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 634, Mon May 1 10:03:50 2000 UTC revision 1494, Thu May 20 03:09:17 2004 UTC
# Line 8  Line 8 
8  The form of an entry should be:  The form of an entry should be:
9    
10  Name:  Name:
11  Date:  Date: yyyy/mm/dd
12  Tag: <post-commit CVS tag>  Tag: <post-commit CVS tag>
13  Description:  Description:
14    
15    ----------------------------------------------------------------------
16    Name: Matthias Blume (blume (at) tti - c (dot) org)
17    Date: 2004/05/19 22:10:00 CDT
18    Tag: blume-20040519-ml-build
19    Description:
20    
21    Arranged for ml-build to clean up after itself a little bit better.
22    The script generates a temporary SML source file and compiles it using
23    CM, so CM generates metadata (GUID, SKEL, objectfile) for it.  It now
24    gets rid of those at the end, so they don't accumulate under .cm.
25    
26    This required a minor change to install.sh because the name of the
27    metadata directory (default: .cm) is actually configurable at
28    installation time.
29    
30    ----------------------------------------------------------------------
31    Name: Matthias Blume (blume (at) tti - c (dot) org)
32    Date: 2004/05/18 15:50:00 CDT
33    Tag: blume-20040518-mkreader
34    Description:
35    
36    Added Posix.IO.mk{Bin,Text}{Reader,Writer} by lifting their respective
37    implementations from internal modules PosixBinPrimIO and PosixTextPrimIO.
38    
39    ----------------------------------------------------------------------
40    Name: Matthias Blume (blume (at) tti - c (dot) org)
41    Date: 2004/05/11 14:35:00 CDT
42    Tag: blume-20040511-win32sock
43    Description:
44    
45    Added previously missing support for many socket-related functions
46    under win32.  Thanks to David Hansel <hansel@reactive-systems.com>
47    for the voluminous patch!
48    
49    (I have not tested this patch under win32 yet.)
50    
51    Here is David's e-mail:
52    
53    Hi,
54    
55    Attached to this email you find a diff against sml/nj 110.45
56    that will enable socket support under Windows.
57    
58    To apply the patch (using unix or cygwin)
59    1) gunzip runtime.diff.gz
60    2) "cd" into "src/runtime" in the source tree of a fresh
61       110.45 installation.
62    3) patch -p 1 < [your/path/to]runtime.diff
63    
64    The code compiles fine but has NOT yet been extensively tested.
65    I only ran a few tests for basic socket client functionality
66    (which worked fine).  Especially the functions that use ioctl
67    are not tested at all and might not work (see below).
68    
69    I implemented this since we want to move to a newer version of sml/nj
70    but need socket support in order to use it.  This is the first time I
71    even had a look at the sml/nj source,  so please review my changes
72    before making this part of the distribution!  Here are a few issues
73    that I think might be better for someone to solve who is more
74    familiar with the sml/nj source (and socket programming):
75    
76    - getnetbyaddr.c and getnetbyname.c will raise a "not implemented"
77      exception since I could not figure out what the windows equivalent
78      of these functions is
79    
80    - In sockets-osdep.h there are a some #include statements that are
81      only used in a few files that include sockets-osdep.h
82    
83    - In smlnj-sock-lib.c, function init_fn() calls WSAStartup() but
84      does not process its return value since I don't know how to
85      report an error upwards.
86    
87    - It would probably be good to have a call to WSACleanup() when
88      the library is unloaded (if there is such a possibility).
89      Otherwise I think Windows will take care of this automatically
90      when the process finishes.
91    
92    - I used ioctlsocket() as a replacement for ioctl() but I have
93      no idea if that is actually the proper replacement on Windows.
94    
95    - All these issues are marked in the code by "FIXME" comments.
96    
97    We use sml/nj extensively in our products and are quite happy
98    with it.  I hope this contribution will help you.
99    
100    Keep up the good work!
101    
102    David
103    
104    ----------------------------------------------------------------------
105    Name: Matthias Blume (blume (at) tti - c (dot) org)
106    Date: 2004/05/11 14:20:00 CDT
107    Tag: blume-20040511-installml
108    Description:
109    
110    Fixed two bugs in installml script.  (Thanks to Vesa A. Norrman for
111    the patch.)
112    
113    ----------------------------------------------------------------------
114    Name: Matthias Blume (blume (at) tti - c (dot) org)
115    Date: 2004/05/11 14:05:00 CDT
116    Tag: blume-20040511-nlffi-netbsd
117    Description:
118    
119    Added support for nlffi under netbsd.  (Thanks to Vesa A. Norrman for
120    the patch.)
121    
122    ----------------------------------------------------------------------
123    Name: Matthias Blume (blume (at) tti - c (dot) org)
124    Date: 2004/05/11 12:05:00 CDT
125    Tag: blume-20040511-exports
126    Description:
127    
128    As per request by Adam Chlipala <adam@hcoop.net>, extended various
129    export lists in compiler-related .cm-files.
130    
131    ----------------------------------------------------------------------
132    Name: Matthias Blume (blume (at) tti - c (dot) org)
133    Date: 2004/05/11 11:35:00 CDT
134    Tag: blume-20040511-allsource
135    Description:
136    
137    The installer now honors the "src-smlnj" target again, although its meaning
138    has changed from "all sources required for the compiler" to "all sources
139    the installer knows about".  In other words, if you enable "src-smlnj"
140    in the "targets" file, then the installer will pull in sources for
141    everything.  (Notice that this refers to source code only.  Compiled
142    code is still only installed for modules that were requested explicitly
143    or which are required for other modules that were requested explicitly.)
144    
145    ----------------------------------------------------------------------
146    Name: Matthias Blume (blume (at) tti - c (dot) org)
147    Date: 2004/04/23 17:40:00 CDT
148    Tag: blume-20040423-ieee-scan
149    Description:
150    
151    Fixed IEEEReal.scan (and .fromString) so that if there is an overflow
152    in the exponent calculation we get INF or ZERO (depending on the mantissa
153    and the sign of the exponent).
154    
155    ----------------------------------------------------------------------
156    Name: Matthias Blume (blume (at) tti - c (dot) org)
157    Date: 2004/04/23 10:40:00 CDT
158    Tag: blume-20040423-ml-build
159    Description:
160    
161    The ml-build script now terminates with a non-0 status when something
162    goes wrong.
163    
164    ----------------------------------------------------------------------
165    Name: Matthias Blume (blume (at) tti - c (dot) org)
166    Date: 2004/04/22 16:35:00 CDT
167    Tag: blume-20040422-Option
168    Description:
169    
170    Made exception Option to be the same as exception Option.Option
171    (as it should be).
172    
173    ----------------------------------------------------------------------
174    Name: Allen Leung (leunga (at) reservoir (dot) com)
175    Date: 2004/03/19 14:40:00 EST
176    Tag: leunga-20040319-cygwin-nlffi
177    Description:
178    
179    Fixed the runtime so that ml-nlffi-lib runs on the cygwin version
180    of SML/NJ.  The problem is that
181    
182       lib = dlopen(NULL, ...)
183       f   = dlsym(lib, "malloc");
184    
185    does not work on Windows unless we explicitly export symbols
186    such as 'malloc' during linking.  We fixed this by explicitly
187    exporting the required symbols with the magic gcc incantation:
188    
189        -Wl,--export-all cygwin.def
190    
191    where cygwin.def is a file containing all the symbols that
192    we wish to export.
193    
194    I suspect this is a Windows problem and we'll have to
195    do the same (somehow with windows compilers) when
196    we build the native win32 version with the system
197    calls LoadLibrary/GetProcAddress.
198    
199    ----------------------------------------------------------------------
200    Name: Matthias Blume (blume (at) tti - c (dot) org)
201    Date: 2004/03/04 16:35:00 CST
202    Tag: blume-20040304-intinf-fmt
203    Description:
204    
205    Fixed problem with IntInf.fmt (sign would show up on the right instead
206    of on the left for BIN, OCT, and HEX).
207    
208    ----------------------------------------------------------------------
209    Name: Matthias Blume (blume (at) tti - c (dot) org)
210    Date: 2004/03/04 11:25:00 CST
211    Tag: blume-20040304-symlinks
212    Description:
213    
214    Fixed problem with installer script (unix only) where bin/ml-yacc and
215    friends pointed (via symlinks) to absolute locations instead of just
216    .run-sml.  This was reported by Vesa A Norrman.
217    
218    ----------------------------------------------------------------------
219    Name: Matthias Blume (blume (at) tti - c (dot) org)
220    Date: 2004/02/13 14:50:00 CST
221    Tag: Release_110_45
222    Description:
223    
224    New working version (110.45). New bootfiles.
225    
226    ----------------------------------------------------------------------
227    Name: Matthias Blume (blume (at) tti - c (dot) org)
228    Date: 2004/01/26 15:15:15 CST
229    Tag: blume-20040126-toplevel
230    Description:
231    
232    Improved handling of exceptions at the interactive toplevel.
233    
234    ----------------------------------------------------------------------
235    Name: Matthias Blume (blume (at) tti - c (dot) org)
236    Date: 2004/01/26 11:25:00
237    Tag: blume-20040126-app
238    Description:
239    
240    Type of top-level "app" corrected.
241    Added code for setting vp_limitPtrMask to Win32-specific runtime.
242    
243    ----------------------------------------------------------------------
244    Name: Matthias Blume (blume (at) tti - c (dot) org)
245    Date: 2003/11/18 17:10 CST
246    Tag: blume-20031118-basis-fiddle
247    Description:
248    
249    - changed Timer interface to what might become the spec
250    - POSIX_FLAGS -> BIT_FLAGS according to spec
251    - some other minor discrepancies wrt. spec eliminated
252    
253    ----------------------------------------------------------------------
254    Name: Matthias Blume (blume (at) tti - c (dot) org)
255    Date: 2003/11/06 12:00:00 CST
256    Tag: Release_110_44
257    Description:
258    
259    New working version (110.44). New bootfiles.
260    
261    ----------------------------------------------------------------------
262    Name: Matthias Blume (blume (at) tti - c (dot) org)
263    Date: 2003/11/04 11:50:00 CST
264    Tag: blume-20031104-move-libraries
265    Description:
266    
267    Eliminated the "dont_move_libraries" directive in config/targets.
268    (The mechanism was broken and could not be fixed easily.  Moreover,
269    there does not seem to be any reason not to move all libraries into
270    lib during installation.  I originally implemented this directive as a
271    backward-compatibility feature when I first introduced the new CM.
272    Now that things have been stable for a long time and going back to the
273    old CM is not an option, there is no reason to keep it around.)
274    
275    ----------------------------------------------------------------------
276    Name: Matthias Blume (blume (at) tti - c (dot) org)
277    Date: 2003/11/03 16:00:00 CST
278    Tag: blume-20031103-installdir
279    Description:
280    
281    Made installer honor INSTALLDIR variable again.  (Thanks to Chris
282    Richards for pointing out the problem and providing the solution.)
283    
284    ----------------------------------------------------------------------
285    Name: Matthias Blume (blume (at) tti - c (dot) org)
286    Date: 2003/10/01 17:05:00 CDT
287    Tag: blume-20031001-lal-mlrisc
288    Description:
289    
290    MLRISC bug fix from Lal.
291    
292    ----------------------------------------------------------------------
293    Name: Matthias Blume (blume (at) tti - c (dot) org)
294    Date: 2003/09/30 16:10:00 CDT
295    Tag: blume-20030930-primio-bat
296    Description:
297    
298    1. Added openVector, nullRd, and nullWr to PRIM_IO.
299    2. Improved .bat files (for Win32 port) to make things work under Win95.
300       (thanks to Aaron S. Hawley for this one)
301    
302    ----------------------------------------------------------------------
303    Name: Matthias Blume (blume (at) tti - c (dot) org)
304    Date: 2003/09/26 16:05:00 CDT
305    Tag: blume-20030926-wrappriv
306    Description:
307    
308    Added missing wrapper for privilege "primitive" in $smlnj/viscomp/core.cm.
309    
310    ----------------------------------------------------------------------
311    Name: Matthias Blume (blume (at) tti - c (dot) org)
312    Date: 2003/09/26 15:00:00 CDT
313    Tag: blume-20030926-110_43_3
314    Description:
315    
316    - additional cleanup
317    - version number bump, NEW BOOTFILES
318    
319    ----------------------------------------------------------------------
320    Name: Matthias Blume (blume (at) tti - c (dot) org)
321    Date: 2003/09/26 12:00:00 CDT
322    Tag: blume-20030926-ppautoload
323    Description:
324    
325    I modified the read-eval-print loop so that the autoloader gets
326    invoked whenever the prettyprinter tries to look up a symbol that
327    is not currently defined in the toplevel environment but which
328    appears in CM's autoload registry.  As a result, we see far fewer of
329    those ?.Foo.Bar.xxx names in the prettyprinter's output.
330    
331    In addition to this I tried to clean up some pieces of the Basis
332    implementation (e.g., Socket, Word8Array) in order to prevent other
333    instances of these ?.Foo.Bar.xxx names from being printed.
334    
335    The mechanism that picks names for types still needs some work, though.
336    (Right now it seems that if there is a type A.t which is defined to
337    be B.u, but B is unavailable at toplevel, then A.t gets printed as
338    "?.B.u" although the perhaps more sensible solution would be to use
339    "A.t" in this case.  In other words, the prettyprinter should follow
340    a chain of DEFtycs not farther than there are corresponding toplevel
341    names in the current environment.)
342    
343    ----------------------------------------------------------------------
344    Name: Matthias Blume (blume (at) tti - c (dot) org)
345    Date: 2003/09/24 16:31:00 CDT
346    Tag: blume-20030924-installer
347    Description:
348    
349    Another installer tweak:  All the ML code for the installer is now
350    compiled during CMB.make and put into a little library called
351    $smlnj/installer.cm.  The installation then simply invokes
352    
353       sml -m $smlnj/installer.cm
354    
355    and everything happens automagically.
356    
357    Win32:  ML code senses value of environment variable SMLNJ_HOME.
358    Unix:   ML code senses values of environment variables ROOT, CONFIGDIR,
359            and BINDIR.
360    
361    The new scheme guarantees that the ML code responsible for the installation
362    is in sync with the APIs of the main system.  Also, the installer is
363    somewhat faster because the installer script is precompiled.
364    
365    ----------------------------------------------------------------------
366    Name: Matthias Blume (blume (at) tti - c (dot) org)
367    Date: 2003/09/24 15:35:00 CDT
368    Tag: blume-20030924-synsock
369    Description:
370    
371    Added a signature SYNCHRONOUS_SOCKET to basis.cm.  This is like SOCKET
372    but excludes all non-blocking operations.  Defined SOCKET (in Basis)
373    and CML_SOCKET in terms of SYNCHRONOUS_SOCKET.  Removed superfluous
374    implementations of non-blocking operations from CML's Socket
375    structure.
376    
377    ----------------------------------------------------------------------
378    Name: Matthias Blume (blume (at) tti - c (dot) org)
379    Date: 2003/09/24 15:10:05 CDT
380    Tag: blume-20030924-sockets
381    Description:
382    
383    1. Fixed SOCKET API and implementation to match Basis spec.
384       This required changing the internal representation of sockets to one
385       that remembers (for each socket file descriptor) whether it is currently
386       blocking or non-blocking.  This state is maintained lazily (i.e., a system
387       call is made only if the state actually needs to change).
388    
389    2. OS-specific details of sockets were moved into separate files, thus
390       making it possible to unify the bulk of the socket implementations
391       between Unix and Win32.
392    
393    3. CML's socket API changed accordingly.
394       (Note that we need to remove non-blocking functions from this API
395       since they are redundant in the case of CML!)
396    
397    4. CML's socket implementation now makes use of non-blocking functions
398       provided by Basis, thus removing all OS-dependent code from this part
399       of CML.
400    
401    5. Changed Real64.precision from 52 to 53.  Minor cleanup in Real64 code.
402    
403    ----------------------------------------------------------------------
404    Name: Matthias Blume (blume (at) tti - c (dot) org)
405    Date: 2003/09/22 12:10:00 CDT
406    Tag: blume-20030922-110_43_2
407    Description:
408    
409    Made a new interim version and bootfiles for developer's bootstrapping
410    convenience.
411    
412       110.43.2 -- NEW BOOTFILES
413    
414    ----------------------------------------------------------------------
415    Name: Matthias Blume (blume (at) tti - c (dot) org)
416    Date: 2003/09/19 15:55:00 CDT
417    Tag: blume-20030919-cmdir
418    Description:
419    
420    1. new-install.sh -> install.sh
421    2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
422    3. tweaked installer so that another name instead of .cm can be chosen
423       at install time (by setting the CM_DIR_ARC environment variable
424       during installation); once installation is complete, the name is
425       fixed
426    
427    ----------------------------------------------------------------------
428    Name: Matthias Blume (blume (at) tti - c (dot) org)
429    Date: 2003/09/18 16:00:00 CDT
430    Tag: blume-20030918-110_43_1
431    Description:
432    
433    Made a new interim version and bootfiles for developer's bootstrapping
434    convenience.
