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 977, Wed Nov 14 16:53:16 2001 UTC revision 1515, Fri Jun 18 19:29:02 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: 2004/06/18 14:30:00 CDT
18    Tag: blume-20040618-unix
19    Description:
20    
21    Changed the implementation of structure Unix so that the same stream
22    is returned every time one of the {text,bin}{In,Out}streamOf functions
23    is invoked on the same proc.  This is not what the spec currently
24    says -- although IMO it arguably should.  (See discussion below.)
25    
26    ----------------------------------------------------------------------
27    Name: Matthias Blume (blume (at) tti - c (dot) org)
28    Date: 2004/06/17 18:15:00 CDT
29    Tag: Release_110_46
30    Description:
31    
32    New working version (110.46).  NEW BOOTFILES!
33    
34    ----------------------------------------------------------------------
35    Name: Matthias Blume (blume (at) tti - c (dot) org)
36    Date: 2004/06/17 17:20:00 CDT
37    Tag: blume-20040617-timer-unix
38    Description:
39    
40    Changed the interface of structures Timer and Unix to match the most
41    recent Basis spec.
42    
43    In the case of Unix there still seems to be an open/weird issue:
44    
45       The {text,bin}{In,Out}streamOf functions are supposed to create
46       fresh streams whenever they are called -- as opposed to have them
47       return the same stream every time.  This design is supposed to
48       prevent space leaks caused by proc values hanging on to streams.
49    
50       The reap function, on the other hand, is supposed to close the
51       streams.  This cannot be done without having a handle on the
52       stream in proc after all...
53    
54    I took the liberty to implement the following stopgap solution:
55    
56      The proc value hangs on to the most recently created stream(s).
57      Reap closes those.  If either or both of the two streams hadn't
58      been created at all yet, then reap will close the corresponding
59      file descriptors directly.
60    
61    PS: I don't understand the original space leak argument anymore.  If
62    a proc hangs on to the imperative stream, then I/O operations on those
63    will advance the state of the cached stream and avoid the space leak.
64    
65    ----------------------------------------------------------------------
66    Name: Matthias Blume (blume (at) tti - c (dot) org)
67    Date: 2004/05/28 16:45:00 CDT
68    Tag: blume-20040528-basis
69    Description:
70    
71    Added signature PACK_REAL and exported functor PrimIO.
72    
73    ----------------------------------------------------------------------
74    Name: Matthias Blume (blume (at) tti - c (dot) org)
75    Date: 2004/05/25 16:00:00 CDT
76    Tag: blume-20040525-group-owner
77    Description:
78    
79    CM now ignores (but still accepts) the "owner" information in group
80    descriptions.  The owner of a group is its next enclosing
81    library. Each group must have a unique owner.  (There is a virtual
82    "toplevel" library that owns groups which are not nested within a real
83    library.)  Previously, each group had to explicitly declare its owner,
84    and CM would check that such a declaration is correct.  The new scheme
85    is to have CM check that for each group there is precisely one owning
86    library.
87    
88    The advantage of the new scheme is that the programmer no longer needs
89    to maintain the somewhat annoying owner information.  The downside is
90    that CM cannot enforce the ownership rule across multiple runs of
91    CM.make.  Fortunately, enclosing the same group in two different
92    libraries A and B which are not part of the same program does not
93    cause real problems.
94    
95    ----------------------------------------------------------------------
96    Name: Matthias Blume (blume (at) tti - c (dot) org)
97    Date: 2004/05/20 16:00:00 CDT
98    Tag: blume-20040520-win32
99    Description:
100    
101    Made the win32 version work again.  (Strangely, a misplaced comma had
102    slipped into win32-process.c which prevented the runtime from being
103    compiled correctly.)
104    
105    Also, included a minor addition to ml-build.bat analogous to what was
106    done in blume-20040519-ml-build.
107    
108    ----------------------------------------------------------------------
109    Name: Matthias Blume (blume (at) tti - c (dot) org)
110    Date: 2004/05/19 22:10:00 CDT
111    Tag: blume-20040519-ml-build
112    Description:
113    
114    Arranged for ml-build to clean up after itself a little bit better.
115    The script generates a temporary SML source file and compiles it using
116    CM, so CM generates metadata (GUID, SKEL, objectfile) for it.  It now
117    gets rid of those at the end, so they don't accumulate under .cm.
118    
119    This required a minor change to install.sh because the name of the
120    metadata directory (default: .cm) is actually configurable at
121    installation time.
122    
123    ----------------------------------------------------------------------
124    Name: Matthias Blume (blume (at) tti - c (dot) org)
125    Date: 2004/05/18 15:50:00 CDT
126    Tag: blume-20040518-mkreader
127    Description:
128    
129    Added Posix.IO.mk{Bin,Text}{Reader,Writer} by lifting their respective
130    implementations from internal modules PosixBinPrimIO and PosixTextPrimIO.
131    
132    ----------------------------------------------------------------------
133    Name: Matthias Blume (blume (at) tti - c (dot) org)
134    Date: 2004/05/11 14:35:00 CDT
135    Tag: blume-20040511-win32sock
136    Description:
137    
138    Added previously missing support for many socket-related functions
139    under win32.  Thanks to David Hansel <hansel@reactive-systems.com>
140    for the voluminous patch!
141    
142    (I have not tested this patch under win32 yet.)
143    
144    Here is David's e-mail:
145    
146    Hi,
147    
148    Attached to this email you find a diff against sml/nj 110.45
149    that will enable socket support under Windows.
150    
151    To apply the patch (using unix or cygwin)
152    1) gunzip runtime.diff.gz
153    2) "cd" into "src/runtime" in the source tree of a fresh
154       110.45 installation.
155    3) patch -p 1 < [your/path/to]runtime.diff
156    
157    The code compiles fine but has NOT yet been extensively tested.
158    I only ran a few tests for basic socket client functionality
159    (which worked fine).  Especially the functions that use ioctl
160    are not tested at all and might not work (see below).
161    
162    I implemented this since we want to move to a newer version of sml/nj
163    but need socket support in order to use it.  This is the first time I
164    even had a look at the sml/nj source,  so please review my changes
165    before making this part of the distribution!  Here are a few issues
166    that I think might be better for someone to solve who is more
167    familiar with the sml/nj source (and socket programming):
168    
169    - getnetbyaddr.c and getnetbyname.c will raise a "not implemented"
170      exception since I could not figure out what the windows equivalent
171      of these functions is
172    
173    - In sockets-osdep.h there are a some #include statements that are
174      only used in a few files that include sockets-osdep.h
175    
176    - In smlnj-sock-lib.c, function init_fn() calls WSAStartup() but
177      does not process its return value since I don't know how to
178      report an error upwards.
179    
180    - It would probably be good to have a call to WSACleanup() when
181      the library is unloaded (if there is such a possibility).
182      Otherwise I think Windows will take care of this automatically
183      when the process finishes.
184    
185    - I used ioctlsocket() as a replacement for ioctl() but I have
186      no idea if that is actually the proper replacement on Windows.
187    
188    - All these issues are marked in the code by "FIXME" comments.
189    
190    We use sml/nj extensively in our products and are quite happy
191    with it.  I hope this contribution will help you.
192    
193    Keep up the good work!
194    
195    David
196    
197    ----------------------------------------------------------------------
198    Name: Matthias Blume (blume (at) tti - c (dot) org)
199    Date: 2004/05/11 14:20:00 CDT
200    Tag: blume-20040511-installml
201    Description:
202    
203    Fixed two bugs in installml script.  (Thanks to Vesa A. Norrman for
204    the patch.)
205    
206    ----------------------------------------------------------------------
207    Name: Matthias Blume (blume (at) tti - c (dot) org)
208    Date: 2004/05/11 14:05:00 CDT
209    Tag: blume-20040511-nlffi-netbsd
210    Description:
211    
212    Added support for nlffi under netbsd.  (Thanks to Vesa A. Norrman for
213    the patch.)
214    
215    ----------------------------------------------------------------------
216    Name: Matthias Blume (blume (at) tti - c (dot) org)
217    Date: 2004/05/11 12:05:00 CDT
218    Tag: blume-20040511-exports
219    Description:
220    
221    As per request by Adam Chlipala <adam@hcoop.net>, extended various
222    export lists in compiler-related .cm-files.
223    
224    ----------------------------------------------------------------------
225    Name: Matthias Blume (blume (at) tti - c (dot) org)
226    Date: 2004/05/11 11:35:00 CDT
227    Tag: blume-20040511-allsource
228    Description:
229    
230    The installer now honors the "src-smlnj" target again, although its meaning
231    has changed from "all sources required for the compiler" to "all sources
232    the installer knows about".  In other words, if you enable "src-smlnj"
233    in the "targets" file, then the installer will pull in sources for
234    everything.  (Notice that this refers to source code only.  Compiled
235    code is still only installed for modules that were requested explicitly
236    or which are required for other modules that were requested explicitly.)
237    
238    ----------------------------------------------------------------------
239    Name: Matthias Blume (blume (at) tti - c (dot) org)
240    Date: 2004/04/23 17:40:00 CDT
241    Tag: blume-20040423-ieee-scan
242    Description:
243    
244    Fixed IEEEReal.scan (and .fromString) so that if there is an overflow
245    in the exponent calculation we get INF or ZERO (depending on the mantissa
246    and the sign of the exponent).
247    
248    ----------------------------------------------------------------------
249    Name: Matthias Blume (blume (at) tti - c (dot) org)
250    Date: 2004/04/23 10:40:00 CDT
251    Tag: blume-20040423-ml-build
252    Description:
253    
254    The ml-build script now terminates with a non-0 status when something
255    goes wrong.
256    
257    ----------------------------------------------------------------------
258    Name: Matthias Blume (blume (at) tti - c (dot) org)
259    Date: 2004/04/22 16:35:00 CDT
260    Tag: blume-20040422-Option
261    Description:
262    
263    Made exception Option to be the same as exception Option.Option
264    (as it should be).
265    
266    ----------------------------------------------------------------------
267    Name: Allen Leung (leunga (at) reservoir (dot) com)
268    Date: 2004/03/19 14:40:00 EST
269    Tag: leunga-20040319-cygwin-nlffi
270    Description:
271    
272    Fixed the runtime so that ml-nlffi-lib runs on the cygwin version
273    of SML/NJ.  The problem is that
274    
275       lib = dlopen(NULL, ...)
276       f   = dlsym(lib, "malloc");
277    
278    does not work on Windows unless we explicitly export symbols
279    such as 'malloc' during linking.  We fixed this by explicitly
280    exporting the required symbols with the magic gcc incantation:
281    
282        -Wl,--export-all cygwin.def
283    
284    where cygwin.def is a file containing all the symbols that
285    we wish to export.
286    
287    I suspect this is a Windows problem and we'll have to
288    do the same (somehow with windows compilers) when
289    we build the native win32 version with the system
290    calls LoadLibrary/GetProcAddress.
291    
292    ----------------------------------------------------------------------
293    Name: Matthias Blume (blume (at) tti - c (dot) org)
294    Date: 2004/03/04 16:35:00 CST
295    Tag: blume-20040304-intinf-fmt
296    Description:
297    
298    Fixed problem with IntInf.fmt (sign would show up on the right instead
299    of on the left for BIN, OCT, and HEX).
300    
301    ----------------------------------------------------------------------
302    Name: Matthias Blume (blume (at) tti - c (dot) org)
303    Date: 2004/03/04 11:25:00 CST
304    Tag: blume-20040304-symlinks
305    Description:
306    
307    Fixed problem with installer script (unix only) where bin/ml-yacc and
308    friends pointed (via symlinks) to absolute locations instead of just
309    .run-sml.  This was reported by Vesa A Norrman.
310    
311    ----------------------------------------------------------------------
312    Name: Matthias Blume (blume (at) tti - c (dot) org)
313    Date: 2004/02/13 14:50:00 CST
314    Tag: Release_110_45
315    Description:
316    
317    New working version (110.45). New bootfiles.
318    
319    ----------------------------------------------------------------------
320    Name: Matthias Blume (blume (at) tti - c (dot) org)
321    Date: 2004/01/26 15:15:15 CST
322    Tag: blume-20040126-toplevel
323    Description:
324    
325    Improved handling of exceptions at the interactive toplevel.
326    
327    ----------------------------------------------------------------------
328    Name: Matthias Blume (blume (at) tti - c (dot) org)
329    Date: 2004/01/26 11:25:00
330    Tag: blume-20040126-app
331    Description:
332    
333    Type of top-level "app" corrected.
334    Added code for setting vp_limitPtrMask to Win32-specific runtime.
335    
336    ----------------------------------------------------------------------
337    Name: Matthias Blume (blume (at) tti - c (dot) org)
338    Date: 2003/11/18 17:10 CST
339    Tag: blume-20031118-basis-fiddle
340    Description:
341    
342    - changed Timer interface to what might become the spec
343    - POSIX_FLAGS -> BIT_FLAGS according to spec
344    - some other minor discrepancies wrt. spec eliminated
345    
346    ----------------------------------------------------------------------
347    Name: Matthias Blume (blume (at) tti - c (dot) org)
348    Date: 2003/11/06 12:00:00 CST
349    Tag: Release_110_44
350    Description:
351    
352    New working version (110.44). New bootfiles.
353    
354    ----------------------------------------------------------------------
355    Name: Matthias Blume (blume (at) tti - c (dot) org)
356    Date: 2003/11/04 11:50:00 CST
357    Tag: blume-20031104-move-libraries
358    Description:
359    
360    Eliminated the "dont_move_libraries" directive in config/targets.
361    (The mechanism was broken and could not be fixed easily.  Moreover,
362    there does not seem to be any reason not to move all libraries into
363    lib during installation.  I originally implemented this directive as a
364    backward-compatibility feature when I first introduced the new CM.
365    Now that things have been stable for a long time and going back to the
366    old CM is not an option, there is no reason to keep it around.)
