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