299      number of times, after which the size must be monotonically
3300      increasing.
3301    
3302      This table shows the number of span-dependent instructions
3303      whose size was over-estimated as a function of maxIter, for the
3304      file Parse/parse/ml.grm.sml:
3305    
3306         maxIter            # of instructions:
3307            10                      687
3308            20                      438
3309            30                      198
3310            40                        0
3311    
3312      In compiling the compiler, there is no significant difference in
3313      compilation speed between maxIter=10 and maxIter=40. Actually,
3314      my measurements showed that maxIter=40 was a tad faster than
3315      maxIter=10! Also 96% of the  files in the compiler reach a fix
3316      point within 13 iterations, so fixing maxIter at 40, while high,
3317      is okay.
3318    
3319    ----------------------------------------------------------------------
3320    Name: Matthias Blume
3321    Date: 2001/10/31 15:25:00 EST
3322    Tag: blume-20011031-pgraph
3323    Description:
3324    
3325    CKIT:
3326    * Changed the "Function" constructor of type Ast.ctype to carry optional
3327      argument identifiers.
3328    * Changed the return type of TypeUtil.getFunction accordingly.
3329    * Type equality ignores the argument names.
3330    * TypeUtil.composite tries to preserve argument names but gives up quickly
3331      if there is a mismatch.
3332    
3333    installation script:
3334    * attempts to use "curl" if available (unless "wget" is available as well)
3335    
3336    CM:
3337    * has an experimental implementation of "portable graphs" which I will
3338      soon propose as an implementation-independent library format
3339    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
3340    
3341    NLFFI-LIB:
3342    * some cleanup (all cosmetic)
3343    
3344    NLFFIGEN:
3345    * temporarily disabled the mechanism that suppresses ML output for
3346      C definitions whose identifiers start with an underscore character
3347    * generate val bindings for enum constants
3348    * user can request that only one style (light or heavy) is being used;
3349      default is to use both (command-line arguments: -heavy and -light)
3350    * fixed bug in handling of function types involving incomplete pointers
3351    * generate ML entry points that take record arguments (i.e., using
3352      named arguments) for C functions that have a prototype with named
3353      arguments
3354      (see changes to CKIT)
3355    
3356    ----------------------------------------------------------------------
3357    Name: Allen Leung
3358    Date: 2001/10/27 20:34:00 EDT
3359    Tag: leunga-20011027-x86-fast-fp-call
3360    Description:
3361    
3362       Fixed the bug described in blume-20010920-slowfp.
3363    
3364       The fix involves
3365          1. generating FCOPYs in FSTP in ia32-svid
3366          2. marking a CALL with the appropriate annotation
3367    
3368    ----------------------------------------------------------------------
3369    Name: Matthias Blume
3370    Date: 2001/10/16 11:32:00 EDT
3371    Tag: blume-20011016-netbsd
3372    Description:
3373    
3374    Underscore patch from Chris Richards (fixing problem with compiling
3375    runtime system under recent NetBSD).
3376    
3377    ----------------------------------------------------------------------
3378    Name: Allen Leung
3379    Date: 2001/10/12 17:18:32 EDT 2001
3380    Tag: leung-20011012-x86-printflowgraph
3381    Description:
3382    
3383    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
3384    
3385    ----------------------------------------------------------------------
3386    Name: Lal George
3387    Date: 2001/10/11 23:51:34 EDT
3388    Tag: george-20011011-too-many-instrs
3389    Description:
3390    
3391    The representation of a program point never expected to see more
3392    than 65536 instructions in a basic block!
3393    
3394    ----------------------------------------------------------------------
3395    Name: Lal George
3396    Date: 2001/10/09 09:41:37 EDT
3397    Tag: george-20011008-mlrisc-labels
3398    Description:
3399    
3400    Changed the machine description files to support printing of
3401    local and global labels in assembly code, based on host assembler
3402    conventions.
3403    
3404    ----------------------------------------------------------------------
3405    Name: Matthias Blume
3406    Date: 2001/09/25 15:25:00 EDT
3407    Tag: blume-20010925-exninfo
3408    Description:
3409    
3410    I provided a non-hook implementation of exnName (at the toplevel) and
3411    made the "dummy" implementation of exnMessage (at the toplevel) more
3412    useful: if nothing gets "hooked in", then at least you are going to
3413    see the exception name and a message indicating why you don't see more.