367    
368    ----------------------------------------------------------------------
369    Name: Matthias Blume (blume (at) tti - c (dot) org)
370    Date: 2003/11/03 16:00:00 CST
371    Tag: blume-20031103-installdir
372    Description:
373    
374    Made installer honor INSTALLDIR variable again.  (Thanks to Chris
375    Richards for pointing out the problem and providing the solution.)
376    
377    ----------------------------------------------------------------------
378    Name: Matthias Blume (blume (at) tti - c (dot) org)
379    Date: 2003/10/01 17:05:00 CDT
380    Tag: blume-20031001-lal-mlrisc
381    Description:
382    
383    MLRISC bug fix from Lal.
384    
385    ----------------------------------------------------------------------
386    Name: Matthias Blume (blume (at) tti - c (dot) org)
387    Date: 2003/09/30 16:10:00 CDT
388    Tag: blume-20030930-primio-bat
389    Description:
390    
391    1. Added openVector, nullRd, and nullWr to PRIM_IO.
392    2. Improved .bat files (for Win32 port) to make things work under Win95.
393       (thanks to Aaron S. Hawley for this one)
394    
395    ----------------------------------------------------------------------
396    Name: Matthias Blume (blume (at) tti - c (dot) org)
397    Date: 2003/09/26 16:05:00 CDT
398    Tag: blume-20030926-wrappriv
399    Description:
400    
401    Added missing wrapper for privilege "primitive" in $smlnj/viscomp/core.cm.
402    
403    ----------------------------------------------------------------------
404    Name: Matthias Blume (blume (at) tti - c (dot) org)
405    Date: 2003/09/26 15:00:00 CDT
406    Tag: blume-20030926-110_43_3
407    Description:
408    
409    - additional cleanup
410    - version number bump, NEW BOOTFILES
411    
412    ----------------------------------------------------------------------
413    Name: Matthias Blume (blume (at) tti - c (dot) org)
414    Date: 2003/09/26 12:00:00 CDT
415    Tag: blume-20030926-ppautoload
416    Description:
417    
418    I modified the read-eval-print loop so that the autoloader gets
419    invoked whenever the prettyprinter tries to look up a symbol that
420    is not currently defined in the toplevel environment but which
421    appears in CM's autoload registry.  As a result, we see far fewer of
422    those ?.Foo.Bar.xxx names in the prettyprinter's output.
423    
424    In addition to this I tried to clean up some pieces of the Basis
425    implementation (e.g., Socket, Word8Array) in order to prevent other
426    instances of these ?.Foo.Bar.xxx names from being printed.
427    
428    The mechanism that picks names for types still needs some work, though.
429    (Right now it seems that if there is a type A.t which is defined to
430    be B.u, but B is unavailable at toplevel, then A.t gets printed as
431    "?.B.u" although the perhaps more sensible solution would be to use
432    "A.t" in this case.  In other words, the prettyprinter should follow
433    a chain of DEFtycs not farther than there are corresponding toplevel
434    names in the current environment.)
435    
436    ----------------------------------------------------------------------
437    Name: Matthias Blume (blume (at) tti - c (dot) org)
438    Date: 2003/09/24 16:31:00 CDT
439    Tag: blume-20030924-installer
440    Description:
441    
442    Another installer tweak:  All the ML code for the installer is now
443    compiled during CMB.make and put into a little library called
444    $smlnj/installer.cm.  The installation then simply invokes
445    
446       sml -m $smlnj/installer.cm
447    
448    and everything happens automagically.
449    
450    Win32:  ML code senses value of environment variable SMLNJ_HOME.
451    Unix:   ML code senses values of environment variables ROOT, CONFIGDIR,
452            and BINDIR.
453    
454    The new scheme guarantees that the ML code responsible for the installation
455    is in sync with the APIs of the main system.  Also, the installer is
456    somewhat faster because the installer script is precompiled.
457    
458    ----------------------------------------------------------------------
459    Name: Matthias Blume (blume (at) tti - c (dot) org)
460    Date: 2003/09/24 15:35:00 CDT
461    Tag: blume-20030924-synsock
462    Description:
463    
464    Added a signature SYNCHRONOUS_SOCKET to basis.cm.  This is like SOCKET
465    but excludes all non-blocking operations.  Defined SOCKET (in Basis)
466    and CML_SOCKET in terms of SYNCHRONOUS_SOCKET.  Removed superfluous
467    implementations of non-blocking operations from CML's Socket
468    structure.
469    
470    ----------------------------------------------------------------------
471    Name: Matthias Blume (blume (at) tti - c (dot) org)
472    Date: 2003/09/24 15:10:05 CDT
473    Tag: blume-20030924-sockets
474    Description:
475    
476    1. Fixed SOCKET API and implementation to match Basis spec.
477       This required changing the internal representation of sockets to one
478       that remembers (for each socket file descriptor) whether it is currently
479       blocking or non-blocking.  This state is maintained lazily (i.e., a system
480       call is made only if the state actually needs to change).
481    
482    2. OS-specific details of sockets were moved into separate files, thus
483       making it possible to unify the bulk of the socket implementations
484       between Unix and Win32.
485    
486    3. CML's socket API changed accordingly.
487       (Note that we need to remove non-blocking functions from this API
488       since they are redundant in the case of CML!)
489    
490    4. CML's socket implementation now makes use of non-blocking functions
491       provided by Basis, thus removing all OS-dependent code from this part
492       of CML.
493    
494    5. Changed Real64.precision from 52 to 53.  Minor cleanup in Real64 code.
495    
496    ----------------------------------------------------------------------
497    Name: Matthias Blume (blume (at) tti - c (dot) org)
498    Date: 2003/09/22 12:10:00 CDT
499    Tag: blume-20030922-110_43_2
500    Description:
501    
502    Made a new interim version and bootfiles for developer's bootstrapping
503    convenience.
504    
505       110.43.2 -- NEW BOOTFILES
506    
507    ----------------------------------------------------------------------
508    Name: Matthias Blume (blume (at) tti - c (dot) org)
509    Date: 2003/09/19 15:55:00 CDT
510    Tag: blume-20030919-cmdir
511    Description:
512    
513    1. new-install.sh -> install.sh
514    2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
515    3. tweaked installer so that another name instead of .cm can be chosen
516       at install time (by setting the CM_DIR_ARC environment variable
517       during installation); once installation is complete, the name is
518       fixed
519    
520    ----------------------------------------------------------------------
521    Name: Matthias Blume (blume (at) tti - c (dot) org)
522    Date: 2003/09/18 16:00:00 CDT
523    Tag: blume-20030918-110_43_1
524    Description:
525    
526    Made a new interim version and bootfiles for developer's bootstrapping
527    convenience.
528    
529       110.43.1 -- NEW BOOTFILES
530    
531    ----------------------------------------------------------------------
532    Name: Matthias Blume (blume (at) tti - c (dot) org)
533    Date: 2003/09/18 15:20:00 CDT
534    Tag: blume-20030918-misc
535    Description:
536    
537    1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
538       this seems to be controversial at the moment) and used those in
539       Posix.ProcEnv.times.
540    
541    2. Added Time.{from,to}Nanoseconds to Time.
542    
543    3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
544       For example, fromLargeInt never needs to look at more than 3 "big
545       digits" to get its 53 bits of precision.
546    
547    ----------------------------------------------------------------------
548    Name: Matthias Blume (blume (at) tti - c (dot) org)
549    Date: 2003/09/17 16:30:00 CDT
550    Tag: blume-20030917-real32-slices
551    Description:
552    
553    Added an entry to the primitive environment
554    (compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
555    added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
556    
557    Removed some of the "magic" constants in real64.sml and replaced them
558    with code that generates these values from their corresponding
559    integer counterparts.
560    
561    Made all(?) the slice-related changes to the Basis and made everything
562    compile again...
563    
564    ----------------------------------------------------------------------
565    Name: Matthias Blume (blume (at) tti - c (dot) org)
566    Date: 2003/09/15 17:45:00 CDT
567    Tag: blume-20030915-rbase
568    Description:
569    
570    Fixed bug in Real.fromLargeInt.
571    
572    ----------------------------------------------------------------------
573    Name: Matthias Blume (blume (at) tti - c (dot) org)
574    Date: 2003/09/13 18:11:00 CDT
575    Tag: blume-20030913-libinstall
576    Description:
577    
578    Minor bugfix in config/libinstall (set anchor with path to
579    standalone tool after installing it, otherwise libraries that
580    need ml-lex or ml-yacc won't compile the first time the installer
581    runs).
582    
583    ----------------------------------------------------------------------
584    Name: Matthias Blume (blume (at) tti - c (dot) org)
585    Date: 2003/09/12 11:45:00 CDT
586    Tag: blume-20030912-various
587    Description:
588    
589     - fixed bug in Real.toLargeInt
590     - fixed bug in Posix.ProcEnv.times
591     - changed inputLine functions to return an option
592     - minor installer improvements / bugfixes
593     - changed default @SMLalloc parameter for x86/celeron to 64k
594    
595    ----------------------------------------------------------------------
596    Name: Matthias Blume (blume (at) tti - c (dot) org)
597    Date: 2003/09/09 22:00:00 CDT
598    Tag: Release_110_43
599    Description:
600    
601        New working release 110.43.  New bootfiles.
602    
603    ----------------------------------------------------------------------
604    Name: Matthias Blume (blume (at) tti - c (dot) org)
605    Date: 2003/09/09 19:20:00 CDT
606    Tag: blume-20030909-installer
607    Description:
608    
609    Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
610    Modified config/install.bat to take advantage of it.  Also modified
611    config/install.sh (and called it config/new-install.sh) to take advantage
612    of it on Unix systems.  (The SML code is (supposed to be) platform-
613    independent.)
614    
615    The installer can now install everything under Win32
616    as well as under *nix as long as it compiles.
617    
618    Other changes:
619    
620      - made CML compile again under Win32
621      - made eXene compile under Win32 (by providing a fake structure UnixSock
622        and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
623      - fixed a bug in nowhere: it assumed that type OS.Process.status is the
624        same as type int;  under Win32 it isn't
625      - fixed some slice-related problems in the win32-specific parts of CML
626      - added a functor argument "sameVol" to os-path-fn.sml in the Basis
627        (under Win32, the volume name is case-insensitive, and the
628        OS.Path code compares volume names for equality)
629    
630    ----------------------------------------------------------------------
631    Name: Matthias Blume (blume (at) tti - c (dot) org)
632    Date: 2003/09/08 11:55:00 CDT
633    Tag: blume-20030908-fullpath
634    Description:
635    
636    Made Win32 version of OS.FileSys.fullPath return current directory
637    when given an empty string.  This is what the spec says, and incidentally,
638    CM depends on it.  (CM otherwise goes into an infinite loop in certain
639    cases when presented with the name of a non-existing .cm file.)
640    
641    ----------------------------------------------------------------------
642    Name: Matthias Blume (blume (at) tti - c (dot) org)
643    Date: 2003/09/04 16:30:00 CDT
644    Tag: blume-20030905-slices-etc
645    Description:
646    
647      1. Changed interface to vectors and arrays in Basis to match
648         (draft) Basis spec.
649      2. Added signatures and implementations of slices according to
650         Basis spec.
651      3. Edited source code throughout the system to make it compile again
652         under 1. and 2.  (In some cases code had to be added to have it
653         match the new signatures.)
654      4. MLRISC should be backward-compatible: the copies of the originals
655         of files that needed to change under 3. were retained, the .cm files
656         check the compiler version number and use old versions when
657         appropriate.
658      5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
659         dirstream -> string option (in accordance with Basis spec).
660      6. When generating code that counts lines, ml-lex used function
661         CharVector.foldli, taking advantage of its old interface.
662         This has been replaced with the corresponding code from
663         CharVectorSlice.  (html-lex must be re-lexed!)
664      7. BitArray in smlnj-lib/Util has been extended/modified to match the
665         new MONO_ARRAY signature.  (Do we need BitArraySlice?)
666      8. Removed temporary additions (fromInternal, toInternal) from the
667         (now obsolete) IntInf in smlnj-lib/Util.
668      9. Cleaned up structure Byte.
669     10. Added localOffset, scan, and fromString to Date (according to spec).
670         Cleaned/corrected implementation of Date.
671         (Still need to check for correctness; implement better canonicalizeDate.)
672     11. Added "scan" to signature IEEE_REAL.
673     12. Some improvements to IntInf [in particular: efficiency-hack for
674         mod and rem when second operand is 2 (for parity checks).]
675     13. Changed representation of type Time.time, using a single IntInf.int
676         value counting microseconds.  This considerably simplified the
677         implementation of structure Time.  We now support negative time
678         values; scan and fromString handle signs.
679     14. Functor PrimIO now takes two additional arguments (VectorSlice and
680         ArraySlice).
681    
682    ----------------------------------------------------------------------
683    Name: Matthias Blume (blume (at) tti - c (dot) org)
684    Date: 2003/08/28 17:00:00 CDT
685    Tag: blume-20030828-intinf
686    Description:
687    
688    This is a major update which comes with a version number bump
689    (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
690    and an implementation of IntInf in the Basis.
691    
692    There are a fairly large number of related changes and updates throughout
693    the system:
694    
695    Basis:
696            - Implemented IntInf.
697            - Made LargeInt a projection of IntInf (by filtering through INTEGER).
698            - Added some missing Real64 operations, most notably Real.toLargeInt.
699            - Added FixedInt as a synonym for Int32.
700    
701    compiler:
702            * Added support for a built-in intinf type.
703              - literals
704              - pattern matching
705              - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
706              - overloading on literals and operations
707    
708            This required adding a primitive type intinf, some additional
709            primops, and implementations for several non-trivial intinf
710            operations in Core.  (The intinf type is completely abstract
711            to the compiler; all operations get delegated back to the Core.)
712    
713            * Intinf equality is handled by polyequal.  However, the compiler
714            does not print its usual warning in this case (since polyequal
715            is the right thing to do there).
716    
717            * Improved the organization of structure InlineT.