435    
436       110.43.1 -- NEW BOOTFILES
437    
438    ----------------------------------------------------------------------
439    Name: Matthias Blume (blume (at) tti - c (dot) org)
440    Date: 2003/09/18 15:20:00 CDT
441    Tag: blume-20030918-misc
442    Description:
443    
444    1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
445       this seems to be controversial at the moment) and used those in
446       Posix.ProcEnv.times.
447    
448    2. Added Time.{from,to}Nanoseconds to Time.
449    
450    3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
451       For example, fromLargeInt never needs to look at more than 3 "big
452       digits" to get its 53 bits of precision.
453    
454    ----------------------------------------------------------------------
455    Name: Matthias Blume (blume (at) tti - c (dot) org)
456    Date: 2003/09/17 16:30:00 CDT
457    Tag: blume-20030917-real32-slices
458    Description:
459    
460    Added an entry to the primitive environment
461    (compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
462    added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
463    
464    Removed some of the "magic" constants in real64.sml and replaced them
465    with code that generates these values from their corresponding
466    integer counterparts.
467    
468    Made all(?) the slice-related changes to the Basis and made everything
469    compile again...
470    
471    ----------------------------------------------------------------------
472    Name: Matthias Blume (blume (at) tti - c (dot) org)
473    Date: 2003/09/15 17:45:00 CDT
474    Tag: blume-20030915-rbase
475    Description:
476    
477    Fixed bug in Real.fromLargeInt.
478    
479    ----------------------------------------------------------------------
480    Name: Matthias Blume (blume (at) tti - c (dot) org)
481    Date: 2003/09/13 18:11:00 CDT
482    Tag: blume-20030913-libinstall
483    Description:
484    
485    Minor bugfix in config/libinstall (set anchor with path to
486    standalone tool after installing it, otherwise libraries that
487    need ml-lex or ml-yacc won't compile the first time the installer
488    runs).
489    
490    ----------------------------------------------------------------------
491    Name: Matthias Blume (blume (at) tti - c (dot) org)
492    Date: 2003/09/12 11:45:00 CDT
493    Tag: blume-20030912-various
494    Description:
495    
496     - fixed bug in Real.toLargeInt
497     - fixed bug in Posix.ProcEnv.times
498     - changed inputLine functions to return an option
499     - minor installer improvements / bugfixes
500     - changed default @SMLalloc parameter for x86/celeron to 64k
501    
502    ----------------------------------------------------------------------
503    Name: Matthias Blume (blume (at) tti - c (dot) org)
504    Date: 2003/09/09 22:00:00 CDT
505    Tag: Release_110_43
506    Description:
507    
508        New working release 110.43.  New bootfiles.
509    
510    ----------------------------------------------------------------------
511    Name: Matthias Blume (blume (at) tti - c (dot) org)
512    Date: 2003/09/09 19:20:00 CDT
513    Tag: blume-20030909-installer
514    Description:
515    
516    Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
517    Modified config/install.bat to take advantage of it.  Also modified
518    config/install.sh (and called it config/new-install.sh) to take advantage
519    of it on Unix systems.  (The SML code is (supposed to be) platform-
520    independent.)
521    
522    The installer can now install everything under Win32
523    as well as under *nix as long as it compiles.
524    
525    Other changes:
526    
527      - made CML compile again under Win32
528      - made eXene compile under Win32 (by providing a fake structure UnixSock
529        and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
530      - fixed a bug in nowhere: it assumed that type OS.Process.status is the
531        same as type int;  under Win32 it isn't
532      - fixed some slice-related problems in the win32-specific parts of CML
533      - added a functor argument "sameVol" to os-path-fn.sml in the Basis
534        (under Win32, the volume name is case-insensitive, and the
535        OS.Path code compares volume names for equality)
536    
537    ----------------------------------------------------------------------
538    Name: Matthias Blume (blume (at) tti - c (dot) org)
539    Date: 2003/09/08 11:55:00 CDT
540    Tag: blume-20030908-fullpath
541    Description:
542    
543    Made Win32 version of OS.FileSys.fullPath return current directory
544    when given an empty string.  This is what the spec says, and incidentally,
545    CM depends on it.  (CM otherwise goes into an infinite loop in certain
546    cases when presented with the name of a non-existing .cm file.)
547    
548    ----------------------------------------------------------------------
549    Name: Matthias Blume (blume (at) tti - c (dot) org)
550    Date: 2003/09/04 16:30:00 CDT
551    Tag: blume-20030905-slices-etc
552    Description:
553    
554      1. Changed interface to vectors and arrays in Basis to match
555         (draft) Basis spec.
556      2. Added signatures and implementations of slices according to
557         Basis spec.
558      3. Edited source code throughout the system to make it compile again
559         under 1. and 2.  (In some cases code had to be added to have it
560         match the new signatures.)
561      4. MLRISC should be backward-compatible: the copies of the originals
562         of files that needed to change under 3. were retained, the .cm files
563         check the compiler version number and use old versions when
564         appropriate.
565      5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
566         dirstream -> string option (in accordance with Basis spec).
567      6. When generating code that counts lines, ml-lex used function
568         CharVector.foldli, taking advantage of its old interface.
569         This has been replaced with the corresponding code from
570         CharVectorSlice.  (html-lex must be re-lexed!)
571      7. BitArray in smlnj-lib/Util has been extended/modified to match the
572         new MONO_ARRAY signature.  (Do we need BitArraySlice?)
573      8. Removed temporary additions (fromInternal, toInternal) from the
574         (now obsolete) IntInf in smlnj-lib/Util.
575      9. Cleaned up structure Byte.
576     10. Added localOffset, scan, and fromString to Date (according to spec).
577         Cleaned/corrected implementation of Date.
578         (Still need to check for correctness; implement better canonicalizeDate.)
579     11. Added "scan" to signature IEEE_REAL.
580     12. Some improvements to IntInf [in particular: efficiency-hack for
581         mod and rem when second operand is 2 (for parity checks).]
582     13. Changed representation of type Time.time, using a single IntInf.int
583         value counting microseconds.  This considerably simplified the
584         implementation of structure Time.  We now support negative time
585         values; scan and fromString handle signs.
586     14. Functor PrimIO now takes two additional arguments (VectorSlice and
587         ArraySlice).
588    
589    ----------------------------------------------------------------------
590    Name: Matthias Blume (blume (at) tti - c (dot) org)
591    Date: 2003/08/28 17:00:00 CDT
592    Tag: blume-20030828-intinf
593    Description:
594    
595    This is a major update which comes with a version number bump
596    (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
597    and an implementation of IntInf in the Basis.
598    
599    There are a fairly large number of related changes and updates throughout
600    the system:
601    
602    Basis:
603            - Implemented IntInf.
604            - Made LargeInt a projection of IntInf (by filtering through INTEGER).
605            - Added some missing Real64 operations, most notably Real.toLargeInt.
606            - Added FixedInt as a synonym for Int32.
607    
608    compiler:
609            * Added support for a built-in intinf type.
610              - literals
611              - pattern matching
612              - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
613              - overloading on literals and operations
614    
615            This required adding a primitive type intinf, some additional
616            primops, and implementations for several non-trivial intinf
617            operations in Core.  (The intinf type is completely abstract
618            to the compiler; all operations get delegated back to the Core.)
619    
620            * Intinf equality is handled by polyequal.  However, the compiler
621            does not print its usual warning in this case (since polyequal
622            is the right thing to do there).
623    
624            * Improved the organization of structure InlineT.
625    
626            * A word about conversion primops:
627            If conversions involving intinf do not cancel out during
628            CPS contract, then the compiler must insert calls to Core functions.
629            Since all core access must be resolved already during the FLINT
630            translate phase, it would be too late a the time of CPS contract
631            to add new Core calls.  For this reason, conversion primops
632            for intinf carry two arguments:  1. the numeric argument that
633            they are supposed to convert, and 2. the Core function that
634            can help with this conversion if necessary.  If CPS contract
635            eliminates a primop, then the associated Core function becomes
636            dead and goes away.  Intinf conversion primops that do not get
637            eliminated by CPS contract get rewritten into calls of their
638            core functions by a separate, new phase.
639    
640    interactive system:
641            - Control.Print.intinfDepth controls max length of intinf constants
642            being printed. (Analogous to Control.Print.stringDepth.)
643            - Cleanup in printutil and pputil: got rid of unused stuff and
644            duplicates; replaced some of the code with code that makes better
645            use of library functionality.
646    
647    CM:
648            Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
649            are no longer silent.
650    
651    CKIT:
652            Fixed mismatched uses of Int32 and LargeInt.  I always decided
653            in favor of LargeInt -- which is now the same as IntInf.
654            CKIT-knowledgable people should check whether this is what's
655            intended and otherwise change things back to using Int32 or
656            FixedInt.
657    
658    Throughout the code:
659            Started using IntInf.int literals and built-in operations
660            (e.g., comparison with 0) where this seems appropriate.
661    
662    
663    ----------------------------------------------------------------------
664    Name: Dave MacQueen (dbm@cs.uchicago.edu)
665    Date: 2003/08/13 11:36:00 CDT
666    Tag: dbm-20030813-mcz-merge1
667    Description:
668    
669    Merging changes from the mcz-branch development branch into trunk.
670    These changes involve replacement of the emulated old prettyprinter
671    interface with direct use of the SML/NJ Lib PP library, and fixing
672    of a couple of bugs (895, 1186) relating to error messages.  A new
673    prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
674    has been added.
675    
676    ----------------------------------------------------------------------
677    Name: Matthias Blume (blume (at) tti - c (dot) org)
678    Date: 2003/08/11 15:45:00 CDT
679    Tag: blume-20030811-windows
680    Description:
681    
682    Version number bumped to 110.42.9.  NEW BOOTFILES!!!
683    
684        http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
685    
686    This patch restores SML/NJ's ability to run under win32.  There are a
687    number of changes, including fixes for several bugs that had gone
688    unnoticed until now:
689    
690      - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
691        trying to run the win32 version from within cygwin.)
692    
693      - There are a number of simple .bat scripts that substitute for their
694        corresponding Unix shell-scripts.  (See below.)
695    
696      - The internals of ml-build have been modified slightly.  The main
697        difference is that instead of calling ".link-sml" (or link-sml.bat)
698        using OS.Process.system, the ML process delegates this task back
699        to the script.  Otherwise problems arise in mixed environments such
700        as Cygwin where scripts look and work like Unix scripts, but
701        where OS.Process.system cannot run them.
702    
703      - In CM, the srcpath pickler used native pathname syntax -- which
704        is incorrect in the case of cross-compilation.  The new pickle format
705        is independent of platform-specific naming conventions.
706    
707      - Path configuration files (such as lib/pathconfig) can now choose
708        between native and standard syntax.  Placing a line of the form
709    
710            standard!
711    
712        into the file causes all subsequent paths to be interpreted using
713        CM standard pathname syntax (= Unix conventions); a line
714    
715            native!
716    
717        switches back to native style.  This was needed so that
718        path config files can be written portably, see src/system/pathconfig.
719    
720      - Runtime system:
721    
722          - win32-filesys.c:  get_file_time and set_file_time now
723            access modification time, not creation time.
724    
725          - I/O code made aware of new array representation.
726    
727          - Bug fixes in X86.prim.masm.
728    
729       - src/system/makeml made aware of win32. (For use under cygwin
730            and other Unix-environments for windows.)
731    
732       - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
733         which caused BinIO.inputAll to fail consistently.
734    
735    .bat scripts:
736    
737       Windows .bat scripts assume that SMLNJ_HOME is defined.
738    
739       - sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
740         applications (sml, ml-yacc, ml-lex).
741       - ml-build.bat: analogous to ml-build.
742       - config\install.bat:  Analogous to config/install.sh.  This requires
743         that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
744         (nmake etc. must be on the path, and vcvars32 must have been run.)
745         Moreover, sources for ml-lex and ml-yacc need to exist under src,
746         and the bootfile hierarchy must have been unpacked under
747         sml.boot.x86-win32.
748         The script is very primitive and does a poor job at error checking.
749         It only installs the base system, ml-lex, and ml-yacc.  No other
750         libraries are being installed (i.e., you get only those that
751         are part of the compiler.)
752       - link-sml.bat:  analogous to .link-sml, but not currently used
753    
754    Unrelated bug fixes:
755    
756      - ml-nlffigen now exports structures ST_* corresponding to incomplete
757        types.
758      - Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
759        otherwise.)
760    
761    ----------------------------------------------------------------------
762    Name: Dave MacQueen (macqueen@cs.uchicago.edu)
763    Date: 2003/06/17
764    Tag: macqueen-20030617-bug895
765    Description:
766    
767    Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
768    Tag will be used for new development branch (mcz-branch) for
769    use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
770    
771    ----------------------------------------------------------------------
772    Name: Matthias Blume (blume (at) tti - c (dot) org)
773    Date: 2003/05/27 16:55:00 CDT
774    Tag: blume-20030527-polyeq
775    Description:
776    
777    Tried to eliminated most cases of polymorphic equality.
778    
779    ----------------------------------------------------------------------
780    Name: Matthias Blume (blume (at) tti - c (dot) org)
781    Date: 2003/05/21 17:45:00 CDT
782    Tag: blume-20030517-complete
783    Description:
784    
785    Two changes:
786    
787      1. Added a flag for controlling whether non-exhaustive bindings will
788         be treated as errors (default is false).
789      2. Cleaned up the *entire* source tree so that CMB.make goes through
790         without a single non-exhaustive match- or bind warning.
791    
792    ----------------------------------------------------------------------
793    Name: Matthias Blume (blume (at) tti - c (dot) org)
794    Date: 2003/05/17 10:20:00 CDT
795    Tag: blume-20030517-absyn
796    Description:
797    
798    1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
799    
800       This mainly affects the quality of error messages.  However, some
801       of the code is now more straightforward than before. (Treatment of
802       the above four constructs in translate.sml is much simpler than
803       the "macro-expansion" that was going on before.  Plus, the mach-
804       compiler no longer gets invoked just to be able to compile an
805       if-expression.)
806    
807    2. The ErrorMsg.Error exception is now caught and absorbed by the
808       interactive loop.
809    
810    ----------------------------------------------------------------------
811    Name: Allen Leung
812    Date: 2003/05/16 13:05:00 CDT
813    Tag: leunga-20030516-cygwin-runtime
814    Description:
815    
816        Ported the runtime system to cygwin, which uses the unix
817    x86-unix bin files.  Missing/buggy features:
818    
819       o getnetbyname, getnetbyaddr: these functions seem to be missing in
820         the Cygwin library.
821       o Ctrl-C handling may be flaky.
822       o Windows system calls and Windows I/O are not supported.
823    
824       A new set of binfiles is located at:
825    
826         http://www.dorsai.org/~leunga/boot.x86-unix.tgz
827    
828       This is only needed for bootstrapping the cygwin version of smlnj.
829       Other x86 versions can use the existing binfiles.
830    
831    ----------------------------------------------------------------------
832    Name: Matthias Blume
833    Date: 2003/04/08 15:42:00 CDT
834    Tag: blume-20030408-listpair
835    Description:
836    
837    1. Added a target 'mlrisc' to installer.
838    
839    2. Added missing elements to structure ListPair.
840    
841    ----------------------------------------------------------------------
842    Name: Allen Leung
843    Date: 2003/01/07 10:40:00 EST
844    Tag: leunga-20030107-int-rem
845    Description:
846    
847    Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
848    The arguments to the SUBL instruction were swapped.
849    
850    ----------------------------------------------------------------------
851    Name: Matthias Blume
852    Date: 2002/12/12 16:25:00 EST
853    Tag: blume-20021212-risc-ra
854    Description:
855    
856    Fixed a serious bug in the rewrite code for FP spilling/reloading that
857    sent the RA into an infinite loop when floating point registers get
858    spilled.  (Because of this bug, e.g., nucleic stopped compiling between
859    110.37 and 110.38.)
860    There was another set of potential problems related to the handling of
861    MLRISC annotations (but those did not yet cause real problems, apparently).
862    
863    ----------------------------------------------------------------------
864    Name: Matthias Blume
865    Date: 2002/12/06 22:40:00 EST
866    Tag: blume-20021206-cm-fileid
867    Description:
868    
869    Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
870    This fixes the problem of CM getting confused by files that suddenly
871    change their identity (e.g., by getting unlinked and recreated by some
872    text editor such as vi).  There might be a better/cheaper/cleaner way
873    of doing this, but for now this will have to do.