3414    
3415    [For the time being, programs that need exnMessage and want to use
3416    ml-build should either use General.exnMessage (strongly recommended) or
3417    refer to structure General at some other point so that CM sees a
3418    static dependency.]
3419    
3420    [Similar remarks go for "print" and "use":  If you want to use their
3421    functionality in stand-alone programs generated by ml-build, then use
3422    TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
3423    
3424    ----------------------------------------------------------------------
3425    Name: Matthias Blume
3426    Date: 2001/09/20 17:28:00 EDT
3427    Tag: blume-20010920-slowfp
3428    Description:
3429    
3430    Allen says that x86-fast-fp is not safe yet, so I turned it off again...
3431    
3432    ----------------------------------------------------------------------
3433    Name: Matthias Blume
3434    Date: 2001/09/20 17:20:00 EDT
3435    Tag: blume-20010920-canonicalpaths
3436    Description:
3437    
3438    0. Updated the BOOT file (something that I forgot to do earlier).
3439    
3440    1. Small internal change to CM so that it avoids "/../" in filenames
3441       as much as possible (but only where it is safe).
3442    
3443    2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
3444       that arguments that contain delimiters are passed through correctly.
3445       This change also means that all "special" arguments of the form
3446       @SMLxxx... must come first.
3447    
3448    3. Changed install script to put relative anchor names for tool commands
3449       into pathconfig.
3450    
3451    ----------------------------------------------------------------------
3452    Name: Matthias Blume
3453    Date: 2001/09/18 15:35:00 EDT
3454    Tag: blume-20010918-readme11036
3455    Description:
3456    
3457    Added README files.
3458    
3459    ----------------------------------------------------------------------
3460    Name: Matthias Blume
3461    Date: 2001/09/18 11:45:00 EDT
3462    Tag: Release_110_36 (retag)
3463    Description:
3464    
3465    Fixed mistake in config/preloads. Retagged as 110.36.
3466    
3467    ----------------------------------------------------------------------
3468    Name: Matthias Blume
3469    Date: 2001/09/18 09:40:00 EDT
3470    Tag: Release_110_36_orig (tag changed)
3471    Description:
3472    
3473    New version (110.36).  New bootfiles.
3474    
3475    ----------------------------------------------------------------------
3476    Name: Matthias Blume
3477    Date: 2001/09/14 16:15:00 EDT
3478    Tag: blume-20010914-x86fastfp
3479    Description:
3480    
3481    John committed some changes that Allen made, in particular a (hopefully)
3482    correctly working version of the x86-fp module.
3483    
3484    I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
3485    flag to "true".  Everything seems to compile to a fixpoint ok, and
3486    "mandelbrot" speeds up by about 15%.
3487    
3488    ----------------------------------------------------------------------
3489    Name: Matthias Blume
3490    Date: 2001/09/13 11:20:00 EDT
3491    Tag: blume-20010913-minimal
3492    Description:
3493    
3494    1. Stefan Monnier's patch to fix a miscompilation problem that
3495       was brought to light by John Reppy's work on Moby.
3496    
3497    2. Implemented a minimal "structure Compiler" that contains just
3498       "version" and "architecture".  The minimal version will be
3499       available when the full version is not.  This is for backward-
3500       compatibility with code that wants to test Compiler.version.
3501    
3502    ----------------------------------------------------------------------
3503    Name: Matthias Blume
3504    Date: 2001/08/28 14:03:00 EDT
3505    Tag: blume-20010828-ml-lex
3506    Description:
3507    
3508    Fix for bug 1581, received from Neophytos Michael.
3509    
3510    ----------------------------------------------------------------------
3511    Name: Matthias Blume
3512    Date: 2001/08/27 11:20:00 EDT
3513    Tag: blume-20010827-readme11035
3514    Description:
3515    
3516    Fleshed out the README file for 110.35.
3517    
3518    ----------------------------------------------------------------------
3519    Name: Matthias Blume
3520    Date: 2001/08/24 17:10:00 EDT
3521    Tag: Release_110_35
3522    Description:
3523    
3524    New version number (110.35).  New bootfiles.
3525    
3526    ----------------------------------------------------------------------
3527    Name: Lal George
3528    Date: 2001/08/24 13:47:18 EDT 2001
3529    Tag: george-20010824-MLRISC-graphs
3530    Description:
3531    
3532     removed clusters from MLRISC completely and replaced with graphs.