718    
719            * A word about conversion primops:
720            If conversions involving intinf do not cancel out during
721            CPS contract, then the compiler must insert calls to Core functions.
722            Since all core access must be resolved already during the FLINT
723            translate phase, it would be too late a the time of CPS contract
724            to add new Core calls.  For this reason, conversion primops
725            for intinf carry two arguments:  1. the numeric argument that
726            they are supposed to convert, and 2. the Core function that
727            can help with this conversion if necessary.  If CPS contract
728            eliminates a primop, then the associated Core function becomes
729            dead and goes away.  Intinf conversion primops that do not get
730            eliminated by CPS contract get rewritten into calls of their
731            core functions by a separate, new phase.
732    
733    interactive system:
734            - Control.Print.intinfDepth controls max length of intinf constants
735            being printed. (Analogous to Control.Print.stringDepth.)
736            - Cleanup in printutil and pputil: got rid of unused stuff and
737            duplicates; replaced some of the code with code that makes better
738            use of library functionality.
739    
740    CM:
741            Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
742            are no longer silent.
743    
744    CKIT:
745            Fixed mismatched uses of Int32 and LargeInt.  I always decided
746            in favor of LargeInt -- which is now the same as IntInf.
747            CKIT-knowledgable people should check whether this is what's
748            intended and otherwise change things back to using Int32 or
749            FixedInt.
750    
751    Throughout the code:
752            Started using IntInf.int literals and built-in operations
753            (e.g., comparison with 0) where this seems appropriate.
754    
755    
756    ----------------------------------------------------------------------
757    Name: Dave MacQueen (dbm@cs.uchicago.edu)
758    Date: 2003/08/13 11:36:00 CDT
759    Tag: dbm-20030813-mcz-merge1
760    Description:
761    
762    Merging changes from the mcz-branch development branch into trunk.
763    These changes involve replacement of the emulated old prettyprinter
764    interface with direct use of the SML/NJ Lib PP library, and fixing
765    of a couple of bugs (895, 1186) relating to error messages.  A new
766    prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
767    has been added.
768    
769    ----------------------------------------------------------------------
770    Name: Matthias Blume (blume (at) tti - c (dot) org)
771    Date: 2003/08/11 15:45:00 CDT
772    Tag: blume-20030811-windows
773    Description:
774    
775    Version number bumped to 110.42.9.  NEW BOOTFILES!!!
776    
777        http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
778    
779    This patch restores SML/NJ's ability to run under win32.  There are a
780    number of changes, including fixes for several bugs that had gone
781    unnoticed until now:
782    
783      - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
784        trying to run the win32 version from within cygwin.)
785    
786      - There are a number of simple .bat scripts that substitute for their
787        corresponding Unix shell-scripts.  (See below.)
788    
789      - The internals of ml-build have been modified slightly.  The main
790        difference is that instead of calling ".link-sml" (or link-sml.bat)
791        using OS.Process.system, the ML process delegates this task back
792        to the script.  Otherwise problems arise in mixed environments such
793        as Cygwin where scripts look and work like Unix scripts, but
794        where OS.Process.system cannot run them.
795    
796      - In CM, the srcpath pickler used native pathname syntax -- which
797        is incorrect in the case of cross-compilation.  The new pickle format
798        is independent of platform-specific naming conventions.
799    
800      - Path configuration files (such as lib/pathconfig) can now choose
801        between native and standard syntax.  Placing a line of the form
802    
803            standard!
804    
805        into the file causes all subsequent paths to be interpreted using
806        CM standard pathname syntax (= Unix conventions); a line
807    
808            native!
809    
810        switches back to native style.  This was needed so that
811        path config files can be written portably, see src/system/pathconfig.
812    
813      - Runtime system:
814    
815          - win32-filesys.c:  get_file_time and set_file_time now
816            access modification time, not creation time.
817    
818          - I/O code made aware of new array representation.
819    
820          - Bug fixes in X86.prim.masm.
821    
822       - src/system/makeml made aware of win32. (For use under cygwin
823            and other Unix-environments for windows.)
824    
825       - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
826         which caused BinIO.inputAll to fail consistently.
827    
828    .bat scripts:
829    
830       Windows .bat scripts assume that SMLNJ_HOME is defined.
831    
832       - sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
833         applications (sml, ml-yacc, ml-lex).
834       - ml-build.bat: analogous to ml-build.
835       - config\install.bat:  Analogous to config/install.sh.  This requires
836         that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
837         (nmake etc. must be on the path, and vcvars32 must have been run.)
838         Moreover, sources for ml-lex and ml-yacc need to exist under src,
839         and the bootfile hierarchy must have been unpacked under
840         sml.boot.x86-win32.
841         The script is very primitive and does a poor job at error checking.
842         It only installs the base system, ml-lex, and ml-yacc.  No other
843         libraries are being installed (i.e., you get only those that
844         are part of the compiler.)
845       - link-sml.bat:  analogous to .link-sml, but not currently used
846    
847    Unrelated bug fixes:
848    
849      - ml-nlffigen now exports structures ST_* corresponding to incomplete
850        types.
851      - Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
852        otherwise.)
853    
854    ----------------------------------------------------------------------
855    Name: Dave MacQueen (macqueen@cs.uchicago.edu)
856    Date: 2003/06/17
857    Tag: macqueen-20030617-bug895
858    Description:
859    
860    Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
861    Tag will be used for new development branch (mcz-branch) for
862    use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
863    
864    ----------------------------------------------------------------------
865    Name: Matthias Blume (blume (at) tti - c (dot) org)
866    Date: 2003/05/27 16:55:00 CDT
867    Tag: blume-20030527-polyeq
868    Description:
869    
870    Tried to eliminated most cases of polymorphic equality.
871    
872    ----------------------------------------------------------------------
873    Name: Matthias Blume (blume (at) tti - c (dot) org)
874    Date: 2003/05/21 17:45:00 CDT
875    Tag: blume-20030517-complete
876    Description:
877    
878    Two changes:
879    
880      1. Added a flag for controlling whether non-exhaustive bindings will
881         be treated as errors (default is false).
882      2. Cleaned up the *entire* source tree so that CMB.make goes through
883         without a single non-exhaustive match- or bind warning.
884    
885    ----------------------------------------------------------------------
886    Name: Matthias Blume (blume (at) tti - c (dot) org)
887    Date: 2003/05/17 10:20:00 CDT
888    Tag: blume-20030517-absyn
889    Description:
890    
891    1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
892    
893       This mainly affects the quality of error messages.  However, some
894       of the code is now more straightforward than before. (Treatment of
895       the above four constructs in translate.sml is much simpler than
896       the "macro-expansion" that was going on before.  Plus, the mach-
897       compiler no longer gets invoked just to be able to compile an
898       if-expression.)
899    
900    2. The ErrorMsg.Error exception is now caught and absorbed by the
901       interactive loop.
902    
903    ----------------------------------------------------------------------
904    Name: Allen Leung
905    Date: 2003/05/16 13:05:00 CDT
906    Tag: leunga-20030516-cygwin-runtime
907    Description:
908    
909        Ported the runtime system to cygwin, which uses the unix
910    x86-unix bin files.  Missing/buggy features:
911    
912       o getnetbyname, getnetbyaddr: these functions seem to be missing in
913         the Cygwin library.
914       o Ctrl-C handling may be flaky.
915       o Windows system calls and Windows I/O are not supported.
916    
917       A new set of binfiles is located at:
918    
919         http://www.dorsai.org/~leunga/boot.x86-unix.tgz
920    
921       This is only needed for bootstrapping the cygwin version of smlnj.
922       Other x86 versions can use the existing binfiles.
923    
924    ----------------------------------------------------------------------
925    Name: Matthias Blume
926    Date: 2003/04/08 15:42:00 CDT
927    Tag: blume-20030408-listpair
928    Description:
929    
930    1. Added a target 'mlrisc' to installer.
931    
932    2. Added missing elements to structure ListPair.
933    
934    ----------------------------------------------------------------------
935    Name: Allen Leung
936    Date: 2003/01/07 10:40:00 EST
937    Tag: leunga-20030107-int-rem
938    Description:
939    
940    Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
941    The arguments to the SUBL instruction were swapped.
942    
943    ----------------------------------------------------------------------
944    Name: Matthias Blume
945    Date: 2002/12/12 16:25:00 EST
946    Tag: blume-20021212-risc-ra
947    Description:
948    
949    Fixed a serious bug in the rewrite code for FP spilling/reloading that
950    sent the RA into an infinite loop when floating point registers get
951    spilled.  (Because of this bug, e.g., nucleic stopped compiling between
952    110.37 and 110.38.)
953    There was another set of potential problems related to the handling of
954    MLRISC annotations (but those did not yet cause real problems, apparently).
955    
956    ----------------------------------------------------------------------
957    Name: Matthias Blume
958    Date: 2002/12/06 22:40:00 EST
959    Tag: blume-20021206-cm-fileid
960    Description:
961    
962    Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
963    This fixes the problem of CM getting confused by files that suddenly
964    change their identity (e.g., by getting unlinked and recreated by some
965    text editor such as vi).  There might be a better/cheaper/cleaner way
966    of doing this, but for now this will have to do.
967    
968    ----------------------------------------------------------------------
969    Name: Matthias Blume
970    Date: 2002/10/28 09:50:00 EST
971    Tag: blume-20021028-typecheck
972    Description:
973    
974    Exported structure Typecheck from $smlnj/viscomp/core.cm.
975    
976    ----------------------------------------------------------------------
977    Name: Matthias Blume
978    Date: 2002/10/17 09:10:00 EDT
979    Tag: Release_110_42
980    Description:
981    
982    In good old tradition, there has been a slight hiccup so that we have
983    to patch 110.42 after the fact.   The old release tag has been replaced
984    (see below).
985    
986    The change solves a problem with two competing approaches the
987    configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
988    each other's way.
989    
990    This change only affects the runtime system code and the installer script.
991    (No new bootfiles.)
992    
993    ----------------------------------------------------------------------
994    Name: Matthias Blume
995    Date: 2002/10/16 12:00:00 EDT
996    Tag: Release_110_42_removed
997    Description:
998    
999    New working release.  New bootfiles.
1000    
1001    ----------------------------------------------------------------------
1002    Name: Matthias Blume
1003    Date: 2002/10/10 13:10:00 EDT
1004    Tag: blume-20021010-ppc-divs
1005    Description:
1006    
1007    The mltree operator DIVS must be implemented with an overflow check on
1008    the PPC because the hardware indicates divide-by-zero using "overflow" as
1009    well.
1010    
1011    ----------------------------------------------------------------------
1012    Name: Matthias Blume
1013    Date: 2002/07/23 11:45:00
1014    Tag: blume-20020723-smlnj-home
1015    Description:
1016    
1017    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
1018    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
1019    CM_PATHCONFIG is also set), the path configuration file is assumed
1020    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
1021    the entire tree to some other place and everything will "just work".
1022    
1023    (Companion commands such as ml-build and ml-makedepend also sense this
1024    variable.)
1025    
1026    ----------------------------------------------------------------------
1027    Name: Matthias Blume
1028    Date: 2002/07/12 21:19:00 EDT
1029    Tag: blume-20020712-liveness
1030    Description:
1031    
1032    Exported two useful "step" functions from liveness module (MLRISC).
1033    
1034    ----------------------------------------------------------------------
1035    Name: Matthias Blume
1036    Date: 2002/07/05 16:00 EDT
1037    Tag: Release_110_41
1038    Description:
1039    
1040    New working release.  New bootfiles.
1041    
1042    ----------------------------------------------------------------------
1043    Name: Matthias Blume
1044    Date: 2002/07/05 10:25:00 EDT
1045    Tag: blume-20020705-btimp
1046    Description:
1047    
1048    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
1049    other clients can set up backtracing support.
1050    
1051    ----------------------------------------------------------------------
1052    Name: Matthias Blume
1053    Date: 2002/06/25 17:23:00 EDT
1054    Tag: blume-20020625-fpmax
1055    Description:
1056    
1057    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
1058    case.  (The sense of the isNaN test was reversed -- which made min and
1059    max always return their first argument.)
1060    
1061    ----------------------------------------------------------------------
1062    Name: Matthias Blume
1063    Date: 2002/06/11
1064    Tag: blume-20020611-unixpath
1065    Description:
1066    
1067    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
1068    
1069    ----------------------------------------------------------------------
1070    Name: Matthias Blume
1071    Date: 2002/06/10 16:35:00 EDT
1072    Tag: blume-20020610-ieeereal
1073    Description:
1074    
1075    I back-ported my implementation of IEEEReal.fromString from the
1076    idlbasis-devel branch so that we can test it.
1077    
1078    Another small change is that ppDec tries to give more information
1079    than just "<sig>" in the case of functors.  However, this code is
1080    broken in some mysterious way if the functor's body's signature
1081    has not been declared by ascription but gets inferred from the
1082    implementation.  This needs fixing...
1083    
1084    ----------------------------------------------------------------------
1085    Name: Matthias Blume
1086    Date: 2002/05/31
1087    Tag: blume-20020531-btrace-mode
1088    Description:
1089    
1090    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
1091    the wayside when I switched over to using Controls everywhere.)
1092    
1093    ----------------------------------------------------------------------
1094    Name: Lal George
1095    Date: 2002/05/23 12:21:40 EDT
1096    Tag: george-20020523-visual-labels
1097    Description:
1098    
1099    Labels are now displayed in the graphical output to make
1100    the fall-through and target blocks obvious.
1101    
1102    ----------------------------------------------------------------------
1103    Name: Matthias Blume
1104    Date: 2002/05/22 11:03:00 EDT
1105    Tag: blume-20020522-shrink
1106    Description:
1107    
1108    John tweaked yesterday's fix for 1131 to handle an out-of-memory
1109    situation that comes up when allocating huge arrays.
1110    
1111    ----------------------------------------------------------------------
1112    Name: Matthias Blume
1113    Date: 2002/05/21 16:00:00 EDT
1114    Tag: Release_110_40
1115    Description:
1116    
1117    New working release (110.40).  New bootfiles.
1118    
1119    [Also: John Reppy fixed GC bug 1131.]
