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 873, Mon Jul 9 15:26:20 2001 UTC revision 1496, Thu May 20 20:58:32 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: 2004/05/20 16:00:00 CDT
18    Tag: blume-20040520-win32
19    Description:
20    
21    Made the win32 version work again.  (Strangely, a misplaced comma had
22    slipped into win32-process.c which prevented the runtime from being
23    compiled correctly.)
24    
25    Also, included a minor addition to ml-build.bat analogous to what was
26    done in blume-20040519-ml-build.
27    
28    ----------------------------------------------------------------------
29    Name: Matthias Blume (blume (at) tti - c (dot) org)
30    Date: 2004/05/19 22:10:00 CDT
31    Tag: blume-20040519-ml-build
32    Description:
33    
34    Arranged for ml-build to clean up after itself a little bit better.
35    The script generates a temporary SML source file and compiles it using
36    CM, so CM generates metadata (GUID, SKEL, objectfile) for it.  It now
37    gets rid of those at the end, so they don't accumulate under .cm.
38    
39    This required a minor change to install.sh because the name of the
40    metadata directory (default: .cm) is actually configurable at
41    installation time.
42    
43    ----------------------------------------------------------------------
44    Name: Matthias Blume (blume (at) tti - c (dot) org)
45    Date: 2004/05/18 15:50:00 CDT
46    Tag: blume-20040518-mkreader
47    Description:
48    
49    Added Posix.IO.mk{Bin,Text}{Reader,Writer} by lifting their respective
50    implementations from internal modules PosixBinPrimIO and PosixTextPrimIO.
51    
52    ----------------------------------------------------------------------
53    Name: Matthias Blume (blume (at) tti - c (dot) org)
54    Date: 2004/05/11 14:35:00 CDT
55    Tag: blume-20040511-win32sock
56    Description:
57    
58    Added previously missing support for many socket-related functions
59    under win32.  Thanks to David Hansel <hansel@reactive-systems.com>
60    for the voluminous patch!
61    
62    (I have not tested this patch under win32 yet.)
63    
64    Here is David's e-mail:
65    
66    Hi,
67    
68    Attached to this email you find a diff against sml/nj 110.45
69    that will enable socket support under Windows.
70    
71    To apply the patch (using unix or cygwin)
72    1) gunzip runtime.diff.gz
73    2) "cd" into "src/runtime" in the source tree of a fresh
74       110.45 installation.
75    3) patch -p 1 < [your/path/to]runtime.diff
76    
77    The code compiles fine but has NOT yet been extensively tested.
78    I only ran a few tests for basic socket client functionality
79    (which worked fine).  Especially the functions that use ioctl
80    are not tested at all and might not work (see below).
81    
82    I implemented this since we want to move to a newer version of sml/nj
83    but need socket support in order to use it.  This is the first time I
84    even had a look at the sml/nj source,  so please review my changes
85    before making this part of the distribution!  Here are a few issues
86    that I think might be better for someone to solve who is more
87    familiar with the sml/nj source (and socket programming):
88    
89    - getnetbyaddr.c and getnetbyname.c will raise a "not implemented"
90      exception since I could not figure out what the windows equivalent
91      of these functions is
92    
93    - In sockets-osdep.h there are a some #include statements that are
94      only used in a few files that include sockets-osdep.h
95    
96    - In smlnj-sock-lib.c, function init_fn() calls WSAStartup() but
97      does not process its return value since I don't know how to
98      report an error upwards.
99    
100    - It would probably be good to have a call to WSACleanup() when
101      the library is unloaded (if there is such a possibility).
102      Otherwise I think Windows will take care of this automatically
103      when the process finishes.
104    
105    - I used ioctlsocket() as a replacement for ioctl() but I have
106      no idea if that is actually the proper replacement on Windows.
107    
108    - All these issues are marked in the code by "FIXME" comments.
109    
110    We use sml/nj extensively in our products and are quite happy
111    with it.  I hope this contribution will help you.
112    
113    Keep up the good work!
114    
115    David
116    
117    ----------------------------------------------------------------------
118    Name: Matthias Blume (blume (at) tti - c (dot) org)
119    Date: 2004/05/11 14:20:00 CDT
120    Tag: blume-20040511-installml
121    Description:
122    
123    Fixed two bugs in installml script.  (Thanks to Vesa A. Norrman for
124    the patch.)
125    
126    ----------------------------------------------------------------------
127    Name: Matthias Blume (blume (at) tti - c (dot) org)
128    Date: 2004/05/11 14:05:00 CDT
129    Tag: blume-20040511-nlffi-netbsd
130    Description:
131    
132    Added support for nlffi under netbsd.  (Thanks to Vesa A. Norrman for
133    the patch.)
134    
135    ----------------------------------------------------------------------
136    Name: Matthias Blume (blume (at) tti - c (dot) org)
137    Date: 2004/05/11 12:05:00 CDT
138    Tag: blume-20040511-exports
139    Description:
140    
141    As per request by Adam Chlipala <adam@hcoop.net>, extended various
142    export lists in compiler-related .cm-files.
143    
144    ----------------------------------------------------------------------
145    Name: Matthias Blume (blume (at) tti - c (dot) org)
146    Date: 2004/05/11 11:35:00 CDT
147    Tag: blume-20040511-allsource
148    Description:
149    
150    The installer now honors the "src-smlnj" target again, although its meaning
151    has changed from "all sources required for the compiler" to "all sources
152    the installer knows about".  In other words, if you enable "src-smlnj"
153    in the "targets" file, then the installer will pull in sources for
154    everything.  (Notice that this refers to source code only.  Compiled
155    code is still only installed for modules that were requested explicitly
156    or which are required for other modules that were requested explicitly.)
157    
158    ----------------------------------------------------------------------
159    Name: Matthias Blume (blume (at) tti - c (dot) org)
160    Date: 2004/04/23 17:40:00 CDT
161    Tag: blume-20040423-ieee-scan
162    Description:
163    
164    Fixed IEEEReal.scan (and .fromString) so that if there is an overflow
165    in the exponent calculation we get INF or ZERO (depending on the mantissa
166    and the sign of the exponent).
167    
168    ----------------------------------------------------------------------
169    Name: Matthias Blume (blume (at) tti - c (dot) org)
170    Date: 2004/04/23 10:40:00 CDT
171    Tag: blume-20040423-ml-build
172    Description:
173    
174    The ml-build script now terminates with a non-0 status when something
175    goes wrong.
176    
177    ----------------------------------------------------------------------
178    Name: Matthias Blume (blume (at) tti - c (dot) org)
179    Date: 2004/04/22 16:35:00 CDT
180    Tag: blume-20040422-Option
181    Description:
182    
183    Made exception Option to be the same as exception Option.Option
184    (as it should be).
185    
186    ----------------------------------------------------------------------
187    Name: Allen Leung (leunga (at) reservoir (dot) com)
188    Date: 2004/03/19 14:40:00 EST
189    Tag: leunga-20040319-cygwin-nlffi
190    Description:
191    
192    Fixed the runtime so that ml-nlffi-lib runs on the cygwin version
193    of SML/NJ.  The problem is that
194    
195       lib = dlopen(NULL, ...)
196       f   = dlsym(lib, "malloc");
197    
198    does not work on Windows unless we explicitly export symbols
199    such as 'malloc' during linking.  We fixed this by explicitly
200    exporting the required symbols with the magic gcc incantation:
201    
202        -Wl,--export-all cygwin.def
203    
204    where cygwin.def is a file containing all the symbols that
205    we wish to export.
206    
207    I suspect this is a Windows problem and we'll have to
208    do the same (somehow with windows compilers) when
209    we build the native win32 version with the system
210    calls LoadLibrary/GetProcAddress.
211    
212    ----------------------------------------------------------------------
213    Name: Matthias Blume (blume (at) tti - c (dot) org)
214    Date: 2004/03/04 16:35:00 CST
215    Tag: blume-20040304-intinf-fmt
216    Description:
217    
218    Fixed problem with IntInf.fmt (sign would show up on the right instead
219    of on the left for BIN, OCT, and HEX).
220    
221    ----------------------------------------------------------------------
222    Name: Matthias Blume (blume (at) tti - c (dot) org)
223    Date: 2004/03/04 11:25:00 CST
224    Tag: blume-20040304-symlinks
225    Description:
226    
227    Fixed problem with installer script (unix only) where bin/ml-yacc and
228    friends pointed (via symlinks) to absolute locations instead of just
229    .run-sml.  This was reported by Vesa A Norrman.
230    
231    ----------------------------------------------------------------------
232    Name: Matthias Blume (blume (at) tti - c (dot) org)
233    Date: 2004/02/13 14:50:00 CST
234    Tag: Release_110_45
235    Description:
236    
237    New working version (110.45). New bootfiles.
238    
239    ----------------------------------------------------------------------
240    Name: Matthias Blume (blume (at) tti - c (dot) org)
241    Date: 2004/01/26 15:15:15 CST
242    Tag: blume-20040126-toplevel
243    Description:
244    
245    Improved handling of exceptions at the interactive toplevel.
246    
247    ----------------------------------------------------------------------
248    Name: Matthias Blume (blume (at) tti - c (dot) org)
249    Date: 2004/01/26 11:25:00
250    Tag: blume-20040126-app
251    Description:
252    
253    Type of top-level "app" corrected.
254    Added code for setting vp_limitPtrMask to Win32-specific runtime.
255    
256    ----------------------------------------------------------------------
257    Name: Matthias Blume (blume (at) tti - c (dot) org)
258    Date: 2003/11/18 17:10 CST
259    Tag: blume-20031118-basis-fiddle
260    Description:
261    
262    - changed Timer interface to what might become the spec
263    - POSIX_FLAGS -> BIT_FLAGS according to spec
264    - some other minor discrepancies wrt. spec eliminated
265    
266    ----------------------------------------------------------------------
267    Name: Matthias Blume (blume (at) tti - c (dot) org)
268    Date: 2003/11/06 12:00:00 CST
269    Tag: Release_110_44
270    Description:
271    
272    New working version (110.44). New bootfiles.
273    
274    ----------------------------------------------------------------------
275    Name: Matthias Blume (blume (at) tti - c (dot) org)
276    Date: 2003/11/04 11:50:00 CST
277    Tag: blume-20031104-move-libraries
278    Description:
279    
280    Eliminated the "dont_move_libraries" directive in config/targets.
281    (The mechanism was broken and could not be fixed easily.  Moreover,
282    there does not seem to be any reason not to move all libraries into
283    lib during installation.  I originally implemented this directive as a
284    backward-compatibility feature when I first introduced the new CM.
285    Now that things have been stable for a long time and going back to the
286    old CM is not an option, there is no reason to keep it around.)
287    
288    ----------------------------------------------------------------------
289    Name: Matthias Blume (blume (at) tti - c (dot) org)
290    Date: 2003/11/03 16:00:00 CST
291    Tag: blume-20031103-installdir
292    Description:
293    
294    Made installer honor INSTALLDIR variable again.  (Thanks to Chris
295    Richards for pointing out the problem and providing the solution.)
296    
297    ----------------------------------------------------------------------
298    Name: Matthias Blume (blume (at) tti - c (dot) org)
299    Date: 2003/10/01 17:05:00 CDT
300    Tag: blume-20031001-lal-mlrisc
301    Description:
302    
303    MLRISC bug fix from Lal.
304    
305    ----------------------------------------------------------------------
306    Name: Matthias Blume (blume (at) tti - c (dot) org)
307    Date: 2003/09/30 16:10:00 CDT
308    Tag: blume-20030930-primio-bat
309    Description:
310    
311    1. Added openVector, nullRd, and nullWr to PRIM_IO.
312    2. Improved .bat files (for Win32 port) to make things work under Win95.
313       (thanks to Aaron S. Hawley for this one)
314    
315    ----------------------------------------------------------------------
316    Name: Matthias Blume (blume (at) tti - c (dot) org)
317    Date: 2003/09/26 16:05:00 CDT
318    Tag: blume-20030926-wrappriv
319    Description:
320    
321    Added missing wrapper for privilege "primitive" in $smlnj/viscomp/core.cm.
322    
323    ----------------------------------------------------------------------
324    Name: Matthias Blume (blume (at) tti - c (dot) org)
325    Date: 2003/09/26 15:00:00 CDT
326    Tag: blume-20030926-110_43_3
327    Description:
328    
329    - additional cleanup
330    - version number bump, NEW BOOTFILES
331    
332    ----------------------------------------------------------------------
333    Name: Matthias Blume (blume (at) tti - c (dot) org)
334    Date: 2003/09/26 12:00:00 CDT
335    Tag: blume-20030926-ppautoload
336    Description:
337    
338    I modified the read-eval-print loop so that the autoloader gets
339    invoked whenever the prettyprinter tries to look up a symbol that
340    is not currently defined in the toplevel environment but which
341    appears in CM's autoload registry.  As a result, we see far fewer of
342    those ?.Foo.Bar.xxx names in the prettyprinter's output.
343    
344    In addition to this I tried to clean up some pieces of the Basis
345    implementation (e.g., Socket, Word8Array) in order to prevent other
346    instances of these ?.Foo.Bar.xxx names from being printed.
347    
348    The mechanism that picks names for types still needs some work, though.
349    (Right now it seems that if there is a type A.t which is defined to
350    be B.u, but B is unavailable at toplevel, then A.t gets printed as
351    "?.B.u" although the perhaps more sensible solution would be to use
352    "A.t" in this case.  In other words, the prettyprinter should follow
353    a chain of DEFtycs not farther than there are corresponding toplevel
354    names in the current environment.)
355    
356    ----------------------------------------------------------------------
357    Name: Matthias Blume (blume (at) tti - c (dot) org)
358    Date: 2003/09/24 16:31:00 CDT
359    Tag: blume-20030924-installer
360    Description:
361    
362    Another installer tweak:  All the ML code for the installer is now
363    compiled during CMB.make and put into a little library called
364    $smlnj/installer.cm.  The installation then simply invokes
365    
366       sml -m $smlnj/installer.cm
367    
368    and everything happens automagically.
369    
370    Win32:  ML code senses value of environment variable SMLNJ_HOME.
371    Unix:   ML code senses values of environment variables ROOT, CONFIGDIR,
372            and BINDIR.
373    
374    The new scheme guarantees that the ML code responsible for the installation
375    is in sync with the APIs of the main system.  Also, the installer is
376    somewhat faster because the installer script is precompiled.
377    
378    ----------------------------------------------------------------------
379    Name: Matthias Blume (blume (at) tti - c (dot) org)
380    Date: 2003/09/24 15:35:00 CDT
381    Tag: blume-20030924-synsock
382    Description:
383    
384    Added a signature SYNCHRONOUS_SOCKET to basis.cm.  This is like SOCKET
385    but excludes all non-blocking operations.  Defined SOCKET (in Basis)
386    and CML_SOCKET in terms of SYNCHRONOUS_SOCKET.  Removed superfluous
387    implementations of non-blocking operations from CML's Socket
388    structure.
389    
390    ----------------------------------------------------------------------
391    Name: Matthias Blume (blume (at) tti - c (dot) org)
392    Date: 2003/09/24 15:10:05 CDT
393    Tag: blume-20030924-sockets
394    Description:
395    
396    1. Fixed SOCKET API and implementation to match Basis spec.
397       This required changing the internal representation of sockets to one
398       that remembers (for each socket file descriptor) whether it is currently
399       blocking or non-blocking.  This state is maintained lazily (i.e., a system
400       call is made only if the state actually needs to change).
401    
402    2. OS-specific details of sockets were moved into separate files, thus
403       making it possible to unify the bulk of the socket implementations
404       between Unix and Win32.
405    
406    3. CML's socket API changed accordingly.