3533    
3534    ----------------------------------------------------------------------
3535    Name: Matthias Blume
3536    Date: 2001/08/23 17:50:00 EDT
3537    Tag: blume-20010823-toplevel
3538    Description:
3539    
3540    - some reorganization of the code that implements various kinds of
3541      environments in the compiler (static, dynamic, symbolic, combined)
3542    - re-implemented the EnvRef module so that evalStream works properly
3543      (if the stream contains references to "use", "CM.make", etc.)
3544    - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
3545    
3546    ----------------------------------------------------------------------
3547    Name: Matthias Blume
3548    Date: 2001/08/20 15:50 EDT
3549    Tag: blume20010820-slipup
3550    Description:
3551    
3552    I forgot to commit a few files.  Here they are...
3553    
3554    ----------------------------------------------------------------------
3555    Name: Matthias Blume
3556    Date: 2001/08/20 15:35:00 EDT
3557    Tag: blume-20010820-debugprof
3558    Description:
3559    
3560    !!!! NEW BOOTFILES !!!!
3561    
3562    This is another round of reorganizing the compiler sources.  This
3563    time the main goal was to factor out all the "instrumentation"
3564    passes (for profiling and backtracing) into their own library.
3565    The difficulty was to do it in such a way that it does not depend
3566    on elaborate.cm but only on elabdata.cm.
3567    
3568    Therefore there have been further changes to both elaborate.cm and
3569    elabdata.cm -- more "generic" things have been moved from the former
3570    to the latter.  As a result, I was forced to split the assignment
3571    of numbers indicating "primtyc"s into two portions: SML-generic and
3572    SML/NJ-specific.  Since it would have been awkward to maintain,
3573    I bit the bullet and actually _changed_ the mapping between these
3574    numbers and primtycs.  The bottom line of this is that you need
3575    a new set of bin- and bootfiles.
3576    
3577    I have built new bootfiles for all architectures, so doing a fresh
3578    checkout and config/install.sh should be all you need.
3579    
3580    The newly created library's name is
3581    
3582        $smlnj/viscomp/debugprof.cm
3583    
3584    and its sources live under
3585    
3586        src/compiler/DebugProf
3587    
3588    ----------------------------------------------------------------------
3589    Name: Matthias Blume
3590    Date: 2001/08/15 17:15:00 EDT
3591    Tag: blume-20010815-compreorg
3592    Description:
3593    
3594    This is a first cut at reorganizing the CM libraries that make up the
3595    core of the compiler.  The idea is to separate out pieces that could
3596    be used independently by tools, e.g., the parser, the typechecker, etc.
3597    
3598    The current status is a step in this direction, but it is not quite
3599    satisfactory yet.  Expect more changes in the future.
3600    
3601    Here is the current (new) organization...
3602    
3603        What used to be $smlnj/viscomp/core.cm is now divided into
3604        six CM libraries:
3605    
3606             $smlnj/viscomp/basics.cm
3607                           /parser.cm
3608                           /elabdata.cm
3609                           /elaborate.cm
3610                           /execute.cm
3611                           /core.cm
3612    
3613        The CM files for these libraries live under src/system/smlnj/viscomp.
3614        All these libraries are proxy libraries that contain precisely
3615        one CM library component.  Here are the locations of the components
3616        (all within the src/compiler tree):
3617    
3618             Basics/basics.cm
3619             Parse/parser.cm
3620             ElabData/elabdata.cm
3621             Elaborator/elaborate.cm
3622             Execution/execute.cm
3623             core.cm
3624    
3625         [This organization is the same that has been used already
3626         for a while for the architecture-specific parts of the visible
3627         compiler and for the old version of core.cm.]
3628    
3629         As you will notice, many source files have been moved from their
3630         respective original locations to a new home in one of the above
3631         subtrees.