1120    
1121    ----------------------------------------------------------------------
1122    Name: Matthias Blume
1123    Date: 2002/05/21 12:35:00 EDT
1124    Tag: blume-20020521-cmdoc
1125    Description:
1126    
1127    CM documentation update.
1128    
1129    ----------------------------------------------------------------------
1130    Name: Matthias Blume
1131    Date: 2002/05/21 10:55:00 EDT
1132    Tag: blume-20020521-misc
1133    Description:
1134    
1135    - John tweaked runtime to be silent on heap export (except when
1136      GC messages are on).
1137    - I added a few more things (cross-compiling versions of CMB) to
1138      config/preloads (as suggestions).
1139    
1140    ----------------------------------------------------------------------
1141    Name: Matthias Blume
1142    Date: 2002/05/20 22:25:00 EDT
1143    Tag: blume-20020520-controls
1144    Description:
1145    
1146    - Added ControlUtil structure to control-lib.cm.
1147    - Use it throughout.
1148    - Used Controls facility to define MLRISC controls (as opposed to
1149      registering MLRISC control ref cells with Controls after the fact)
1150    - Fixed messed-up controls priorities.
1151    
1152    * Removed again all the stuff from config/preloads that one wouldn't
1153      be able to preload at the time the initial heap image is built.
1154      (Many libraries, e.g., CML, do not exist yet at this time.  The
1155       only libraries that can be preloaded via config/preloads are those
1156       that come bundled with the bootfiles.)
1157    
1158    ----------------------------------------------------------------------
1159    Name: Matthias Blume
1160    Date: 2002/05/20 10:59:00 EDT
1161    Tag: blume-20020520-preloads
1162    Description:
1163    
1164    Added a lot of commented-out suggestions for things to be included
1165    in config/preloads.
1166    
1167    ----------------------------------------------------------------------
1168    Name: Allen Leung
1169    Date: 2002/05/18 14:20:00 EDT
1170    Tag: leunga-20020518-mdl
1171    Description:
1172    
1173    o Made the mdl tool stuff compile and run again.
1174    o I've disabled all the stuff that depends on RTL specifications; they
1175      are all badly broken anyway.
1176    
1177    ----------------------------------------------------------------------
1178    Name: Matthias Blume
1179    Date: 2002/05/17 16:49:00 EDT
1180    Tag: blume-20020517-controls
1181    Description:
1182    
1183    0. John Reppy made several modifications to the SML/NJ library.
1184       In particular, there is a shiny new controls-lib.cm.
1185    
1186    1. Pushed new controls interface through compiler so that everything
1187       compiles again.
1188    
1189    2. Added FormatComb and FORMAT_COMB to the CML version of the
1190       SML/NJ library (so that CML compiles again).
1191    
1192    3. Modified init scripts because XXX_DEFAULT environment variables
1193       are no longer with us.  (Boot-time initialization is now done
1194       using the same environment variables that are also used for
1195       startup-time initialization of controls.)
1196    
1197    ----------------------------------------------------------------------
1198    Name: Lal George
1199    Date: 2002/05/15 09:20:10 EDT
1200    Tag: george-20020515-pseudo-op-decls
1201    Description:
1202    
1203            All pseudo-ops emitted before the first segment declaration
1204    such as TEXT, DATA, and BSS directives are assumed to be global
1205    declarations and are emitted first in the assembly file. This is
1206    useful in a number of situations where one has pseudo-ops that are not
1207    specific to any segment, and also works around the constraint that one
1208    cannot have client pseudo-ops in the TEXT segment.
1209    
1210            Because no segment is associated with these declarations it is
1211    an error to allocate any space or objects before the first segment
1212    directive and an exception will be raised. However, we cannot make
1213    this check for client pseudo-ops.
1214    
1215            These top level declarations are a field in the CFG graph_info.
1216    In theory you can continue to add to this field after the CFG has been
1217    built -- provided you know what you are doing;-)
1218    
1219    ----------------------------------------------------------------------
1220    Name: Matthias Blume
1221    Date: 2002/05/13 16:40:00 EDT
1222    Tag: blume-20020513-pp-etc
1223    Description:
1224    
1225    A few minor bugfixes:
1226    
1227      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
1228        (Bogus printouts for redefined bindings still occur.  Compiler
1229        bug should no longer occur now.  We need to redo the prettyprinter
1230        from scratch.)
1231    
1232      - CM pathname printer now also adds escape sequences for ( and )
1233    
1234      - commend and docu fixes for ml-nlffi
1235    
1236    ----------------------------------------------------------------------
1237    Name: Matthias Blume
1238    Date: 2002/05/10 16:40:00 EDT
1239    Tag: blume-20020510-erg-textio
1240    Description:
1241    
1242    Applied the following bugfix provided by Emden Gansner:
1243    
1244        Output is corrupted when outputSubstr is used rather than output.
1245    
1246        The problem occurs when a substring
1247    
1248            ss = (s, dataStart, dataLen)
1249    
1250        where dataStart > 0, fills a stream buffer with avail bytes left.
1251        avail bytes of s, starting at index dataStart, are copied into the
1252        buffer, the buffer is flushed, and then the remaining dataLen-avail
1253        bytes of ss are copied into the beginning of the buffer. Instead of
1254        starting this copy at index dataStart+avail in s, the current code
1255        starts the copy at index avail.
1256    
1257        Fix:
1258        In text-io-fn.sml, change line 695 from
1259             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
1260        to
1261             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
1262    
1263    ----------------------------------------------------------------------
1264    Name: Matthias Blume
1265    Date: 2002/04/12 13:55:00 EDT
1266    Tag: blume-20020412-assyntax
1267    Description:
1268    
1269    1. Grabbed newer assyntax.h from the XFree86 project.
1270    2. Fiddled with how to compile X86.prim.asm without warnings.
1271    3. (Very) Minor cleanup in CM.
1272    
1273    ----------------------------------------------------------------------
1274    Name: Matthias Blume
1275    Date: 2002/04/01 (no joke!) 17:07:00 EST
1276    Tag: blume-20020401-x86div
1277    Description:
1278    
1279    Added full support for div/mod/rem/quot on the x86, using the machine
1280    instruction's two results (without clumsily recomputing the remainder)
1281    directly where appropriate.
1282    
1283    Some more extensive power-of-two support was added to the x86 instruction
1284    selector (avoiding expensive divs, mods, and muls where they can be
1285    replaced with cheaper shifts and masks).  However, this sort of thing
1286    ought to be done earlier, e.g., within the CPS optimizer so that
1287    all architectures benefit from it.
1288    
1289    The compiler compiles to a fixed point, but changes might be somewhat
1290    fragile nevertheless.  Please, report any strange things that you might
1291    see wrt. div/mod/quot/rem...
1292    
1293    ----------------------------------------------------------------------
1294    Name: Matthias Blume
1295    Date: 2002/03/29 17:22:00
1296    Tag: blume-20020329-div
1297    Description:
1298    
1299    Fixed my broken div/mod logic.  Unfortunately, this means that the
1300    inline code for div/mod now has one more comparison than before.
1301    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
1302    The problem was with quotient = 0, because that alone does not tell
1303    us which way the rounding went.  One then has to look at whether
1304    remainder and divisor have the same sign...  :(
1305    
1306    Anyway, I replaced the bootfiles with fresh ones...
1307    
1308    ----------------------------------------------------------------------
1309    Name: Matthias Blume
1310    Date: 2002/03/29 14:10:00 EST
1311    Tag: blume-20020329-inlprims
1312    Description:
1313    
1314    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
1315    
1316    Primops have changed. This means that the bin/boot-file formats have
1317    changed as well.
1318    
1319    To make sure that there is no confusion, I made a new version.
1320    
1321    
1322    CHANGES:
1323    
1324    * removed REMT from mltree (remainder should never overflow).
1325    
1326    * added primops to deal with divisions of all flavors to the frontend
1327    
1328    * handled these primops all the way through so they map to their respective
1329      MLRISC support
1330    
1331    * used these primops in the implementation of Int, Int32, Word, Word32
1332    
1333    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
1334    
1335    * parameterized INLMIN, INLMAX, and INLABS by a numkind
1336    
1337    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
1338      floating point
1339    
1340    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
1341      and Real (but Real.abs maps to a separate floating-point-only primop)
1342    
1343    
1344    TODO items:
1345    
1346    * Hacked Alpha32 instruction selection, disabling the selection of REMx
1347      instructions because the machine instruction encoder cannot handle
1348      them.  (Hppa, PPC, and Sparc instruction selection did not handle
1349      REM in the first place, and REM is supported by the x86 machine coder.)
1350    
1351    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
1352      selection phase.  (The two can be streamlined because the hardware
1353      delivers both quotient and remainder at the same time anyway.)
1354    
1355    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
1356      (Currently the behavior is inconsistent both across architectures and
1357      wrt. the draft Basis spec.)
1358    
1359    * Word8 should eventually be handled natively, too.
1360    
1361    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
1362      as if there currently is no execution path that could trigger it in
1363      SML/NJ.  (The assumptions underlying functions arith and promotable do not
1364      hold for things like multiplication and division.)
1365    
1366    ----------------------------------------------------------------------
1367    Name: Matthias Blume
1368    Date: 2002/03/27 16:27:00 EST
1369    Tag: blume-20020327-mlrisc-divisions
1370    Description:
1371    
1372    Added support for all four division operations (ML's div, mod, quot,
1373    and rem) to MLRISC.  In the course of doing so, I also rationalized
1374    the naming (no more annoying switch-around of DIV and QUOT), by
1375    parameterizing the operation by div_rounding_mode (which can be either
1376    DIV_TO_ZERO or DIV_TO_NEGINF).
1377    
1378    The generic MLTreeGen functor takes care of compiling all four
1379    operations down to only round-to-zero div.
1380    
1381    Missing pieces:
1382    
1383      * Doing something smarter than relying on MLTreeGen on architectures
1384        like, e.g., the x86 where hardware division delivers both quotient and
1385        remainder at the same time.  With this, the implementation of the
1386        round-to-neginf operations could be further streamlined.
1387    
1388      * Remove inlining support for div/mod/rem from the frontend and replace it
1389        with primops that get carried through to the backend.  Do this for all
1390        int and word types.
1391    
1392    ----------------------------------------------------------------------
1393    Name: Matthias Blume
1394    Date: 2002/03/25 17:25:00 EST
1395    Tag: blume-20020325-divmod
1396    Description:
1397    
1398    I improved (hopefully without breaking them) the implementation of Int.div,
1399    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
1400    advantage of the following observations:
1401    
1402      Let  q = x quot y      r = x rem y
1403           d = x div  y      m = x mod y
1404    
1405    where "quot" is the round-to-zero version of integer division that
1406    hardware usually provides.  Then we have:
1407    
1408         r = x - q * y        where neither the * nor the - will overflow
1409         d = if q >= 0 orelse x = q * y then q else q - 1
1410                              where neither the * nor the - will overflow
1411         m = if q >= 0 orelse r = 0 then r else r + y
1412                              where the + will not overflow
1413    
1414    This results in substantial simplification of the generated code.
1415    The following table shows the number of CFG nodes and edges generated
1416    for
1417            fun f (x, y) = x OPER y
1418            (* with OPER \in div, mod, quot, rem *)
1419    
1420    
1421        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
1422        --------------------------------------------------------
1423         div |         24 |         39 |         12 |         16
1424         mod |         41 |         71 |         12 |         16
1425        quot |          8 |         10 |          8 |         10
1426         rem |         10 |         14 |          8 |         10
1427    
1428    
1429    ----------------------------------------------------------------------
1430    Name: Matthias Blume
1431    Date: 2002/03/25 22:06:00 EST
1432    Tag: blume-20020325-cprotobug
1433    Description:
1434    
1435    Fixed a bug in cproto (c prototype decoder).
1436    
1437    ----------------------------------------------------------------------
1438    Name: Matthias Blume
1439    Date: 2002/03/25 16:00:00 EST
1440    Tag: blume-20020325-raw-primops
1441    Description:
1442    
1443    I did some cleanup to Allen's new primop code and
1444    replaced yesterday's bootfiles with new ones.
1445    (But they are stored in the same place.)
1446    
1447    ----------------------------------------------------------------------
1448    Name: Matthias Blume
1449    Date: 2002/03/24 22:40:00 EST
1450    Tag: blume-20020324-bootfiles
1451    Description:
1452    
1453    Made the bootfiles that Allen asked for.
1454    
1455    ----------------------------------------------------------------------
1456    Name: Allen Leung
1457    Date: 2002/03/23 15:50:00 EST
1458    Tag: leunga-20020323-flint-cps-rcc-primops
1459    Description:
1460    
1461      1. Changes to FLINT primops:
1462    
1463        (* make a call to a C-function;
1464         * The primop carries C function prototype information and specifies
1465         * which of its (ML-) arguments are floating point. C prototype
1466         * information is for use by the backend, ML information is for
1467         * use by the CPS converter. *)
1468      | RAW_CCALL of { c_proto: CTypes.c_proto,
1469                       ml_args: ccall_type list,
1470                       ml_res_opt: ccall_type option,
1471                       reentrant : bool
1472                     } option
1473       (* Allocate uninitialized storage on the heap.
1474        * The record is meant to hold short-lived C objects, i.e., they
1475        * are not ML pointers.  With the tag, the representation is
1476        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
1477        *)
1478      | RAW_RECORD of {tag:bool,sz:int}
1479      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
1480    
1481      2.  These CPS primops are now overloaded:
1482    
1483           rawload of {kind:numkind}
1484           rawstore of {kind:numkind}
1485    
1486          The one argument form is:
1487    
1488             rawload {kind} address
1489    
1490          The two argument form is:
1491    
1492             rawload {kind} [ml object, byte-offset]
1493    
1494      3. RAW_CCALL/RCC now takes two extra arguments:
1495    
1496         a. The first is whether the C call is reentrant, i.e., whether
1497            ML state should be saved and restored.
1498         b. The second argument is a string argument specifying the name of
1499            library and the C function.