407       (Note that we need to remove non-blocking functions from this API
408       since they are redundant in the case of CML!)
409    
410    4. CML's socket implementation now makes use of non-blocking functions
411       provided by Basis, thus removing all OS-dependent code from this part
412       of CML.
413    
414    5. Changed Real64.precision from 52 to 53.  Minor cleanup in Real64 code.
415    
416    ----------------------------------------------------------------------
417    Name: Matthias Blume (blume (at) tti - c (dot) org)
418    Date: 2003/09/22 12:10:00 CDT
419    Tag: blume-20030922-110_43_2
420    Description:
421    
422    Made a new interim version and bootfiles for developer's bootstrapping
423    convenience.
424    
425       110.43.2 -- NEW BOOTFILES
426    
427    ----------------------------------------------------------------------
428    Name: Matthias Blume (blume (at) tti - c (dot) org)
429    Date: 2003/09/19 15:55:00 CDT
430    Tag: blume-20030919-cmdir
431    Description:
432    
433    1. new-install.sh -> install.sh
434    2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
435    3. tweaked installer so that another name instead of .cm can be chosen
436       at install time (by setting the CM_DIR_ARC environment variable
437       during installation); once installation is complete, the name is
438       fixed
439    
440    ----------------------------------------------------------------------
441    Name: Matthias Blume (blume (at) tti - c (dot) org)
442    Date: 2003/09/18 16:00:00 CDT
443    Tag: blume-20030918-110_43_1
444    Description:
445    
446    Made a new interim version and bootfiles for developer's bootstrapping
447    convenience.
448    
449       110.43.1 -- NEW BOOTFILES
450    
451    ----------------------------------------------------------------------
452    Name: Matthias Blume (blume (at) tti - c (dot) org)
453    Date: 2003/09/18 15:20:00 CDT
454    Tag: blume-20030918-misc
455    Description:
456    
457    1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
458       this seems to be controversial at the moment) and used those in
459       Posix.ProcEnv.times.
460    
461    2. Added Time.{from,to}Nanoseconds to Time.
462    
463    3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
464       For example, fromLargeInt never needs to look at more than 3 "big
465       digits" to get its 53 bits of precision.
466    
467    ----------------------------------------------------------------------
468    Name: Matthias Blume (blume (at) tti - c (dot) org)
469    Date: 2003/09/17 16:30:00 CDT
470    Tag: blume-20030917-real32-slices
471    Description:
472    
473    Added an entry to the primitive environment
474    (compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
475    added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
476    
477    Removed some of the "magic" constants in real64.sml and replaced them
478    with code that generates these values from their corresponding
479    integer counterparts.
480    
481    Made all(?) the slice-related changes to the Basis and made everything
482    compile again...
483    
484    ----------------------------------------------------------------------
485    Name: Matthias Blume (blume (at) tti - c (dot) org)
486    Date: 2003/09/15 17:45:00 CDT
487    Tag: blume-20030915-rbase
488    Description:
489    
490    Fixed bug in Real.fromLargeInt.
491    
492    ----------------------------------------------------------------------
493    Name: Matthias Blume (blume (at) tti - c (dot) org)
494    Date: 2003/09/13 18:11:00 CDT
495    Tag: blume-20030913-libinstall
496    Description:
497    
498    Minor bugfix in config/libinstall (set anchor with path to
499    standalone tool after installing it, otherwise libraries that
500    need ml-lex or ml-yacc won't compile the first time the installer
501    runs).
502    
503    ----------------------------------------------------------------------
504    Name: Matthias Blume (blume (at) tti - c (dot) org)
505    Date: 2003/09/12 11:45:00 CDT
506    Tag: blume-20030912-various
507    Description:
508    
509     - fixed bug in Real.toLargeInt
510     - fixed bug in Posix.ProcEnv.times
511     - changed inputLine functions to return an option
512     - minor installer improvements / bugfixes
513     - changed default @SMLalloc parameter for x86/celeron to 64k
514    
515    ----------------------------------------------------------------------
516    Name: Matthias Blume (blume (at) tti - c (dot) org)
517    Date: 2003/09/09 22:00:00 CDT
518    Tag: Release_110_43
519    Description:
520    
521        New working release 110.43.  New bootfiles.
522    
523    ----------------------------------------------------------------------
524    Name: Matthias Blume (blume (at) tti - c (dot) org)
525    Date: 2003/09/09 19:20:00 CDT
526    Tag: blume-20030909-installer
527    Description:
528    
529    Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
530    Modified config/install.bat to take advantage of it.  Also modified
531    config/install.sh (and called it config/new-install.sh) to take advantage
532    of it on Unix systems.  (The SML code is (supposed to be) platform-
533    independent.)
534    
535    The installer can now install everything under Win32
536    as well as under *nix as long as it compiles.
537    
538    Other changes:
539    
540      - made CML compile again under Win32
541      - made eXene compile under Win32 (by providing a fake structure UnixSock
542        and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
543      - fixed a bug in nowhere: it assumed that type OS.Process.status is the
544        same as type int;  under Win32 it isn't
545      - fixed some slice-related problems in the win32-specific parts of CML
546      - added a functor argument "sameVol" to os-path-fn.sml in the Basis
547        (under Win32, the volume name is case-insensitive, and the
548        OS.Path code compares volume names for equality)
549    
550    ----------------------------------------------------------------------
551    Name: Matthias Blume (blume (at) tti - c (dot) org)
552    Date: 2003/09/08 11:55:00 CDT
553    Tag: blume-20030908-fullpath
554    Description:
555    
556    Made Win32 version of OS.FileSys.fullPath return current directory
557    when given an empty string.  This is what the spec says, and incidentally,
558    CM depends on it.  (CM otherwise goes into an infinite loop in certain
559    cases when presented with the name of a non-existing .cm file.)
560    
561    ----------------------------------------------------------------------
562    Name: Matthias Blume (blume (at) tti - c (dot) org)
563    Date: 2003/09/04 16:30:00 CDT
564    Tag: blume-20030905-slices-etc
565    Description:
566    
567      1. Changed interface to vectors and arrays in Basis to match
568         (draft) Basis spec.
569      2. Added signatures and implementations of slices according to
570         Basis spec.
571      3. Edited source code throughout the system to make it compile again
572         under 1. and 2.  (In some cases code had to be added to have it
573         match the new signatures.)
574      4. MLRISC should be backward-compatible: the copies of the originals
575         of files that needed to change under 3. were retained, the .cm files
576         check the compiler version number and use old versions when
577         appropriate.
578      5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
579         dirstream -> string option (in accordance with Basis spec).
580      6. When generating code that counts lines, ml-lex used function
581         CharVector.foldli, taking advantage of its old interface.
582         This has been replaced with the corresponding code from
583         CharVectorSlice.  (html-lex must be re-lexed!)
584      7. BitArray in smlnj-lib/Util has been extended/modified to match the
585         new MONO_ARRAY signature.  (Do we need BitArraySlice?)
586      8. Removed temporary additions (fromInternal, toInternal) from the
587         (now obsolete) IntInf in smlnj-lib/Util.
588      9. Cleaned up structure Byte.
589     10. Added localOffset, scan, and fromString to Date (according to spec).
590         Cleaned/corrected implementation of Date.
591         (Still need to check for correctness; implement better canonicalizeDate.)
592     11. Added "scan" to signature IEEE_REAL.
593     12. Some improvements to IntInf [in particular: efficiency-hack for
594         mod and rem when second operand is 2 (for parity checks).]
595     13. Changed representation of type Time.time, using a single IntInf.int
596         value counting microseconds.  This considerably simplified the
597         implementation of structure Time.  We now support negative time
598         values; scan and fromString handle signs.
599     14. Functor PrimIO now takes two additional arguments (VectorSlice and
600         ArraySlice).
601    
602    ----------------------------------------------------------------------
603    Name: Matthias Blume (blume (at) tti - c (dot) org)
604    Date: 2003/08/28 17:00:00 CDT
605    Tag: blume-20030828-intinf
606    Description:
607    
608    This is a major update which comes with a version number bump
609    (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
610    and an implementation of IntInf in the Basis.
611    
612    There are a fairly large number of related changes and updates throughout
613    the system:
614    
615    Basis:
616            - Implemented IntInf.
617            - Made LargeInt a projection of IntInf (by filtering through INTEGER).
618            - Added some missing Real64 operations, most notably Real.toLargeInt.
619            - Added FixedInt as a synonym for Int32.
620    
621    compiler:
622            * Added support for a built-in intinf type.
623              - literals
624              - pattern matching
625              - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
626              - overloading on literals and operations
627    
628            This required adding a primitive type intinf, some additional
629            primops, and implementations for several non-trivial intinf
630            operations in Core.  (The intinf type is completely abstract
631            to the compiler; all operations get delegated back to the Core.)
632    
633            * Intinf equality is handled by polyequal.  However, the compiler
634            does not print its usual warning in this case (since polyequal
635            is the right thing to do there).
636    
637            * Improved the organization of structure InlineT.
638    
639            * A word about conversion primops:
640            If conversions involving intinf do not cancel out during
641            CPS contract, then the compiler must insert calls to Core functions.
642            Since all core access must be resolved already during the FLINT
643            translate phase, it would be too late a the time of CPS contract
644            to add new Core calls.  For this reason, conversion primops
645            for intinf carry two arguments:  1. the numeric argument that
646            they are supposed to convert, and 2. the Core function that
647            can help with this conversion if necessary.  If CPS contract
648            eliminates a primop, then the associated Core function becomes
649            dead and goes away.  Intinf conversion primops that do not get
650            eliminated by CPS contract get rewritten into calls of their
651            core functions by a separate, new phase.
652    
653    interactive system:
654            - Control.Print.intinfDepth controls max length of intinf constants
655            being printed. (Analogous to Control.Print.stringDepth.)
656            - Cleanup in printutil and pputil: got rid of unused stuff and
657            duplicates; replaced some of the code with code that makes better
658            use of library functionality.
659    
660    CM:
661            Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
662            are no longer silent.
663    
664    CKIT:
665            Fixed mismatched uses of Int32 and LargeInt.  I always decided
666            in favor of LargeInt -- which is now the same as IntInf.
667            CKIT-knowledgable people should check whether this is what's
668            intended and otherwise change things back to using Int32 or
669            FixedInt.
670    
671    Throughout the code:
672            Started using IntInf.int literals and built-in operations
673            (e.g., comparison with 0) where this seems appropriate.
674    
675    
676    ----------------------------------------------------------------------
677    Name: Dave MacQueen (dbm@cs.uchicago.edu)
678    Date: 2003/08/13 11:36:00 CDT
679    Tag: dbm-20030813-mcz-merge1
680    Description:
681    
682    Merging changes from the mcz-branch development branch into trunk.
683    These changes involve replacement of the emulated old prettyprinter
684    interface with direct use of the SML/NJ Lib PP library, and fixing
685    of a couple of bugs (895, 1186) relating to error messages.  A new
686    prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
687    has been added.
688    
689    ----------------------------------------------------------------------
690    Name: Matthias Blume (blume (at) tti - c (dot) org)
691    Date: 2003/08/11 15:45:00 CDT
692    Tag: blume-20030811-windows
693    Description:
694    
695    Version number bumped to 110.42.9.  NEW BOOTFILES!!!
696    
697        http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
698    
699    This patch restores SML/NJ's ability to run under win32.  There are a
700    number of changes, including fixes for several bugs that had gone
701    unnoticed until now:
702    
703      - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
704        trying to run the win32 version from within cygwin.)
705    
706      - There are a number of simple .bat scripts that substitute for their
707        corresponding Unix shell-scripts.  (See below.)
708    
709      - The internals of ml-build have been modified slightly.  The main
710        difference is that instead of calling ".link-sml" (or link-sml.bat)
711        using OS.Process.system, the ML process delegates this task back
712        to the script.  Otherwise problems arise in mixed environments such
713        as Cygwin where scripts look and work like Unix scripts, but
714        where OS.Process.system cannot run them.
715    
716      - In CM, the srcpath pickler used native pathname syntax -- which
717        is incorrect in the case of cross-compilation.  The new pickle format
718        is independent of platform-specific naming conventions.
719    
720      - Path configuration files (such as lib/pathconfig) can now choose
721        between native and standard syntax.  Placing a line of the form
722    
723            standard!
724    
725        into the file causes all subsequent paths to be interpreted using
726        CM standard pathname syntax (= Unix conventions); a line
727    
728            native!
729    
730        switches back to native style.  This was needed so that
731        path config files can be written portably, see src/system/pathconfig.
732    
733      - Runtime system:
734    
735          - win32-filesys.c:  get_file_time and set_file_time now
736            access modification time, not creation time.
737    
738          - I/O code made aware of new array representation.
739    
740          - Bug fixes in X86.prim.masm.
741    
742       - src/system/makeml made aware of win32. (For use under cygwin
743            and other Unix-environments for windows.)
744    
745       - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
746         which caused BinIO.inputAll to fail consistently.
747    
748    .bat scripts:
749    
750       Windows .bat scripts assume that SMLNJ_HOME is defined.
751    
752       - sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
753         applications (sml, ml-yacc, ml-lex).
754       - ml-build.bat: analogous to ml-build.
755       - config\install.bat:  Analogous to config/install.sh.  This requires
756         that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
757         (nmake etc. must be on the path, and vcvars32 must have been run.)
758         Moreover, sources for ml-lex and ml-yacc need to exist under src,
759         and the bootfile hierarchy must have been unpacked under
760         sml.boot.x86-win32.
761         The script is very primitive and does a poor job at error checking.
762         It only installs the base system, ml-lex, and ml-yacc.  No other
763         libraries are being installed (i.e., you get only those that
764         are part of the compiler.)
765       - link-sml.bat:  analogous to .link-sml, but not currently used
766    
767    Unrelated bug fixes:
768    
769      - ml-nlffigen now exports structures ST_* corresponding to incomplete
770        types.
771      - Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
772        otherwise.)
773    
774    ----------------------------------------------------------------------
775    Name: Dave MacQueen (macqueen@cs.uchicago.edu)
776    Date: 2003/06/17
777    Tag: macqueen-20030617-bug895
778    Description:
779    
780    Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
781    Tag will be used for new development branch (mcz-branch) for
782    use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
783    
784    ----------------------------------------------------------------------
785    Name: Matthias Blume (blume (at) tti - c (dot) org)
786    Date: 2003/05/27 16:55:00 CDT
787    Tag: blume-20030527-polyeq
788    Description:
789    
790    Tried to eliminated most cases of polymorphic equality.
791    
792    ----------------------------------------------------------------------
793    Name: Matthias Blume (blume (at) tti - c (dot) org)
794    Date: 2003/05/21 17:45:00 CDT
795    Tag: blume-20030517-complete
796    Description:
797    
798    Two changes:
799    
800      1. Added a flag for controlling whether non-exhaustive bindings will
801         be treated as errors (default is false).
802      2. Cleaned up the *entire* source tree so that CMB.make goes through
803         without a single non-exhaustive match- or bind warning.
804    
805    ----------------------------------------------------------------------
806    Name: Matthias Blume (blume (at) tti - c (dot) org)
807    Date: 2003/05/17 10:20:00 CDT
808    Tag: blume-20030517-absyn
809    Description:
810    
811    1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
812    
813       This mainly affects the quality of error messages.  However, some
814       of the code is now more straightforward than before. (Treatment of
815       the above four constructs in translate.sml is much simpler than
816       the "macro-expansion" that was going on before.  Plus, the mach-
817       compiler no longer gets invoked just to be able to compile an
818       if-expression.)
819    
820    2. The ErrorMsg.Error exception is now caught and absorbed by the
821       interactive loop.