874    
875    ----------------------------------------------------------------------
876    Name: Matthias Blume
877    Date: 2002/10/28 09:50:00 EST
878    Tag: blume-20021028-typecheck
879    Description:
880    
881    Exported structure Typecheck from $smlnj/viscomp/core.cm.
882    
883    ----------------------------------------------------------------------
884    Name: Matthias Blume
885    Date: 2002/10/17 09:10:00 EDT
886    Tag: Release_110_42
887    Description:
888    
889    In good old tradition, there has been a slight hiccup so that we have
890    to patch 110.42 after the fact.   The old release tag has been replaced
891    (see below).
892    
893    The change solves a problem with two competing approaches the
894    configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
895    each other's way.
896    
897    This change only affects the runtime system code and the installer script.
898    (No new bootfiles.)
899    
900    ----------------------------------------------------------------------
901    Name: Matthias Blume
902    Date: 2002/10/16 12:00:00 EDT
903    Tag: Release_110_42_removed
904    Description:
905    
906    New working release.  New bootfiles.
907    
908    ----------------------------------------------------------------------
909    Name: Matthias Blume
910    Date: 2002/10/10 13:10:00 EDT
911    Tag: blume-20021010-ppc-divs
912    Description:
913    
914    The mltree operator DIVS must be implemented with an overflow check on
915    the PPC because the hardware indicates divide-by-zero using "overflow" as
916    well.
917    
918    ----------------------------------------------------------------------
919    Name: Matthias Blume
920    Date: 2002/07/23 11:45:00
921    Tag: blume-20020723-smlnj-home
922    Description:
923    
924    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
925    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
926    CM_PATHCONFIG is also set), the path configuration file is assumed
927    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
928    the entire tree to some other place and everything will "just work".
929    
930    (Companion commands such as ml-build and ml-makedepend also sense this
931    variable.)
932    
933    ----------------------------------------------------------------------
934    Name: Matthias Blume
935    Date: 2002/07/12 21:19:00 EDT
936    Tag: blume-20020712-liveness
937    Description:
938    
939    Exported two useful "step" functions from liveness module (MLRISC).
940    
941    ----------------------------------------------------------------------
942    Name: Matthias Blume
943    Date: 2002/07/05 16:00 EDT
944    Tag: Release_110_41
945    Description:
946    
947    New working release.  New bootfiles.
948    
949    ----------------------------------------------------------------------
950    Name: Matthias Blume
951    Date: 2002/07/05 10:25:00 EDT
952    Tag: blume-20020705-btimp
953    Description:
954    
955    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
956    other clients can set up backtracing support.
957    
958    ----------------------------------------------------------------------
959    Name: Matthias Blume
960    Date: 2002/06/25 17:23:00 EDT
961    Tag: blume-20020625-fpmax
962    Description:
963    
964    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
965    case.  (The sense of the isNaN test was reversed -- which made min and
966    max always return their first argument.)
967    
968    ----------------------------------------------------------------------
969    Name: Matthias Blume
970    Date: 2002/06/11
971    Tag: blume-20020611-unixpath
972    Description:
973    
974    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
975    
976    ----------------------------------------------------------------------
977    Name: Matthias Blume
978    Date: 2002/06/10 16:35:00 EDT
979    Tag: blume-20020610-ieeereal
980    Description:
981    
982    I back-ported my implementation of IEEEReal.fromString from the
983    idlbasis-devel branch so that we can test it.
984    
985    Another small change is that ppDec tries to give more information
986    than just "<sig>" in the case of functors.  However, this code is
987    broken in some mysterious way if the functor's body's signature
988    has not been declared by ascription but gets inferred from the
989    implementation.  This needs fixing...
990    
991    ----------------------------------------------------------------------
992    Name: Matthias Blume
993    Date: 2002/05/31
994    Tag: blume-20020531-btrace-mode
995    Description:
996    
997    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
998    the wayside when I switched over to using Controls everywhere.)
999    
1000    ----------------------------------------------------------------------
1001    Name: Lal George
1002    Date: 2002/05/23 12:21:40 EDT
1003    Tag: george-20020523-visual-labels
1004    Description:
1005    
1006    Labels are now displayed in the graphical output to make
1007    the fall-through and target blocks obvious.
1008    
1009    ----------------------------------------------------------------------
1010    Name: Matthias Blume
1011    Date: 2002/05/22 11:03:00 EDT
1012    Tag: blume-20020522-shrink
1013    Description:
1014    
1015    John tweaked yesterday's fix for 1131 to handle an out-of-memory
1016    situation that comes up when allocating huge arrays.
1017    
1018    ----------------------------------------------------------------------
1019    Name: Matthias Blume
1020    Date: 2002/05/21 16:00:00 EDT
1021    Tag: Release_110_40
1022    Description:
1023    
1024    New working release (110.40).  New bootfiles.
1025    
1026    [Also: John Reppy fixed GC bug 1131.]
1027    
1028    ----------------------------------------------------------------------
1029    Name: Matthias Blume
1030    Date: 2002/05/21 12:35:00 EDT
1031    Tag: blume-20020521-cmdoc
1032    Description:
1033    
1034    CM documentation update.
1035    
1036    ----------------------------------------------------------------------
1037    Name: Matthias Blume
1038    Date: 2002/05/21 10:55:00 EDT
1039    Tag: blume-20020521-misc
1040    Description:
1041    
1042    - John tweaked runtime to be silent on heap export (except when
1043      GC messages are on).
1044    - I added a few more things (cross-compiling versions of CMB) to
1045      config/preloads (as suggestions).
1046    
1047    ----------------------------------------------------------------------
1048    Name: Matthias Blume
1049    Date: 2002/05/20 22:25:00 EDT
1050    Tag: blume-20020520-controls
1051    Description:
1052    
1053    - Added ControlUtil structure to control-lib.cm.
1054    - Use it throughout.
1055    - Used Controls facility to define MLRISC controls (as opposed to
1056      registering MLRISC control ref cells with Controls after the fact)
1057    - Fixed messed-up controls priorities.
1058    
1059    * Removed again all the stuff from config/preloads that one wouldn't
1060      be able to preload at the time the initial heap image is built.
1061      (Many libraries, e.g., CML, do not exist yet at this time.  The
1062       only libraries that can be preloaded via config/preloads are those
1063       that come bundled with the bootfiles.)
1064    
1065    ----------------------------------------------------------------------
1066    Name: Matthias Blume
1067    Date: 2002/05/20 10:59:00 EDT
1068    Tag: blume-20020520-preloads
1069    Description:
1070    
1071    Added a lot of commented-out suggestions for things to be included
1072    in config/preloads.
1073    
1074    ----------------------------------------------------------------------
1075    Name: Allen Leung
1076    Date: 2002/05/18 14:20:00 EDT
1077    Tag: leunga-20020518-mdl
1078    Description:
1079    
1080    o Made the mdl tool stuff compile and run again.
1081    o I've disabled all the stuff that depends on RTL specifications; they
1082      are all badly broken anyway.
1083    
1084    ----------------------------------------------------------------------
1085    Name: Matthias Blume
1086    Date: 2002/05/17 16:49:00 EDT
1087    Tag: blume-20020517-controls
1088    Description:
1089    
1090    0. John Reppy made several modifications to the SML/NJ library.
1091       In particular, there is a shiny new controls-lib.cm.
1092    
1093    1. Pushed new controls interface through compiler so that everything
1094       compiles again.
1095    
1096    2. Added FormatComb and FORMAT_COMB to the CML version of the
1097       SML/NJ library (so that CML compiles again).
1098    
1099    3. Modified init scripts because XXX_DEFAULT environment variables
1100       are no longer with us.  (Boot-time initialization is now done
1101       using the same environment variables that are also used for
1102       startup-time initialization of controls.)
1103    
1104    ----------------------------------------------------------------------
1105    Name: Lal George
1106    Date: 2002/05/15 09:20:10 EDT
1107    Tag: george-20020515-pseudo-op-decls
1108    Description:
1109    
1110            All pseudo-ops emitted before the first segment declaration
1111    such as TEXT, DATA, and BSS directives are assumed to be global
1112    declarations and are emitted first in the assembly file. This is
1113    useful in a number of situations where one has pseudo-ops that are not
1114    specific to any segment, and also works around the constraint that one
1115    cannot have client pseudo-ops in the TEXT segment.
1116    
1117            Because no segment is associated with these declarations it is
1118    an error to allocate any space or objects before the first segment
1119    directive and an exception will be raised. However, we cannot make
1120    this check for client pseudo-ops.
1121    
1122            These top level declarations are a field in the CFG graph_info.
1123    In theory you can continue to add to this field after the CFG has been
1124    built -- provided you know what you are doing;-)
1125    
1126    ----------------------------------------------------------------------
1127    Name: Matthias Blume
1128    Date: 2002/05/13 16:40:00 EDT
1129    Tag: blume-20020513-pp-etc
1130    Description:
1131    
1132    A few minor bugfixes:
1133    
1134      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
1135        (Bogus printouts for redefined bindings still occur.  Compiler
1136        bug should no longer occur now.  We need to redo the prettyprinter
1137        from scratch.)
1138    
1139      - CM pathname printer now also adds escape sequences for ( and )
1140    
1141      - commend and docu fixes for ml-nlffi
1142    
1143    ----------------------------------------------------------------------
1144    Name: Matthias Blume
1145    Date: 2002/05/10 16:40:00 EDT
1146    Tag: blume-20020510-erg-textio
1147    Description:
1148    
1149    Applied the following bugfix provided by Emden Gansner:
1150    
1151        Output is corrupted when outputSubstr is used rather than output.
1152    
1153        The problem occurs when a substring
1154    
1155            ss = (s, dataStart, dataLen)
1156    
1157        where dataStart > 0, fills a stream buffer with avail bytes left.
1158        avail bytes of s, starting at index dataStart, are copied into the
1159        buffer, the buffer is flushed, and then the remaining dataLen-avail
1160        bytes of ss are copied into the beginning of the buffer. Instead of
1161        starting this copy at index dataStart+avail in s, the current code
1162        starts the copy at index avail.
1163    
1164        Fix:
1165        In text-io-fn.sml, change line 695 from
1166             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
1167        to
1168             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
1169    
1170    ----------------------------------------------------------------------
1171    Name: Matthias Blume
1172    Date: 2002/04/12 13:55:00 EDT
1173    Tag: blume-20020412-assyntax
1174    Description:
1175    
1176    1. Grabbed newer assyntax.h from the XFree86 project.
1177    2. Fiddled with how to compile X86.prim.asm without warnings.
1178    3. (Very) Minor cleanup in CM.
1179    
1180    ----------------------------------------------------------------------
1181    Name: Matthias Blume
1182    Date: 2002/04/01 (no joke!) 17:07:00 EST
1183    Tag: blume-20020401-x86div
1184    Description:
1185    
1186    Added full support for div/mod/rem/quot on the x86, using the machine
1187    instruction's two results (without clumsily recomputing the remainder)
1188    directly where appropriate.
1189    
1190    Some more extensive power-of-two support was added to the x86 instruction
1191    selector (avoiding expensive divs, mods, and muls where they can be
1192    replaced with cheaper shifts and masks).  However, this sort of thing
1193    ought to be done earlier, e.g., within the CPS optimizer so that
1194    all architectures benefit from it.
1195    
1196    The compiler compiles to a fixed point, but changes might be somewhat
1197    fragile nevertheless.  Please, report any strange things that you might
1198    see wrt. div/mod/quot/rem...
1199    
1200    ----------------------------------------------------------------------
1201    Name: Matthias Blume
1202    Date: 2002/03/29 17:22:00
1203    Tag: blume-20020329-div
1204    Description:
1205    
1206    Fixed my broken div/mod logic.  Unfortunately, this means that the
1207    inline code for div/mod now has one more comparison than before.
1208    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
1209    The problem was with quotient = 0, because that alone does not tell
1210    us which way the rounding went.  One then has to look at whether
1211    remainder and divisor have the same sign...  :(
1212    
1213    Anyway, I replaced the bootfiles with fresh ones...
1214    
1215    ----------------------------------------------------------------------
1216    Name: Matthias Blume
1217    Date: 2002/03/29 14:10:00 EST
1218    Tag: blume-20020329-inlprims
1219    Description:
1220    
1221    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
1222    
1223    Primops have changed. This means that the bin/boot-file formats have
1224    changed as well.
1225    
1226    To make sure that there is no confusion, I made a new version.
1227    
1228    
1229    CHANGES:
1230    
1231    * removed REMT from mltree (remainder should never overflow).
1232    
1233    * added primops to deal with divisions of all flavors to the frontend
1234    
1235    * handled these primops all the way through so they map to their respective
1236      MLRISC support
1237    
1238    * used these primops in the implementation of Int, Int32, Word, Word32
1239    
1240    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
1241    
1242    * parameterized INLMIN, INLMAX, and INLABS by a numkind
1243    
1244    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
1245      floating point
1246    
1247    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
1248      and Real (but Real.abs maps to a separate floating-point-only primop)
1249    
1250    
1251    TODO items:
1252    
1253    * Hacked Alpha32 instruction selection, disabling the selection of REMx
1254      instructions because the machine instruction encoder cannot handle
1255      them.  (Hppa, PPC, and Sparc instruction selection did not handle
1256      REM in the first place, and REM is supported by the x86 machine coder.)
1257    
1258    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
1259      selection phase.  (The two can be streamlined because the hardware
1260      delivers both quotient and remainder at the same time anyway.)
1261    
1262    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
1263      (Currently the behavior is inconsistent both across architectures and
1264      wrt. the draft Basis spec.)
1265    
1266    * Word8 should eventually be handled natively, too.
1267    
1268    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
1269      as if there currently is no execution path that could trigger it in
1270      SML/NJ.  (The assumptions underlying functions arith and promotable do not
1271      hold for things like multiplication and division.)
1272    
1273    ----------------------------------------------------------------------
1274    Name: Matthias Blume
1275    Date: 2002/03/27 16:27:00 EST
1276    Tag: blume-20020327-mlrisc-divisions
1277    Description:
1278    
1279    Added support for all four division operations (ML's div, mod, quot,
1280    and rem) to MLRISC.  In the course of doing so, I also rationalized
1281    the naming (no more annoying switch-around of DIV and QUOT), by
1282    parameterizing the operation by div_rounding_mode (which can be either
1283    DIV_TO_ZERO or DIV_TO_NEGINF).
1284    
1285    The generic MLTreeGen functor takes care of compiling all four
1286    operations down to only round-to-zero div.
1287    
1288    Missing pieces:
1289    
1290      * Doing something smarter than relying on MLTreeGen on architectures
1291        like, e.g., the x86 where hardware division delivers both quotient and
1292        remainder at the same time.  With this, the implementation of the
1293        round-to-neginf operations could be further streamlined.
1294    
1295      * Remove inlining support for div/mod/rem from the frontend and replace it
1296        with primops that get carried through to the backend.  Do this for all
1297        int and word types.
1298    
1299    ----------------------------------------------------------------------
1300    Name: Matthias Blume
1301    Date: 2002/03/25 17:25:00 EST
1302    Tag: blume-20020325-divmod
1303    Description:
1304    
1305    I improved (hopefully without breaking them) the implementation of Int.div,
1306    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
1307    advantage of the following observations:
1308    
1309      Let  q = x quot y      r = x rem y
1310           d = x div  y      m = x mod y
1311    
1312    where "quot" is the round-to-zero version of integer division that
1313    hardware usually provides.  Then we have:
1314    
1315         r = x - q * y        where neither the * nor the - will overflow
1316         d = if q >= 0 orelse x = q * y then q else q - 1
1317                              where neither the * nor the - will overflow
1318         m = if q >= 0 orelse r = 0 then r else r + y
1319                              where the + will not overflow
1320    
1321    This results in substantial simplification of the generated code.
1322    The following table shows the number of CFG nodes and edges generated
1323    for
1324            fun f (x, y) = x OPER y
1325            (* with OPER \in div, mod, quot, rem *)
1326    
1327    
1328        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
1329        --------------------------------------------------------
1330         div |         24 |         39 |         12 |         16
1331         mod |         41 |         71 |         12 |         16
1332        quot |          8 |         10 |          8 |         10
1333         rem |         10 |         14 |          8 |         10
1334    
1335    
1336    ----------------------------------------------------------------------
1337    Name: Matthias Blume
1338    Date: 2002/03/25 22:06:00 EST
1339    Tag: blume-20020325-cprotobug
1340    Description:
1341    
1342    Fixed a bug in cproto (c prototype decoder).
1343    
1344    ----------------------------------------------------------------------
1345    Name: Matthias Blume
1346    Date: 2002/03/25 16:00:00 EST
1347    Tag: blume-20020325-raw-primops
1348    Description:
1349    
1350    I did some cleanup to Allen's new primop code and
1351    replaced yesterday's bootfiles with new ones.