1500    
1501         These things are currently not handled in the code generator, yet.
1502    
1503      4. In CProto,
1504    
1505         An encoding type of "bool" means "ml object" and is mapped into
1506         C prototype of PTR.  Note that "bool" is different than "string",
1507         even though "string" is also mapped into PTR, because "bool"
1508         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
1509    
1510      5. Pickler/unpicker
1511    
1512         Changed to handle RAW_RECORD and newest RAW_CCALL
1513    
1514      6. MLRiscGen,
1515    
1516         1. Changed to handle the new rawload/rawstore/rawrecord operators.
1517         2. Code for handling C Calls has been moved to a new module CPSCCalls,
1518            in the file CodeGen/cpscompile/cps-c-calls.sml
1519    
1520      7. Added the conditional move operator
1521    
1522             condmove of branch
1523    
1524         to cps.  Generation of this is still buggy so it is currently
1525         disabled.
1526    
1527    ----------------------------------------------------------------------
1528    Name: Lal George
1529    Date: 2002/03/22 14:18:25 EST
1530    Tag: george-20020322-cps-branch-prob
1531    Description:
1532    
1533    Implemented the Ball-Larus branch prediction-heuristics, and
1534    incorporated graphical viewers for control flow graphs.
1535    
1536    Ball-Larus Heuristics:
1537    ---------------------
1538    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
1539    
1540    By design it uses the Dempster-Shafer theory for combining
1541    probabilities.  For example, in the function:
1542    
1543        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
1544    
1545    the ball-larus heuristics predicts that the n=0 is unlikely
1546    (OH-heuristic), and the 'then' branch is unlikely because of the
1547    RH-heuristic -- giving the 'then' branch an even lower combined
1548    probability using the Dempster-Shafer theory.
1549    
1550    Finally, John Reppy's loop analysis in MLRISC, further lowers the
1551    probability of the 'then' branch because of the loop in the else
1552    branch.
1553    
1554    
1555    Graphical Viewing:
1556    ------------------
1557    I merely plugged in Allen's graphical viewers into the compiler. The
1558    additional code is not much. At the top level, saying:
1559    
1560            Control.MLRISC.getFlag "cfg-graphical-view" := true;
1561    
1562    will display the graphical view of the control flow graph just before
1563    back-patching.  daVinci must be in your path for this to work. If
1564    daVinci is not available, then the default viewer can be changed
1565    using:
1566    
1567            Control.MLRISC.getString "viewer"
1568    
1569    which can be set to "dot" or "vcg" for the corresponding viewers. Of
1570    course, these viewers must be in your path.
1571    
1572    The above will display the compilation unit at the level of clusters,
1573    many of which are small, boring, and un-interesting. Also setting:
1574    
1575            Control.MLRISC.getInt "cfg-graphical-view_size"
1576    
1577    will display clusters that are larger than the value set by the above.
1578    
1579    
1580    ----------------------------------------------------------------------
1581    Name: Matthias Blume
1582    Date: 2002/03/21 22:20:00 EST
1583    Tag: blume-20020321-kmp-bugfix
1584    Description:
1585    
1586    Changed the interface to the KMP routine in PreString and fixed
1587    a minor bug in one place where it was used.
1588    
1589    ----------------------------------------------------------------------
1590    Name: Allen Leung
1591    Date: 2002/03/21 20:30:00 EST
1592    Tag: leunga-20020321-cfg
1593    Description:
1594    
1595      Fixed a potential problem in cfg edge splitting.
1596    
1597    ----------------------------------------------------------------------
1598    Name: Allen Leung
1599    Date: 2002/03/21 17:15:00 EST
1600    Tag: leunga-20020321-x86-fp-cfg
1601    Description:
1602    
1603      1. Recoded the buggy parts of x86-fp.
1604    
1605         a. All the block reordering code has been removed.
1606            We now depend on the block placement phases to do this work.
1607    
1608         b. Critical edge splitting code has been simplified and moved into the
1609            CFG modules, as where they belong.
1610    
1611         Both of these were quite buggy and complex.  The code is now much, much
1612         simpler.
1613    
1614      2. X86 backend.
1615    
1616         a. Added instructions for 64-bit support.  Instruction selection for
1617            64-bit has not been committed, however, since that
1618            requires changes to MLTREE which haven't been approved by
1619            Lal and John.
1620    
1621         b. Added support for FUCOMI and FUCOMIP when generating code for
1622            PentiumPro and above.  We only generate these instructions in
1623            the fast-fp mode.
1624    
1625         c. Added cases for JP and JNP in X86FreqProps.
1626    
1627      3. CFG
1628    
1629         CFG now has a bunch of methods for edge splitting and merging.
1630    
1631      4. Machine description.
1632    
1633         John's simplification of MLTREE_BASIS.fcond broke a few machine
1634         description things:
1635    
1636         rtl-build.{sig,sml} and hppa.mdl fixed.
1637    
1638         NOTE: the machine description stuff in the repository is still broken.
1639               Again, I can't put my fixes in because that involves
1640               changes to MLTREE.
1641    
1642    ----------------------------------------------------------------------
1643    Name: Matthias Blume
1644    Date: 2002/03/20 15:55:00 EST
1645    Tag: blume-20020320-kmp
1646    Description:
1647    
1648    Implemented Knuth-Morris-Pratt string matching in PreString and used
1649    it for String.isSubstring, Substring.isSubstring, and
1650    Substring.position.
1651    
1652    (Might need some stress-testing.  Simple examples worked fine.)
1653    
1654    ----------------------------------------------------------------------
1655    Name: Matthias Blume
1656    Date: 2002/03/19 16:37:00 EST
1657    Tag: blume-20020319-witnesses
1658    Description:
1659    
1660    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
1661    
1662    This implements a generic mechanism for changing constness qualifiers
1663    anywhere within big C types without resorting to outright "casts".
1664    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
1665    modify the constness at the outermost level.)
1666    The implementation of "convert" is based on the idea of "witness"
1667    values -- values that are not used by the operation but whose types
1668    "testify" to their applicability.  On the implementation side, "convert"
1669    is simply a projection (returning its second curried argument).  With
1670    cross-module inlining, it should not result in any machine code being
1671    generated.
1672    
1673    ----------------------------------------------------------------------
1674    Name: Matthias Blume
1675    Date: 2002/03/15 16:40:00 EST
1676    Tag: blume-20020315-basis
1677    Description:
1678    
1679    Provided (preliminary?) implementations for
1680    
1681      {String,Substring}.{concatWith,isSuffix,isSubstring}
1682    
1683    and
1684    
1685      Substring.full
1686    
1687    Those are in the Basis spec but they were missing in SML/NJ.
1688    
1689    ----------------------------------------------------------------------
1690    Name: Matthias Blume
1691    Date: 2002/03/14 21:30:00 EST
1692    Tag: blume-20020314-controls
1693    Description:
1694    
1695    Controls:
1696    ---------
1697    
1698    1. Factored out the recently-added Controls : CONTROLS stuff and put
1699       it into its own library $/controls-lib.cm.  The source tree for
1700       this is under src/smlnj-lib/Controls.
1701    
1702    2. Changed the names of types and functions in this interface, so they
1703       make a bit more "sense":
1704    
1705          module -> registry
1706          'a registry -> 'a group
1707    
1708    3. The interface now deals in ref cells only.  The getter/setter interface
1709       is (mostly) gone.
1710    
1711    4. Added a function that lets one register an already-existing ref cell.
1712    
1713    5. Made the corresponding modifications to the rest of the code so that
1714       everything compiles again.
1715    
1716    6. Changed the implementation of Controls.MLRISC back to something closer
1717       to the original.  In particular, this module (and therefore MLRISC)
1718       does not depend on Controls.  There now is some link-time code in
1719       int-sys.sml that registers the MLRISC controls with the Controls
1720       module.
1721    
1722    CM:
1723    ---
1724    
1725      * One can now specify the lambda-split aggressiveness in init.cmi.
1726    
1727    ----------------------------------------------------------------------
1728    Name: Allen Leung
1729    Date: 2002/03/13 17:30:00 EST
1730    Tag: leunga-20020313-x86-fp-unary
1731    Description:
1732    
1733    Bug fix for:
1734    
1735    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
1736    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
1737    > - fun f(x,(y,z)) = Real.~ y;
1738    > [autoloading]
1739    > [autoloading done]
1740    >       fchsl   (%eax), 184(%esp)
1741    > Error: MLRisc bug: X86MCEmitter.emitInstr
1742    >
1743    > uncaught exception Error
1744    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
1745    
1746    The problem was that the code generator did not generate any fp registers
1747    in this case, and the ra didn't know that it needed to run the X86FP phase to
1748    translate the pseudo fp instruction.   This only happened with unary fp
1749    operators in certain situations.
1750    
1751    ----------------------------------------------------------------------
1752    Name: Matthias Blume
1753    Date: 2002/03/13 14:00:00 EST
1754    Tag: blume-20020313-overload-etc
1755    Description:
1756    
1757    1. Added _overload as a synonym for overload for backward compatibility.
1758       (Control.overloadKW must be true for either version to be accepted.)
1759    
1760    2. Fixed bug in install script that caused more things to be installed
1761       than what was requested in config/targets.
1762    
1763    3. Made CM aware of the (_)overload construct so that autoloading
1764       works.
1765    
1766    ----------------------------------------------------------------------
1767    Name: Matthias Blume
1768    Date: 2002/03/12 22:03:00 EST
1769    Tag: blume-20020312-url
1770    Description:
1771    
1772    Forgot to update BOOT and srcarchiveurl.
1773    
1774    ----------------------------------------------------------------------
1775    Name: Matthias Blume
1776    Date: 2002/03/12 17:30:00 EST
1777    Tag: blume-20020312-version110392
1778    Description:
1779    
1780    Yet another version number bump (because of small changes to the
1781    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
1782    
1783    Changes:
1784    
1785      The new pid generation scheme described a few weeks ago was overly
1786      complicated.  I implemented a new mechanism that is simpler and
1787      provides a bit more "stability":  Once CM has seen a compilation
1788      unit, it keeps its identity constant (as long as you do not delete
1789      those crucial CM/GUID/* files).  This means that when you change
1790      an interface, compile, then go back to the old interface, and
1791      compile again, you arrive at the original pid.
1792    
1793      There now also is a mechanism that instructs CM to use the plain
1794      environment hash as a module's pid (effectively making its GUID
1795      the empty string).  For this, "noguid" must be specified as an
1796      option to the .sml file in question within its .cm file.
1797      This is most useful for code that is being generated by tools such
1798      as ml-nlffigen (because during development programmers tend to
1799      erase the tool's entire output directory tree including CM's cached
1800      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
1801      revert to the old, broken behavior of SML/NJ, but in specific cases
1802      where there is no danger of interface confusion, its use is ok
1803      (I think).
1804    
1805      ml-nlffigen by default generates "noguid" annotations.  They can be
1806      turned off by specifying -guid in its command line.
1807    
1808    ----------------------------------------------------------------------
1809    Name: Lal George
1810    Date: 2002/03/12 12 14:42:36 EST
1811    Tag: george-20020312-frequency-computation
1812    Description:
1813    
1814    Integrated jump chaining and static block frequency into the
1815    compiler. More details and numbers later.
1816    
1817    ----------------------------------------------------------------------
1818    Name: Lal George
1819    Date: 2002/03/11 11 22:38:53 EST
1820    Tag: george-20020311-jump-chain-elim
1821    Description:
1822    
1823    Tested the jump chain elimination on all architectures (except the
1824    hppa).  This is on by default right now and is profitable for the
1825    alpha and x86, however, it may not be profitable for the sparc and ppc
1826    when compiling the compiler.
1827    
1828    The gc test will typically jump to a label at the end of the cluster,
1829    where there is another jump to an external cluster containing the actual
1830    code to invoke gc. This is to allow factoring of common gc invocation
1831    sequences. That is to say, we generate:
1832    
1833            f:
1834               testgc
1835               ja   L1      % jump if above to L1
1836    
1837            L1:
1838               jmp L2
1839    
1840    
1841    After jump chain elimination the 'ja L1' instructions is converted to
1842    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
1843    up being implemented in their long form (if L2 is far away) using:
1844    
1845            jbe     L3      % jump if below or equal to L3
1846            jmp     L2
1847         L3:
1848            ...
1849    
1850    
1851    For large compilation units L2  may be far away.
1852    
1853    
1854    ----------------------------------------------------------------------
1855    Name: Matthias Blume
1856    Date: 2002/03/11 13:30:00 EST
1857    Tag: blume-20020311-mltreeeval
1858    Description:
1859    
1860    A functor parameter was missing.
1861    
1862    ----------------------------------------------------------------------
1863    Name: Allen Leung
1864    Date: 2002/03/11 10:30:00 EST
1865    Tag: leunga-20020311-runtime-string0
1866    Description:
1867    
1868       The representation of the empty string now points to a
1869    legal null terminated C string instead of unit.  It is now possible
1870    to convert an ML string into C string with InlineT.CharVector.getData.
1871    This compiles into one single machine instruction.
1872    
1873    ----------------------------------------------------------------------
1874    Name: Allen Leung
1875    Date: 2002/03/10 23:55:00 EST
1876    Tag: leunga-20020310-x86-call
1877    Description:
1878    
1879       Added machine generation for CALL instruction (relative displacement mode)
1880    
1881    ----------------------------------------------------------------------
1882    Name: Matthias Blume
1883    Date: 2002/03/08 16:05:00
1884    Tag: blume-20020308-entrypoints
1885    Description:
1886    
1887    Version number bumped to 110.39.1.  NEW BOOTFILES!
1888    
1889    Entrypoints: non-zero offset into a code object where execution should begin.
1890    
1891    - Added the notion of an entrypoint to CodeObj.
1892    - Added reading/writing of entrypoint info to Binfile.