822    
823    ----------------------------------------------------------------------
824    Name: Allen Leung
825    Date: 2003/05/16 13:05:00 CDT
826    Tag: leunga-20030516-cygwin-runtime
827    Description:
828    
829        Ported the runtime system to cygwin, which uses the unix
830    x86-unix bin files.  Missing/buggy features:
831    
832       o getnetbyname, getnetbyaddr: these functions seem to be missing in
833         the Cygwin library.
834       o Ctrl-C handling may be flaky.
835       o Windows system calls and Windows I/O are not supported.
836    
837       A new set of binfiles is located at:
838    
839         http://www.dorsai.org/~leunga/boot.x86-unix.tgz
840    
841       This is only needed for bootstrapping the cygwin version of smlnj.
842       Other x86 versions can use the existing binfiles.
843    
844    ----------------------------------------------------------------------
845    Name: Matthias Blume
846    Date: 2003/04/08 15:42:00 CDT
847    Tag: blume-20030408-listpair
848    Description:
849    
850    1. Added a target 'mlrisc' to installer.
851    
852    2. Added missing elements to structure ListPair.
853    
854    ----------------------------------------------------------------------
855    Name: Allen Leung
856    Date: 2003/01/07 10:40:00 EST
857    Tag: leunga-20030107-int-rem
858    Description:
859    
860    Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
861    The arguments to the SUBL instruction were swapped.
862    
863    ----------------------------------------------------------------------
864    Name: Matthias Blume
865    Date: 2002/12/12 16:25:00 EST
866    Tag: blume-20021212-risc-ra
867    Description:
868    
869    Fixed a serious bug in the rewrite code for FP spilling/reloading that
870    sent the RA into an infinite loop when floating point registers get
871    spilled.  (Because of this bug, e.g., nucleic stopped compiling between
872    110.37 and 110.38.)
873    There was another set of potential problems related to the handling of
874    MLRISC annotations (but those did not yet cause real problems, apparently).
875    
876    ----------------------------------------------------------------------
877    Name: Matthias Blume
878    Date: 2002/12/06 22:40:00 EST
879    Tag: blume-20021206-cm-fileid
880    Description:
881    
882    Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
883    This fixes the problem of CM getting confused by files that suddenly
884    change their identity (e.g., by getting unlinked and recreated by some
885    text editor such as vi).  There might be a better/cheaper/cleaner way
886    of doing this, but for now this will have to do.
887    
888    ----------------------------------------------------------------------
889    Name: Matthias Blume
890    Date: 2002/10/28 09:50:00 EST
891    Tag: blume-20021028-typecheck
892    Description:
893    
894    Exported structure Typecheck from $smlnj/viscomp/core.cm.
895    
896    ----------------------------------------------------------------------
897    Name: Matthias Blume
898    Date: 2002/10/17 09:10:00 EDT
899    Tag: Release_110_42
900    Description:
901    
902    In good old tradition, there has been a slight hiccup so that we have
903    to patch 110.42 after the fact.   The old release tag has been replaced
904    (see below).
905    
906    The change solves a problem with two competing approaches the
907    configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
908    each other's way.
909    
910    This change only affects the runtime system code and the installer script.
911    (No new bootfiles.)
912    
913    ----------------------------------------------------------------------
914    Name: Matthias Blume
915    Date: 2002/10/16 12:00:00 EDT
916    Tag: Release_110_42_removed
917    Description:
918    
919    New working release.  New bootfiles.
920    
921    ----------------------------------------------------------------------
922    Name: Matthias Blume
923    Date: 2002/10/10 13:10:00 EDT
924    Tag: blume-20021010-ppc-divs
925    Description:
926    
927    The mltree operator DIVS must be implemented with an overflow check on
928    the PPC because the hardware indicates divide-by-zero using "overflow" as
929    well.
930    
931    ----------------------------------------------------------------------
932    Name: Matthias Blume
933    Date: 2002/07/23 11:45:00
934    Tag: blume-20020723-smlnj-home
935    Description:
936    
937    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
938    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
939    CM_PATHCONFIG is also set), the path configuration file is assumed
940    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
941    the entire tree to some other place and everything will "just work".
942    
943    (Companion commands such as ml-build and ml-makedepend also sense this
944    variable.)
945    
946    ----------------------------------------------------------------------
947    Name: Matthias Blume
948    Date: 2002/07/12 21:19:00 EDT
949    Tag: blume-20020712-liveness
950    Description:
951    
952    Exported two useful "step" functions from liveness module (MLRISC).
953    
954    ----------------------------------------------------------------------
955    Name: Matthias Blume
956    Date: 2002/07/05 16:00 EDT
957    Tag: Release_110_41
958    Description:
959    
960    New working release.  New bootfiles.
961    
962    ----------------------------------------------------------------------
963    Name: Matthias Blume
964    Date: 2002/07/05 10:25:00 EDT
965    Tag: blume-20020705-btimp
966    Description:
967    
968    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
969    other clients can set up backtracing support.
970    
971    ----------------------------------------------------------------------
972    Name: Matthias Blume
973    Date: 2002/06/25 17:23:00 EDT
974    Tag: blume-20020625-fpmax
975    Description:
976    
977    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
978    case.  (The sense of the isNaN test was reversed -- which made min and
979    max always return their first argument.)
980    
981    ----------------------------------------------------------------------
982    Name: Matthias Blume
983    Date: 2002/06/11
984    Tag: blume-20020611-unixpath
985    Description:
986    
987    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
988    
989    ----------------------------------------------------------------------
990    Name: Matthias Blume
991    Date: 2002/06/10 16:35:00 EDT
992    Tag: blume-20020610-ieeereal
993    Description:
994    
995    I back-ported my implementation of IEEEReal.fromString from the
996    idlbasis-devel branch so that we can test it.
997    
998    Another small change is that ppDec tries to give more information
999    than just "<sig>" in the case of functors.  However, this code is
1000    broken in some mysterious way if the functor's body's signature
1001    has not been declared by ascription but gets inferred from the
1002    implementation.  This needs fixing...
1003    
1004    ----------------------------------------------------------------------
1005    Name: Matthias Blume
1006    Date: 2002/05/31
1007    Tag: blume-20020531-btrace-mode
1008    Description:
1009    
1010    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
1011    the wayside when I switched over to using Controls everywhere.)
1012    
1013    ----------------------------------------------------------------------
1014    Name: Lal George
1015    Date: 2002/05/23 12:21:40 EDT
1016    Tag: george-20020523-visual-labels
1017    Description:
1018    
1019    Labels are now displayed in the graphical output to make
1020    the fall-through and target blocks obvious.
1021    
1022    ----------------------------------------------------------------------
1023    Name: Matthias Blume
1024    Date: 2002/05/22 11:03:00 EDT
1025    Tag: blume-20020522-shrink
1026    Description:
1027    
1028    John tweaked yesterday's fix for 1131 to handle an out-of-memory
1029    situation that comes up when allocating huge arrays.
1030    
1031    ----------------------------------------------------------------------
1032    Name: Matthias Blume
1033    Date: 2002/05/21 16:00:00 EDT
1034    Tag: Release_110_40
1035    Description:
1036    
1037    New working release (110.40).  New bootfiles.
1038    
1039    [Also: John Reppy fixed GC bug 1131.]
1040    
1041    ----------------------------------------------------------------------
1042    Name: Matthias Blume
1043    Date: 2002/05/21 12:35:00 EDT
1044    Tag: blume-20020521-cmdoc
1045    Description:
1046    
1047    CM documentation update.
1048    
1049    ----------------------------------------------------------------------
1050    Name: Matthias Blume
1051    Date: 2002/05/21 10:55:00 EDT
1052    Tag: blume-20020521-misc
1053    Description:
1054    
1055    - John tweaked runtime to be silent on heap export (except when
1056      GC messages are on).
1057    - I added a few more things (cross-compiling versions of CMB) to
1058      config/preloads (as suggestions).
1059    
1060    ----------------------------------------------------------------------
1061    Name: Matthias Blume
1062    Date: 2002/05/20 22:25:00 EDT
1063    Tag: blume-20020520-controls
1064    Description:
1065    
1066    - Added ControlUtil structure to control-lib.cm.
1067    - Use it throughout.
1068    - Used Controls facility to define MLRISC controls (as opposed to
1069      registering MLRISC control ref cells with Controls after the fact)
1070    - Fixed messed-up controls priorities.
1071    
1072    * Removed again all the stuff from config/preloads that one wouldn't
1073      be able to preload at the time the initial heap image is built.
1074      (Many libraries, e.g., CML, do not exist yet at this time.  The
1075       only libraries that can be preloaded via config/preloads are those
1076       that come bundled with the bootfiles.)
1077    
1078    ----------------------------------------------------------------------
1079    Name: Matthias Blume
1080    Date: 2002/05/20 10:59:00 EDT
1081    Tag: blume-20020520-preloads
1082    Description:
1083    
1084    Added a lot of commented-out suggestions for things to be included
1085    in config/preloads.
1086    
1087    ----------------------------------------------------------------------
1088    Name: Allen Leung
1089    Date: 2002/05/18 14:20:00 EDT
1090    Tag: leunga-20020518-mdl
1091    Description:
1092    
1093    o Made the mdl tool stuff compile and run again.
1094    o I've disabled all the stuff that depends on RTL specifications; they
1095      are all badly broken anyway.
1096    
1097    ----------------------------------------------------------------------
1098    Name: Matthias Blume
1099    Date: 2002/05/17 16:49:00 EDT
1100    Tag: blume-20020517-controls
1101    Description:
1102    
1103    0. John Reppy made several modifications to the SML/NJ library.
1104       In particular, there is a shiny new controls-lib.cm.
1105    
1106    1. Pushed new controls interface through compiler so that everything
1107       compiles again.
1108    
1109    2. Added FormatComb and FORMAT_COMB to the CML version of the
1110       SML/NJ library (so that CML compiles again).
1111    
1112    3. Modified init scripts because XXX_DEFAULT environment variables
1113       are no longer with us.  (Boot-time initialization is now done
1114       using the same environment variables that are also used for
1115       startup-time initialization of controls.)
1116    
1117    ----------------------------------------------------------------------
1118    Name: Lal George
1119    Date: 2002/05/15 09:20:10 EDT
1120    Tag: george-20020515-pseudo-op-decls
1121    Description:
1122    
1123            All pseudo-ops emitted before the first segment declaration
1124    such as TEXT, DATA, and BSS directives are assumed to be global
1125    declarations and are emitted first in the assembly file. This is
1126    useful in a number of situations where one has pseudo-ops that are not
1127    specific to any segment, and also works around the constraint that one
1128    cannot have client pseudo-ops in the TEXT segment.
1129    
1130            Because no segment is associated with these declarations it is
1131    an error to allocate any space or objects before the first segment
1132    directive and an exception will be raised. However, we cannot make
1133    this check for client pseudo-ops.
1134    
1135            These top level declarations are a field in the CFG graph_info.
1136    In theory you can continue to add to this field after the CFG has been
1137    built -- provided you know what you are doing;-)
1138    
1139    ----------------------------------------------------------------------
1140    Name: Matthias Blume
1141    Date: 2002/05/13 16:40:00 EDT
1142    Tag: blume-20020513-pp-etc
1143    Description:
1144    
1145    A few minor bugfixes:
1146    
1147      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
1148        (Bogus printouts for redefined bindings still occur.  Compiler
1149        bug should no longer occur now.  We need to redo the prettyprinter
1150        from scratch.)
1151    
1152      - CM pathname printer now also adds escape sequences for ( and )
1153    
1154      - commend and docu fixes for ml-nlffi
1155    
1156    ----------------------------------------------------------------------
1157    Name: Matthias Blume
1158    Date: 2002/05/10 16:40:00 EDT
1159    Tag: blume-20020510-erg-textio
1160    Description:
1161    
1162    Applied the following bugfix provided by Emden Gansner:
1163    
1164        Output is corrupted when outputSubstr is used rather than output.
1165    
1166        The problem occurs when a substring
1167    
1168            ss = (s, dataStart, dataLen)
1169    
1170        where dataStart > 0, fills a stream buffer with avail bytes left.
1171        avail bytes of s, starting at index dataStart, are copied into the
1172        buffer, the buffer is flushed, and then the remaining dataLen-avail
1173        bytes of ss are copied into the beginning of the buffer. Instead of
1174        starting this copy at index dataStart+avail in s, the current code
1175        starts the copy at index avail.
1176    
1177        Fix:
1178        In text-io-fn.sml, change line 695 from
1179             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
1180        to
1181             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
1182    
1183    ----------------------------------------------------------------------
1184    Name: Matthias Blume
1185    Date: 2002/04/12 13:55:00 EDT
1186    Tag: blume-20020412-assyntax
1187    Description:
1188    
1189    1. Grabbed newer assyntax.h from the XFree86 project.
1190    2. Fiddled with how to compile X86.prim.asm without warnings.
1191    3. (Very) Minor cleanup in CM.
1192    
1193    ----------------------------------------------------------------------
1194    Name: Matthias Blume
1195    Date: 2002/04/01 (no joke!) 17:07:00 EST
1196    Tag: blume-20020401-x86div
1197    Description:
1198    
1199    Added full support for div/mod/rem/quot on the x86, using the machine
1200    instruction's two results (without clumsily recomputing the remainder)
1201    directly where appropriate.
1202    
1203    Some more extensive power-of-two support was added to the x86 instruction
1204    selector (avoiding expensive divs, mods, and muls where they can be
1205    replaced with cheaper shifts and masks).  However, this sort of thing
1206    ought to be done earlier, e.g., within the CPS optimizer so that
1207    all architectures benefit from it.
1208    
1209    The compiler compiles to a fixed point, but changes might be somewhat
1210    fragile nevertheless.  Please, report any strange things that you might
1211    see wrt. div/mod/quot/rem...
1212    
1213    ----------------------------------------------------------------------
1214    Name: Matthias Blume
1215    Date: 2002/03/29 17:22:00
1216    Tag: blume-20020329-div
1217    Description:
1218    
1219    Fixed my broken div/mod logic.  Unfortunately, this means that the
1220    inline code for div/mod now has one more comparison than before.
1221    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
1222    The problem was with quotient = 0, because that alone does not tell
1223    us which way the rounding went.  One then has to look at whether
1224    remainder and divisor have the same sign...  :(
1225    
1226    Anyway, I replaced the bootfiles with fresh ones...
1227    
1228    ----------------------------------------------------------------------
1229    Name: Matthias Blume
1230    Date: 2002/03/29 14:10:00 EST
1231    Tag: blume-20020329-inlprims
1232    Description:
1233    
1234    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
1235    
1236    Primops have changed. This means that the bin/boot-file formats have
1237    changed as well.
1238    
1239    To make sure that there is no confusion, I made a new version.
1240    
1241    
1242    CHANGES:
1243    
1244    * removed REMT from mltree (remainder should never overflow).
1245    
1246    * added primops to deal with divisions of all flavors to the frontend
1247    
1248    * handled these primops all the way through so they map to their respective
1249      MLRISC support
1250    
1251    * used these primops in the implementation of Int, Int32, Word, Word32
1252    
1253    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
1254    
1255    * parameterized INLMIN, INLMAX, and INLABS by a numkind
1256    
1257    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
1258      floating point
1259    
1260    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
1261      and Real (but Real.abs maps to a separate floating-point-only primop)
1262    
1263    
1264    TODO items:
1265    
1266    * Hacked Alpha32 instruction selection, disabling the selection of REMx
1267      instructions because the machine instruction encoder cannot handle
1268      them.  (Hppa, PPC, and Sparc instruction selection did not handle
1269      REM in the first place, and REM is supported by the x86 machine coder.)
1270    
1271    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
1272      selection phase.  (The two can be streamlined because the hardware
1273      delivers both quotient and remainder at the same time anyway.)