3632    
3633         The division of labor between the new libraries is the following:
3634    
3635             basics.cm:
3636                - Simple, basic definitions that pertain to many (or all) of
3637                  the other libraries.
3638             parser.cm:
3639                - The SML parser, producing output of type Ast.dec.
3640                - The type family for Ast is also defined and exported here.
3641             elabdata.cm:
3642                - The datatypes that describe input and output of the elaborator.
3643                  This includes types, absyn, and static environments.
3644             elaborator.cm:
3645                - The SML/NJ type checker and elaborator.
3646                  This maps an Ast.dec (with a given static environment) to
3647                  an Absyn.dec (with a new static environment).
3648                - This libraries implements certain modules that used to be
3649                  structures as functors (to remove dependencies on FLINT).
3650             execute.cm:
3651                - Everything having to do with executing binary code objects.
3652                - Dynamic environments.
3653             core.cm:
3654                - SML/NJ-specific instantiations of the elaborator and MLRISC.
3655                - Top-level modules.
3656                - FLINT (this should eventually become its own library)
3657    
3658    Notes:
3659    
3660    I am not 100% happy with the way I separated the elaborator (and its
3661    data structures) from FLINT.  Two instances of the same problem:
3662    
3663        1. Data structures contain certain fields that carry FLINT-specific
3664           information.  I hacked around this using exn and the property list
3665           module from smlnj-lib.  But the fact that there are middle-end
3666           specific fields around at all is a bit annoying.
3667    
3668        2. The elaborator calculates certain FLINT-related information.  I tried
3669           to make this as abstract as I could using functorization, but, again,
3670           the fact that the elaborator has to perform calculations on behalf
3671           of the middle-end at all is not nice.
3672    
3673        3. Having to used exn and property lists is unfortunate because it
3674           weakens type checking.  The other alternative (parameterizing
3675           nearly *everything*) is not appealing, though.
3676    
3677    I removed the "rebinding =" warning hack because due to the new organization
3678    it was awkward to maintain it.  As a result, the compiler now issues some of
3679    these warnings when compiling init.cmi during bootstrap compilation. On
3680    the plus side, you also get a warning when you do, for example:
3681       val op = = Int32.+
3682    which was not the case up to now.
3683    
3684    I placed "assign" and "deref" into the _Core structure so that the
3685    code that deals with the "lazy" keyword can find them there.  This
3686    removes the need for having access to the primitive environment
3687    during elaboration.
3688    
3689    ----------------------------------------------------------------------
3690    Name: Matthias Blume
3691    Date: 2001/08/13
3692    Tag: blume-20010813-closures
3693    Description:
3694    
3695    This fix was sent to us by Zhong Shao.  It is supposed to improve the
3696    performance of certain loops by avoiding needless closure allocation.
3697    
3698    ----------------------------------------------------------------------
3699    Name: Lal George
3700    Date: 2001/07/31 10:03:23 EDT 2001
3701    Tag: george-20010731-x86-fmalloc
3702    Description: Fixed bug in x86 calls
3703    
3704        There was a bug where call instructions would mysteriously
3705        vanish. The call instruction had to be one that returned
3706        a floating point value.
3707    
3708    ----------------------------------------------------------------------
3709    Name: Lal George
3710    Date: 2001/07/19 16:36:29 EDT 2001
3711    Tag: george-20010719-simple-cells
3712    Description:
3713    
3714    I have dramatically simplified the interface for CELLS in MLRISC.
3715    
3716    In summary, the cells interface is broken up into three parts:
3717    
3718      1. CellsBasis : CELLS_BASIS
3719    
3720            CellsBasis is a top level structure and common for all
3721            architectures.  it contains the definitions of basic datatypes
3722            and utility  functions over these types.
3723    
3724      2. functor Cells() : CELLS
3725    
3726            Cells generates an interface for CELLS that incorporates the
3727            specific resources on the target architecture, such as the
3728            presence of special register classes, their number and size,
3729            and various useful substructures.
