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