1274    
1275    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
1276      (Currently the behavior is inconsistent both across architectures and
1277      wrt. the draft Basis spec.)
1278    
1279    * Word8 should eventually be handled natively, too.
1280    
1281    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
1282      as if there currently is no execution path that could trigger it in
1283      SML/NJ.  (The assumptions underlying functions arith and promotable do not
1284      hold for things like multiplication and division.)
1285    
1286    ----------------------------------------------------------------------
1287    Name: Matthias Blume
1288    Date: 2002/03/27 16:27:00 EST
1289    Tag: blume-20020327-mlrisc-divisions
1290    Description:
1291    
1292    Added support for all four division operations (ML's div, mod, quot,
1293    and rem) to MLRISC.  In the course of doing so, I also rationalized
1294    the naming (no more annoying switch-around of DIV and QUOT), by
1295    parameterizing the operation by div_rounding_mode (which can be either
1296    DIV_TO_ZERO or DIV_TO_NEGINF).
1297    
1298    The generic MLTreeGen functor takes care of compiling all four
1299    operations down to only round-to-zero div.
1300    
1301    Missing pieces:
1302    
1303      * Doing something smarter than relying on MLTreeGen on architectures
1304        like, e.g., the x86 where hardware division delivers both quotient and
1305        remainder at the same time.  With this, the implementation of the
1306        round-to-neginf operations could be further streamlined.
1307    
1308      * Remove inlining support for div/mod/rem from the frontend and replace it
1309        with primops that get carried through to the backend.  Do this for all
1310        int and word types.
1311    
1312    ----------------------------------------------------------------------
1313    Name: Matthias Blume
1314    Date: 2002/03/25 17:25:00 EST
1315    Tag: blume-20020325-divmod
1316    Description:
1317    
1318    I improved (hopefully without breaking them) the implementation of Int.div,
1319    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
1320    advantage of the following observations:
1321    
1322      Let  q = x quot y      r = x rem y
1323           d = x div  y      m = x mod y
1324    
1325    where "quot" is the round-to-zero version of integer division that
1326    hardware usually provides.  Then we have:
1327    
1328         r = x - q * y        where neither the * nor the - will overflow
1329         d = if q >= 0 orelse x = q * y then q else q - 1
1330                              where neither the * nor the - will overflow
1331         m = if q >= 0 orelse r = 0 then r else r + y
1332                              where the + will not overflow
1333    
1334    This results in substantial simplification of the generated code.
1335    The following table shows the number of CFG nodes and edges generated
1336    for
1337            fun f (x, y) = x OPER y
1338            (* with OPER \in div, mod, quot, rem *)
1339    
1340    
1341        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
1342        --------------------------------------------------------
1343         div |         24 |         39 |         12 |         16
1344         mod |         41 |         71 |         12 |         16
1345        quot |          8 |         10 |          8 |         10
1346         rem |         10 |         14 |          8 |         10
1347    
1348    
1349    ----------------------------------------------------------------------
1350    Name: Matthias Blume
1351    Date: 2002/03/25 22:06:00 EST
1352    Tag: blume-20020325-cprotobug
1353    Description:
1354    
1355    Fixed a bug in cproto (c prototype decoder).
1356    
1357    ----------------------------------------------------------------------
1358    Name: Matthias Blume
1359    Date: 2002/03/25 16:00:00 EST
1360    Tag: blume-20020325-raw-primops
1361    Description:
1362    
1363    I did some cleanup to Allen's new primop code and
1364    replaced yesterday's bootfiles with new ones.
1365    (But they are stored in the same place.)
1366    
1367    ----------------------------------------------------------------------
1368    Name: Matthias Blume
1369    Date: 2002/03/24 22:40:00 EST
1370    Tag: blume-20020324-bootfiles
1371    Description:
1372    
1373    Made the bootfiles that Allen asked for.
1374    
1375    ----------------------------------------------------------------------
1376    Name: Allen Leung
1377    Date: 2002/03/23 15:50:00 EST
1378    Tag: leunga-20020323-flint-cps-rcc-primops
1379    Description:
1380    
1381      1. Changes to FLINT primops:
1382    
1383        (* make a call to a C-function;
1384         * The primop carries C function prototype information and specifies
1385         * which of its (ML-) arguments are floating point. C prototype
1386         * information is for use by the backend, ML information is for
1387         * use by the CPS converter. *)
1388      | RAW_CCALL of { c_proto: CTypes.c_proto,
1389                       ml_args: ccall_type list,
1390                       ml_res_opt: ccall_type option,
1391                       reentrant : bool
1392                     } option
1393       (* Allocate uninitialized storage on the heap.
1394        * The record is meant to hold short-lived C objects, i.e., they
1395        * are not ML pointers.  With the tag, the representation is
1396        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
1397        *)
1398      | RAW_RECORD of {tag:bool,sz:int}
1399      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
1400    
1401      2.  These CPS primops are now overloaded:
1402    
1403           rawload of {kind:numkind}
1404           rawstore of {kind:numkind}
1405    
1406          The one argument form is:
1407    
1408             rawload {kind} address
1409    
1410          The two argument form is:
1411    
1412             rawload {kind} [ml object, byte-offset]
1413    
1414      3. RAW_CCALL/RCC now takes two extra arguments:
1415    
1416         a. The first is whether the C call is reentrant, i.e., whether
1417            ML state should be saved and restored.
1418         b. The second argument is a string argument specifying the name of
1419            library and the C function.
1420    
1421         These things are currently not handled in the code generator, yet.
1422    
1423      4. In CProto,
1424    
1425         An encoding type of "bool" means "ml object" and is mapped into
1426         C prototype of PTR.  Note that "bool" is different than "string",
1427         even though "string" is also mapped into PTR, because "bool"
1428         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
1429    
1430      5. Pickler/unpicker
1431    
1432         Changed to handle RAW_RECORD and newest RAW_CCALL
1433    
1434      6. MLRiscGen,
1435    
1436         1. Changed to handle the new rawload/rawstore/rawrecord operators.
1437         2. Code for handling C Calls has been moved to a new module CPSCCalls,
1438            in the file CodeGen/cpscompile/cps-c-calls.sml
1439    
1440      7. Added the conditional move operator
1441    
1442             condmove of branch
1443    
1444         to cps.  Generation of this is still buggy so it is currently
1445         disabled.
1446    
1447    ----------------------------------------------------------------------
1448    Name: Lal George
1449    Date: 2002/03/22 14:18:25 EST
1450    Tag: george-20020322-cps-branch-prob
1451    Description:
1452    
1453    Implemented the Ball-Larus branch prediction-heuristics, and
1454    incorporated graphical viewers for control flow graphs.
1455    
1456    Ball-Larus Heuristics:
1457    ---------------------
1458    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
1459    
1460    By design it uses the Dempster-Shafer theory for combining
1461    probabilities.  For example, in the function:
1462    
1463        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
1464    
1465    the ball-larus heuristics predicts that the n=0 is unlikely
1466    (OH-heuristic), and the 'then' branch is unlikely because of the
1467    RH-heuristic -- giving the 'then' branch an even lower combined
1468    probability using the Dempster-Shafer theory.
1469    
1470    Finally, John Reppy's loop analysis in MLRISC, further lowers the
1471    probability of the 'then' branch because of the loop in the else
1472    branch.
1473    
1474    
1475    Graphical Viewing:
1476    ------------------
1477    I merely plugged in Allen's graphical viewers into the compiler. The
1478    additional code is not much. At the top level, saying:
1479    
1480            Control.MLRISC.getFlag "cfg-graphical-view" := true;
1481    
1482    will display the graphical view of the control flow graph just before
1483    back-patching.  daVinci must be in your path for this to work. If
1484    daVinci is not available, then the default viewer can be changed
1485    using:
1486    
1487            Control.MLRISC.getString "viewer"
1488    
1489    which can be set to "dot" or "vcg" for the corresponding viewers. Of
1490    course, these viewers must be in your path.
1491    
1492    The above will display the compilation unit at the level of clusters,
1493    many of which are small, boring, and un-interesting. Also setting:
1494    
1495            Control.MLRISC.getInt "cfg-graphical-view_size"
1496    
1497    will display clusters that are larger than the value set by the above.
1498    
1499    
1500    ----------------------------------------------------------------------
1501    Name: Matthias Blume
1502    Date: 2002/03/21 22:20:00 EST
1503    Tag: blume-20020321-kmp-bugfix
1504    Description:
1505    
1506    Changed the interface to the KMP routine in PreString and fixed
1507    a minor bug in one place where it was used.
1508    
1509    ----------------------------------------------------------------------
1510    Name: Allen Leung
1511    Date: 2002/03/21 20:30:00 EST
1512    Tag: leunga-20020321-cfg
1513    Description:
1514    
1515      Fixed a potential problem in cfg edge splitting.
1516    
1517    ----------------------------------------------------------------------
1518    Name: Allen Leung
1519    Date: 2002/03/21 17:15:00 EST
1520    Tag: leunga-20020321-x86-fp-cfg
1521    Description:
1522    
1523      1. Recoded the buggy parts of x86-fp.
1524    
1525         a. All the block reordering code has been removed.
1526            We now depend on the block placement phases to do this work.
1527    
1528         b. Critical edge splitting code has been simplified and moved into the
1529            CFG modules, as where they belong.
1530    
1531         Both of these were quite buggy and complex.  The code is now much, much
1532         simpler.
1533    
1534      2. X86 backend.
1535    
1536         a. Added instructions for 64-bit support.  Instruction selection for
1537            64-bit has not been committed, however, since that
1538            requires changes to MLTREE which haven't been approved by
1539            Lal and John.
1540    
1541         b. Added support for FUCOMI and FUCOMIP when generating code for
1542            PentiumPro and above.  We only generate these instructions in
1543            the fast-fp mode.
1544    
1545         c. Added cases for JP and JNP in X86FreqProps.
1546    
1547      3. CFG
1548    
1549         CFG now has a bunch of methods for edge splitting and merging.
1550    
1551      4. Machine description.
1552    
1553         John's simplification of MLTREE_BASIS.fcond broke a few machine
1554         description things:
1555    
1556         rtl-build.{sig,sml} and hppa.mdl fixed.
1557    
1558         NOTE: the machine description stuff in the repository is still broken.
1559               Again, I can't put my fixes in because that involves
1560               changes to MLTREE.
1561    
1562    ----------------------------------------------------------------------
1563    Name: Matthias Blume
1564    Date: 2002/03/20 15:55:00 EST
1565    Tag: blume-20020320-kmp
1566    Description:
1567    
1568    Implemented Knuth-Morris-Pratt string matching in PreString and used
1569    it for String.isSubstring, Substring.isSubstring, and
1570    Substring.position.
1571    
1572    (Might need some stress-testing.  Simple examples worked fine.)
1573    
1574    ----------------------------------------------------------------------
1575    Name: Matthias Blume
1576    Date: 2002/03/19 16:37:00 EST
1577    Tag: blume-20020319-witnesses
1578    Description:
1579    
1580    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
1581    
1582    This implements a generic mechanism for changing constness qualifiers
1583    anywhere within big C types without resorting to outright "casts".
1584    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
1585    modify the constness at the outermost level.)
1586    The implementation of "convert" is based on the idea of "witness"
1587    values -- values that are not used by the operation but whose types
1588    "testify" to their applicability.  On the implementation side, "convert"
1589    is simply a projection (returning its second curried argument).  With
1590    cross-module inlining, it should not result in any machine code being
1591    generated.
1592    
1593    ----------------------------------------------------------------------
1594    Name: Matthias Blume
1595    Date: 2002/03/15 16:40:00 EST
1596    Tag: blume-20020315-basis
1597    Description:
1598    
1599    Provided (preliminary?) implementations for
1600    
1601      {String,Substring}.{concatWith,isSuffix,isSubstring}
1602    
1603    and
1604    
1605      Substring.full
1606    
1607    Those are in the Basis spec but they were missing in SML/NJ.
1608    
1609    ----------------------------------------------------------------------
1610    Name: Matthias Blume
1611    Date: 2002/03/14 21:30:00 EST
1612    Tag: blume-20020314-controls
1613    Description:
1614    
1615    Controls:
1616    ---------
1617    
1618    1. Factored out the recently-added Controls : CONTROLS stuff and put
1619       it into its own library $/controls-lib.cm.  The source tree for
1620       this is under src/smlnj-lib/Controls.
1621    
1622    2. Changed the names of types and functions in this interface, so they
1623       make a bit more "sense":
1624    
1625          module -> registry
1626          'a registry -> 'a group
1627    
1628    3. The interface now deals in ref cells only.  The getter/setter interface
1629       is (mostly) gone.
1630    
1631    4. Added a function that lets one register an already-existing ref cell.
1632    
1633    5. Made the corresponding modifications to the rest of the code so that
1634       everything compiles again.
1635    
1636    6. Changed the implementation of Controls.MLRISC back to something closer
1637       to the original.  In particular, this module (and therefore MLRISC)
1638       does not depend on Controls.  There now is some link-time code in
1639       int-sys.sml that registers the MLRISC controls with the Controls
1640       module.
1641    
1642    CM:
1643    ---
1644    
1645      * One can now specify the lambda-split aggressiveness in init.cmi.
1646    
1647    ----------------------------------------------------------------------
1648    Name: Allen Leung
1649    Date: 2002/03/13 17:30:00 EST
1650    Tag: leunga-20020313-x86-fp-unary
1651    Description:
1652    
1653    Bug fix for:
1654    
1655    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
1656    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
1657    > - fun f(x,(y,z)) = Real.~ y;
1658    > [autoloading]
1659    > [autoloading done]
1660    >       fchsl   (%eax), 184(%esp)
1661    > Error: MLRisc bug: X86MCEmitter.emitInstr
1662    >
1663    > uncaught exception Error
1664    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
1665    
1666    The problem was that the code generator did not generate any fp registers
1667    in this case, and the ra didn't know that it needed to run the X86FP phase to
1668    translate the pseudo fp instruction.   This only happened with unary fp
1669    operators in certain situations.
1670    
1671    ----------------------------------------------------------------------
1672    Name: Matthias Blume
1673    Date: 2002/03/13 14:00:00 EST
1674    Tag: blume-20020313-overload-etc
1675    Description:
1676    
1677    1. Added _overload as a synonym for overload for backward compatibility.
1678       (Control.overloadKW must be true for either version to be accepted.)
1679    
1680    2. Fixed bug in install script that caused more things to be installed
1681       than what was requested in config/targets.
1682    
1683    3. Made CM aware of the (_)overload construct so that autoloading
1684       works.
1685    
1686    ----------------------------------------------------------------------
1687    Name: Matthias Blume
1688    Date: 2002/03/12 22:03:00 EST
1689    Tag: blume-20020312-url
1690    Description:
1691    
1692    Forgot to update BOOT and srcarchiveurl.
1693    
1694    ----------------------------------------------------------------------
1695    Name: Matthias Blume
1696    Date: 2002/03/12 17:30:00 EST
1697    Tag: blume-20020312-version110392
1698    Description:
1699    
1700    Yet another version number bump (because of small changes to the
1701    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
1702    
1703    Changes:
1704    
1705      The new pid generation scheme described a few weeks ago was overly
1706      complicated.  I implemented a new mechanism that is simpler and
1707      provides a bit more "stability":  Once CM has seen a compilation
1708      unit, it keeps its identity constant (as long as you do not delete
1709      those crucial CM/GUID/* files).  This means that when you change
1710      an interface, compile, then go back to the old interface, and
1711      compile again, you arrive at the original pid.