1893    - Made runtime system bootloader aware of entrypoints.
1894    - Use the address of the label of the first function given to mlriscGen
1895      as the entrypoint.  This address is currently always 0, but it will
1896      not be 0 once we turn on block placement.
1897    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
1898      with entry points) from mlriscGen.
1899    
1900    ----------------------------------------------------------------------
1901    Name: Allen Leung
1902    Date: 2002/03/07 20:45:00 EST
1903    Tag: leunga-20020307-x86-cmov
1904    Description:
1905    
1906       Bug fixes for CMOVcc on x86.
1907    
1908       1. Added machine code generation for CMOVcc
1909       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
1910       3. CMOVcc cannot have an immediate operand as argument.
1911    
1912    ----------------------------------------------------------------------
1913    Name: Matthias Blume
1914    Date: 2002/03/07 16:15:00 EST
1915    Tag: blume-20020307-controls
1916    Description:
1917    
1918    This is a very large but mostly boring patch which makes (almost)
1919    every tuneable compiler knob (i.e., pretty much everything under
1920    Control.* plus a few other things) configurable via both the command
1921    line and environment variables in the style CM did its configuration
1922    until now.
1923    
1924    Try starting sml with '-h' (or, if you are brave, '-H')
1925    
1926    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
1927    implements the underlying generic mechanism.
1928    
1929    The interface to some of the existing such facilities has changed somewhat.
1930    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
1931    (The getFoo interface is still there for backward-compatibility, but its
1932    use is deprecated.)
1933    
1934    The ml-build script passes -Cxxx=yyy command-line arguments through so
1935    that one can now twiddle the compiler settings when using this "batch"
1936    compiler.
1937    
1938    TODO items:
1939    
1940    We should go through and throw out all controls that are no longer
1941    connected to anything.  Moreover, we should go through and provide
1942    meaningful (and correct!) documentation strings for those controls
1943    that still are connected.
1944    
1945    Currently, multiple calls to Controls.new are accepted (only the first
1946    has any effect).  Eventually we should make sure that every control
1947    is being made (via Controls.new) exactly once.  Future access can then
1948    be done using Controls.acc.
1949    
1950    Finally, it would probably be a good idea to use the getter-setter
1951    interface to controls rather than ref cells.  For the time being, both
1952    styles are provided by the Controls module, but getter-setter pairs are
1953    better if thread-safety is of any concern because they can be wrapped.
1954    
1955    *****************************************
1956    
1957    One bug fix: The function blockPlacement in three of the MLRISC
1958    backpatch files used to be hard-wired to one of two possibilities at
1959    link time (according to the value of the placementFlag).  But (I
1960    think) it should rather sense the flag every time.
1961    
1962    *****************************************
1963    
1964    Other assorted changes (by other people who did not supply a HISTORY entry):
1965    
1966    1. the cross-module inliner now works much better (Monnier)
1967    2. representation of weights, frequencies, and probabilities in MLRISC
1968       changed in preparation of using those for weighted block placement
1969       (Reppy, George)
1970    
1971    ----------------------------------------------------------------------
1972    Name: Lal George
1973    Date: 2002/03/07 14:44:24 EST 2002
1974    Tag: george-20020307-weighted-block-placement
1975    
1976    Tested the weighted block placement optimization on all architectures
1977    (except the hppa) using AMPL to generate the block and edge frequencies.
1978    Changes were required in the machine properties to correctly
1979    categorize trap instructions. There is an MLRISC flag
1980    "weighted-block-placement" that can be used to enable weighted block
1981    placement, but this will be ineffective without block/edge
1982    frequencies (coming soon).
1983    
1984    
1985    ----------------------------------------------------------------------
1986    Name: Lal George
1987    Date: 2002/03/05 17:24:48 EST
1988    Tag: george-20020305-linkage-cluster
1989    
1990    In order to support the block placement optimization, a new cluster
1991    is generated as the very first cluster (called the linkage cluster).
1992    It contains a single jump to the 'real' entry point for the compilation
1993    unit. Block placement has no effect on the linkage cluster itself, but
1994    all the other clusters  have full freedom in the manner in which they
1995    reorder blocks or functions.
1996    
1997    On the x86 the typical linkage code that is generated is:
1998       ----------------------
1999            .align 2
2000       L0:
2001            addl    $L1-L0, 72(%esp)
2002            jmp     L1
2003    
2004    
2005            .align  2
2006       L1:
2007       ----------------------
2008    
2009    72(%esp) is the memory location for the stdlink register. This
2010    must contain the address of the CPS function being called. In the
2011    above example, it contains the address of  L0; before
2012    calling L1 (the real entry point for the compilation unit), it
2013    must contain the address for L1, and hence
2014    
2015            addl $L1-L0, 72(%esp)
2016    
2017    I have tested this on all architectures except the hppa.The increase
2018    in code size is of course negligible
2019    
2020    ----------------------------------------------------------------------
2021    Name: Allen Leung
2022    Date: 2002/03/03 13:20:00 EST
2023    Tag: leunga-20020303-mlrisc-tools
2024    
2025      Added #[ ... ] expressions to mlrisc tools
2026    
2027    ----------------------------------------------------------------------
2028    Name: Matthias Blume
2029    Date: 2002/02/27 12:29:00 EST
2030    Tag: blume-20020227-cdebug
2031    Description:
2032    
2033    - made types in structure C and C_Debug to be equal
2034    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
2035    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
2036    
2037    ----------------------------------------------------------------------
2038    Name: Matthias Blume
2039    Date: 2002/02/26 12:00:00 EST
2040    Tag: blume-20020226-ffi
2041    Description:
2042    
2043    1. Fixed a minor bug in CM's "noweb" tool:
2044       If numbering is turned off, then truly don't number (i.e., do not
2045       supply the -L option to noweb).  The previous behavior was to supply
2046       -L'' -- which caused noweb to use the "default" line numbering scheme.
2047       Thanks to Chris Richards for pointing this out (and supplying the fix).
2048    
2049    2. Once again, I reworked some aspects of the FFI:
2050    
2051       A. The incomplete/complete type business:
2052    
2053       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
2054         gone!
2055       - ML types representing an incomplete type are now *equal* to
2056         ML types representing their corresponding complete types (just like
2057         in C).  This is still safe because ml-nlffigen will not generate
2058         RTTI for incomplete types, nor will it generate functions that
2059         require access to such RTTI.   But when ML code generated from both
2060         incomplete and complete versions of the C type meet, the ML types
2061         are trivially interoperable.
2062    
2063         NOTE:  These changes restore the full generality of the translation
2064         (which was previously lost when I eliminated functorization)!
2065    
2066       B. Enum types:
2067    
2068       - Structure C now has a type constructor "enum" that is similar to
2069         how the "su" constructor works.  However, "enum" is not a phantom
2070         type because each "T enum" has values (and is isomorphic to
2071         MLRep.Signed.int).
2072       - There are generic access operations for enum objects (using
2073         MLRep.Signed.int).
2074       - ml-nlffigen will generate a structure E_foo for each "enum foo".
2075         * The structure contains the definition of type "mlrep" (the ML-side
2076         representation type of the enum).  Normally, mlrep is the same
2077         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
2078         then mlrep will be defined as a datatype -- thus facilitating
2079         pattern matching on mlrep values.
2080         ("-ec" will be suppressed if there are duplicate values in an
2081          enumeration.)
2082         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
2083         will be generated for each C enum constant xxx.
2084         * Conversion functions m2i and i2m convert between mlrep and
2085         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
2086         * Coversion functions c and ml convert between mlrep and "tag enum".
2087         * Access functions (get/set) fetch and store mlrep values.
2088       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
2089         enumerations are merged into one single enumeration represented by
2090         structure E_'.
2091    
2092    ----------------------------------------------------------------------
2093    Name: Allen Leung
2094    Date: 2002/02/25 04:45:00 EST
2095    Tag: leunga-20020225-cps-spill
2096    
2097    This is a new implementation of the CPS spill phase.
2098    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
2099    In case of problems, replace it with the old file spill.sml
2100    
2101    The current compiler runs into some serious performance problems when
2102    constructing a large record.  This can happen when we try to compile a
2103    structure with many items.  Even a very simple structure like the following
2104    makes the compiler slow down.
2105    
2106        structure Foo = struct
2107           val x_1 = 0w1 : Word32.int
2108           val x_2 = 0w2 : Word32.int
2109           val x_3 = 0w3 : Word32.int
2110           ...
2111           val x_N = 0wN : Word32.int
2112        end
2113    
2114    The following table shows the compile time, from N=1000 to N=4000,
2115    with the old compiler:
2116    
2117    N
2118    1000   CPS 100 spill                           0.04u  0.00s  0.00g
2119           MLRISC ra                               0.06u  0.00s  0.05g
2120              (spills = 0 reloads = 0)
2121           TOTAL                                   0.63u  0.07s  0.21g
2122    
2123    1100   CPS 100 spill                           8.25u  0.32s  0.64g
2124           MLRISC ra                               5.68u  0.59s  3.93g
2125              (spills = 0 reloads = 0)
2126           TOTAL                                   14.71u  0.99s  4.81g
2127    
2128    1500   CPS 100 spill                           58.55u  2.34s  1.74g
2129           MLRISC ra                               5.54u  0.65s  3.91g
2130              (spills = 543 reloads = 1082)
2131           TOTAL                                   65.40u  3.13s  6.00g
2132    
2133    2000   CPS 100 spill                           126.69u  4.84s  3.08g
2134           MLRISC ra                               0.80u  0.10s  0.55g
2135              (spills = 42 reloads = 84)
2136           TOTAL                                   129.42u  5.10s  4.13g
2137    
2138    3000   CPS 100 spill                           675.59u  19.03s  11.64g
2139           MLRISC ra                               2.69u  0.27s  1.38g
2140              (spills = 62 reloads = 124)
2141           TOTAL                                   682.48u  19.61s  13.99g
2142    
2143    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
2144           MLRISC ra                               4.96u  0.27s  2.72g
2145              (spills = 85 reloads = 170)
2146           TOTAL                                   2375.26u  57.21s  48.00g
2147    
2148    As you can see the old cps spill module suffers from some serious
2149    performance problem.  But since I cannot decipher the old code fully,
2150    instead of patching the problems up, I'm reimplementing it
2151    with a different algorithm.  The new code is more modular,
2152    smaller when compiled, and substantially faster
2153    (O(n log n) time and O(n) space).  Timing of the new spill module:
2154    
2155    4000  CPS 100 spill                           0.02u  0.00s  0.00g
2156          MLRISC ra                               0.25u  0.02s  0.15g
2157             (spills=1 reloads=3)
2158          TOTAL                                   7.74u  0.34s  1.62g
2159    
2160    Implementation details:
2161    
2162    As far as I can tell, the purpose of the CPS spill module is to make sure the
2163    number of live variables at any program point (the bandwidth)
2164    does not exceed a certain limit, which is determined by the
2165    size of the spill area.
2166    
2167    When the bandwidth is too large, we decrease the register pressure by
2168    packing live variables into spill records.  How we achieve this is
2169    completely different than what we did in the old code.
2170    
2171    First, there is something about the MLRiscGen code generator
2172    that we should be aware of:
2173    
2174    o MLRiscGen performs code motion!
2175    
2176       In particular, it will move floating point computations and
2177       address computations involving only the heap pointer to
2178       their use sites (if there is only a single use).
2179       What this means is that if we have a CPS record construction
2180       statement
2181    
2182           RECORD(k,vl,w,e)
2183    
2184       we should never count the new record address w as live if w
2185       has only one use (which is often the case).
2186    
2187       We should do something similar to floating point, but the transformation
2188       there is much more complex, so I won't deal with that.
2189    
2190    Secondly, there are now two new cps primops at our disposal:
2191    
2192     1. rawrecord of record_kind option
2193        This pure operator allocates some uninitialized storage from the heap.
2194        There are two forms:
2195    
2196         rawrecord NONE [INT n]  allocates a tagless record of length n
2197         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
2198                                     and initializes the tag.
2199    
2200     2. rawupdate of cty
2201          rawupdate cty (v,i,x)
2202          Assigns to x to the ith component of record v.
2203          The storelist is not updated.
2204    
2205    We use these new primops for both spilling and increment record construction.
2206    
2207     1. Spilling.
2208    
2209        This is implemented with a linear scan algorithm (but generalized
2210        to trees).  The algorithm will create a single spill record at the
2211        beginning of the cps function and use rawupdate to spill to it,
2212        and SELECT or SELp to reload from it.  So both spills and reloads
2213        are fine-grain operations.  In contrast, in the old algorithm
2214        "spills" have to be bundled together in records.
2215    
2216        Ideally, we should sink the spill record construction to where
2217        it is needed.  We can even split the spill record into multiple ones
2218        at the places where they are needed.  But CPS is not a good
2219        representation for global code motion, so I'll keep it simple and
2220        am not attempting this.
2221    
2222     2. Incremental record construction (aka record splitting).
2223    
2224        Long records with many component values which are simulatenously live
2225        (recall that single use record addresses are not considered to
2226         be live) are constructed with rawrecord and rawupdate.
2227        We allocate space on the heap with rawrecord first, then gradually
2228        fill it in with rawupdate.  This is the technique suggested to me
2229        by Matthias.
2230    
2231        Some restrictions on when this is applicable:
2232        1. It is not a VECTOR record.  The code generator currently does not handle
2233           this case. VECTOR record uses double indirection like arrays.
2234        2. All the record component values are defined in the same "basic block"
2235           as the record constructor.  This is to prevent speculative
2236           record construction.
2237    
2238    ----------------------------------------------------------------------
2239    Name: Allen Leung
2240    Date: 2002/02/22 01:02:00 EST
2241    Tag: leunga-20020222-mlrisc-tools
2242    
2243    Minor bug fixes in the parser and rewriter
2244    
2245    ----------------------------------------------------------------------
2246    Name: Allen Leung
2247    Date: 2002/02/21 20:20:00 EST
2248    Tag: leunga-20020221-peephole
2249    
2250    Regenerated the peephole files.  Some contained typos in the specification
2251    and some didn't compile because of pretty printing bugs in the old version
2252    of 'nowhere'.