1352    (But they are stored in the same place.)
1353    
1354    ----------------------------------------------------------------------
1355    Name: Matthias Blume
1356    Date: 2002/03/24 22:40:00 EST
1357    Tag: blume-20020324-bootfiles
1358    Description:
1359    
1360    Made the bootfiles that Allen asked for.
1361    
1362    ----------------------------------------------------------------------
1363    Name: Allen Leung
1364    Date: 2002/03/23 15:50:00 EST
1365    Tag: leunga-20020323-flint-cps-rcc-primops
1366    Description:
1367    
1368      1. Changes to FLINT primops:
1369    
1370        (* make a call to a C-function;
1371         * The primop carries C function prototype information and specifies
1372         * which of its (ML-) arguments are floating point. C prototype
1373         * information is for use by the backend, ML information is for
1374         * use by the CPS converter. *)
1375      | RAW_CCALL of { c_proto: CTypes.c_proto,
1376                       ml_args: ccall_type list,
1377                       ml_res_opt: ccall_type option,
1378                       reentrant : bool
1379                     } option
1380       (* Allocate uninitialized storage on the heap.
1381        * The record is meant to hold short-lived C objects, i.e., they
1382        * are not ML pointers.  With the tag, the representation is
1383        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
1384        *)
1385      | RAW_RECORD of {tag:bool,sz:int}
1386      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
1387    
1388      2.  These CPS primops are now overloaded:
1389    
1390           rawload of {kind:numkind}
1391           rawstore of {kind:numkind}
1392    
1393          The one argument form is:
1394    
1395             rawload {kind} address
1396    
1397          The two argument form is:
1398    
1399             rawload {kind} [ml object, byte-offset]
1400    
1401      3. RAW_CCALL/RCC now takes two extra arguments:
1402    
1403         a. The first is whether the C call is reentrant, i.e., whether
1404            ML state should be saved and restored.
1405         b. The second argument is a string argument specifying the name of
1406            library and the C function.
1407    
1408         These things are currently not handled in the code generator, yet.
1409    
1410      4. In CProto,
1411    
1412         An encoding type of "bool" means "ml object" and is mapped into
1413         C prototype of PTR.  Note that "bool" is different than "string",
1414         even though "string" is also mapped into PTR, because "bool"
1415         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
1416    
1417      5. Pickler/unpicker
1418    
1419         Changed to handle RAW_RECORD and newest RAW_CCALL
1420    
1421      6. MLRiscGen,
1422    
1423         1. Changed to handle the new rawload/rawstore/rawrecord operators.
1424         2. Code for handling C Calls has been moved to a new module CPSCCalls,
1425            in the file CodeGen/cpscompile/cps-c-calls.sml
1426    
1427      7. Added the conditional move operator
1428    
1429             condmove of branch
1430    
1431         to cps.  Generation of this is still buggy so it is currently
1432         disabled.
1433    
1434    ----------------------------------------------------------------------
1435    Name: Lal George
1436    Date: 2002/03/22 14:18:25 EST
1437    Tag: george-20020322-cps-branch-prob
1438    Description:
1439    
1440    Implemented the Ball-Larus branch prediction-heuristics, and
1441    incorporated graphical viewers for control flow graphs.
1442    
1443    Ball-Larus Heuristics:
1444    ---------------------
1445    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
1446    
1447    By design it uses the Dempster-Shafer theory for combining
1448    probabilities.  For example, in the function:
1449    
1450        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
1451    
1452    the ball-larus heuristics predicts that the n=0 is unlikely
1453    (OH-heuristic), and the 'then' branch is unlikely because of the
1454    RH-heuristic -- giving the 'then' branch an even lower combined
1455    probability using the Dempster-Shafer theory.
1456    
1457    Finally, John Reppy's loop analysis in MLRISC, further lowers the
1458    probability of the 'then' branch because of the loop in the else
1459    branch.
1460    
1461    
1462    Graphical Viewing:
1463    ------------------
1464    I merely plugged in Allen's graphical viewers into the compiler. The
1465    additional code is not much. At the top level, saying:
1466    
1467            Control.MLRISC.getFlag "cfg-graphical-view" := true;
1468    
1469    will display the graphical view of the control flow graph just before
1470    back-patching.  daVinci must be in your path for this to work. If
1471    daVinci is not available, then the default viewer can be changed
1472    using:
1473    
1474            Control.MLRISC.getString "viewer"
1475    
1476    which can be set to "dot" or "vcg" for the corresponding viewers. Of
1477    course, these viewers must be in your path.
1478    
1479    The above will display the compilation unit at the level of clusters,
1480    many of which are small, boring, and un-interesting. Also setting:
1481    
1482            Control.MLRISC.getInt "cfg-graphical-view_size"
1483    
1484    will display clusters that are larger than the value set by the above.
1485    
1486    
1487    ----------------------------------------------------------------------
1488    Name: Matthias Blume
1489    Date: 2002/03/21 22:20:00 EST
1490    Tag: blume-20020321-kmp-bugfix
1491    Description:
1492    
1493    Changed the interface to the KMP routine in PreString and fixed
1494    a minor bug in one place where it was used.
1495    
1496    ----------------------------------------------------------------------
1497    Name: Allen Leung
1498    Date: 2002/03/21 20:30:00 EST
1499    Tag: leunga-20020321-cfg
1500    Description:
1501    
1502      Fixed a potential problem in cfg edge splitting.
1503    
1504    ----------------------------------------------------------------------
1505    Name: Allen Leung
1506    Date: 2002/03/21 17:15:00 EST
1507    Tag: leunga-20020321-x86-fp-cfg
1508    Description:
1509    
1510      1. Recoded the buggy parts of x86-fp.
1511    
1512         a. All the block reordering code has been removed.
1513            We now depend on the block placement phases to do this work.
1514    
1515         b. Critical edge splitting code has been simplified and moved into the
1516            CFG modules, as where they belong.
1517    
1518         Both of these were quite buggy and complex.  The code is now much, much
1519         simpler.
1520    
1521      2. X86 backend.
1522    
1523         a. Added instructions for 64-bit support.  Instruction selection for
1524            64-bit has not been committed, however, since that
1525            requires changes to MLTREE which haven't been approved by
1526            Lal and John.
1527    
1528         b. Added support for FUCOMI and FUCOMIP when generating code for
1529            PentiumPro and above.  We only generate these instructions in
1530            the fast-fp mode.
1531    
1532         c. Added cases for JP and JNP in X86FreqProps.
1533    
1534      3. CFG
1535    
1536         CFG now has a bunch of methods for edge splitting and merging.
1537    
1538      4. Machine description.
1539    
1540         John's simplification of MLTREE_BASIS.fcond broke a few machine
1541         description things:
1542    
1543         rtl-build.{sig,sml} and hppa.mdl fixed.
1544    
1545         NOTE: the machine description stuff in the repository is still broken.
1546               Again, I can't put my fixes in because that involves
1547               changes to MLTREE.
1548    
1549    ----------------------------------------------------------------------
1550    Name: Matthias Blume
1551    Date: 2002/03/20 15:55:00 EST
1552    Tag: blume-20020320-kmp
1553    Description:
1554    
1555    Implemented Knuth-Morris-Pratt string matching in PreString and used
1556    it for String.isSubstring, Substring.isSubstring, and
1557    Substring.position.
1558    
1559    (Might need some stress-testing.  Simple examples worked fine.)
1560    
1561    ----------------------------------------------------------------------
1562    Name: Matthias Blume
1563    Date: 2002/03/19 16:37:00 EST
1564    Tag: blume-20020319-witnesses
1565    Description:
1566    
1567    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
1568    
1569    This implements a generic mechanism for changing constness qualifiers
1570    anywhere within big C types without resorting to outright "casts".
1571    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
1572    modify the constness at the outermost level.)
1573    The implementation of "convert" is based on the idea of "witness"
1574    values -- values that are not used by the operation but whose types
1575    "testify" to their applicability.  On the implementation side, "convert"
1576    is simply a projection (returning its second curried argument).  With
1577    cross-module inlining, it should not result in any machine code being
1578    generated.
1579    
1580    ----------------------------------------------------------------------
1581    Name: Matthias Blume
1582    Date: 2002/03/15 16:40:00 EST
1583    Tag: blume-20020315-basis
1584    Description:
1585    
1586    Provided (preliminary?) implementations for
1587    
1588      {String,Substring}.{concatWith,isSuffix,isSubstring}
1589    
1590    and
1591    
1592      Substring.full
1593    
1594    Those are in the Basis spec but they were missing in SML/NJ.
1595    
1596    ----------------------------------------------------------------------
1597    Name: Matthias Blume
1598    Date: 2002/03/14 21:30:00 EST
1599    Tag: blume-20020314-controls
1600    Description:
1601    
1602    Controls:
1603    ---------
1604    
1605    1. Factored out the recently-added Controls : CONTROLS stuff and put
1606       it into its own library $/controls-lib.cm.  The source tree for
1607       this is under src/smlnj-lib/Controls.
1608    
1609    2. Changed the names of types and functions in this interface, so they
1610       make a bit more "sense":
1611    
1612          module -> registry
1613          'a registry -> 'a group
1614    
1615    3. The interface now deals in ref cells only.  The getter/setter interface
1616       is (mostly) gone.
1617    
1618    4. Added a function that lets one register an already-existing ref cell.
1619    
1620    5. Made the corresponding modifications to the rest of the code so that
1621       everything compiles again.
1622    
1623    6. Changed the implementation of Controls.MLRISC back to something closer
1624       to the original.  In particular, this module (and therefore MLRISC)
1625       does not depend on Controls.  There now is some link-time code in
1626       int-sys.sml that registers the MLRISC controls with the Controls
1627       module.
1628    
1629    CM:
1630    ---
1631    
1632      * One can now specify the lambda-split aggressiveness in init.cmi.
1633    
1634    ----------------------------------------------------------------------
1635    Name: Allen Leung
1636    Date: 2002/03/13 17:30:00 EST
1637    Tag: leunga-20020313-x86-fp-unary
1638    Description:
1639    
1640    Bug fix for:
1641    
1642    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
1643    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
1644    > - fun f(x,(y,z)) = Real.~ y;
1645    > [autoloading]
1646    > [autoloading done]
1647    >       fchsl   (%eax), 184(%esp)
1648    > Error: MLRisc bug: X86MCEmitter.emitInstr
1649    >
1650    > uncaught exception Error
1651    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
1652    
1653    The problem was that the code generator did not generate any fp registers
1654    in this case, and the ra didn't know that it needed to run the X86FP phase to
1655    translate the pseudo fp instruction.   This only happened with unary fp
1656    operators in certain situations.
1657    
1658    ----------------------------------------------------------------------
1659    Name: Matthias Blume
1660    Date: 2002/03/13 14:00:00 EST
1661    Tag: blume-20020313-overload-etc
1662    Description:
1663    
1664    1. Added _overload as a synonym for overload for backward compatibility.
1665       (Control.overloadKW must be true for either version to be accepted.)
1666    
1667    2. Fixed bug in install script that caused more things to be installed
1668       than what was requested in config/targets.
1669    
1670    3. Made CM aware of the (_)overload construct so that autoloading
1671       works.
1672    
1673    ----------------------------------------------------------------------
1674    Name: Matthias Blume
1675    Date: 2002/03/12 22:03:00 EST
1676    Tag: blume-20020312-url
1677    Description:
1678    
1679    Forgot to update BOOT and srcarchiveurl.
1680    
1681    ----------------------------------------------------------------------
1682    Name: Matthias Blume
1683    Date: 2002/03/12 17:30:00 EST
1684    Tag: blume-20020312-version110392
1685    Description:
1686    
1687    Yet another version number bump (because of small changes to the
1688    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
1689    
1690    Changes:
1691    
1692      The new pid generation scheme described a few weeks ago was overly
1693      complicated.  I implemented a new mechanism that is simpler and
1694      provides a bit more "stability":  Once CM has seen a compilation
1695      unit, it keeps its identity constant (as long as you do not delete
1696      those crucial CM/GUID/* files).  This means that when you change
1697      an interface, compile, then go back to the old interface, and
1698      compile again, you arrive at the original pid.
1699    
1700      There now also is a mechanism that instructs CM to use the plain
1701      environment hash as a module's pid (effectively making its GUID
1702      the empty string).  For this, "noguid" must be specified as an
1703      option to the .sml file in question within its .cm file.
1704      This is most useful for code that is being generated by tools such
1705      as ml-nlffigen (because during development programmers tend to
1706      erase the tool's entire output directory tree including CM's cached
1707      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
1708      revert to the old, broken behavior of SML/NJ, but in specific cases
1709      where there is no danger of interface confusion, its use is ok
1710      (I think).
1711    
1712      ml-nlffigen by default generates "noguid" annotations.  They can be
1713      turned off by specifying -guid in its command line.
1714    
1715    ----------------------------------------------------------------------
1716    Name: Lal George
1717    Date: 2002/03/12 12 14:42:36 EST
1718    Tag: george-20020312-frequency-computation
1719    Description:
1720    
1721    Integrated jump chaining and static block frequency into the
1722    compiler. More details and numbers later.
1723    
1724    ----------------------------------------------------------------------
1725    Name: Lal George
1726    Date: 2002/03/11 11 22:38:53 EST
1727    Tag: george-20020311-jump-chain-elim
1728    Description:
1729    
1730    Tested the jump chain elimination on all architectures (except the
1731    hppa).  This is on by default right now and is profitable for the
1732    alpha and x86, however, it may not be profitable for the sparc and ppc
1733    when compiling the compiler.
1734    
1735    The gc test will typically jump to a label at the end of the cluster,
1736    where there is another jump to an external cluster containing the actual
1737    code to invoke gc. This is to allow factoring of common gc invocation
1738    sequences. That is to say, we generate:
1739    
1740            f:
1741               testgc
1742               ja   L1      % jump if above to L1
1743    
1744            L1:
1745               jmp L2
1746    
1747    
1748    After jump chain elimination the 'ja L1' instructions is converted to
1749    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
1750    up being implemented in their long form (if L2 is far away) using:
1751    
1752            jbe     L3      % jump if below or equal to L3
1753            jmp     L2
1754         L3:
1755            ...
1756    
1757    
1758    For large compilation units L2  may be far away.
1759    
1760    
1761    ----------------------------------------------------------------------
1762    Name: Matthias Blume
1763    Date: 2002/03/11 13:30:00 EST
1764    Tag: blume-20020311-mltreeeval
1765    Description:
1766    
1767    A functor parameter was missing.
1768    
1769    ----------------------------------------------------------------------
1770    Name: Allen Leung
1771    Date: 2002/03/11 10:30:00 EST
1772    Tag: leunga-20020311-runtime-string0
1773    Description:
1774    
1775       The representation of the empty string now points to a
1776    legal null terminated C string instead of unit.  It is now possible
1777    to convert an ML string into C string with InlineT.CharVector.getData.
1778    This compiles into one single machine instruction.
1779    
1780    ----------------------------------------------------------------------
1781    Name: Allen Leung
1782    Date: 2002/03/10 23:55:00 EST
1783    Tag: leunga-20020310-x86-call
1784    Description:
1785    
1786       Added machine generation for CALL instruction (relative displacement mode)
1787    
1788    ----------------------------------------------------------------------
1789    Name: Matthias Blume
1790    Date: 2002/03/08 16:05:00
1791    Tag: blume-20020308-entrypoints
1792    Description:
1793    
1794    Version number bumped to 110.39.1.  NEW BOOTFILES!
1795    
1796    Entrypoints: non-zero offset into a code object where execution should begin.
1797    
1798    - Added the notion of an entrypoint to CodeObj.
1799    - Added reading/writing of entrypoint info to Binfile.
1800    - Made runtime system bootloader aware of entrypoints.
1801    - Use the address of the label of the first function given to mlriscGen
1802      as the entrypoint.  This address is currently always 0, but it will
1803      not be 0 once we turn on block placement.
1804    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
1805      with entry points) from mlriscGen.
1806    
1807    ----------------------------------------------------------------------
1808    Name: Allen Leung
1809    Date: 2002/03/07 20:45:00 EST
1810    Tag: leunga-20020307-x86-cmov
1811    Description:
1812    
1813       Bug fixes for CMOVcc on x86.
1814    
1815       1. Added machine code generation for CMOVcc
1816       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
1817       3. CMOVcc cannot have an immediate operand as argument.