1712    
1713      There now also is a mechanism that instructs CM to use the plain
1714      environment hash as a module's pid (effectively making its GUID
1715      the empty string).  For this, "noguid" must be specified as an
1716      option to the .sml file in question within its .cm file.
1717      This is most useful for code that is being generated by tools such
1718      as ml-nlffigen (because during development programmers tend to
1719      erase the tool's entire output directory tree including CM's cached
1720      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
1721      revert to the old, broken behavior of SML/NJ, but in specific cases
1722      where there is no danger of interface confusion, its use is ok
1723      (I think).
1724    
1725      ml-nlffigen by default generates "noguid" annotations.  They can be
1726      turned off by specifying -guid in its command line.
1727    
1728    ----------------------------------------------------------------------
1729    Name: Lal George
1730    Date: 2002/03/12 12 14:42:36 EST
1731    Tag: george-20020312-frequency-computation
1732    Description:
1733    
1734    Integrated jump chaining and static block frequency into the
1735    compiler. More details and numbers later.
1736    
1737    ----------------------------------------------------------------------
1738    Name: Lal George
1739    Date: 2002/03/11 11 22:38:53 EST
1740    Tag: george-20020311-jump-chain-elim
1741    Description:
1742    
1743    Tested the jump chain elimination on all architectures (except the
1744    hppa).  This is on by default right now and is profitable for the
1745    alpha and x86, however, it may not be profitable for the sparc and ppc
1746    when compiling the compiler.
1747    
1748    The gc test will typically jump to a label at the end of the cluster,
1749    where there is another jump to an external cluster containing the actual
1750    code to invoke gc. This is to allow factoring of common gc invocation
1751    sequences. That is to say, we generate:
1752    
1753            f:
1754               testgc
1755               ja   L1      % jump if above to L1
1756    
1757            L1:
1758               jmp L2
1759    
1760    
1761    After jump chain elimination the 'ja L1' instructions is converted to
1762    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
1763    up being implemented in their long form (if L2 is far away) using:
1764    
1765            jbe     L3      % jump if below or equal to L3
1766            jmp     L2
1767         L3:
1768            ...
1769    
1770    
1771    For large compilation units L2  may be far away.
1772    
1773    
1774    ----------------------------------------------------------------------
1775    Name: Matthias Blume
1776    Date: 2002/03/11 13:30:00 EST
1777    Tag: blume-20020311-mltreeeval
1778    Description:
1779    
1780    A functor parameter was missing.
1781    
1782    ----------------------------------------------------------------------
1783    Name: Allen Leung
1784    Date: 2002/03/11 10:30:00 EST
1785    Tag: leunga-20020311-runtime-string0
1786    Description:
1787    
1788       The representation of the empty string now points to a
1789    legal null terminated C string instead of unit.  It is now possible
1790    to convert an ML string into C string with InlineT.CharVector.getData.
1791    This compiles into one single machine instruction.
1792    
1793    ----------------------------------------------------------------------
1794    Name: Allen Leung
1795    Date: 2002/03/10 23:55:00 EST
1796    Tag: leunga-20020310-x86-call
1797    Description:
1798    
1799       Added machine generation for CALL instruction (relative displacement mode)
1800    
1801    ----------------------------------------------------------------------
1802    Name: Matthias Blume
1803    Date: 2002/03/08 16:05:00
1804    Tag: blume-20020308-entrypoints
1805    Description:
1806    
1807    Version number bumped to 110.39.1.  NEW BOOTFILES!
1808    
1809    Entrypoints: non-zero offset into a code object where execution should begin.
1810    
1811    - Added the notion of an entrypoint to CodeObj.
1812    - Added reading/writing of entrypoint info to Binfile.
1813    - Made runtime system bootloader aware of entrypoints.
1814    - Use the address of the label of the first function given to mlriscGen
1815      as the entrypoint.  This address is currently always 0, but it will
1816      not be 0 once we turn on block placement.
1817    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
1818      with entry points) from mlriscGen.
1819    
1820    ----------------------------------------------------------------------
1821    Name: Allen Leung
1822    Date: 2002/03/07 20:45:00 EST
1823    Tag: leunga-20020307-x86-cmov
1824    Description:
1825    
1826       Bug fixes for CMOVcc on x86.
1827    
1828       1. Added machine code generation for CMOVcc
1829       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
1830       3. CMOVcc cannot have an immediate operand as argument.
1831    
1832    ----------------------------------------------------------------------
1833    Name: Matthias Blume
1834    Date: 2002/03/07 16:15:00 EST
1835    Tag: blume-20020307-controls
1836    Description:
1837    
1838    This is a very large but mostly boring patch which makes (almost)
1839    every tuneable compiler knob (i.e., pretty much everything under
1840    Control.* plus a few other things) configurable via both the command
1841    line and environment variables in the style CM did its configuration
1842    until now.
1843    
1844    Try starting sml with '-h' (or, if you are brave, '-H')
1845    
1846    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
1847    implements the underlying generic mechanism.
1848    
1849    The interface to some of the existing such facilities has changed somewhat.
1850    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
1851    (The getFoo interface is still there for backward-compatibility, but its
1852    use is deprecated.)
1853    
1854    The ml-build script passes -Cxxx=yyy command-line arguments through so
1855    that one can now twiddle the compiler settings when using this "batch"
1856    compiler.
1857    
1858    TODO items:
1859    
1860    We should go through and throw out all controls that are no longer
1861    connected to anything.  Moreover, we should go through and provide
1862    meaningful (and correct!) documentation strings for those controls
1863    that still are connected.
1864    
1865    Currently, multiple calls to Controls.new are accepted (only the first
1866    has any effect).  Eventually we should make sure that every control
1867    is being made (via Controls.new) exactly once.  Future access can then
1868    be done using Controls.acc.
1869    
1870    Finally, it would probably be a good idea to use the getter-setter
1871    interface to controls rather than ref cells.  For the time being, both
1872    styles are provided by the Controls module, but getter-setter pairs are
1873    better if thread-safety is of any concern because they can be wrapped.
1874    
1875    *****************************************
1876    
1877    One bug fix: The function blockPlacement in three of the MLRISC
1878    backpatch files used to be hard-wired to one of two possibilities at
1879    link time (according to the value of the placementFlag).  But (I
1880    think) it should rather sense the flag every time.
1881    
1882    *****************************************
1883    
1884    Other assorted changes (by other people who did not supply a HISTORY entry):
1885    
1886    1. the cross-module inliner now works much better (Monnier)
1887    2. representation of weights, frequencies, and probabilities in MLRISC
1888       changed in preparation of using those for weighted block placement
1889       (Reppy, George)
1890    
1891    ----------------------------------------------------------------------
1892    Name: Lal George
1893    Date: 2002/03/07 14:44:24 EST 2002
1894    Tag: george-20020307-weighted-block-placement
1895    
1896    Tested the weighted block placement optimization on all architectures
1897    (except the hppa) using AMPL to generate the block and edge frequencies.
1898    Changes were required in the machine properties to correctly
1899    categorize trap instructions. There is an MLRISC flag
1900    "weighted-block-placement" that can be used to enable weighted block
1901    placement, but this will be ineffective without block/edge
1902    frequencies (coming soon).
1903    
1904    
1905    ----------------------------------------------------------------------
1906    Name: Lal George
1907    Date: 2002/03/05 17:24:48 EST
1908    Tag: george-20020305-linkage-cluster
1909    
1910    In order to support the block placement optimization, a new cluster
1911    is generated as the very first cluster (called the linkage cluster).
1912    It contains a single jump to the 'real' entry point for the compilation
1913    unit. Block placement has no effect on the linkage cluster itself, but
1914    all the other clusters  have full freedom in the manner in which they
1915    reorder blocks or functions.
1916    
1917    On the x86 the typical linkage code that is generated is:
1918       ----------------------
1919            .align 2
1920       L0:
1921            addl    $L1-L0, 72(%esp)
1922            jmp     L1
1923    
1924    
1925            .align  2
1926       L1:
1927       ----------------------
1928    
1929    72(%esp) is the memory location for the stdlink register. This
1930    must contain the address of the CPS function being called. In the
1931    above example, it contains the address of  L0; before
1932    calling L1 (the real entry point for the compilation unit), it
1933    must contain the address for L1, and hence
1934    
1935            addl $L1-L0, 72(%esp)
1936    
1937    I have tested this on all architectures except the hppa.The increase
1938    in code size is of course negligible
1939    
1940    ----------------------------------------------------------------------
1941    Name: Allen Leung
1942    Date: 2002/03/03 13:20:00 EST
1943    Tag: leunga-20020303-mlrisc-tools
1944    
1945      Added #[ ... ] expressions to mlrisc tools
1946    
1947    ----------------------------------------------------------------------
1948    Name: Matthias Blume
1949    Date: 2002/02/27 12:29:00 EST
1950    Tag: blume-20020227-cdebug
1951    Description:
1952    
1953    - made types in structure C and C_Debug to be equal
1954    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
1955    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
1956    
1957    ----------------------------------------------------------------------
1958    Name: Matthias Blume
1959    Date: 2002/02/26 12:00:00 EST
1960    Tag: blume-20020226-ffi
1961    Description:
1962    
1963    1. Fixed a minor bug in CM's "noweb" tool:
1964       If numbering is turned off, then truly don't number (i.e., do not
1965       supply the -L option to noweb).  The previous behavior was to supply
1966       -L'' -- which caused noweb to use the "default" line numbering scheme.
1967       Thanks to Chris Richards for pointing this out (and supplying the fix).
1968    
1969    2. Once again, I reworked some aspects of the FFI:
1970    
1971       A. The incomplete/complete type business:
1972    
1973       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
1974         gone!
1975       - ML types representing an incomplete type are now *equal* to
1976         ML types representing their corresponding complete types (just like
1977         in C).  This is still safe because ml-nlffigen will not generate
1978         RTTI for incomplete types, nor will it generate functions that
1979         require access to such RTTI.   But when ML code generated from both
1980         incomplete and complete versions of the C type meet, the ML types
1981         are trivially interoperable.
1982    
1983         NOTE:  These changes restore the full generality of the translation
1984         (which was previously lost when I eliminated functorization)!
1985    
1986       B. Enum types:
1987    
1988       - Structure C now has a type constructor "enum" that is similar to
1989         how the "su" constructor works.  However, "enum" is not a phantom
1990         type because each "T enum" has values (and is isomorphic to
1991         MLRep.Signed.int).
1992       - There are generic access operations for enum objects (using
1993         MLRep.Signed.int).
1994       - ml-nlffigen will generate a structure E_foo for each "enum foo".
1995         * The structure contains the definition of type "mlrep" (the ML-side
1996         representation type of the enum).  Normally, mlrep is the same
1997         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
1998         then mlrep will be defined as a datatype -- thus facilitating
1999         pattern matching on mlrep values.
2000         ("-ec" will be suppressed if there are duplicate values in an
2001          enumeration.)
2002         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
2003         will be generated for each C enum constant xxx.
2004         * Conversion functions m2i and i2m convert between mlrep and
2005         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
2006         * Coversion functions c and ml convert between mlrep and "tag enum".
2007         * Access functions (get/set) fetch and store mlrep values.
2008       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
2009         enumerations are merged into one single enumeration represented by
2010         structure E_'.
2011    
2012    ----------------------------------------------------------------------
2013    Name: Allen Leung
2014    Date: 2002/02/25 04:45:00 EST
2015    Tag: leunga-20020225-cps-spill
2016    
2017    This is a new implementation of the CPS spill phase.
2018    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
2019    In case of problems, replace it with the old file spill.sml
2020    
2021    The current compiler runs into some serious performance problems when
2022    constructing a large record.  This can happen when we try to compile a
2023    structure with many items.  Even a very simple structure like the following
2024    makes the compiler slow down.
2025    
2026        structure Foo = struct
2027           val x_1 = 0w1 : Word32.int
2028           val x_2 = 0w2 : Word32.int
2029           val x_3 = 0w3 : Word32.int
2030           ...
2031           val x_N = 0wN : Word32.int
2032        end
2033    
2034    The following table shows the compile time, from N=1000 to N=4000,
2035    with the old compiler:
2036    
2037    N
2038    1000   CPS 100 spill                           0.04u  0.00s  0.00g
2039           MLRISC ra                               0.06u  0.00s  0.05g
2040              (spills = 0 reloads = 0)
2041           TOTAL                                   0.63u  0.07s  0.21g
2042    
2043    1100   CPS 100 spill                           8.25u  0.32s  0.64g
2044           MLRISC ra                               5.68u  0.59s  3.93g
2045              (spills = 0 reloads = 0)
2046           TOTAL                                   14.71u  0.99s  4.81g
2047    
2048    1500   CPS 100 spill                           58.55u  2.34s  1.74g
2049           MLRISC ra                               5.54u  0.65s  3.91g
2050              (spills = 543 reloads = 1082)
2051           TOTAL                                   65.40u  3.13s  6.00g
2052    
2053    2000   CPS 100 spill                           126.69u  4.84s  3.08g
2054           MLRISC ra                               0.80u  0.10s  0.55g
2055              (spills = 42 reloads = 84)
2056           TOTAL                                   129.42u  5.10s  4.13g
2057    
2058    3000   CPS 100 spill                           675.59u  19.03s  11.64g
2059           MLRISC ra                               2.69u  0.27s  1.38g
2060              (spills = 62 reloads = 124)
2061           TOTAL                                   682.48u  19.61s  13.99g
2062    
2063    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
2064           MLRISC ra                               4.96u  0.27s  2.72g
2065              (spills = 85 reloads = 170)
2066           TOTAL                                   2375.26u  57.21s  48.00g
2067    
2068    As you can see the old cps spill module suffers from some serious
2069    performance problem.  But since I cannot decipher the old code fully,
2070    instead of patching the problems up, I'm reimplementing it
2071    with a different algorithm.  The new code is more modular,
2072    smaller when compiled, and substantially faster
2073    (O(n log n) time and O(n) space).  Timing of the new spill module:
2074    
2075    4000  CPS 100 spill                           0.02u  0.00s  0.00g
2076          MLRISC ra                               0.25u  0.02s  0.15g
2077             (spills=1 reloads=3)
2078          TOTAL                                   7.74u  0.34s  1.62g
2079    
2080    Implementation details:
2081    
2082    As far as I can tell, the purpose of the CPS spill module is to make sure the
2083    number of live variables at any program point (the bandwidth)
2084    does not exceed a certain limit, which is determined by the
2085    size of the spill area.
2086    
2087    When the bandwidth is too large, we decrease the register pressure by
2088    packing live variables into spill records.  How we achieve this is
2089    completely different than what we did in the old code.
2090    
2091    First, there is something about the MLRiscGen code generator
2092    that we should be aware of:
2093    
2094    o MLRiscGen performs code motion!
2095    
2096       In particular, it will move floating point computations and
2097       address computations involving only the heap pointer to
2098       their use sites (if there is only a single use).
2099       What this means is that if we have a CPS record construction
2100       statement
2101    
2102           RECORD(k,vl,w,e)
2103    
2104       we should never count the new record address w as live if w
2105       has only one use (which is often the case).
2106    
2107       We should do something similar to floating point, but the transformation
2108       there is much more complex, so I won't deal with that.
2109    
2110    Secondly, there are now two new cps primops at our disposal:
2111    
2112     1. rawrecord of record_kind option
2113        This pure operator allocates some uninitialized storage from the heap.
2114        There are two forms:
2115    
2116         rawrecord NONE [INT n]  allocates a tagless record of length n
2117         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
2118                                     and initializes the tag.
2119    
2120     2. rawupdate of cty
2121          rawupdate cty (v,i,x)
2122          Assigns to x to the ith component of record v.
2123          The storelist is not updated.
2124    
2125    We use these new primops for both spilling and increment record construction.
2126    
2127     1. Spilling.