3730    
3731      3. <ARCH>CELLS
3732    
3733            e.g. SparcCells: SPARCCELLS
3734    
3735            <ARCH>CELLS usually contains additional bindings for special
3736            registers  on the architecture, such as:
3737    
3738                    val r0 : cell           (* register zero *)
3739                    val y : cell            (* Y register *)
3740                    val psr : cell          (* processor status register *)
3741                    ...
3742    
3743            The structure returned by applying the Cells functor is opened
3744            in this interface.
3745    
3746    The main implication of all this is that the datatypes for cells is
3747    split between CellsBasis and CELLS -- a fairly simple change for user
3748    code.
3749    
3750    In the old scheme the CELLS interface had a definitional binding of
3751    the form:
3752    
3753            signature CELLS = sig
3754    
3755               structure CellsBasis = CellsBasis
3756    
3757               ...
3758    
3759            end
3760    
3761    With all the sharing constraints that goes on in MLRISC, this old
3762    design  quickly leads to errors such as:
3763    
3764            "structure definition spec inside of sharing ... "
3765    
3766    
3767    and appears to require an unacceptable amount of sharing and where
3768    constraint hackery.
3769    
3770    I think this error message (the interaction of definitional specs and
3771    sharing) requires more explanation on our web page.
3772    
3773    ----------------------------------------------------------------------
3774    Name: Matthias Blume
3775    Date: 2001/07/19 15:00:00 EDT
3776    Tag: blume-20010719-libreorg
3777    Description:
3778    
3779    This update puts together a fairly extensive but straightforward change
3780    to the way the libraries that implement the interactive system are
3781    organized:
3782    
3783       The biggest change is the elimination of structure Compiler.  As a
3784       replacement for this structure, there is now a CM library
3785       (known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
3786       that exports all the substructures of the original structure Compiler
3787       directly.  So instead of saying Compiler.Foo.bar one now simply
3788       says Foo.bar.  (The CM libraries actually export a collection of
3789       structures that is richer than the collection of substructures of
3790       structure Compiler.)
3791    
3792       To make the transition smooth, there is a separate library called
3793       $smlnj/compiler/compiler.cm which puts together and exports the
3794       original structure Compiler (or at least something very close to it).
3795    
3796       There are five members of the original structure Compiler
3797       that are not exported directly but which instead became members
3798       of a new structure Backend (described by signature BACKEND).  These are:
3799       structure Profile (: PROFILE), structure Compile (: COMPILE), structure
3800       Interact (: INTERACT), structure Machine (: MACHINE), and val
3801       architecture (: string).
3802    
3803       Structure Compiler.Version has become structure CompilerVersion.
3804    
3805       Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
3806       by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
3807       or x86, respectively.
3808       Each of these exports the same frontend structures that
3809       $smlnj/compiler.cm exports.  But they do not have a structure Backend
3810       and instead export some structure <Arch>Backend where <Arch> is Alpha32,
3811       Hppa, PPC, Sparc, or X86, respectively.
3812    
3813       Library $smlnj/compiler/all.cm exports the union of the exports of
3814       $smlnj/compiler/<arch>.cm
3815    
3816       There are no structures <Arch>Compiler anymore, use
3817       $smlnj/compiler/<arch>.cm instead.
3818    
3819       Library host-compiler-0.cm is gone.  Instead, the internal library
3820       that instantiates CM is now called cm0.cm.  Selection of the host
3821       compiler (backend) is no longer done here but. (Responsibility for it
3822       now lies with $smlnj/compiler/current.cm.  This seems to be more
3823       logical.)
3824    
3825       Many individual files have been moved or renamed.  Some files have
3826       been split into multiple files, and some "dead" files have been deleted.
3827    
3828    Aside from these changes to library organization, there are also changes
3829    to the way the code itself is organized:
3830    
3831       Structure Binfile has been re-implemented in such a way that it no
3832       longer needs any knowledge of the compiler.  It exclusively deals
3833       with the details of binfile layout.  It no longer invokes the
3834       compiler (for the purpose of creating new prospective binfile
3835       content), and it no longer has any knowledge of how to interpret
3836       pickles.
3837    
3838       Structure Compile (: COMPILE) has been stripped down to the bare
3839       essentials of compilation.  It no longer deals with linking/execution.
3840       The interface has been cleaned up considerably.
3841    
3842       Utility routines for dealing with linking and execution have been