2253    
2254    ----------------------------------------------------------------------
2255    Name: Allen Leung
2256    Date: 2002/02/19 20:20:00 EST
2257    Tag: leunga-20020219-mlrisc-tools
2258    Description:
2259    
2260       Minor bug fixes to the mlrisc-tools library:
2261    
2262       1.  Fixed up parsing colon suffixed keywords
2263       2.  Added the ability to shut the error messages up
2264       3.  Reimplemented the pretty printer and fixed up/improved
2265           the pretty printing of handle and -> types.
2266       4.  Fixed up generation of literal symbols in the nowhere tool.
2267       5.  Added some SML keywords to to sml.sty
2268    
2269    ----------------------------------------------------------------------
2270    Name: Matthias Blume
2271    Date: 2002/02/19 16:20:00 EST
2272    Tag: blume-20020219-cmffi
2273    Description:
2274    
2275    A wild mix of changes, some minor, some major:
2276    
2277    * All C FFI-related libraries are now anchored under $c:
2278        $/c.cm      --> $c/c.cm
2279        $/c-int.cm  --> $c/internals/c-int.cm
2280        $/memory.cm --> $c/memory/memory.cm
2281    
2282    * "make" tool (in CM) now treats its argument pathname slightly
2283      differently:
2284        1. If the native expansion is an absolute name, then before invoking
2285           the "make" command on it, CM will apply OS.Path.mkRelative
2286           (with relativeTo = OS.FileSys.getDir()) to it.
2287        2. The argument will be passed through to subsequent phases of CM
2288           processing without "going native".  In particular, if the argument
2289           was an anchored path, then "make" will not lose track of that anchor.
2290    
2291    * Compiler backends now "know" their respective C calling conventions
2292      instead of having to be told about it by ml-nlffigen.  This relieves
2293      ml-nlffigen from one of its burdens.
2294    
2295    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
2296    
2297    * Export C_DEBUG and C_Debug from $c/c.cm.
2298    
2299    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
2300      subtyping relationship between incomplete pointers and their complete
2301      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
2302      with the convention of instantiating 'o with ('t, 'c) obj whenever
2303      the pointer target type is complete.  In the incomplete case, 'o
2304      will be instantiated with some "'c iobj" -- a type obtained by
2305      using one of the functors PointerToIncompleteType or PointerToCompleteType.
2306    
2307      Operations that work on both incomplete and complete pointer types are
2308      typed as taking an 'o ptr while operations that require the target to
2309      be known are typed as taking some ('t, 'c) obj ptr.
2310    
2311      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
2312      where void is an eqtype without any values.  This makes it possible
2313      to work on voidptr values using functions meant to operate on light
2314      incomplete pointers.
2315    
2316    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
2317      been vastly simplified.
2318    
2319    ----------------------------------------------------------------------
2320    Name: Matthias Blume
2321    Date: 2002/02/19 10:48:00 EST
2322    Tag: blume-20020219-pqfix
2323    Description:
2324    
2325    Applied Chris Okasaki's bug fix for priority queues.
2326    
2327    ----------------------------------------------------------------------
2328    Name: Matthias Blume
2329    Date: 2002/02/15 17:05:00
2330    Tag: Release_110_39
2331    Description:
2332    
2333    Last-minute retagging is becoming a tradition... :-(
2334    
2335    This is the working release 110.39.
2336    
2337    ----------------------------------------------------------------------
2338    Name: Matthias Blume
2339    Date: 2002/02/15 16:00:00 EST
2340    Tag: Release_110_39-orig
2341    Description:
2342    
2343    Working release 110.39.  New bootfiles.
2344    
2345    (Update: There was a small bug in the installer so it wouldn't work
2346    with all shells.  So I retagged. -Matthias)
2347    
2348    ----------------------------------------------------------------------
2349    Name: Matthias Blume
2350    Date: 2002/02/15 14:17:00 EST
2351    Tag: blume-20020215-showbindings
2352    Description:
2353    
2354    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
2355    the latter can be useful for exploring what bindings are available at
2356    the interactive prompt.  (The first function returns only the list
2357    of symbols that are really bound, the second prints those but also the
2358    ones that CM's autoloading mechanism knows about.)
2359    
2360    ----------------------------------------------------------------------
2361    Name: Matthias Blume
2362    Date: 2002/02/15 12:08:00 EST
2363    Tag: blume-20020215-iptrs
2364    Description:
2365    
2366    Two improvements to ml-nlffigen:
2367    
2368      1. Write files only if they do not exist or if their current contents
2369         do not coincide with what's being written.  (That is, avoid messing
2370         with the time stamps unless absolutely necessary.)
2371    
2372      2. Implement a "repository" mechanism for generated files related
2373         to "incomplete pointer types".   See the README file for details.
2374    
2375    ----------------------------------------------------------------------
2376    Name: Matthias Blume
2377    Date: 2002/02/14 11:50:00 EST
2378    Tag: blume-20020214-quote
2379    Description:
2380    
2381    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
2382    because of the new and improved tag generation scheme.  (Thanks to Allen
2383    Leung for pointing it out.)
2384    
2385    ----------------------------------------------------------------------
2386    Name: Lal George
2387    Date: 2002/02/14 09:55:27 EST 2002
2388    Tag: george-20020214-isabelle-bug
2389    Description:
2390    
2391    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
2392    of Isabelle on the x86.
2393    
2394    From Allen:
2395    -----------
2396     I've found the problem:
2397    
2398         in ra-core.sml, I use the counter "blocked" to keep track of the
2399         true number of elements in the freeze queue.  When the counter goes
2400         to zero, I skip examining the queue.  But I've messed up the
2401         bookkeeping in combine():
2402    
2403             else ();
2404             case !ucol of
2405               PSEUDO => (if !cntv > 0 then
2406                     (if !cntu > 0 then blocked := !blocked - 1 else ();
2407                                        ^^^^^^^^^^^^^^^^^^^^^^^
2408                      moveu := mergeMoveList(!movev, !moveu)
2409                     )
2410                  else ();
2411    
2412         combine() is called to coalesce two nodes u and v.
2413         I think I was thinking that if the move counts of u and v are both
2414         greater than zero then after they are coalesced then one node is
2415         removed from the freeze queue.  Apparently I was thinking that
2416         both u and v are of low degree, but that's clearly not necessarily true.
2417    
2418    
2419    02/12/2002:
2420        Here's the patch.  HOL now compiles.
2421    
2422        I don't know how this impact on performance (compile
2423        time or runtime).  This bug caused the RA (especially on the x86)
2424        to go thru the potential spill phase when there are still nodes on the
2425        freeze queue.
2426    
2427    
2428    
2429    
2430    ----------------------------------------------------------------------
2431    Name: Matthias Blume
2432    Date: 2002/02/13 22:40:00 EST
2433    Tag: blume-20020213-fptr-rtti
2434    Description:
2435    
2436    Fixed a bug in ml-nlffigen that was introduced with one of the previous
2437    updates.
2438    
2439    ----------------------------------------------------------------------
2440    Name: Matthias Blume
2441    Date: 2002/02/13 16:41:00 EST
2442    Tag: blume-20020213-cmlpq
2443    Description:
2444    
2445    Added new priority queue export symbols (which have just been added to
2446    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
2447    not compile and the installer would choke.)
2448    
2449    ----------------------------------------------------------------------
2450    Name: Matthias Blume
2451    Date: 2002/02/13 16:15:00 EST
2452    Tag: blume-20020213-various
2453    Description:
2454    
2455    1. More tweaks to ml-nlffigen:
2456    
2457       - better internal datastructures (resulting in slight speedup)
2458       - "-match" option requires exact match
2459       - "localized" gensym counters (untagged structs/unions nested within
2460         other structs/unions or within typedefs get a fresh counter; their
2461         tag will be prefixed by a concatenation of their parents' tags)
2462       - bug fixes (related to calculation of transitive closure of types
2463         to be included in the output)
2464    
2465    2. Minor Basis updates:
2466    
2467       - added implementations for List.collate and Option.app
2468    
2469    ----------------------------------------------------------------------
2470    Name: Matthias Blume
2471    Date: 2002/02/11 15:55:00 EST
2472    Tag: blume-20020211-gensym
2473    Description:
2474    
2475    Added a "-gensym" option to command line of ml-nlffigen.  This can be
2476    used to specify a "stem" -- a string that is inserted in all "gensym'd"
2477    names (ML structure names that correspond to unnamed C structs, unions,
2478    and enums), so that separate runs of ml-nlffigen do not clash.
2479    
2480    ----------------------------------------------------------------------
2481    Name: Matthias Blume
2482    Date: 2002/02/11 12:05:00 EST
2483    Tag: blume-20020211-gensml
2484    Description:
2485    
2486    A quick fix for a problem with GenSML (in the pgraph-util library):
2487    Make generation of toplevel "local" optional.  (Strictly speaking,
2488    signature definitions within "local" are not legal SML.)
2489    
2490    Other than that: updates to INSTALL and cm/TODO.
2491    
2492    ----------------------------------------------------------------------
2493    Name: Matthias Blume
2494    Date: 2002/02/08 15:00:00 EST
2495    Tag: blume-20020208-uniquepid
2496    Description:
2497    
2498    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
2499    
2500    1. The installer (config/install.sh) has gotten smarter:
2501    
2502         - Configuration options are a bit easier to specify now
2503           (in config/targets).
2504         - Bug in recognizing .tar.bz2 files fixed.
2505         - Installer automatically resolves dependencies between
2506           configuration options (e.g., if you ask for eXene, you will
2507           also get cml -- regardless whether you asked for it or not).
2508         - Installer can run in "quieter mode" by setting the environment
2509           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
2510           "completely silent", though.
2511         - Build HashCons library as part of smlnj-lib.
2512    
2513    2. A new scheme for assigning persistent identifiers to compilation
2514       units (and, by extension, to types etc.) has been put into place.
2515       This fixes a long-standing bug where types and even dynamic values
2516       can get internally confused, thereby compromising type safety
2517       (abstraction) and dynamic correctness.  See
2518    
2519         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
2520    
2521       for an example of how things could go wrong until now.
2522    
2523       The downside of the new scheme is that pids are not quite as
2524       persistent as they used to be: CM will generate a fresh pid
2525       for every compilation unit that it thinks it sees for the first
2526       time.  That means that if you compile starting from a clean, fresh
2527       source tree at two different times, you end up with different
2528       binaries.
2529    
2530       Cutoff recompilation, however, has not been compromised because
2531       CM keeps pid information in special caches between runs.
2532    
2533    ----------------------------------------------------------------------
2534    Name: Lal George
2535    Date: 2002/02/07 15:34:13 EST 2002
2536    Tag: <none>
2537    Description:
2538    
2539    Compilers that generate assembly code may produce  global labels
2540    whose value is resolved at link time. The various peephole optimization
2541    modules did not take this in account.
2542    
2543    TODO. The Labels.addrOf function should really return an option
2544    type so that clients are forced to deal with this issue, rather
2545    than an exception being raised.
2546    
2547    ----------------------------------------------------------------------
2548    Name: Lal George
2549    Date: 2002/02/06 13:55:02 EST
2550    Tag: george-20020206-ra-breakup
2551    Description:
2552    
2553    1. A bug fix from Allen.
2554    
2555        A typo causes extra fstp %st(0)'s to be generated at compensation
2556        edges, which might cause stack underflow traps at runtime.  This
2557        occurs in fft where there are extraneous fstps right before the 'into'
2558        trap instruction (in this case they are harmless since none of the
2559        integers overflow.)
2560    
2561    2. Pulled out various utility modules that were embedded in the modules
2562       of the register allocator. I need these modules for other purposes, but
2563       they are not complete enough to put into a library (just yet).
2564    ----------------------------------------------------------------------
2565    Name: Matthias Blume
2566    Date: 2002/01/31 16:05:00 EST
2567    Tag: blume-20020131-sparc-ccalls
2568    Description:
2569    
2570    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
2571       of extra stack space by mistake.  Fixed.
2572    
2573    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
2574    
2575    ----------------------------------------------------------------------
2576    Name: Allen Leung
2577    Date: 2002/01/30
2578    Tag: leunga-20020130-nowhere-bug-fix
2579    Description:
2580    
2581       MLRISC bug fixes:
2582       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
2583       2. MachineInt.fromString was negating its value.
2584    
2585    ----------------------------------------------------------------------
2586    Name: Matthias Blume
2587    Date: 2002/01/29
2588    Tag: blume-20020129-INSTALL
2589    Description:
2590    
2591    - Added somewhat detailed installation instructions (file INSTALL).
2592    - Fixed curl-detection bug in config/install.sh.
2593    - It is now possible to select the URL getter using the URLGETTER
2594      environment variable:
2595    
2596          not set / "unknown"      --> automatic detection (script tries wget,
2597                                       curl, and lynx)
2598          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
2599                                       how to properly invoke them)
2600          other                    --> use $URLGETTER directly, it must take
2601                                       precisely two command-line arguments
2602                                       (source URL and destination file name)
2603    
2604    ----------------------------------------------------------------------
2605    Name: Matthias Blume
2606    Date: 2002/01/28
2607    Tag: blume-20020128-sparc-ccalls
2608    Description:
2609    
2610    - Fixed problem with calculation of "used" registers in sparc-c-calls.
2611    - Make use of the allocParam argument in sparc-c-calls.
2612    
2613    ----------------------------------------------------------------------
2614    Name: Matthias Blume
2615    Date: 2002/01/28
2616    Tag: blume-20020128-allocParam
2617    Description:
2618    
2619    John Reppy:  Changes c-calls API to accept client-callback for
2620    allocating extra stack space.
2621    me: Corresponding changes to mlriscGen (using a dummy argument that
2622        does not change the current behavior).