2128    
2129        This is implemented with a linear scan algorithm (but generalized
2130        to trees).  The algorithm will create a single spill record at the
2131        beginning of the cps function and use rawupdate to spill to it,
2132        and SELECT or SELp to reload from it.  So both spills and reloads
2133        are fine-grain operations.  In contrast, in the old algorithm
2134        "spills" have to be bundled together in records.
2135    
2136        Ideally, we should sink the spill record construction to where
2137        it is needed.  We can even split the spill record into multiple ones
2138        at the places where they are needed.  But CPS is not a good
2139        representation for global code motion, so I'll keep it simple and
2140        am not attempting this.
2141    
2142     2. Incremental record construction (aka record splitting).
2143    
2144        Long records with many component values which are simulatenously live
2145        (recall that single use record addresses are not considered to
2146         be live) are constructed with rawrecord and rawupdate.
2147        We allocate space on the heap with rawrecord first, then gradually
2148        fill it in with rawupdate.  This is the technique suggested to me
2149        by Matthias.
2150    
2151        Some restrictions on when this is applicable:
2152        1. It is not a VECTOR record.  The code generator currently does not handle
2153           this case. VECTOR record uses double indirection like arrays.
2154        2. All the record component values are defined in the same "basic block"
2155           as the record constructor.  This is to prevent speculative
2156           record construction.
2157    
2158    ----------------------------------------------------------------------
2159    Name: Allen Leung
2160    Date: 2002/02/22 01:02:00 EST
2161    Tag: leunga-20020222-mlrisc-tools
2162    
2163    Minor bug fixes in the parser and rewriter
2164    
2165    ----------------------------------------------------------------------
2166    Name: Allen Leung
2167    Date: 2002/02/21 20:20:00 EST
2168    Tag: leunga-20020221-peephole
2169    
2170    Regenerated the peephole files.  Some contained typos in the specification
2171    and some didn't compile because of pretty printing bugs in the old version
2172    of 'nowhere'.
2173    
2174    ----------------------------------------------------------------------
2175    Name: Allen Leung
2176    Date: 2002/02/19 20:20:00 EST
2177    Tag: leunga-20020219-mlrisc-tools
2178    Description:
2179    
2180       Minor bug fixes to the mlrisc-tools library:
2181    
2182       1.  Fixed up parsing colon suffixed keywords
2183       2.  Added the ability to shut the error messages up
2184       3.  Reimplemented the pretty printer and fixed up/improved
2185           the pretty printing of handle and -> types.
2186       4.  Fixed up generation of literal symbols in the nowhere tool.
2187       5.  Added some SML keywords to to sml.sty
2188    
2189    ----------------------------------------------------------------------
2190    Name: Matthias Blume
2191    Date: 2002/02/19 16:20:00 EST
2192    Tag: blume-20020219-cmffi
2193    Description:
2194    
2195    A wild mix of changes, some minor, some major:
2196    
2197    * All C FFI-related libraries are now anchored under $c:
2198        $/c.cm      --> $c/c.cm
2199        $/c-int.cm  --> $c/internals/c-int.cm
2200        $/memory.cm --> $c/memory/memory.cm
2201    
2202    * "make" tool (in CM) now treats its argument pathname slightly
2203      differently:
2204        1. If the native expansion is an absolute name, then before invoking
2205           the "make" command on it, CM will apply OS.Path.mkRelative
2206           (with relativeTo = OS.FileSys.getDir()) to it.
2207        2. The argument will be passed through to subsequent phases of CM
2208           processing without "going native".  In particular, if the argument
2209           was an anchored path, then "make" will not lose track of that anchor.
2210    
2211    * Compiler backends now "know" their respective C calling conventions
2212      instead of having to be told about it by ml-nlffigen.  This relieves
2213      ml-nlffigen from one of its burdens.
2214    
2215    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
2216    
2217    * Export C_DEBUG and C_Debug from $c/c.cm.
2218    
2219    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
2220      subtyping relationship between incomplete pointers and their complete
2221      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
2222      with the convention of instantiating 'o with ('t, 'c) obj whenever
2223      the pointer target type is complete.  In the incomplete case, 'o
2224      will be instantiated with some "'c iobj" -- a type obtained by
2225      using one of the functors PointerToIncompleteType or PointerToCompleteType.
2226    
2227      Operations that work on both incomplete and complete pointer types are
2228      typed as taking an 'o ptr while operations that require the target to
2229      be known are typed as taking some ('t, 'c) obj ptr.
2230    
2231      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
2232      where void is an eqtype without any values.  This makes it possible
2233      to work on voidptr values using functions meant to operate on light
2234      incomplete pointers.
2235    
2236    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
2237      been vastly simplified.
2238    
2239    ----------------------------------------------------------------------
2240    Name: Matthias Blume
2241    Date: 2002/02/19 10:48:00 EST
2242    Tag: blume-20020219-pqfix
2243    Description:
2244    
2245    Applied Chris Okasaki's bug fix for priority queues.
2246    
2247    ----------------------------------------------------------------------
2248    Name: Matthias Blume
2249    Date: 2002/02/15 17:05:00
2250    Tag: Release_110_39
2251    Description:
2252    
2253    Last-minute retagging is becoming a tradition... :-(
2254    
2255    This is the working release 110.39.
2256    
2257    ----------------------------------------------------------------------
2258    Name: Matthias Blume
2259    Date: 2002/02/15 16:00:00 EST
2260    Tag: Release_110_39-orig
2261    Description:
2262    
2263    Working release 110.39.  New bootfiles.
2264    
2265    (Update: There was a small bug in the installer so it wouldn't work
2266    with all shells.  So I retagged. -Matthias)
2267    
2268    ----------------------------------------------------------------------
2269    Name: Matthias Blume
2270    Date: 2002/02/15 14:17:00 EST
2271    Tag: blume-20020215-showbindings
2272    Description:
2273    
2274    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
2275    the latter can be useful for exploring what bindings are available at
2276    the interactive prompt.  (The first function returns only the list
2277    of symbols that are really bound, the second prints those but also the
2278    ones that CM's autoloading mechanism knows about.)
2279    
2280    ----------------------------------------------------------------------
2281    Name: Matthias Blume
2282    Date: 2002/02/15 12:08:00 EST
2283    Tag: blume-20020215-iptrs
2284    Description:
2285    
2286    Two improvements to ml-nlffigen:
2287    
2288      1. Write files only if they do not exist or if their current contents
2289         do not coincide with what's being written.  (That is, avoid messing
2290         with the time stamps unless absolutely necessary.)
2291    
2292      2. Implement a "repository" mechanism for generated files related
2293         to "incomplete pointer types".   See the README file for details.
2294    
2295    ----------------------------------------------------------------------
2296    Name: Matthias Blume
2297    Date: 2002/02/14 11:50:00 EST
2298    Tag: blume-20020214-quote
2299    Description:
2300    
2301    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
2302    because of the new and improved tag generation scheme.  (Thanks to Allen
2303    Leung for pointing it out.)
2304    
2305    ----------------------------------------------------------------------
2306    Name: Lal George
2307    Date: 2002/02/14 09:55:27 EST 2002
2308    Tag: george-20020214-isabelle-bug
2309    Description:
2310    
2311    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
2312    of Isabelle on the x86.
2313    
2314    From Allen:
2315    -----------
2316     I've found the problem:
2317    
2318         in ra-core.sml, I use the counter "blocked" to keep track of the
2319         true number of elements in the freeze queue.  When the counter goes
2320         to zero, I skip examining the queue.  But I've messed up the
2321         bookkeeping in combine():
2322    
2323             else ();
2324             case !ucol of
2325               PSEUDO => (if !cntv > 0 then
2326                     (if !cntu > 0 then blocked := !blocked - 1 else ();
2327                                        ^^^^^^^^^^^^^^^^^^^^^^^
2328                      moveu := mergeMoveList(!movev, !moveu)
2329                     )
2330                  else ();
2331    
2332         combine() is called to coalesce two nodes u and v.
2333         I think I was thinking that if the move counts of u and v are both
2334         greater than zero then after they are coalesced then one node is
2335         removed from the freeze queue.  Apparently I was thinking that
2336         both u and v are of low degree, but that's clearly not necessarily true.
2337    
2338    
2339    02/12/2002:
2340        Here's the patch.  HOL now compiles.
2341    
2342        I don't know how this impact on performance (compile
2343        time or runtime).  This bug caused the RA (especially on the x86)
2344        to go thru the potential spill phase when there are still nodes on the
2345        freeze queue.
2346    
2347    
2348    
2349    
2350    ----------------------------------------------------------------------
2351    Name: Matthias Blume
2352    Date: 2002/02/13 22:40:00 EST
2353    Tag: blume-20020213-fptr-rtti
2354    Description:
2355    
2356    Fixed a bug in ml-nlffigen that was introduced with one of the previous
2357    updates.
2358    
2359    ----------------------------------------------------------------------
2360    Name: Matthias Blume
2361    Date: 2002/02/13 16:41:00 EST
2362    Tag: blume-20020213-cmlpq
2363    Description:
2364    
2365    Added new priority queue export symbols (which have just been added to
2366    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
2367    not compile and the installer would choke.)
2368    
2369    ----------------------------------------------------------------------
2370    Name: Matthias Blume
2371    Date: 2002/02/13 16:15:00 EST
2372    Tag: blume-20020213-various
2373    Description:
2374    
2375    1. More tweaks to ml-nlffigen:
2376    
2377       - better internal datastructures (resulting in slight speedup)
2378       - "-match" option requires exact match
2379       - "localized" gensym counters (untagged structs/unions nested within
2380         other structs/unions or within typedefs get a fresh counter; their
2381         tag will be prefixed by a concatenation of their parents' tags)
2382       - bug fixes (related to calculation of transitive closure of types
2383         to be included in the output)
2384    
2385    2. Minor Basis updates:
2386    
2387       - added implementations for List.collate and Option.app
2388    
2389    ----------------------------------------------------------------------
2390    Name: Matthias Blume
2391    Date: 2002/02/11 15:55:00 EST
2392    Tag: blume-20020211-gensym
2393    Description:
2394    
2395    Added a "-gensym" option to command line of ml-nlffigen.  This can be
2396    used to specify a "stem" -- a string that is inserted in all "gensym'd"
2397    names (ML structure names that correspond to unnamed C structs, unions,
2398    and enums), so that separate runs of ml-nlffigen do not clash.
2399    
2400    ----------------------------------------------------------------------
2401    Name: Matthias Blume
2402    Date: 2002/02/11 12:05:00 EST
2403    Tag: blume-20020211-gensml
2404    Description:
2405    
2406    A quick fix for a problem with GenSML (in the pgraph-util library):
2407    Make generation of toplevel "local" optional.  (Strictly speaking,
2408    signature definitions within "local" are not legal SML.)
2409    
2410    Other than that: updates to INSTALL and cm/TODO.
2411    
2412    ----------------------------------------------------------------------
2413    Name: Matthias Blume
2414    Date: 2002/02/08 15:00:00 EST
2415    Tag: blume-20020208-uniquepid
2416    Description:
2417    
2418    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
2419    
2420    1. The installer (config/install.sh) has gotten smarter:
2421    
2422         - Configuration options are a bit easier to specify now
2423           (in config/targets).
2424         - Bug in recognizing .tar.bz2 files fixed.
2425         - Installer automatically resolves dependencies between
2426           configuration options (e.g., if you ask for eXene, you will
2427           also get cml -- regardless whether you asked for it or not).
2428         - Installer can run in "quieter mode" by setting the environment
2429           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
2430           "completely silent", though.
2431         - Build HashCons library as part of smlnj-lib.
2432    
2433    2. A new scheme for assigning persistent identifiers to compilation
2434       units (and, by extension, to types etc.) has been put into place.
2435       This fixes a long-standing bug where types and even dynamic values
2436       can get internally confused, thereby compromising type safety
2437       (abstraction) and dynamic correctness.  See
2438    
2439         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
2440    
2441       for an example of how things could go wrong until now.
2442    
2443       The downside of the new scheme is that pids are not quite as
2444       persistent as they used to be: CM will generate a fresh pid
2445       for every compilation unit that it thinks it sees for the first
2446       time.  That means that if you compile starting from a clean, fresh
2447       source tree at two different times, you end up with different
2448       binaries.
2449    
2450       Cutoff recompilation, however, has not been compromised because
2451       CM keeps pid information in special caches between runs.
2452    
2453    ----------------------------------------------------------------------
2454    Name: Lal George
2455    Date: 2002/02/07 15:34:13 EST 2002
2456    Tag: <none>
2457    Description:
2458    
2459    Compilers that generate assembly code may produce  global labels
2460    whose value is resolved at link time. The various peephole optimization
2461    modules did not take this in account.
2462    
2463    TODO. The Labels.addrOf function should really return an option
2464    type so that clients are forced to deal with this issue, rather
2465    than an exception being raised.
2466    
2467    ----------------------------------------------------------------------
2468    Name: Lal George
2469    Date: 2002/02/06 13:55:02 EST
2470    Tag: george-20020206-ra-breakup
2471    Description:
2472    
2473    1. A bug fix from Allen.
2474    
2475        A typo causes extra fstp %st(0)'s to be generated at compensation
2476        edges, which might cause stack underflow traps at runtime.  This
2477        occurs in fft where there are extraneous fstps right before the 'into'
2478        trap instruction (in this case they are harmless since none of the
2479        integers overflow.)
2480    
2481    2. Pulled out various utility modules that were embedded in the modules
2482       of the register allocator. I need these modules for other purposes, but
2483       they are not complete enough to put into a library (just yet).
2484    ----------------------------------------------------------------------
2485    Name: Matthias Blume
2486    Date: 2002/01/31 16:05:00 EST
2487    Tag: blume-20020131-sparc-ccalls
2488    Description:
2489    
2490    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
2491       of extra stack space by mistake.  Fixed.
2492    
2493    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
2494    
2495    ----------------------------------------------------------------------
2496    Name: Allen Leung
2497    Date: 2002/01/30
2498    Tag: leunga-20020130-nowhere-bug-fix
2499    Description:
2500    
2501       MLRISC bug fixes:
2502       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
2503       2. MachineInt.fromString was negating its value.
2504    
2505    ----------------------------------------------------------------------
2506    Name: Matthias Blume
2507    Date: 2002/01/29
2508    Tag: blume-20020129-INSTALL
2509    Description:
2510    
2511    - Added somewhat detailed installation instructions (file INSTALL).
2512    - Fixed curl-detection bug in config/install.sh.
2513    - It is now possible to select the URL getter using the URLGETTER
2514      environment variable:
2515    
2516          not set / "unknown"      --> automatic detection (script tries wget,
2517                                       curl, and lynx)
2518          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
2519                                       how to properly invoke them)
2520          other                    --> use $URLGETTER directly, it must take
2521                                       precisely two command-line arguments
2522                                       (source URL and destination file name)
2523    
2524    ----------------------------------------------------------------------
2525    Name: Matthias Blume
2526    Date: 2002/01/28
2527    Tag: blume-20020128-sparc-ccalls
2528    Description:
2529    
2530    - Fixed problem with calculation of "used" registers in sparc-c-calls.
2531    - Make use of the allocParam argument in sparc-c-calls.
2532    
2533    ----------------------------------------------------------------------
2534    Name: Matthias Blume
2535    Date: 2002/01/28
2536    Tag: blume-20020128-allocParam
2537    Description:
2538    
2539    John Reppy:  Changes c-calls API to accept client-callback for
2540    allocating extra stack space.
2541    me: Corresponding changes to mlriscGen (using a dummy argument that
2542        does not change the current behavior).