1818    
1819    ----------------------------------------------------------------------
1820    Name: Matthias Blume
1821    Date: 2002/03/07 16:15:00 EST
1822    Tag: blume-20020307-controls
1823    Description:
1824    
1825    This is a very large but mostly boring patch which makes (almost)
1826    every tuneable compiler knob (i.e., pretty much everything under
1827    Control.* plus a few other things) configurable via both the command
1828    line and environment variables in the style CM did its configuration
1829    until now.
1830    
1831    Try starting sml with '-h' (or, if you are brave, '-H')
1832    
1833    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
1834    implements the underlying generic mechanism.
1835    
1836    The interface to some of the existing such facilities has changed somewhat.
1837    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
1838    (The getFoo interface is still there for backward-compatibility, but its
1839    use is deprecated.)
1840    
1841    The ml-build script passes -Cxxx=yyy command-line arguments through so
1842    that one can now twiddle the compiler settings when using this "batch"
1843    compiler.
1844    
1845    TODO items:
1846    
1847    We should go through and throw out all controls that are no longer
1848    connected to anything.  Moreover, we should go through and provide
1849    meaningful (and correct!) documentation strings for those controls
1850    that still are connected.
1851    
1852    Currently, multiple calls to Controls.new are accepted (only the first
1853    has any effect).  Eventually we should make sure that every control
1854    is being made (via Controls.new) exactly once.  Future access can then
1855    be done using Controls.acc.
1856    
1857    Finally, it would probably be a good idea to use the getter-setter
1858    interface to controls rather than ref cells.  For the time being, both
1859    styles are provided by the Controls module, but getter-setter pairs are
1860    better if thread-safety is of any concern because they can be wrapped.
1861    
1862    *****************************************
1863    
1864    One bug fix: The function blockPlacement in three of the MLRISC
1865    backpatch files used to be hard-wired to one of two possibilities at
1866    link time (according to the value of the placementFlag).  But (I
1867    think) it should rather sense the flag every time.
1868    
1869    *****************************************
1870    
1871    Other assorted changes (by other people who did not supply a HISTORY entry):
1872    
1873    1. the cross-module inliner now works much better (Monnier)
1874    2. representation of weights, frequencies, and probabilities in MLRISC
1875       changed in preparation of using those for weighted block placement
1876       (Reppy, George)
1877    
1878    ----------------------------------------------------------------------
1879    Name: Lal George
1880    Date: 2002/03/07 14:44:24 EST 2002
1881    Tag: george-20020307-weighted-block-placement
1882    
1883    Tested the weighted block placement optimization on all architectures
1884    (except the hppa) using AMPL to generate the block and edge frequencies.
1885    Changes were required in the machine properties to correctly
1886    categorize trap instructions. There is an MLRISC flag
1887    "weighted-block-placement" that can be used to enable weighted block
1888    placement, but this will be ineffective without block/edge
1889    frequencies (coming soon).
1890    
1891    
1892    ----------------------------------------------------------------------
1893    Name: Lal George
1894    Date: 2002/03/05 17:24:48 EST
1895    Tag: george-20020305-linkage-cluster
1896    
1897    In order to support the block placement optimization, a new cluster
1898    is generated as the very first cluster (called the linkage cluster).
1899    It contains a single jump to the 'real' entry point for the compilation
1900    unit. Block placement has no effect on the linkage cluster itself, but
1901    all the other clusters  have full freedom in the manner in which they
1902    reorder blocks or functions.
1903    
1904    On the x86 the typical linkage code that is generated is:
1905       ----------------------
1906            .align 2
1907       L0:
1908            addl    $L1-L0, 72(%esp)
1909            jmp     L1
1910    
1911    
1912            .align  2
1913       L1:
1914       ----------------------
1915    
1916    72(%esp) is the memory location for the stdlink register. This
1917    must contain the address of the CPS function being called. In the
1918    above example, it contains the address of  L0; before
1919    calling L1 (the real entry point for the compilation unit), it
1920    must contain the address for L1, and hence
1921    
1922            addl $L1-L0, 72(%esp)
1923    
1924    I have tested this on all architectures except the hppa.The increase
1925    in code size is of course negligible
1926    
1927    ----------------------------------------------------------------------
1928    Name: Allen Leung
1929    Date: 2002/03/03 13:20:00 EST
1930    Tag: leunga-20020303-mlrisc-tools
1931    
1932      Added #[ ... ] expressions to mlrisc tools
1933    
1934    ----------------------------------------------------------------------
1935    Name: Matthias Blume
1936    Date: 2002/02/27 12:29:00 EST
1937    Tag: blume-20020227-cdebug
1938    Description:
1939    
1940    - made types in structure C and C_Debug to be equal
1941    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
1942    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
1943    
1944    ----------------------------------------------------------------------
1945    Name: Matthias Blume
1946    Date: 2002/02/26 12:00:00 EST
1947    Tag: blume-20020226-ffi
1948    Description:
1949    
1950    1. Fixed a minor bug in CM's "noweb" tool:
1951       If numbering is turned off, then truly don't number (i.e., do not
1952       supply the -L option to noweb).  The previous behavior was to supply
1953       -L'' -- which caused noweb to use the "default" line numbering scheme.
1954       Thanks to Chris Richards for pointing this out (and supplying the fix).
1955    
1956    2. Once again, I reworked some aspects of the FFI:
1957    
1958       A. The incomplete/complete type business:
1959    
1960       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
1961         gone!
1962       - ML types representing an incomplete type are now *equal* to
1963         ML types representing their corresponding complete types (just like
1964         in C).  This is still safe because ml-nlffigen will not generate
1965         RTTI for incomplete types, nor will it generate functions that
1966         require access to such RTTI.   But when ML code generated from both
1967         incomplete and complete versions of the C type meet, the ML types
1968         are trivially interoperable.
1969    
1970         NOTE:  These changes restore the full generality of the translation
1971         (which was previously lost when I eliminated functorization)!
1972    
1973       B. Enum types:
1974    
1975       - Structure C now has a type constructor "enum" that is similar to
1976         how the "su" constructor works.  However, "enum" is not a phantom
1977         type because each "T enum" has values (and is isomorphic to
1978         MLRep.Signed.int).
1979       - There are generic access operations for enum objects (using
1980         MLRep.Signed.int).
1981       - ml-nlffigen will generate a structure E_foo for each "enum foo".
1982         * The structure contains the definition of type "mlrep" (the ML-side
1983         representation type of the enum).  Normally, mlrep is the same
1984         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
1985         then mlrep will be defined as a datatype -- thus facilitating
1986         pattern matching on mlrep values.
1987         ("-ec" will be suppressed if there are duplicate values in an
1988          enumeration.)
1989         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
1990         will be generated for each C enum constant xxx.
1991         * Conversion functions m2i and i2m convert between mlrep and
1992         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
1993         * Coversion functions c and ml convert between mlrep and "tag enum".
1994         * Access functions (get/set) fetch and store mlrep values.
1995       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
1996         enumerations are merged into one single enumeration represented by
1997         structure E_'.
1998    
1999    ----------------------------------------------------------------------
2000    Name: Allen Leung
2001    Date: 2002/02/25 04:45:00 EST
2002    Tag: leunga-20020225-cps-spill
2003    
2004    This is a new implementation of the CPS spill phase.
2005    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
2006    In case of problems, replace it with the old file spill.sml
2007    
2008    The current compiler runs into some serious performance problems when
2009    constructing a large record.  This can happen when we try to compile a
2010    structure with many items.  Even a very simple structure like the following
2011    makes the compiler slow down.
2012    
2013        structure Foo = struct
2014           val x_1 = 0w1 : Word32.int
2015           val x_2 = 0w2 : Word32.int
2016           val x_3 = 0w3 : Word32.int
2017           ...
2018           val x_N = 0wN : Word32.int
2019        end
2020    
2021    The following table shows the compile time, from N=1000 to N=4000,
2022    with the old compiler:
2023    
2024    N
2025    1000   CPS 100 spill                           0.04u  0.00s  0.00g
2026           MLRISC ra                               0.06u  0.00s  0.05g
2027              (spills = 0 reloads = 0)
2028           TOTAL                                   0.63u  0.07s  0.21g
2029    
2030    1100   CPS 100 spill                           8.25u  0.32s  0.64g
2031           MLRISC ra                               5.68u  0.59s  3.93g
2032              (spills = 0 reloads = 0)
2033           TOTAL                                   14.71u  0.99s  4.81g
2034    
2035    1500   CPS 100 spill                           58.55u  2.34s  1.74g
2036           MLRISC ra                               5.54u  0.65s  3.91g
2037              (spills = 543 reloads = 1082)
2038           TOTAL                                   65.40u  3.13s  6.00g
2039    
2040    2000   CPS 100 spill                           126.69u  4.84s  3.08g
2041           MLRISC ra                               0.80u  0.10s  0.55g
2042              (spills = 42 reloads = 84)
2043           TOTAL                                   129.42u  5.10s  4.13g
2044    
2045    3000   CPS 100 spill                           675.59u  19.03s  11.64g
2046           MLRISC ra                               2.69u  0.27s  1.38g
2047              (spills = 62 reloads = 124)
2048           TOTAL                                   682.48u  19.61s  13.99g
2049    
2050    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
2051           MLRISC ra                               4.96u  0.27s  2.72g
2052              (spills = 85 reloads = 170)
2053           TOTAL                                   2375.26u  57.21s  48.00g
2054    
2055    As you can see the old cps spill module suffers from some serious
2056    performance problem.  But since I cannot decipher the old code fully,
2057    instead of patching the problems up, I'm reimplementing it
2058    with a different algorithm.  The new code is more modular,
2059    smaller when compiled, and substantially faster
2060    (O(n log n) time and O(n) space).  Timing of the new spill module:
2061    
2062    4000  CPS 100 spill                           0.02u  0.00s  0.00g
2063          MLRISC ra                               0.25u  0.02s  0.15g
2064             (spills=1 reloads=3)
2065          TOTAL                                   7.74u  0.34s  1.62g
2066    
2067    Implementation details:
2068    
2069    As far as I can tell, the purpose of the CPS spill module is to make sure the
2070    number of live variables at any program point (the bandwidth)
2071    does not exceed a certain limit, which is determined by the
2072    size of the spill area.
2073    
2074    When the bandwidth is too large, we decrease the register pressure by
2075    packing live variables into spill records.  How we achieve this is
2076    completely different than what we did in the old code.
2077    
2078    First, there is something about the MLRiscGen code generator
2079    that we should be aware of:
2080    
2081    o MLRiscGen performs code motion!
2082    
2083       In particular, it will move floating point computations and
2084       address computations involving only the heap pointer to
2085       their use sites (if there is only a single use).
2086       What this means is that if we have a CPS record construction
2087       statement
2088    
2089           RECORD(k,vl,w,e)
2090    
2091       we should never count the new record address w as live if w
2092       has only one use (which is often the case).
2093    
2094       We should do something similar to floating point, but the transformation
2095       there is much more complex, so I won't deal with that.
2096    
2097    Secondly, there are now two new cps primops at our disposal:
2098    
2099     1. rawrecord of record_kind option
2100        This pure operator allocates some uninitialized storage from the heap.
2101        There are two forms:
2102    
2103         rawrecord NONE [INT n]  allocates a tagless record of length n
2104         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
2105                                     and initializes the tag.
2106    
2107     2. rawupdate of cty
2108          rawupdate cty (v,i,x)
2109          Assigns to x to the ith component of record v.
2110          The storelist is not updated.
2111    
2112    We use these new primops for both spilling and increment record construction.
2113    
2114     1. Spilling.
2115    
2116        This is implemented with a linear scan algorithm (but generalized
2117        to trees).  The algorithm will create a single spill record at the
2118        beginning of the cps function and use rawupdate to spill to it,
2119        and SELECT or SELp to reload from it.  So both spills and reloads
2120        are fine-grain operations.  In contrast, in the old algorithm
2121        "spills" have to be bundled together in records.
2122    
2123        Ideally, we should sink the spill record construction to where
2124        it is needed.  We can even split the spill record into multiple ones
2125        at the places where they are needed.  But CPS is not a good
2126        representation for global code motion, so I'll keep it simple and
2127        am not attempting this.
2128    
2129     2. Incremental record construction (aka record splitting).
2130    
2131        Long records with many component values which are simulatenously live
2132        (recall that single use record addresses are not considered to
2133         be live) are constructed with rawrecord and rawupdate.
2134        We allocate space on the heap with rawrecord first, then gradually
2135        fill it in with rawupdate.  This is the technique suggested to me
2136        by Matthias.
2137    
2138        Some restrictions on when this is applicable:
2139        1. It is not a VECTOR record.  The code generator currently does not handle
2140           this case. VECTOR record uses double indirection like arrays.
2141        2. All the record component values are defined in the same "basic block"
2142           as the record constructor.  This is to prevent speculative
2143           record construction.
2144    
2145    ----------------------------------------------------------------------
2146    Name: Allen Leung
2147    Date: 2002/02/22 01:02:00 EST
2148    Tag: leunga-20020222-mlrisc-tools
2149    
2150    Minor bug fixes in the parser and rewriter
2151    
2152    ----------------------------------------------------------------------
2153    Name: Allen Leung
2154    Date: 2002/02/21 20:20:00 EST
2155    Tag: leunga-20020221-peephole
2156    
2157    Regenerated the peephole files.  Some contained typos in the specification
2158    and some didn't compile because of pretty printing bugs in the old version
2159    of 'nowhere'.
2160    
2161    ----------------------------------------------------------------------
2162    Name: Allen Leung
2163    Date: 2002/02/19 20:20:00 EST
2164    Tag: leunga-20020219-mlrisc-tools
2165    Description:
2166    
2167       Minor bug fixes to the mlrisc-tools library:
2168    
2169       1.  Fixed up parsing colon suffixed keywords
2170       2.  Added the ability to shut the error messages up
2171       3.  Reimplemented the pretty printer and fixed up/improved
2172           the pretty printing of handle and -> types.
2173       4.  Fixed up generation of literal symbols in the nowhere tool.
2174       5.  Added some SML keywords to to sml.sty
2175    
2176    ----------------------------------------------------------------------
2177    Name: Matthias Blume
2178    Date: 2002/02/19 16:20:00 EST
2179    Tag: blume-20020219-cmffi
2180    Description:
2181    
2182    A wild mix of changes, some minor, some major:
2183    
2184    * All C FFI-related libraries are now anchored under $c:
2185        $/c.cm      --> $c/c.cm
2186        $/c-int.cm  --> $c/internals/c-int.cm
2187        $/memory.cm --> $c/memory/memory.cm
2188    
2189    * "make" tool (in CM) now treats its argument pathname slightly
2190      differently:
2191        1. If the native expansion is an absolute name, then before invoking
2192           the "make" command on it, CM will apply OS.Path.mkRelative
2193           (with relativeTo = OS.FileSys.getDir()) to it.
2194        2. The argument will be passed through to subsequent phases of CM
2195           processing without "going native".  In particular, if the argument
2196           was an anchored path, then "make" will not lose track of that anchor.
2197    
2198    * Compiler backends now "know" their respective C calling conventions
2199      instead of having to be told about it by ml-nlffigen.  This relieves
2200      ml-nlffigen from one of its burdens.
2201    
2202    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
2203    
2204    * Export C_DEBUG and C_Debug from $c/c.cm.
2205    
2206    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
2207      subtyping relationship between incomplete pointers and their complete
2208      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
2209      with the convention of instantiating 'o with ('t, 'c) obj whenever
2210      the pointer target type is complete.  In the incomplete case, 'o
2211      will be instantiated with some "'c iobj" -- a type obtained by
2212      using one of the functors PointerToIncompleteType or PointerToCompleteType.
2213    
2214      Operations that work on both incomplete and complete pointer types are
2215      typed as taking an 'o ptr while operations that require the target to
2216      be known are typed as taking some ('t, 'c) obj ptr.
2217    
2218      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
2219      where void is an eqtype without any values.  This makes it possible
2220      to work on voidptr values using functions meant to operate on light
2221      incomplete pointers.
2222    
2223    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
2224      been vastly simplified.