2623    
2624    ----------------------------------------------------------------------
2625    Name: Matthias Blume
2626    Date: 2002/01/28 12:00:00
2627    Tag: Release_110_38
2628    Description:
2629    
2630    This time for real!!!
2631    
2632    ----------------------------------------------------------------------
2633    Name: Matthias Blume
2634    Date: 2002/01/28 10:56:00 EST
2635    Tag: blume-20020128-retraction
2636    Description:
2637    
2638    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
2639       with blume-Release_110_38-retracted.)
2640    
2641    1. Fixed a problem with incorrect rounding modes in real64.sml.
2642       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
2643    
2644    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
2645       and enums fixed.  The naming of corresponding ML identifiers should
2646       now be consistent again.
2647    
2648    ----------------------------------------------------------------------
2649    Name: Allen Leung
2650    Date: 2002/01/27
2651    Tag: leunga-20020127-nowhere
2652    Description:
2653    
2654       Added a target called nowhere in the configuration scripts.
2655       Enabling this will build the MLRISC 'nowhere' tool (for translating
2656       programs with where-clauses into legal SML code) during installation.
2657    
2658    ----------------------------------------------------------------------
2659    Name: Matthias Blume
2660    Date: 2002/01/25 21:27:00 EST
2661    Tag: blume-Release_110_38-retracted
2662    Description:
2663    
2664    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
2665    
2666    README will be added later.
2667    
2668    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
2669    (Release_110_38) removed.  Reason: Last-minute bug fixes.
2670    
2671    ----------------------------------------------------------------------
2672    Name: Matthias Blume
2673    Date: 2002/01/25
2674    Tag: blume-20020125-ffi
2675    Description:
2676    
2677    A large number of tweaks and improvements to ml-nlffi-lib and
2678    ml-nlffigen:
2679    
2680       - ML represenation types have been streamlined
2681       - getter and setter functions work with concrete values, not abstract
2682         ones where possible
2683       - ml-nlffigen command line more flexible (see README file there)
2684       - some bugs have been fixed (hopefully)
2685    
2686    ----------------------------------------------------------------------
2687    Name: Lal George
2688    Date: 2002/01/24
2689    Tag: george-20020124-risc-ra-interface
2690    Description:
2691    
2692       There is a dramatic simplification in the interface to the
2693       register allocator for RISC architectures as a result of making
2694       parallel copy instructions explicit.
2695    
2696    ----------------------------------------------------------------------
2697    Name: Matthias Blume
2698    Date: 2002/01/22
2699    Tag: blume-20020122-x86-ccalls
2700    Description:
2701    
2702    Bug fix for c-calls on x86 (having to do with how char- and
2703    short-arguments are being handled).
2704    
2705    ----------------------------------------------------------------------
2706    Name: Matthias Blume
2707    Date: 2002/01/21
2708    Tag: blume-20020121-ff
2709    Description:
2710    
2711    Another day of fiddling with the FFI...
2712    
2713    1. Bug fix/workaround:  CKIT does not complain about negative array
2714       dimensions, so ml-nlffigen has to guard itself against this possibility.
2715       (Otherwise a negative dimension would send it into an infinite loop.)
2716    
2717    2. Some of the abstract types (light objects, light pointers, most "base"
2718       types) in structure C are now eqtypes.
2719    
2720    3. Added constructors and test functions for NULL function pointers.
2721    
2722    ----------------------------------------------------------------------
2723    Name: Matthias Blume
2724    Date: 2002/01/18
2725    Tag: blume-20020118-ready-for-new-release
2726    Description:
2727    
2728    Made config/srcarchiveurl point to a new place.  (Will provide boot
2729    files shortly.)
2730    
2731    Maybe we christen this to be 110.38?
2732    
2733    ----------------------------------------------------------------------
2734    Name: Matthias Blume
2735    Date: 2002/01/18
2736    Tag: blume-20020118-more-ffifiddle
2737    Description:
2738    
2739    Today's FFI fiddling:
2740    
2741      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
2742        C.Set.* that use concrete (MLRep.*) arguments and results instead
2743        of abstract ones.
2744    
2745      - Provided word-style bit operations etc. for "int" representation
2746        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
2747        and Long.
2748    
2749    ----------------------------------------------------------------------
2750    Name: Matthias Blume
2751    Date: 2002/01/18
2752    Tag: blume-20020118-use-x86-fp
2753    Description:
2754    
2755    Now that x86-fast-fp seems to be working, I turned it back on again
2756    by default.  (Seems to work fine now, even with the FFI.)
2757    
2758    Other than that, I added some documentation about the FFI to
2759    src/ml-nlffigen/README and updated the FFI test examples in
2760    src/ml-nlffi-lib/Tests/*.
2761    
2762    ----------------------------------------------------------------------
2763    Name: Allen Leung
2764    Date: 2002/01/17
2765    Tag: leunga-20020117-x86-fast-fp-call
2766    Description:
2767    
2768       1. Fixed a problem with handling return fp values when x86's fast fp
2769          mode is turned on.
2770    
2771       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
2772          of %f32.
2773    
2774       3. Added a constructor INT32lit to the ast of MLRISC tools.
2775    
2776    ----------------------------------------------------------------------
2777    Name: Matthias Blume
2778    Date: 2002/01/16
2779    Tag: blume-20020116-ffifiddle
2780    Description:
2781    
2782    More fiddling with the FFI interface:
2783    
2784     - Make constness 'c instead of rw wherever possible.  This eliminates
2785       the need for certain explicit coercions.  (However, due to ML's
2786       value polymorphism, there will still be many cases where explicit
2787       coercions are necessary.  Phantom types are not the whole answer
2788       to modeling a subtyping relationship in ML.)
2789    
2790     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
2791    
2792     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
2793       clumsy workaround for varargs problem.)
2794    
2795    ----------------------------------------------------------------------
2796    Name: Lal George
2797    Date: 2002/01/15
2798    Tag: <none>
2799    Description:
2800    
2801    1. Since COPY instructions are no longer native to the architecture,
2802       a generic functor can be used to implement the expandCopies function.
2803    
2804    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
2805       TEXT segment.
2806    
2807    ----------------------------------------------------------------------
2808    Name: Matthias Blume
2809    Date: 2002/01/15
2810    Tag: blume-20020115-ffiupdates
2811    Description:
2812    
2813    1. Fix for bug resulting in single-precision float values being returned
2814       incorrectly from FFI calls.
2815    
2816    2. Small modifications to C FFI API:
2817    
2818        - memory-allocation routines return straight objects (no options)
2819          and raise an exception in out-of-memory situations
2820        - unsafe extensions to cast between function pointers and pointers
2821          from/to ints
2822        - added structure C_Debug as an alternative to structure C where
2823          pointer-dereferencing (|*| and |*!) always check for null-pointers
2824        - added open_lib' to DynLinkage;  open_lib' works like open_lib
2825          but also takes a (possibly empty) list of existing library handles
2826          that the current library depends on
2827    
2828    ----------------------------------------------------------------------
2829    Name: Matthias Blume
2830    Date: 2002/01/10
2831    Tag: blume-20020110-newffigen
2832    Description:
2833    
2834    1. Updates to portable graph code.
2835    
2836    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
2837       more scalable now so that even huge interfaces such as the one
2838       for GTK compile in finite time and space. :-)
2839       See src/ml-nlffigen/README for details on what's new.
2840    
2841    ----------------------------------------------------------------------
2842    Name: Lal George
2843    Date: 2001/01/09 14:31:35 EST 2002
2844    Tag: george-20011206-rm-native-copy
2845    Description:
2846    
2847            Removed the native COPY and FCOPY instructions
2848            from all the architectures and replaced it with the
2849            explicit COPY instruction from the previous commit.
2850    
2851            It is now possible to simplify many of the optimizations
2852            modules that manipulate copies. This has not been
2853            done in this change.
2854    
2855    ----------------------------------------------------------------------
2856    Name: Lal George
2857    Date: 2001/12/06 16:50:13 EST 2001
2858    Tag: george-20011206-mlrisc-instruction
2859    Description:
2860    
2861    Changed the representation of instructions from being fully abstract
2862    to being partially concrete. That is to say:
2863    
2864      from
2865            type instruction
2866    
2867      to
2868            type instr                              (* machine instruction *)
2869    
2870            datatype instruction =
2871                LIVE of {regs: C.cellset, spilled: C.cellset}
2872              | KILL of {regs: C.cellset, spilled: C.cellset}
2873              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
2874              | ANNOTATION of {i: instruction, a: Annotations.annotation}
2875              | INSTR of instr
2876    
2877    This makes the handling of certain special instructions that appear on
2878    all architectures easier and uniform.
2879    
2880    LIVE and KILL say that a list of registers are live or killed at the
2881    program point where they appear. No spill code is generated when an
2882    element of the 'regs' field is spilled, but the register is moved to
2883    the 'spilled' (which is present, more for debugging than anything else).
2884    
2885    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
2886    We used to generate:
2887    
2888            DEFFREG f1
2889            f1 := f2 + f3
2890            trapb
2891    
2892    but now generate:
2893    
2894            f1 := f2 + f3
2895            trapb
2896            LIVE {regs=[f1,f2,f3], spilled=[]}
2897    
2898    Furthermore, the DEFFREG (hack) required that all floating point instruction
2899    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
2900    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
2901    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
2902    this architecture).
2903    
2904    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
2905    all the architectures. This will result in further simplification of the
2906    register allocator that must be aware of them for coalescing purposes, and
2907    will also simplify certain aspects of the machine description that provides
2908    callbacks related to parallel copies.
2909    
2910    ANNOTATION should be obvious, and now INSTR represents the honest to God
2911    machine instruction set!
2912    
2913    The <arch>/instructions/<arch>Instr.sml files define certain utility
2914    functions for making porting easier -- essentially converting upper case
2915    to lower case. All machine instructions (of type instr) are in upper case,
2916    and the lower case form generates an MLRISC instruction. For example on
2917    the alpha we have:
2918    
2919      datatype instr =
2920         LDA of {r:cell, b:cell, d:operand}
2921       | ...
2922    
2923      val lda : {r:cell, b:cell, d:operand} -> instruction
2924        ...
2925    
2926    where lda is just (INSTR o LDA), etc.
2927    
2928    ----------------------------------------------------------------------
2929    Name: Matthias Blume
2930    Date: 2001/11/22 21:40:00 EST
2931    Tag: Release_110_37
2932    Description:
2933    
2934    Release 110.37.  This time for real.
2935    
2936    ----------------------------------------------------------------------
2937    Name: Matthias Blume
2938    Date: 2001/11/21 16:35:00 EST
2939    Tag: blume-20011121-foot-in-mouth
2940    Description:
2941    
2942    Removed the "Release_110_37" tag because of a serious bug.
2943    This will be re-tagged once the bug is fixed.
2944    
2945    ----------------------------------------------------------------------
2946    Name: Matthias Blume
2947    Date: 2001/11/21 16:14:00 EST
2948    Tag: blume-20011121-forgottenfile
2949    Description:
2950    
2951    Forgot to add a file.  (Just a .tex-file -- part of
2952    the CM manual source.)
2953    
2954    ----------------------------------------------------------------------
2955    Name: Matthias Blume
2956    Date: 2001/11/21 16:10:00 EST
2957    Tag: blume-20011121-invalid_110_37
2958    Description:
2959    
2960    Note: I removed the original tag "Release_110_37" from this commit
2961          because we found a serious bug in all non-x86 backends.
2962          - Matthias
2963    
2964    1. Modifications to the SML/NJ code generator and to the runtime system
2965       so that code object name strings are directly inserted into code
2966       objects at code generation time.  The only business the runtime system
2967       has with this is now to read the name strings on occasions.
2968       (The encoding of the name string has also changed somewhat.)
2969    
2970    2. CM now implements a simple "set calculus" for specifying export lists.
2971       In particular, it is now possible to refer to the export lists of
2972       other libraries/groups/sources and form unions as well as differences.
2973       See the latest CM manual for details.
2974    
2975    3. An separate notion of "proxy" libraries has again be eliminated from
2976       CM's model.  (Proxy libraries are now simply a special case of using
2977       the export list calculus.)
2978    
2979    4. Some of the existing libraries now take advantage of the new set
2980       calculus.
2981       (Notice that not all libraries have been converted because some
2982       of the existing .cm-files are supposed to be backward compatible
2983       with 110.0.x.)
2984    
2985    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
2986       "General.exnMessage"!  The former relies on a certain hook to be
2987       initialized, and that often does not happen in the stand-alone case.)
2988    
2989    ----------------------------------------------------------------------
2990    Name: Lal George
2991    Date: 2001/11/21  13:56:18 EST
2992    Tag: george-2001121-pseudo-ops
2993    Description:
2994    
2995      Implemented a complete redesign of MLRISC pseudo-ops. Now there
2996      ought to never be any question of incompatabilities with
2997      pseudo-op syntax expected by host assemblers.
2998    
2999      For now, only modules supporting GAS syntax are implemented
3000      but more should follow, such as MASM, and vendor assembler
3001      syntax, e.g. IBM as, Sun as, etc.
3002    
3003    ----------------------------------------------------------------------
3004  Name: Matthias Blume  Name: Matthias Blume
3005  Date: 2001/11/14 11:52:00 EST  Date: 2001/11/14 11:52:00 EST
3006  Tag: blume-20011114-srcname  Tag: blume-20011114-srcname
# Line 194  Line 3182 
3182    
3183  ----------------------------------------------------------------------  ----------------------------------------------------------------------
3184  Name: Matthias Blume  Name: Matthias Blume
 >>>>>>> 1.169  
3185  Date: 2001/09/18 15:35:00 EDT  Date: 2001/09/18 15:35:00 EDT
3186  Tag: blume-20010918-readme11036  Tag: blume-20010918-readme11036
3187  Description:  Description:

Legend:
Removed from v.977  
changed lines
  Added in v.1515

root@smlnj-gforge.cs.uchicago.edu
ViewVC Help
Powered by ViewVC 1.0.0