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