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