2543    
2544    ----------------------------------------------------------------------
2545    Name: Matthias Blume
2546    Date: 2002/01/28 12:00:00
2547    Tag: Release_110_38
2548    Description:
2549    
2550    This time for real!!!
2551    
2552    ----------------------------------------------------------------------
2553    Name: Matthias Blume
2554    Date: 2002/01/28 10:56:00 EST
2555    Tag: blume-20020128-retraction
2556    Description:
2557    
2558    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
2559       with blume-Release_110_38-retracted.)
2560    
2561    1. Fixed a problem with incorrect rounding modes in real64.sml.
2562       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
2563    
2564    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
2565       and enums fixed.  The naming of corresponding ML identifiers should
2566       now be consistent again.
2567    
2568    ----------------------------------------------------------------------
2569    Name: Allen Leung
2570    Date: 2002/01/27
2571    Tag: leunga-20020127-nowhere
2572    Description:
2573    
2574       Added a target called nowhere in the configuration scripts.
2575       Enabling this will build the MLRISC 'nowhere' tool (for translating
2576       programs with where-clauses into legal SML code) during installation.
2577    
2578    ----------------------------------------------------------------------
2579    Name: Matthias Blume
2580    Date: 2002/01/25 21:27:00 EST
2581    Tag: blume-Release_110_38-retracted
2582    Description:
2583    
2584    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
2585    
2586    README will be added later.
2587    
2588    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
2589    (Release_110_38) removed.  Reason: Last-minute bug fixes.
2590    
2591    ----------------------------------------------------------------------
2592    Name: Matthias Blume
2593    Date: 2002/01/25
2594    Tag: blume-20020125-ffi
2595    Description:
2596    
2597    A large number of tweaks and improvements to ml-nlffi-lib and
2598    ml-nlffigen:
2599    
2600       - ML represenation types have been streamlined
2601       - getter and setter functions work with concrete values, not abstract
2602         ones where possible
2603       - ml-nlffigen command line more flexible (see README file there)
2604       - some bugs have been fixed (hopefully)
2605    
2606    ----------------------------------------------------------------------
2607    Name: Lal George
2608    Date: 2002/01/24
2609    Tag: george-20020124-risc-ra-interface
2610    Description:
2611    
2612       There is a dramatic simplification in the interface to the
2613       register allocator for RISC architectures as a result of making
2614       parallel copy instructions explicit.
2615    
2616    ----------------------------------------------------------------------
2617    Name: Matthias Blume
2618    Date: 2002/01/22
2619    Tag: blume-20020122-x86-ccalls
2620    Description:
2621    
2622    Bug fix for c-calls on x86 (having to do with how char- and
2623    short-arguments are being handled).
2624    
2625    ----------------------------------------------------------------------
2626    Name: Matthias Blume
2627    Date: 2002/01/21
2628    Tag: blume-20020121-ff
2629    Description:
2630    
2631    Another day of fiddling with the FFI...
2632    
2633    1. Bug fix/workaround:  CKIT does not complain about negative array
2634       dimensions, so ml-nlffigen has to guard itself against this possibility.
2635       (Otherwise a negative dimension would send it into an infinite loop.)
2636    
2637    2. Some of the abstract types (light objects, light pointers, most "base"
2638       types) in structure C are now eqtypes.
2639    
2640    3. Added constructors and test functions for NULL function pointers.
2641    
2642    ----------------------------------------------------------------------
2643    Name: Matthias Blume
2644    Date: 2002/01/18
2645    Tag: blume-20020118-ready-for-new-release
2646    Description:
2647    
2648    Made config/srcarchiveurl point to a new place.  (Will provide boot
2649    files shortly.)
2650    
2651    Maybe we christen this to be 110.38?
2652    
2653    ----------------------------------------------------------------------
2654    Name: Matthias Blume
2655    Date: 2002/01/18
2656    Tag: blume-20020118-more-ffifiddle
2657    Description:
2658    
2659    Today's FFI fiddling:
2660    
2661      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
2662        C.Set.* that use concrete (MLRep.*) arguments and results instead
2663        of abstract ones.
2664    
2665      - Provided word-style bit operations etc. for "int" representation
2666        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
2667        and Long.
2668    
2669    ----------------------------------------------------------------------
2670    Name: Matthias Blume
2671    Date: 2002/01/18
2672    Tag: blume-20020118-use-x86-fp
2673    Description:
2674    
2675    Now that x86-fast-fp seems to be working, I turned it back on again
2676    by default.  (Seems to work fine now, even with the FFI.)
2677    
2678    Other than that, I added some documentation about the FFI to
2679    src/ml-nlffigen/README and updated the FFI test examples in
2680    src/ml-nlffi-lib/Tests/*.
2681    
2682    ----------------------------------------------------------------------
2683    Name: Allen Leung
2684    Date: 2002/01/17
2685    Tag: leunga-20020117-x86-fast-fp-call
2686    Description:
2687    
2688       1. Fixed a problem with handling return fp values when x86's fast fp
2689          mode is turned on.
2690    
2691       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
2692          of %f32.
2693    
2694       3. Added a constructor INT32lit to the ast of MLRISC tools.
2695    
2696    ----------------------------------------------------------------------
2697    Name: Matthias Blume
2698    Date: 2002/01/16
2699    Tag: blume-20020116-ffifiddle
2700    Description:
2701    
2702    More fiddling with the FFI interface:
2703    
2704     - Make constness 'c instead of rw wherever possible.  This eliminates
2705       the need for certain explicit coercions.  (However, due to ML's
2706       value polymorphism, there will still be many cases where explicit
2707       coercions are necessary.  Phantom types are not the whole answer
2708       to modeling a subtyping relationship in ML.)
2709    
2710     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
2711    
2712     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
2713       clumsy workaround for varargs problem.)
2714    
2715    ----------------------------------------------------------------------
2716    Name: Lal George
2717    Date: 2002/01/15
2718    Tag: <none>
2719    Description:
2720    
2721    1. Since COPY instructions are no longer native to the architecture,
2722       a generic functor can be used to implement the expandCopies function.
2723    
2724    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
2725       TEXT segment.
2726    
2727    ----------------------------------------------------------------------
2728    Name: Matthias Blume
2729    Date: 2002/01/15
2730    Tag: blume-20020115-ffiupdates
2731    Description:
2732    
2733    1. Fix for bug resulting in single-precision float values being returned
2734       incorrectly from FFI calls.
2735    
2736    2. Small modifications to C FFI API:
2737    
2738        - memory-allocation routines return straight objects (no options)
2739          and raise an exception in out-of-memory situations
2740        - unsafe extensions to cast between function pointers and pointers
2741          from/to ints
2742        - added structure C_Debug as an alternative to structure C where
2743          pointer-dereferencing (|*| and |*!) always check for null-pointers
2744        - added open_lib' to DynLinkage;  open_lib' works like open_lib
2745          but also takes a (possibly empty) list of existing library handles
2746          that the current library depends on
2747    
2748    ----------------------------------------------------------------------
2749    Name: Matthias Blume
2750    Date: 2002/01/10
2751    Tag: blume-20020110-newffigen
2752    Description:
2753    
2754    1. Updates to portable graph code.
2755    
2756    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
2757       more scalable now so that even huge interfaces such as the one
2758       for GTK compile in finite time and space. :-)
2759       See src/ml-nlffigen/README for details on what's new.
2760    
2761    ----------------------------------------------------------------------
2762    Name: Lal George
2763    Date: 2001/01/09 14:31:35 EST 2002
2764    Tag: george-20011206-rm-native-copy
2765    Description:
2766    
2767            Removed the native COPY and FCOPY instructions
2768            from all the architectures and replaced it with the
2769            explicit COPY instruction from the previous commit.
2770    
2771            It is now possible to simplify many of the optimizations
2772            modules that manipulate copies. This has not been
2773            done in this change.
2774    
2775    ----------------------------------------------------------------------
2776    Name: Lal George
2777    Date: 2001/12/06 16:50:13 EST 2001
2778    Tag: george-20011206-mlrisc-instruction
2779    Description:
2780    
2781    Changed the representation of instructions from being fully abstract
2782    to being partially concrete. That is to say:
2783    
2784      from
2785            type instruction
2786    
2787      to
2788            type instr                              (* machine instruction *)
2789    
2790            datatype instruction =
2791                LIVE of {regs: C.cellset, spilled: C.cellset}
2792              | KILL of {regs: C.cellset, spilled: C.cellset}
2793              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
2794              | ANNOTATION of {i: instruction, a: Annotations.annotation}
2795              | INSTR of instr
2796    
2797    This makes the handling of certain special instructions that appear on
2798    all architectures easier and uniform.
2799    
2800    LIVE and KILL say that a list of registers are live or killed at the
2801    program point where they appear. No spill code is generated when an
2802    element of the 'regs' field is spilled, but the register is moved to
2803    the 'spilled' (which is present, more for debugging than anything else).
2804    
2805    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
2806    We used to generate:
2807    
2808            DEFFREG f1
2809            f1 := f2 + f3
2810            trapb
2811    
2812    but now generate:
2813    
2814            f1 := f2 + f3
2815            trapb
2816            LIVE {regs=[f1,f2,f3], spilled=[]}
2817    
2818    Furthermore, the DEFFREG (hack) required that all floating point instruction
2819    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
2820    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
2821    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
2822    this architecture).
2823    
2824    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
2825    all the architectures. This will result in further simplification of the
2826    register allocator that must be aware of them for coalescing purposes, and
2827    will also simplify certain aspects of the machine description that provides
2828    callbacks related to parallel copies.
2829    
2830    ANNOTATION should be obvious, and now INSTR represents the honest to God
2831    machine instruction set!
2832    
2833    The <arch>/instructions/<arch>Instr.sml files define certain utility
2834    functions for making porting easier -- essentially converting upper case
2835    to lower case. All machine instructions (of type instr) are in upper case,
2836    and the lower case form generates an MLRISC instruction. For example on
2837    the alpha we have:
2838    
2839      datatype instr =
2840         LDA of {r:cell, b:cell, d:operand}
2841       | ...
2842    
2843      val lda : {r:cell, b:cell, d:operand} -> instruction
2844        ...
2845    
2846    where lda is just (INSTR o LDA), etc.
2847    
2848    ----------------------------------------------------------------------
2849    Name: Matthias Blume
2850    Date: 2001/11/22 21:40:00 EST
2851    Tag: Release_110_37
2852    Description:
2853    
2854    Release 110.37.  This time for real.
2855    
2856    ----------------------------------------------------------------------
2857    Name: Matthias Blume
2858    Date: 2001/11/21 16:35:00 EST
2859    Tag: blume-20011121-foot-in-mouth
2860    Description:
2861    
2862    Removed the "Release_110_37" tag because of a serious bug.
2863    This will be re-tagged once the bug is fixed.
2864    
2865    ----------------------------------------------------------------------
2866    Name: Matthias Blume
2867    Date: 2001/11/21 16:14:00 EST
2868    Tag: blume-20011121-forgottenfile
2869    Description:
2870    
2871    Forgot to add a file.  (Just a .tex-file -- part of
2872    the CM manual source.)
2873    
2874    ----------------------------------------------------------------------
2875    Name: Matthias Blume
2876    Date: 2001/11/21 16:10:00 EST
2877    Tag: blume-20011121-invalid_110_37
2878    Description:
2879    
2880    Note: I removed the original tag "Release_110_37" from this commit
2881          because we found a serious bug in all non-x86 backends.
2882          - Matthias
2883    
2884    1. Modifications to the SML/NJ code generator and to the runtime system
2885       so that code object name strings are directly inserted into code
2886       objects at code generation time.  The only business the runtime system
2887       has with this is now to read the name strings on occasions.
2888       (The encoding of the name string has also changed somewhat.)
2889    
2890    2. CM now implements a simple "set calculus" for specifying export lists.
2891       In particular, it is now possible to refer to the export lists of
2892       other libraries/groups/sources and form unions as well as differences.
2893       See the latest CM manual for details.
2894    
2895    3. An separate notion of "proxy" libraries has again be eliminated from
2896       CM's model.  (Proxy libraries are now simply a special case of using
2897       the export list calculus.)
2898    
2899    4. Some of the existing libraries now take advantage of the new set
2900       calculus.
2901       (Notice that not all libraries have been converted because some
2902       of the existing .cm-files are supposed to be backward compatible
2903       with 110.0.x.)
2904    
2905    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
2906       "General.exnMessage"!  The former relies on a certain hook to be
2907       initialized, and that often does not happen in the stand-alone case.)
2908    
2909    ----------------------------------------------------------------------
2910    Name: Lal George
2911    Date: 2001/11/21  13:56:18 EST
2912    Tag: george-2001121-pseudo-ops
2913    Description:
2914    
2915      Implemented a complete redesign of MLRISC pseudo-ops. Now there
2916      ought to never be any question of incompatabilities with
2917      pseudo-op syntax expected by host assemblers.
2918    
2919      For now, only modules supporting GAS syntax are implemented
2920      but more should follow, such as MASM, and vendor assembler
2921      syntax, e.g. IBM as, Sun as, etc.
2922    
2923    ----------------------------------------------------------------------
2924    Name: Matthias Blume
2925    Date: 2001/11/14 11:52:00 EST
2926    Tag: blume-20011114-srcname
2927    Description:
2928    
2929    1. Routed the name of the current source file to mlriscgen where it
2930       should be directly emitted into the code object.  (This last part
2931       is yet to be done.)
2932    
2933    2. Some cleanup of the pgraph code to make it match the proposal that
2934       I put out the other day.  (The proposal notwithstanding, things are
2935       still in flux here.)
2936    
2937    ----------------------------------------------------------------------
2938    Name: Lal George
2939    Date: 2001/11/14 09:44:04 EST
2940    Tag:
2941    Description:
2942    
2943      Fix for a backpatching bug reported by Allen.
2944    
2945      Because the boundary between short and long span-dependent
2946      instructions is +/- 128, there are an astounding number of
2947      span-dependent instructions whose size is over estimated.
2948    
2949      Allen came up with the idea of letting the size of span
2950      dependent instructions be non-monotonic, for a maxIter
2951      number of times, after which the size must be monotonically
2952      increasing.
2953    
2954      This table shows the number of span-dependent instructions
2955      whose size was over-estimated as a function of maxIter, for the
2956      file Parse/parse/ml.grm.sml:
2957    
2958         maxIter            # of instructions:
2959            10                      687
2960            20                      438
2961            30                      198
2962            40                        0
2963    
2964      In compiling the compiler, there is no significant difference in
2965      compilation speed between maxIter=10 and maxIter=40. Actually,
2966      my measurements showed that maxIter=40 was a tad faster than
2967      maxIter=10! Also 96% of the  files in the compiler reach a fix
2968      point within 13 iterations, so fixing maxIter at 40, while high,
2969      is okay.
2970    
2971    ----------------------------------------------------------------------
2972    Name: Matthias Blume
2973    Date: 2001/10/31 15:25:00 EST
2974    Tag: blume-20011031-pgraph
2975    Description:
2976    
2977    CKIT:
2978    * Changed the "Function" constructor of type Ast.ctype to carry optional
2979      argument identifiers.
2980    * Changed the return type of TypeUtil.getFunction accordingly.
2981    * Type equality ignores the argument names.
2982    * TypeUtil.composite tries to preserve argument names but gives up quickly
2983      if there is a mismatch.