2225    
2226    ----------------------------------------------------------------------
2227    Name: Matthias Blume
2228    Date: 2002/02/19 10:48:00 EST
2229    Tag: blume-20020219-pqfix
2230    Description:
2231    
2232    Applied Chris Okasaki's bug fix for priority queues.
2233    
2234    ----------------------------------------------------------------------
2235    Name: Matthias Blume
2236    Date: 2002/02/15 17:05:00
2237    Tag: Release_110_39
2238    Description:
2239    
2240    Last-minute retagging is becoming a tradition... :-(
2241    
2242    This is the working release 110.39.
2243    
2244    ----------------------------------------------------------------------
2245    Name: Matthias Blume
2246    Date: 2002/02/15 16:00:00 EST
2247    Tag: Release_110_39-orig
2248    Description:
2249    
2250    Working release 110.39.  New bootfiles.
2251    
2252    (Update: There was a small bug in the installer so it wouldn't work
2253    with all shells.  So I retagged. -Matthias)
2254    
2255    ----------------------------------------------------------------------
2256    Name: Matthias Blume
2257    Date: 2002/02/15 14:17:00 EST
2258    Tag: blume-20020215-showbindings
2259    Description:
2260    
2261    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
2262    the latter can be useful for exploring what bindings are available at
2263    the interactive prompt.  (The first function returns only the list
2264    of symbols that are really bound, the second prints those but also the
2265    ones that CM's autoloading mechanism knows about.)
2266    
2267    ----------------------------------------------------------------------
2268    Name: Matthias Blume
2269    Date: 2002/02/15 12:08:00 EST
2270    Tag: blume-20020215-iptrs
2271    Description:
2272    
2273    Two improvements to ml-nlffigen:
2274    
2275      1. Write files only if they do not exist or if their current contents
2276         do not coincide with what's being written.  (That is, avoid messing
2277         with the time stamps unless absolutely necessary.)
2278    
2279      2. Implement a "repository" mechanism for generated files related
2280         to "incomplete pointer types".   See the README file for details.
2281    
2282    ----------------------------------------------------------------------
2283    Name: Matthias Blume
2284    Date: 2002/02/14 11:50:00 EST
2285    Tag: blume-20020214-quote
2286    Description:
2287    
2288    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
2289    because of the new and improved tag generation scheme.  (Thanks to Allen
2290    Leung for pointing it out.)
2291    
2292    ----------------------------------------------------------------------
2293    Name: Lal George
2294    Date: 2002/02/14 09:55:27 EST 2002
2295    Tag: george-20020214-isabelle-bug
2296    Description:
2297    
2298    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
2299    of Isabelle on the x86.
2300    
2301    From Allen:
2302    -----------
2303     I've found the problem:
2304    
2305         in ra-core.sml, I use the counter "blocked" to keep track of the
2306         true number of elements in the freeze queue.  When the counter goes
2307         to zero, I skip examining the queue.  But I've messed up the
2308         bookkeeping in combine():
2309    
2310             else ();
2311             case !ucol of
2312               PSEUDO => (if !cntv > 0 then
2313                     (if !cntu > 0 then blocked := !blocked - 1 else ();
2314                                        ^^^^^^^^^^^^^^^^^^^^^^^
2315                      moveu := mergeMoveList(!movev, !moveu)
2316                     )
2317                  else ();
2318    
2319         combine() is called to coalesce two nodes u and v.
2320         I think I was thinking that if the move counts of u and v are both
2321         greater than zero then after they are coalesced then one node is
2322         removed from the freeze queue.  Apparently I was thinking that
2323         both u and v are of low degree, but that's clearly not necessarily true.
2324    
2325    
2326    02/12/2002:
2327        Here's the patch.  HOL now compiles.
2328    
2329        I don't know how this impact on performance (compile
2330        time or runtime).  This bug caused the RA (especially on the x86)
2331        to go thru the potential spill phase when there are still nodes on the
2332        freeze queue.
2333    
2334    
2335    
2336    
2337    ----------------------------------------------------------------------
2338    Name: Matthias Blume
2339    Date: 2002/02/13 22:40:00 EST
2340    Tag: blume-20020213-fptr-rtti
2341    Description:
2342    
2343    Fixed a bug in ml-nlffigen that was introduced with one of the previous
2344    updates.
2345    
2346    ----------------------------------------------------------------------
2347    Name: Matthias Blume
2348    Date: 2002/02/13 16:41:00 EST
2349    Tag: blume-20020213-cmlpq
2350    Description:
2351    
2352    Added new priority queue export symbols (which have just been added to
2353    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
2354    not compile and the installer would choke.)
2355    
2356    ----------------------------------------------------------------------
2357    Name: Matthias Blume
2358    Date: 2002/02/13 16:15:00 EST
2359    Tag: blume-20020213-various
2360    Description:
2361    
2362    1. More tweaks to ml-nlffigen:
2363    
2364       - better internal datastructures (resulting in slight speedup)
2365       - "-match" option requires exact match
2366       - "localized" gensym counters (untagged structs/unions nested within
2367         other structs/unions or within typedefs get a fresh counter; their
2368         tag will be prefixed by a concatenation of their parents' tags)
2369       - bug fixes (related to calculation of transitive closure of types
2370         to be included in the output)
2371    
2372    2. Minor Basis updates:
2373    
2374       - added implementations for List.collate and Option.app
2375    
2376    ----------------------------------------------------------------------
2377    Name: Matthias Blume
2378    Date: 2002/02/11 15:55:00 EST
2379    Tag: blume-20020211-gensym
2380    Description:
2381    
2382    Added a "-gensym" option to command line of ml-nlffigen.  This can be
2383    used to specify a "stem" -- a string that is inserted in all "gensym'd"
2384    names (ML structure names that correspond to unnamed C structs, unions,
2385    and enums), so that separate runs of ml-nlffigen do not clash.
2386    
2387    ----------------------------------------------------------------------
2388    Name: Matthias Blume
2389    Date: 2002/02/11 12:05:00 EST
2390    Tag: blume-20020211-gensml
2391    Description:
2392    
2393    A quick fix for a problem with GenSML (in the pgraph-util library):
2394    Make generation of toplevel "local" optional.  (Strictly speaking,
2395    signature definitions within "local" are not legal SML.)
2396    
2397    Other than that: updates to INSTALL and cm/TODO.
2398    
2399    ----------------------------------------------------------------------
2400    Name: Matthias Blume
2401    Date: 2002/02/08 15:00:00 EST
2402    Tag: blume-20020208-uniquepid
2403    Description:
2404    
2405    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
2406    
2407    1. The installer (config/install.sh) has gotten smarter:
2408    
2409         - Configuration options are a bit easier to specify now
2410           (in config/targets).
2411         - Bug in recognizing .tar.bz2 files fixed.
2412         - Installer automatically resolves dependencies between
2413           configuration options (e.g., if you ask for eXene, you will
2414           also get cml -- regardless whether you asked for it or not).
2415         - Installer can run in "quieter mode" by setting the environment
2416           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
2417           "completely silent", though.
2418         - Build HashCons library as part of smlnj-lib.
2419    
2420    2. A new scheme for assigning persistent identifiers to compilation
2421       units (and, by extension, to types etc.) has been put into place.
2422       This fixes a long-standing bug where types and even dynamic values
2423       can get internally confused, thereby compromising type safety
2424       (abstraction) and dynamic correctness.  See
2425    
2426         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
2427    
2428       for an example of how things could go wrong until now.
2429    
2430       The downside of the new scheme is that pids are not quite as
2431       persistent as they used to be: CM will generate a fresh pid
2432       for every compilation unit that it thinks it sees for the first
2433       time.  That means that if you compile starting from a clean, fresh
2434       source tree at two different times, you end up with different
2435       binaries.
2436    
2437       Cutoff recompilation, however, has not been compromised because
2438       CM keeps pid information in special caches between runs.
2439    
2440    ----------------------------------------------------------------------
2441    Name: Lal George
2442    Date: 2002/02/07 15:34:13 EST 2002
2443    Tag: <none>
2444    Description:
2445    
2446    Compilers that generate assembly code may produce  global labels
2447    whose value is resolved at link time. The various peephole optimization
2448    modules did not take this in account.
2449    
2450    TODO. The Labels.addrOf function should really return an option
2451    type so that clients are forced to deal with this issue, rather
2452    than an exception being raised.
2453    
2454    ----------------------------------------------------------------------
2455    Name: Lal George
2456    Date: 2002/02/06 13:55:02 EST
2457    Tag: george-20020206-ra-breakup
2458    Description:
2459    
2460    1. A bug fix from Allen.
2461    
2462        A typo causes extra fstp %st(0)'s to be generated at compensation
2463        edges, which might cause stack underflow traps at runtime.  This
2464        occurs in fft where there are extraneous fstps right before the 'into'
2465        trap instruction (in this case they are harmless since none of the
2466        integers overflow.)
2467    
2468    2. Pulled out various utility modules that were embedded in the modules
2469       of the register allocator. I need these modules for other purposes, but
2470       they are not complete enough to put into a library (just yet).
2471    ----------------------------------------------------------------------
2472    Name: Matthias Blume
2473    Date: 2002/01/31 16:05:00 EST
2474    Tag: blume-20020131-sparc-ccalls
2475    Description:
2476    
2477    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
2478       of extra stack space by mistake.  Fixed.
2479    
2480    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
2481    
2482    ----------------------------------------------------------------------
2483    Name: Allen Leung
2484    Date: 2002/01/30
2485    Tag: leunga-20020130-nowhere-bug-fix
2486    Description:
2487    
2488       MLRISC bug fixes:
2489       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
2490       2. MachineInt.fromString was negating its value.
2491    
2492    ----------------------------------------------------------------------
2493    Name: Matthias Blume
2494    Date: 2002/01/29
2495    Tag: blume-20020129-INSTALL
2496    Description:
2497    
2498    - Added somewhat detailed installation instructions (file INSTALL).
2499    - Fixed curl-detection bug in config/install.sh.
2500    - It is now possible to select the URL getter using the URLGETTER
2501      environment variable:
2502    
2503          not set / "unknown"      --> automatic detection (script tries wget,
2504                                       curl, and lynx)
2505          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
2506                                       how to properly invoke them)
2507          other                    --> use $URLGETTER directly, it must take
2508                                       precisely two command-line arguments
2509                                       (source URL and destination file name)
2510    
2511    ----------------------------------------------------------------------
2512    Name: Matthias Blume
2513    Date: 2002/01/28
2514    Tag: blume-20020128-sparc-ccalls
2515    Description:
2516    
2517    - Fixed problem with calculation of "used" registers in sparc-c-calls.
2518    - Make use of the allocParam argument in sparc-c-calls.
2519    
2520    ----------------------------------------------------------------------
2521    Name: Matthias Blume
2522    Date: 2002/01/28
2523    Tag: blume-20020128-allocParam
2524    Description:
2525    
2526    John Reppy:  Changes c-calls API to accept client-callback for
2527    allocating extra stack space.
2528    me: Corresponding changes to mlriscGen (using a dummy argument that
2529        does not change the current behavior).
2530    
2531    ----------------------------------------------------------------------
2532    Name: Matthias Blume
2533    Date: 2002/01/28 12:00:00
2534    Tag: Release_110_38
2535    Description:
2536    
2537    This time for real!!!
2538    
2539    ----------------------------------------------------------------------
2540    Name: Matthias Blume
2541    Date: 2002/01/28 10:56:00 EST
2542    Tag: blume-20020128-retraction
2543    Description:
2544    
2545    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
2546       with blume-Release_110_38-retracted.)
2547    
2548    1. Fixed a problem with incorrect rounding modes in real64.sml.
2549       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
2550    
2551    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
2552       and enums fixed.  The naming of corresponding ML identifiers should
2553       now be consistent again.
2554    
2555    ----------------------------------------------------------------------
2556    Name: Allen Leung
2557    Date: 2002/01/27
2558    Tag: leunga-20020127-nowhere
2559    Description:
2560    
2561       Added a target called nowhere in the configuration scripts.
2562       Enabling this will build the MLRISC 'nowhere' tool (for translating
2563       programs with where-clauses into legal SML code) during installation.
2564    
2565    ----------------------------------------------------------------------
2566    Name: Matthias Blume
2567    Date: 2002/01/25 21:27:00 EST
2568    Tag: blume-Release_110_38-retracted
2569    Description:
2570    
2571    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
2572    
2573    README will be added later.
2574    
2575    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
2576    (Release_110_38) removed.  Reason: Last-minute bug fixes.
2577    
2578    ----------------------------------------------------------------------
2579    Name: Matthias Blume
2580    Date: 2002/01/25
2581    Tag: blume-20020125-ffi
2582    Description:
2583    
2584    A large number of tweaks and improvements to ml-nlffi-lib and
2585    ml-nlffigen:
2586    
2587       - ML represenation types have been streamlined
2588       - getter and setter functions work with concrete values, not abstract
2589         ones where possible
2590       - ml-nlffigen command line more flexible (see README file there)
2591       - some bugs have been fixed (hopefully)
2592    
2593    ----------------------------------------------------------------------
2594    Name: Lal George
2595    Date: 2002/01/24
2596    Tag: george-20020124-risc-ra-interface
2597    Description:
2598    
2599       There is a dramatic simplification in the interface to the
2600       register allocator for RISC architectures as a result of making
2601       parallel copy instructions explicit.
2602    
2603    ----------------------------------------------------------------------
2604    Name: Matthias Blume
2605    Date: 2002/01/22
2606    Tag: blume-20020122-x86-ccalls
2607    Description:
2608    
2609    Bug fix for c-calls on x86 (having to do with how char- and
2610    short-arguments are being handled).
2611    
2612    ----------------------------------------------------------------------
2613    Name: Matthias Blume
2614    Date: 2002/01/21
2615    Tag: blume-20020121-ff
2616    Description:
2617    
2618    Another day of fiddling with the FFI...
2619    
2620    1. Bug fix/workaround:  CKIT does not complain about negative array
2621       dimensions, so ml-nlffigen has to guard itself against this possibility.
2622       (Otherwise a negative dimension would send it into an infinite loop.)
2623    
2624    2. Some of the abstract types (light objects, light pointers, most "base"
2625       types) in structure C are now eqtypes.
2626    
2627    3. Added constructors and test functions for NULL function pointers.
2628    
2629    ----------------------------------------------------------------------
2630    Name: Matthias Blume
2631    Date: 2002/01/18
2632    Tag: blume-20020118-ready-for-new-release
2633    Description:
2634    
2635    Made config/srcarchiveurl point to a new place.  (Will provide boot
2636    files shortly.)
2637    
2638    Maybe we christen this to be 110.38?
2639    
2640    ----------------------------------------------------------------------
2641    Name: Matthias Blume
2642    Date: 2002/01/18
2643    Tag: blume-20020118-more-ffifiddle
2644    Description:
2645    
2646    Today's FFI fiddling:
2647    
2648      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
2649        C.Set.* that use concrete (MLRep.*) arguments and results instead
2650        of abstract ones.
2651    
2652      - Provided word-style bit operations etc. for "int" representation
2653        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
2654        and Long.
2655    
2656    ----------------------------------------------------------------------
2657    Name: Matthias Blume
2658    Date: 2002/01/18
2659    Tag: blume-20020118-use-x86-fp
2660    Description:
2661    
2662    Now that x86-fast-fp seems to be working, I turned it back on again
2663    by default.  (Seems to work fine now, even with the FFI.)
2664    
2665    Other than that, I added some documentation about the FFI to
2666    src/ml-nlffigen/README and updated the FFI test examples in
2667    src/ml-nlffi-lib/Tests/*.
2668    
2669    ----------------------------------------------------------------------
2670    Name: Allen Leung
2671    Date: 2002/01/17
2672    Tag: leunga-20020117-x86-fast-fp-call
2673    Description:
2674    
2675       1. Fixed a problem with handling return fp values when x86's fast fp
2676          mode is turned on.
2677    
2678       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
2679          of %f32.
2680    
2681       3. Added a constructor INT32lit to the ast of MLRISC tools.
2682    
2683    ----------------------------------------------------------------------
2684    Name: Matthias Blume
2685    Date: 2002/01/16
2686    Tag: blume-20020116-ffifiddle
2687    Description:
2688    
2689    More fiddling with the FFI interface:
2690    
2691     - Make constness 'c instead of rw wherever possible.  This eliminates
2692       the need for certain explicit coercions.  (However, due to ML's
2693       value polymorphism, there will still be many cases where explicit
2694       coercions are necessary.  Phantom types are not the whole answer
2695       to modeling a subtyping relationship in ML.)
2696    
2697     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
2698    
2699     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
2700       clumsy workaround for varargs problem.)
2701    
2702    ----------------------------------------------------------------------
2703    Name: Lal George
2704    Date: 2002/01/15
2705    Tag: <none>
2706    Description:
2707    
2708    1. Since COPY instructions are no longer native to the architecture,
2709       a generic functor can be used to implement the expandCopies function.
2710    
2711    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
2712       TEXT segment.
2713    
2714    ----------------------------------------------------------------------
2715    Name: Matthias Blume
2716    Date: 2002/01/15
2717    Tag: blume-20020115-ffiupdates
2718    Description:
2719    
2720    1. Fix for bug resulting in single-precision float values being returned
2721       incorrectly from FFI calls.
2722    
2723    2. Small modifications to C FFI API:
2724    
2725        - memory-allocation routines return straight objects (no options)
2726          and raise an exception in out-of-memory situations
2727        - unsafe extensions to cast between function pointers and pointers
2728          from/to ints
2729        - added structure C_Debug as an alternative to structure C where
2730          pointer-dereferencing (|*| and |*!) always check for null-pointers
2731        - added open_lib' to DynLinkage;  open_lib' works like open_lib
2732          but also takes a (possibly empty) list of existing library handles
2733          that the current library depends on
2734    
2735    ----------------------------------------------------------------------
2736    Name: Matthias Blume
2737    Date: 2002/01/10
2738    Tag: blume-20020110-newffigen
2739    Description:
2740    
2741    1. Updates to portable graph code.
2742    
2743    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
2744       more scalable now so that even huge interfaces such as the one
2745       for GTK compile in finite time and space. :-)
2746       See src/ml-nlffigen/README for details on what's new.
2747    
2748    ----------------------------------------------------------------------
2749    Name: Lal George
2750    Date: 2001/01/09 14:31:35 EST 2002
2751    Tag: george-20011206-rm-native-copy
2752    Description:
2753    
2754            Removed the native COPY and FCOPY instructions
2755            from all the architectures and replaced it with the
2756            explicit COPY instruction from the previous commit.
2757    
2758            It is now possible to simplify many of the optimizations
2759            modules that manipulate copies. This has not been
2760            done in this change.
2761    
2762    ----------------------------------------------------------------------
2763    Name: Lal George
2764    Date: 2001/12/06 16:50:13 EST 2001
2765    Tag: george-20011206-mlrisc-instruction
2766    Description:
2767    
2768    Changed the representation of instructions from being fully abstract
2769    to being partially concrete. That is to say:
2770    
2771      from
2772            type instruction
2773    
2774      to
2775            type instr                              (* machine instruction *)
2776    
2777            datatype instruction =
2778                LIVE of {regs: C.cellset, spilled: C.cellset}
2779              | KILL of {regs: C.cellset, spilled: C.cellset}
2780              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
2781              | ANNOTATION of {i: instruction, a: Annotations.annotation}
2782              | INSTR of instr
2783    
2784    This makes the handling of certain special instructions that appear on
2785    all architectures easier and uniform.
2786    
2787    LIVE and KILL say that a list of registers are live or killed at the
2788    program point where they appear. No spill code is generated when an
2789    element of the 'regs' field is spilled, but the register is moved to
2790    the 'spilled' (which is present, more for debugging than anything else).
2791    
2792    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
2793    We used to generate:
2794    
2795            DEFFREG f1
2796            f1 := f2 + f3
2797            trapb
2798    
2799    but now generate:
2800    
2801            f1 := f2 + f3
2802            trapb
2803            LIVE {regs=[f1,f2,f3], spilled=[]}
2804    
2805    Furthermore, the DEFFREG (hack) required that all floating point instruction
2806    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
2807    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
2808    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
2809    this architecture).
2810    
2811    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
2812    all the architectures. This will result in further simplification of the
2813    register allocator that must be aware of them for coalescing purposes, and
2814    will also simplify certain aspects of the machine description that provides
2815    callbacks related to parallel copies.
2816    
2817    ANNOTATION should be obvious, and now INSTR represents the honest to God
2818    machine instruction set!
2819    
2820    The <arch>/instructions/<arch>Instr.sml files define certain utility
2821    functions for making porting easier -- essentially converting upper case
2822    to lower case. All machine instructions (of type instr) are in upper case,
2823    and the lower case form generates an MLRISC instruction. For example on
2824    the alpha we have:
2825    
2826      datatype instr =
2827         LDA of {r:cell, b:cell, d:operand}
2828       | ...
2829    
2830      val lda : {r:cell, b:cell, d:operand} -> instruction
2831        ...
2832    
2833    where lda is just (INSTR o LDA), etc.
2834    
2835    ----------------------------------------------------------------------
2836    Name: Matthias Blume
2837    Date: 2001/11/22 21:40:00 EST
2838    Tag: Release_110_37
2839    Description:
2840    
2841    Release 110.37.  This time for real.
2842    
2843    ----------------------------------------------------------------------
2844    Name: Matthias Blume
2845    Date: 2001/11/21 16:35:00 EST
2846    Tag: blume-20011121-foot-in-mouth
2847    Description:
2848    
2849    Removed the "Release_110_37" tag because of a serious bug.
2850    This will be re-tagged once the bug is fixed.
2851    
2852    ----------------------------------------------------------------------
2853    Name: Matthias Blume
2854    Date: 2001/11/21 16:14:00 EST
2855    Tag: blume-20011121-forgottenfile
2856    Description:
2857    
2858    Forgot to add a file.  (Just a .tex-file -- part of
2859    the CM manual source.)
2860    
2861    ----------------------------------------------------------------------
2862    Name: Matthias Blume
2863    Date: 2001/11/21 16:10:00 EST
2864    Tag: blume-20011121-invalid_110_37
2865    Description:
2866    
2867    Note: I removed the original tag "Release_110_37" from this commit
2868          because we found a serious bug in all non-x86 backends.
2869          - Matthias
2870    
2871    1. Modifications to the SML/NJ code generator and to the runtime system
2872       so that code object name strings are directly inserted into code
2873       objects at code generation time.  The only business the runtime system
2874       has with this is now to read the name strings on occasions.
2875       (The encoding of the name string has also changed somewhat.)
2876    
2877    2. CM now implements a simple "set calculus" for specifying export lists.
2878       In particular, it is now possible to refer to the export lists of
2879       other libraries/groups/sources and form unions as well as differences.
2880       See the latest CM manual for details.
2881    
2882    3. An separate notion of "proxy" libraries has again be eliminated from
2883       CM's model.  (Proxy libraries are now simply a special case of using
2884       the export list calculus.)
2885    
2886    4. Some of the existing libraries now take advantage of the new set
2887       calculus.
2888       (Notice that not all libraries have been converted because some
2889       of the existing .cm-files are supposed to be backward compatible
2890       with 110.0.x.)
2891    
2892    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
2893       "General.exnMessage"!  The former relies on a certain hook to be
2894       initialized, and that often does not happen in the stand-alone case.)
2895    
2896    ----------------------------------------------------------------------
2897    Name: Lal George
2898    Date: 2001/11/21  13:56:18 EST
2899    Tag: george-2001121-pseudo-ops
2900    Description:
2901    
2902      Implemented a complete redesign of MLRISC pseudo-ops. Now there
2903      ought to never be any question of incompatabilities with
2904      pseudo-op syntax expected by host assemblers.
2905    
2906      For now, only modules supporting GAS syntax are implemented
2907      but more should follow, such as MASM, and vendor assembler
2908      syntax, e.g. IBM as, Sun as, etc.
2909    
2910    ----------------------------------------------------------------------
2911    Name: Matthias Blume
2912    Date: 2001/11/14 11:52:00 EST
2913    Tag: blume-20011114-srcname
2914    Description:
2915    
2916    1. Routed the name of the current source file to mlriscgen where it
2917       should be directly emitted into the code object.  (This last part
2918       is yet to be done.)
2919    
2920    2. Some cleanup of the pgraph code to make it match the proposal that
2921       I put out the other day.  (The proposal notwithstanding, things are
2922       still in flux here.)
2923    
2924    ----------------------------------------------------------------------
2925    Name: Lal George
2926    Date: 2001/11/14 09:44:04 EST
2927    Tag:
2928    Description:
2929    
2930      Fix for a backpatching bug reported by Allen.
2931    
2932      Because the boundary between short and long span-dependent
2933      instructions is +/- 128, there are an astounding number of
2934      span-dependent instructions whose size is over estimated.
2935    
2936      Allen came up with the idea of letting the size of span
2937      dependent instructions be non-monotonic, for a maxIter
2938      number of times, after which the size must be monotonically
2939      increasing.
2940    
2941      This table shows the number of span-dependent instructions
2942      whose size was over-estimated as a function of maxIter, for the
2943      file Parse/parse/ml.grm.sml:
2944    
2945         maxIter            # of instructions:
2946            10                      687
2947            20                      438
2948            30                      198
2949            40                        0
2950    
2951      In compiling the compiler, there is no significant difference in
2952      compilation speed between maxIter=10 and maxIter=40. Actually,
2953      my measurements showed that maxIter=40 was a tad faster than
2954      maxIter=10! Also 96% of the  files in the compiler reach a fix
2955      point within 13 iterations, so fixing maxIter at 40, while high,
2956      is okay.
2957    
2958    ----------------------------------------------------------------------
2959    Name: Matthias Blume
2960    Date: 2001/10/31 15:25:00 EST
2961    Tag: blume-20011031-pgraph
2962    Description:
2963    
2964    CKIT:
2965    * Changed the "Function" constructor of type Ast.ctype to carry optional
2966      argument identifiers.
2967    * Changed the return type of TypeUtil.getFunction accordingly.
2968    * Type equality ignores the argument names.
2969    * TypeUtil.composite tries to preserve argument names but gives up quickly
2970      if there is a mismatch.
2971    
2972    installation script:
2973    * attempts to use "curl" if available (unless "wget" is available as well)
2974    
2975    CM:
2976    * has an experimental implementation of "portable graphs" which I will
2977      soon propose as an implementation-independent library format
2978    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
2979    
2980    NLFFI-LIB:
2981    * some cleanup (all cosmetic)
2982    
2983    NLFFIGEN:
2984    * temporarily disabled the mechanism that suppresses ML output for
2985      C definitions whose identifiers start with an underscore character
2986    * generate val bindings for enum constants
2987    * user can request that only one style (light or heavy) is being used;
2988      default is to use both (command-line arguments: -heavy and -light)
2989    * fixed bug in handling of function types involving incomplete pointers
2990    * generate ML entry points that take record arguments (i.e., using
2991      named arguments) for C functions that have a prototype with named
2992      arguments
2993      (see changes to CKIT)
2994    
2995    ----------------------------------------------------------------------
2996    Name: Allen Leung
2997    Date: 2001/10/27 20:34:00 EDT
2998    Tag: leunga-20011027-x86-fast-fp-call
2999    Description:
3000    
3001       Fixed the bug described in blume-20010920-slowfp.
3002    
3003       The fix involves
3004          1. generating FCOPYs in FSTP in ia32-svid
3005          2. marking a CALL with the appropriate annotation
3006    
3007    ----------------------------------------------------------------------
3008    Name: Matthias Blume
3009    Date: 2001/10/16 11:32:00 EDT
3010    Tag: blume-20011016-netbsd
3011    Description:
3012    
3013    Underscore patch from Chris Richards (fixing problem with compiling
3014    runtime system under recent NetBSD).
3015    
3016    ----------------------------------------------------------------------
3017    Name: Allen Leung
3018    Date: 2001/10/12 17:18:32 EDT 2001
3019    Tag: leung-20011012-x86-printflowgraph
3020    Description:
3021    
3022    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
3023    
3024    ----------------------------------------------------------------------
3025    Name: Lal George
3026    Date: 2001/10/11 23:51:34 EDT
3027    Tag: george-20011011-too-many-instrs
3028    Description:
3029    
3030    The representation of a program point never expected to see more
3031    than 65536 instructions in a basic block!
3032    
3033    ----------------------------------------------------------------------
3034    Name: Lal George
3035    Date: 2001/10/09 09:41:37 EDT
3036    Tag: george-20011008-mlrisc-labels
3037    Description:
3038    
3039    Changed the machine description files to support printing of
3040    local and global labels in assembly code, based on host assembler
3041    conventions.
3042    
3043    ----------------------------------------------------------------------
3044    Name: Matthias Blume
3045    Date: 2001/09/25 15:25:00 EDT
3046    Tag: blume-20010925-exninfo
3047    Description:
3048    
3049    I provided a non-hook implementation of exnName (at the toplevel) and
3050    made the "dummy" implementation of exnMessage (at the toplevel) more
3051    useful: if nothing gets "hooked in", then at least you are going to
3052    see the exception name and a message indicating why you don't see more.
3053    
3054    [For the time being, programs that need exnMessage and want to use
3055    ml-build should either use General.exnMessage (strongly recommended) or
3056    refer to structure General at some other point so that CM sees a
3057    static dependency.]
3058    
3059    [Similar remarks go for "print" and "use":  If you want to use their
3060    functionality in stand-alone programs generated by ml-build, then use
3061    TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
3062    
3063    ----------------------------------------------------------------------
3064    Name: Matthias Blume
3065    Date: 2001/09/20 17:28:00 EDT
3066    Tag: blume-20010920-slowfp
3067    Description:
3068    
3069    Allen says that x86-fast-fp is not safe yet, so I turned it off again...
3070    
3071    ----------------------------------------------------------------------
3072    Name: Matthias Blume
3073    Date: 2001/09/20 17:20:00 EDT
3074    Tag: blume-20010920-canonicalpaths
3075    Description:
3076    
3077    0. Updated the BOOT file (something that I forgot to do earlier).
3078    
3079    1. Small internal change to CM so that it avoids "/../" in filenames
3080       as much as possible (but only where it is safe).
3081    
3082    2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
3083       that arguments that contain delimiters are passed through correctly.
3084       This change also means that all "special" arguments of the form
3085       @SMLxxx... must come first.
3086    
3087    3. Changed install script to put relative anchor names for tool commands
3088       into pathconfig.
3089    
3090    ----------------------------------------------------------------------
3091    Name: Matthias Blume
3092    Date: 2001/09/18 15:35:00 EDT
3093    Tag: blume-20010918-readme11036
3094    Description:
3095    
3096    Added README files.
3097    
3098    ----------------------------------------------------------------------
3099    Name: Matthias Blume
3100    Date: 2001/09/18 11:45:00 EDT
3101    Tag: Release_110_36 (retag)
3102    Description:
3103    
3104    Fixed mistake in config/preloads. Retagged as 110.36.
3105    
3106    ----------------------------------------------------------------------
3107    Name: Matthias Blume
3108    Date: 2001/09/18 09:40:00 EDT
3109    Tag: Release_110_36_orig (tag changed)
3110    Description:
3111    
3112    New version (110.36).  New bootfiles.
3113    
3114    ----------------------------------------------------------------------
3115    Name: Matthias Blume
3116    Date: 2001/09/14 16:15:00 EDT
3117    Tag: blume-20010914-x86fastfp
3118    Description:
3119    
3120    John committed some changes that Allen made, in particular a (hopefully)
3121    correctly working version of the x86-fp module.
3122    
3123    I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
3124    flag to "true".  Everything seems to compile to a fixpoint ok, and
3125    "mandelbrot" speeds up by about 15%.
3126    
3127    ----------------------------------------------------------------------
3128    Name: Matthias Blume
3129    Date: 2001/09/13 11:20:00 EDT
3130    Tag: blume-20010913-minimal
3131    Description:
3132    
3133    1. Stefan Monnier's patch to fix a miscompilation problem that
3134       was brought to light by John Reppy's work on Moby.
3135    
3136    2. Implemented a minimal "structure Compiler" that contains just
3137       "version" and "architecture".  The minimal version will be
3138       available when the full version is not.  This is for backward-
3139       compatibility with code that wants to test Compiler.version.
3140    
3141    ----------------------------------------------------------------------
3142    Name: Matthias Blume
3143    Date: 2001/08/28 14:03:00 EDT
3144    Tag: blume-20010828-ml-lex
3145    Description:
3146    
3147    Fix for bug 1581, received from Neophytos Michael.
3148    
3149    ----------------------------------------------------------------------
3150    Name: Matthias Blume
3151    Date: 2001/08/27 11:20:00 EDT
3152    Tag: blume-20010827-readme11035
3153    Description:
3154    
3155    Fleshed out the README file for 110.35.
3156    
3157    ----------------------------------------------------------------------
3158    Name: Matthias Blume
3159    Date: 2001/08/24 17:10:00 EDT
3160    Tag: Release_110_35
3161    Description:
3162    
3163    New version number (110.35).  New bootfiles.
3164    
3165    ----------------------------------------------------------------------
3166    Name: Lal George
3167    Date: 2001/08/24 13:47:18 EDT 2001
3168    Tag: george-20010824-MLRISC-graphs
3169    Description:
3170    
3171     removed clusters from MLRISC completely and replaced with graphs.
3172    
3173    ----------------------------------------------------------------------
3174    Name: Matthias Blume
3175    Date: 2001/08/23 17:50:00 EDT
3176    Tag: blume-20010823-toplevel
3177    Description:
3178    
3179    - some reorganization of the code that implements various kinds of
3180      environments in the compiler (static, dynamic, symbolic, combined)
3181    - re-implemented the EnvRef module so that evalStream works properly
3182      (if the stream contains references to "use", "CM.make", etc.)
3183    - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
3184    
3185    ----------------------------------------------------------------------
3186    Name: Matthias Blume
3187    Date: 2001/08/20 15:50 EDT
3188    Tag: blume20010820-slipup
3189    Description:
3190    
3191    I forgot to commit a few files.  Here they are...
3192    
3193    ----------------------------------------------------------------------
3194    Name: Matthias Blume
3195    Date: 2001/08/20 15:35:00 EDT
3196    Tag: blume-20010820-debugprof
3197    Description:
3198    
3199    !!!! NEW BOOTFILES !!!!
3200    
3201    This is another round of reorganizing the compiler sources.  This
3202    time the main goal was to factor out all the "instrumentation"
3203    passes (for profiling and backtracing) into their own library.
3204    The difficulty was to do it in such a way that it does not depend
3205    on elaborate.cm but only on elabdata.cm.
3206    
3207    Therefore there have been further changes to both elaborate.cm and
3208    elabdata.cm -- more "generic" things have been moved from the former
3209    to the latter.  As a result, I was forced to split the assignment
3210    of numbers indicating "primtyc"s into two portions: SML-generic and
3211    SML/NJ-specific.  Since it would have been awkward to maintain,
3212    I bit the bullet and actually _changed_ the mapping between these
3213    numbers and primtycs.  The bottom line of this is that you need
3214    a new set of bin- and bootfiles.
3215    
3216    I have built new bootfiles for all architectures, so doing a fresh
3217    checkout and config/install.sh should be all you need.
3218    
3219    The newly created library's name is
3220    
3221        $smlnj/viscomp/debugprof.cm
3222    
3223    and its sources live under
3224    
3225        src/compiler/DebugProf
3226    
3227    ----------------------------------------------------------------------
3228    Name: Matthias Blume
3229    Date: 2001/08/15 17:15:00 EDT
3230    Tag: blume-20010815-compreorg
3231    Description:
3232    
3233    This is a first cut at reorganizing the CM libraries that make up the
3234    core of the compiler.  The idea is to separate out pieces that could
3235    be used independently by tools, e.g., the parser, the typechecker, etc.
3236    
3237    The current status is a step in this direction, but it is not quite
3238    satisfactory yet.  Expect more changes in the future.
3239    
3240    Here is the current (new) organization...
3241    
3242        What used to be $smlnj/viscomp/core.cm is now divided into
3243        six CM libraries:
3244    
3245             $smlnj/viscomp/basics.cm
3246                           /parser.cm
3247                           /elabdata.cm
3248                           /elaborate.cm
3249                           /execute.cm
3250                           /core.cm
3251    
3252        The CM files for these libraries live under src/system/smlnj/viscomp.
3253        All these libraries are proxy libraries that contain precisely
3254        one CM library component.  Here are the locations of the components
3255        (all within the src/compiler tree):
3256    
3257             Basics/basics.cm
3258             Parse/parser.cm
3259             ElabData/elabdata.cm
3260             Elaborator/elaborate.cm
3261             Execution/execute.cm
3262             core.cm
3263    
3264         [This organization is the same that has been used already
3265         for a while for the architecture-specific parts of the visible
3266         compiler and for the old version of core.cm.]
3267    
3268         As you will notice, many source files have been moved from their
3269         respective original locations to a new home in one of the above
3270         subtrees.
3271    
3272         The division of labor between the new libraries is the following:
3273    
3274             basics.cm:
3275                - Simple, basic definitions that pertain to many (or all) of
3276                  the other libraries.
3277             parser.cm:
3278                - The SML parser, producing output of type Ast.dec.
3279                - The type family for Ast is also defined and exported here.
3280             elabdata.cm:
3281                - The datatypes that describe input and output of the elaborator.
3282                  This includes types, absyn, and static environments.
3283             elaborator.cm:
3284                - The SML/NJ type checker and elaborator.
3285                  This maps an Ast.dec (with a given static environment) to
3286                  an Absyn.dec (with a new static environment).
3287                - This libraries implements certain modules that used to be
3288                  structures as functors (to remove dependencies on FLINT).
3289             execute.cm:
3290                - Everything having to do with executing binary code objects.
3291                - Dynamic environments.
3292             core.cm:
3293                - SML/NJ-specific instantiations of the elaborator and MLRISC.
3294                - Top-level modules.
3295                - FLINT (this should eventually become its own library)
3296    
3297    Notes:
3298    
3299    I am not 100% happy with the way I separated the elaborator (and its
3300    data structures) from FLINT.  Two instances of the same problem:
3301    
3302        1. Data structures contain certain fields that carry FLINT-specific
3303           information.  I hacked around this using exn and the property list
3304           module from smlnj-lib.  But the fact that there are middle-end
3305           specific fields around at all is a bit annoying.
3306    
3307        2. The elaborator calculates certain FLINT-related information.  I tried
3308           to make this as abstract as I could using functorization, but, again,
3309           the fact that the elaborator has to perform calculations on behalf
3310           of the middle-end at all is not nice.
3311    
3312        3. Having to used exn and property lists is unfortunate because it
3313           weakens type checking.  The other alternative (parameterizing
3314           nearly *everything*) is not appealing, though.
3315    
3316    I removed the "rebinding =" warning hack because due to the new organization
3317    it was awkward to maintain it.  As a result, the compiler now issues some of
3318    these warnings when compiling init.cmi during bootstrap compilation. On
3319    the plus side, you also get a warning when you do, for example:
3320       val op = = Int32.+
3321    which was not the case up to now.
3322    
3323    I placed "assign" and "deref" into the _Core structure so that the
3324    code that deals with the "lazy" keyword can find them there.  This
3325    removes the need for having access to the primitive environment
3326    during elaboration.
3327    
3328    ----------------------------------------------------------------------
3329    Name: Matthias Blume
3330    Date: 2001/08/13
3331    Tag: blume-20010813-closures
3332    Description:
3333    
3334    This fix was sent to us by Zhong Shao.  It is supposed to improve the
3335    performance of certain loops by avoiding needless closure allocation.
3336    
3337    ----------------------------------------------------------------------
3338    Name: Lal George
3339    Date: 2001/07/31 10:03:23 EDT 2001
3340    Tag: george-20010731-x86-fmalloc
3341    Description: Fixed bug in x86 calls
3342    
3343        There was a bug where call instructions would mysteriously
3344        vanish. The call instruction had to be one that returned
3345        a floating point value.
3346    
3347    ----------------------------------------------------------------------
3348    Name: Lal George
3349    Date: 2001/07/19 16:36:29 EDT 2001
3350    Tag: george-20010719-simple-cells
3351    Description:
3352    
3353    I have dramatically simplified the interface for CELLS in MLRISC.
3354    
3355    In summary, the cells interface is broken up into three parts:
3356    
3357      1. CellsBasis : CELLS_BASIS
3358    
3359            CellsBasis is a top level structure and common for all
3360            architectures.  it contains the definitions of basic datatypes
3361            and utility  functions over these types.
3362    
3363      2. functor Cells() : CELLS
3364    
3365            Cells generates an interface for CELLS that incorporates the
3366            specific resources on the target architecture, such as the
3367            presence of special register classes, their number and size,
3368            and various useful substructures.
3369    
3370      3. <ARCH>CELLS
3371    
3372            e.g. SparcCells: SPARCCELLS
3373    
3374            <ARCH>CELLS usually contains additional bindings for special
3375            registers  on the architecture, such as:
3376    
3377                    val r0 : cell           (* register zero *)
3378                    val y : cell            (* Y register *)
3379                    val psr : cell          (* processor status register *)
3380                    ...
3381    
3382            The structure returned by applying the Cells functor is opened
3383            in this interface.
3384    
3385    The main implication of all this is that the datatypes for cells is
3386    split between CellsBasis and CELLS -- a fairly simple change for user
3387    code.
3388    
3389    In the old scheme the CELLS interface had a definitional binding of
3390    the form:
3391    
3392            signature CELLS = sig
3393    
3394               structure CellsBasis = CellsBasis
3395    
3396               ...
3397    
3398            end
3399    
3400    With all the sharing constraints that goes on in MLRISC, this old
3401    design  quickly leads to errors such as:
3402    
3403            "structure definition spec inside of sharing ... "
3404    
3405    
3406    and appears to require an unacceptable amount of sharing and where
3407    constraint hackery.
3408    
3409    I think this error message (the interaction of definitional specs and
3410    sharing) requires more explanation on our web page.
3411    
3412    ----------------------------------------------------------------------
3413    Name: Matthias Blume
3414    Date: 2001/07/19 15:00:00 EDT
3415    Tag: blume-20010719-libreorg
3416    Description:
3417    
3418    This update puts together a fairly extensive but straightforward change
3419    to the way the libraries that implement the interactive system are
3420    organized:
3421    
3422       The biggest change is the elimination of structure Compiler.  As a
3423       replacement for this structure, there is now a CM library
3424       (known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
3425       that exports all the substructures of the original structure Compiler
3426       directly.  So instead of saying Compiler.Foo.bar one now simply
3427       says Foo.bar.  (The CM libraries actually export a collection of
3428       structures that is richer than the collection of substructures of
3429       structure Compiler.)
3430    
3431       To make the transition smooth, there is a separate library called
3432       $smlnj/compiler/compiler.cm which puts together and exports the
3433       original structure Compiler (or at least something very close to it).
3434    
3435       There are five members of the original structure Compiler
3436       that are not exported directly but which instead became members
3437       of a new structure Backend (described by signature BACKEND).  These are:
3438       structure Profile (: PROFILE), structure Compile (: COMPILE), structure
3439       Interact (: INTERACT), structure Machine (: MACHINE), and val
3440       architecture (: string).
3441    
3442       Structure Compiler.Version has become structure CompilerVersion.
3443    
3444       Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
3445       by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
3446       or x86, respectively.
3447       Each of these exports the same frontend structures that
3448       $smlnj/compiler.cm exports.  But they do not have a structure Backend
3449       and instead export some structure <Arch>Backend where <Arch> is Alpha32,
3450       Hppa, PPC, Sparc, or X86, respectively.
3451    
3452       Library $smlnj/compiler/all.cm exports the union of the exports of
3453       $smlnj/compiler/<arch>.cm
3454    
3455       There are no structures <Arch>Compiler anymore, use
3456       $smlnj/compiler/<arch>.cm instead.
3457    
3458       Library host-compiler-0.cm is gone.  Instead, the internal library
3459       that instantiates CM is now called cm0.cm.  Selection of the host
3460       compiler (backend) is no longer done here but. (Responsibility for it
3461       now lies with $smlnj/compiler/current.cm.  This seems to be more
3462       logical.)
3463    
3464       Many individual files have been moved or renamed.  Some files have
3465       been split into multiple files, and some "dead" files have been deleted.
3466    
3467    Aside from these changes to library organization, there are also changes
3468    to the way the code itself is organized:
3469    
3470       Structure Binfile has been re-implemented in such a way that it no
3471       longer needs any knowledge of the compiler.  It exclusively deals
3472       with the details of binfile layout.  It no longer invokes the
3473       compiler (for the purpose of creating new prospective binfile
3474       content), and it no longer has any knowledge of how to interpret
3475       pickles.
3476    
3477       Structure Compile (: COMPILE) has been stripped down to the bare
3478       essentials of compilation.  It no longer deals with linking/execution.
3479       The interface has been cleaned up considerably.
3480    
3481       Utility routines for dealing with linking and execution have been
3482       moved into their own substructures.
3483    
3484       (The ultimate goal of these changes is to provide a light-weight
3485       binfile loader/linker (at least for, e.g., stable libraries) that
3486       does not require CM or the compiler to be present.)
3487    
3488    CM documentation has been updated to reflect the changes to library
3489    organization.
3490    
3491    ----------------------------------------------------------------------
3492    Name: Matthias Blume
3493    Date: 2001/07/10 17:30:00 EDT
3494    Tag: Release_110_34
3495    Description:
3496    
3497    Minor tweak to 110.34 (re-tagged):
3498    
3499      - README.html file added to CVS repository
3500      - runtime compiles properly under FreeBSD 3.X and 4.X
3501    
3502    ----------------------------------------------------------------------
3503    Name: Matthias Blume
3504    Date: 2001/07/10 17:30:00 EDT
3505    Tag: Release_110_34
3506    Description:
3507    
3508    New version number (110.34). New bootfiles.
3509    
3510    ----------------------------------------------------------------------
3511    Name: Matthias Blume
3512    Date: 2001/07/09 16:00:00 EDT
3513    Tag: blume-20010709-more-varargs
3514    Description:
3515    
3516    I changed the handling of varargs in ml-nlffigen again:
3517    The ellipsis ... will now simply be ignored (with an accompanying warning).
3518    
3519    The immediate effect is that you can actually call a varargs function
3520    from ML -- but you can't actually supply any arguments beyond the ones
3521    specified explicitly.  (For example, you can call printf with its format
3522    string, but you cannot pass additional arguments.)
3523    
3524    This behavior is only marginally more useful than the one before, but
3525    it has the advantage that a function or, more importantly, a function
3526    type never gets dropped on the floor, thus avoiding follow-up problems with
3527    other types that refer to the offending one.
3528    
3529    ----------------------------------------------------------------------
3530    Name: Matthias Blume
3531    Date: 2001/07/09 11:25:00 EDT
3532    Tag: blume-20010709-varargs
3533    Description:
3534    
3535    1. ckit-lib.cm now exports structure Error
3536    2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
3537       with a warning accompanied by a source location.  Moreover, it
3538       merely skips the offending function or type and proceeds with the
3539       rest of its work.u  As a result, one can safely feed C code containing
3540       "..." to ml-nlffigen.
3541    3. There are some internal improvements to CM, providing slightly
3542       more general string substitutions in the tools subsystem.
3543    
3544    ----------------------------------------------------------------------
3545    Name: Matthias Blume
3546    Date: 2001/06/27 15:10:00 EDT
3547    Tag: blume-20010627-concur
3548    Description:
3549    
3550    Fixed a small bug in CM's handling of parallel compilation.
3551    (You could observe the bug by Control-C-interrupting an ordinary
3552    CMB.make or CM.stabilize and then attaching some compile servers.
3553    The result was that all of a sudden the previously interrupted
3554    compilation would continue on its own.  This was because of
3555    an over-optimization: CM did not bother to clean out certain queues
3556    when no servers were attached "anyway", resulting in the contents
3557    of these queues to grab control when new servers did get attached.)
3558    
3559    There is also another minor update to the CM manual.
3560    
3561    ----------------------------------------------------------------------
3562    Name: Matthias Blume
3563    Date: 2001/06/26 16:15:00 EDT
3564    Tag: blume-20010626-cmdoc
3565    Description:
3566    
3567    Minor typo fixed in CM manual (syntax diagram for libraries).
3568    
3569    ----------------------------------------------------------------------
3570    Name: Matthias Blume
3571    Date: 2001/06/25 22:55:00 EDT
3572    Tag: blume-20010625-x86pc
3573    Description:
3574    
3575    Fixed a nasty bug in the X86 assembly code that caused signal
3576    handlers to fail (crash) randomly.
3577    
3578    ----------------------------------------------------------------------
3579    Name: Matthias Blume
3580    Date: 2001/06/25 12:05:00 EDT
3581    Tag: blume-20010625-nlffigen
3582    Description:
3583    
3584    This update fixes a number of minor bugs in ml-nlffigen as reported by
3585    Nick Carter <nbc@andrew.cmu.edu>.
3586    
3587      1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
3588      2. Default names for generated files are now derived from the name of
3589         the C file *without its directory*.  In particular, this causes generated
3590         files to be placed locally even if the C file is in some system directory.
3591      3. Default names for generated signatures and structures are also derived
3592         from the C file name without its directory.  This avoids silly things
3593         like "structure GL/GL".
3594         (Other silly names are still possible because ml-nlffigen does not do
3595          a thorough check of whether generated names are legal ML identifiers.
3596          When in doubt, use command line arguments to force particular names.)
3597    
3598    ----------------------------------------------------------------------
3599    Name: Matthias Blume
3600    Date: 2001/06/21 12:25:00 EDT
3601    Tag: blume-20010621-eXene
3602    Description:
3603    
3604    eXene now compiles and (sort of) works again.
3605    
3606    The library name (for version > 110.33) is $/eXene.cm.
3607    
3608    I also added an new example in src/eXene/examples/nbody.  See the
3609    README file there for details.
3610    
3611    ----------------------------------------------------------------------
3612    Name: Matthias Blume
3613    Date: 2001/06/20 16:40:00 EDT