2984    
2985    installation script:
2986    * attempts to use "curl" if available (unless "wget" is available as well)
2987    
2988    CM:
2989    * has an experimental implementation of "portable graphs" which I will
2990      soon propose as an implementation-independent library format
2991    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
2992    
2993    NLFFI-LIB:
2994    * some cleanup (all cosmetic)
2995    
2996    NLFFIGEN:
2997    * temporarily disabled the mechanism that suppresses ML output for
2998      C definitions whose identifiers start with an underscore character
2999    * generate val bindings for enum constants
3000    * user can request that only one style (light or heavy) is being used;
3001      default is to use both (command-line arguments: -heavy and -light)
3002    * fixed bug in handling of function types involving incomplete pointers
3003    * generate ML entry points that take record arguments (i.e., using
3004      named arguments) for C functions that have a prototype with named
3005      arguments
3006      (see changes to CKIT)
3007    
3008    ----------------------------------------------------------------------
3009    Name: Allen Leung
3010    Date: 2001/10/27 20:34:00 EDT
3011    Tag: leunga-20011027-x86-fast-fp-call
3012    Description:
3013    
3014       Fixed the bug described in blume-20010920-slowfp.
3015    
3016       The fix involves
3017          1. generating FCOPYs in FSTP in ia32-svid
3018          2. marking a CALL with the appropriate annotation
3019    
3020    ----------------------------------------------------------------------
3021    Name: Matthias Blume
3022    Date: 2001/10/16 11:32:00 EDT
3023    Tag: blume-20011016-netbsd
3024    Description:
3025    
3026    Underscore patch from Chris Richards (fixing problem with compiling
3027    runtime system under recent NetBSD).
3028    
3029    ----------------------------------------------------------------------
3030    Name: Allen Leung
3031    Date: 2001/10/12 17:18:32 EDT 2001
3032    Tag: leung-20011012-x86-printflowgraph
3033    Description:
3034    
3035    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
3036    
3037    ----------------------------------------------------------------------
3038    Name: Lal George
3039    Date: 2001/10/11 23:51:34 EDT
3040    Tag: george-20011011-too-many-instrs
3041    Description:
3042    
3043    The representation of a program point never expected to see more
3044    than 65536 instructions in a basic block!
3045    
3046    ----------------------------------------------------------------------
3047    Name: Lal George
3048    Date: 2001/10/09 09:41:37 EDT
3049    Tag: george-20011008-mlrisc-labels
3050    Description:
3051    
3052    Changed the machine description files to support printing of
3053    local and global labels in assembly code, based on host assembler
3054    conventions.
3055    
3056    ----------------------------------------------------------------------
3057    Name: Matthias Blume
3058    Date: 2001/09/25 15:25:00 EDT
3059    Tag: blume-20010925-exninfo
3060    Description:
3061    
3062    I provided a non-hook implementation of exnName (at the toplevel) and
3063    made the "dummy" implementation of exnMessage (at the toplevel) more
3064    useful: if nothing gets "hooked in", then at least you are going to
3065    see the exception name and a message indicating why you don't see more.
3066    
3067    [For the time being, programs that need exnMessage and want to use
3068    ml-build should either use General.exnMessage (strongly recommended) or
3069    refer to structure General at some other point so that CM sees a
3070    static dependency.]
3071    
3072    [Similar remarks go for "print" and "use":  If you want to use their
3073    functionality in stand-alone programs generated by ml-build, then use
3074    TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
3075    
3076    ----------------------------------------------------------------------
3077    Name: Matthias Blume
3078    Date: 2001/09/20 17:28:00 EDT
3079    Tag: blume-20010920-slowfp
3080    Description:
3081    
3082    Allen says that x86-fast-fp is not safe yet, so I turned it off again...
3083    
3084    ----------------------------------------------------------------------
3085    Name: Matthias Blume
3086    Date: 2001/09/20 17:20:00 EDT
3087    Tag: blume-20010920-canonicalpaths
3088    Description:
3089    
3090    0. Updated the BOOT file (something that I forgot to do earlier).
3091    
3092    1. Small internal change to CM so that it avoids "/../" in filenames
3093       as much as possible (but only where it is safe).
3094    
3095    2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
3096       that arguments that contain delimiters are passed through correctly.
3097       This change also means that all "special" arguments of the form
3098       @SMLxxx... must come first.
3099    
3100    3. Changed install script to put relative anchor names for tool commands
3101       into pathconfig.
3102    
3103    ----------------------------------------------------------------------
3104    Name: Matthias Blume
3105    Date: 2001/09/18 15:35:00 EDT
3106    Tag: blume-20010918-readme11036
3107    Description:
3108    
3109    Added README files.
3110    
3111    ----------------------------------------------------------------------
3112    Name: Matthias Blume
3113    Date: 2001/09/18 11:45:00 EDT
3114    Tag: Release_110_36 (retag)
3115    Description:
3116    
3117    Fixed mistake in config/preloads. Retagged as 110.36.
3118    
3119    ----------------------------------------------------------------------
3120    Name: Matthias Blume
3121    Date: 2001/09/18 09:40:00 EDT
3122    Tag: Release_110_36_orig (tag changed)
3123    Description:
3124    
3125    New version (110.36).  New bootfiles.
3126    
3127    ----------------------------------------------------------------------
3128    Name: Matthias Blume
3129    Date: 2001/09/14 16:15:00 EDT
3130    Tag: blume-20010914-x86fastfp
3131    Description:
3132    
3133    John committed some changes that Allen made, in particular a (hopefully)
3134    correctly working version of the x86-fp module.
3135    
3136    I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
3137    flag to "true".  Everything seems to compile to a fixpoint ok, and
3138    "mandelbrot" speeds up by about 15%.
3139    
3140    ----------------------------------------------------------------------
3141    Name: Matthias Blume
3142    Date: 2001/09/13 11:20:00 EDT
3143    Tag: blume-20010913-minimal
3144    Description:
3145    
3146    1. Stefan Monnier's patch to fix a miscompilation problem that
3147       was brought to light by John Reppy's work on Moby.
3148    
3149    2. Implemented a minimal "structure Compiler" that contains just
3150       "version" and "architecture".  The minimal version will be
3151       available when the full version is not.  This is for backward-
3152       compatibility with code that wants to test Compiler.version.
3153    
3154    ----------------------------------------------------------------------
3155    Name: Matthias Blume
3156    Date: 2001/08/28 14:03:00 EDT
3157    Tag: blume-20010828-ml-lex
3158    Description:
3159    
3160    Fix for bug 1581, received from Neophytos Michael.
3161    
3162    ----------------------------------------------------------------------
3163    Name: Matthias Blume
3164    Date: 2001/08/27 11:20:00 EDT
3165    Tag: blume-20010827-readme11035
3166    Description:
3167    
3168    Fleshed out the README file for 110.35.
3169    
3170    ----------------------------------------------------------------------
3171    Name: Matthias Blume
3172    Date: 2001/08/24 17:10:00 EDT
3173    Tag: Release_110_35
3174    Description:
3175    
3176    New version number (110.35).  New bootfiles.
3177    
3178    ----------------------------------------------------------------------
3179    Name: Lal George
3180    Date: 2001/08/24 13:47:18 EDT 2001
3181    Tag: george-20010824-MLRISC-graphs
3182    Description:
3183    
3184     removed clusters from MLRISC completely and replaced with graphs.
3185    
3186    ----------------------------------------------------------------------
3187    Name: Matthias Blume
3188    Date: 2001/08/23 17:50:00 EDT
3189    Tag: blume-20010823-toplevel
3190    Description:
3191    
3192    - some reorganization of the code that implements various kinds of
3193      environments in the compiler (static, dynamic, symbolic, combined)
3194    - re-implemented the EnvRef module so that evalStream works properly
3195      (if the stream contains references to "use", "CM.make", etc.)
3196    - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
3197    
3198    ----------------------------------------------------------------------
3199    Name: Matthias Blume
3200    Date: 2001/08/20 15:50 EDT
3201    Tag: blume20010820-slipup
3202    Description:
3203    
3204    I forgot to commit a few files.  Here they are...
3205    
3206    ----------------------------------------------------------------------
3207    Name: Matthias Blume
3208    Date: 2001/08/20 15:35:00 EDT
3209    Tag: blume-20010820-debugprof
3210    Description:
3211    
3212    !!!! NEW BOOTFILES !!!!
3213    
3214    This is another round of reorganizing the compiler sources.  This
3215    time the main goal was to factor out all the "instrumentation"
3216    passes (for profiling and backtracing) into their own library.
3217    The difficulty was to do it in such a way that it does not depend
3218    on elaborate.cm but only on elabdata.cm.
3219    
3220    Therefore there have been further changes to both elaborate.cm and
3221    elabdata.cm -- more "generic" things have been moved from the former
3222    to the latter.  As a result, I was forced to split the assignment
3223    of numbers indicating "primtyc"s into two portions: SML-generic and
3224    SML/NJ-specific.  Since it would have been awkward to maintain,
3225    I bit the bullet and actually _changed_ the mapping between these
3226    numbers and primtycs.  The bottom line of this is that you need
3227    a new set of bin- and bootfiles.
3228    
3229    I have built new bootfiles for all architectures, so doing a fresh
3230    checkout and config/install.sh should be all you need.
3231    
3232    The newly created library's name is
3233    
3234        $smlnj/viscomp/debugprof.cm
3235    
3236    and its sources live under
3237    
3238        src/compiler/DebugProf
3239    
3240    ----------------------------------------------------------------------
3241    Name: Matthias Blume
3242    Date: 2001/08/15 17:15:00 EDT
3243    Tag: blume-20010815-compreorg
3244    Description:
3245    
3246    This is a first cut at reorganizing the CM libraries that make up the
3247    core of the compiler.  The idea is to separate out pieces that could
3248    be used independently by tools, e.g., the parser, the typechecker, etc.
3249    
3250    The current status is a step in this direction, but it is not quite
3251    satisfactory yet.  Expect more changes in the future.
3252    
3253    Here is the current (new) organization...
3254    
3255        What used to be $smlnj/viscomp/core.cm is now divided into
3256        six CM libraries:
3257    
3258             $smlnj/viscomp/basics.cm
3259                           /parser.cm
3260                           /elabdata.cm
3261                           /elaborate.cm
3262                           /execute.cm
3263                           /core.cm
3264    
3265        The CM files for these libraries live under src/system/smlnj/viscomp.
3266        All these libraries are proxy libraries that contain precisely
3267        one CM library component.  Here are the locations of the components
3268        (all within the src/compiler tree):
3269    
3270             Basics/basics.cm
3271             Parse/parser.cm
3272             ElabData/elabdata.cm
3273             Elaborator/elaborate.cm
3274             Execution/execute.cm
3275             core.cm
3276    
3277         [This organization is the same that has been used already
3278         for a while for the architecture-specific parts of the visible
3279         compiler and for the old version of core.cm.]
3280    
3281         As you will notice, many source files have been moved from their
3282         respective original locations to a new home in one of the above
3283         subtrees.
3284    
3285         The division of labor between the new libraries is the following:
3286    
3287             basics.cm:
3288                - Simple, basic definitions that pertain to many (or all) of
3289                  the other libraries.
3290             parser.cm:
3291                - The SML parser, producing output of type Ast.dec.
3292                - The type family for Ast is also defined and exported here.
3293             elabdata.cm:
3294                - The datatypes that describe input and output of the elaborator.
3295                  This includes types, absyn, and static environments.
3296             elaborator.cm:
3297                - The SML/NJ type checker and elaborator.
3298                  This maps an Ast.dec (with a given static environment) to
3299                  an Absyn.dec (with a new static environment).
3300                - This libraries implements certain modules that used to be
3301                  structures as functors (to remove dependencies on FLINT).
3302             execute.cm:
3303                - Everything having to do with executing binary code objects.
3304                - Dynamic environments.
3305             core.cm:
3306                - SML/NJ-specific instantiations of the elaborator and MLRISC.
3307                - Top-level modules.
3308                - FLINT (this should eventually become its own library)
3309    
3310    Notes:
3311    
3312    I am not 100% happy with the way I separated the elaborator (and its
3313    data structures) from FLINT.  Two instances of the same problem:
3314    
3315        1. Data structures contain certain fields that carry FLINT-specific
3316           information.  I hacked around this using exn and the property list
3317           module from smlnj-lib.  But the fact that there are middle-end
3318           specific fields around at all is a bit annoying.
3319    
3320        2. The elaborator calculates certain FLINT-related information.  I tried
3321           to make this as abstract as I could using functorization, but, again,
3322           the fact that the elaborator has to perform calculations on behalf
3323           of the middle-end at all is not nice.
3324    
3325        3. Having to used exn and property lists is unfortunate because it
3326           weakens type checking.  The other alternative (parameterizing
3327           nearly *everything*) is not appealing, though.
3328    
3329    I removed the "rebinding =" warning hack because due to the new organization
3330    it was awkward to maintain it.  As a result, the compiler now issues some of
3331    these warnings when compiling init.cmi during bootstrap compilation. On
3332    the plus side, you also get a warning when you do, for example:
3333       val op = = Int32.+
3334    which was not the case up to now.
3335    
3336    I placed "assign" and "deref" into the _Core structure so that the
3337    code that deals with the "lazy" keyword can find them there.  This
3338    removes the need for having access to the primitive environment
3339    during elaboration.
3340    
3341    ----------------------------------------------------------------------
3342    Name: Matthias Blume
3343    Date: 2001/08/13
3344    Tag: blume-20010813-closures
3345    Description:
3346    
3347    This fix was sent to us by Zhong Shao.  It is supposed to improve the
3348    performance of certain loops by avoiding needless closure allocation.
3349    
3350    ----------------------------------------------------------------------
3351    Name: Lal George
3352    Date: 2001/07/31 10:03:23 EDT 2001
3353    Tag: george-20010731-x86-fmalloc
3354    Description: Fixed bug in x86 calls
3355    
3356        There was a bug where call instructions would mysteriously
3357        vanish. The call instruction had to be one that returned
3358        a floating point value.
3359    
3360    ----------------------------------------------------------------------
3361    Name: Lal George
3362    Date: 2001/07/19 16:36:29 EDT 2001
3363    Tag: george-20010719-simple-cells
3364    Description:
3365    
3366    I have dramatically simplified the interface for CELLS in MLRISC.
3367    
3368    In summary, the cells interface is broken up into three parts:
3369    
3370      1. CellsBasis : CELLS_BASIS
3371    
3372            CellsBasis is a top level structure and common for all
3373            architectures.  it contains the definitions of basic datatypes
3374            and utility  functions over these types.
3375    
3376      2. functor Cells() : CELLS
3377    
3378            Cells generates an interface for CELLS that incorporates the
3379            specific resources on the target architecture, such as the
3380            presence of special register classes, their number and size,
3381            and various useful substructures.
3382    
3383      3. <ARCH>CELLS
3384    
3385            e.g. SparcCells: SPARCCELLS
3386    
3387            <ARCH>CELLS usually contains additional bindings for special
3388            registers  on the architecture, such as:
3389    
3390                    val r0 : cell           (* register zero *)
3391                    val y : cell            (* Y register *)
3392                    val psr : cell          (* processor status register *)
3393                    ...
3394    
3395            The structure returned by applying the Cells functor is opened
3396            in this interface.
3397    
3398    The main implication of all this is that the datatypes for cells is
3399    split between CellsBasis and CELLS -- a fairly simple change for user
3400    code.
3401    
3402    In the old scheme the CELLS interface had a definitional binding of
3403    the form:
3404    
3405            signature CELLS = sig
3406    
3407               structure CellsBasis = CellsBasis
3408    
3409               ...
3410    
3411            end
3412    
3413    With all the sharing constraints that goes on in MLRISC, this old
3414    design  quickly leads to errors such as:
3415    
3416            "structure definition spec inside of sharing ... "
3417    
3418    
3419    and appears to require an unacceptable amount of sharing and where
3420    constraint hackery.
3421    
3422    I think this error message (the interaction of definitional specs and
3423    sharing) requires more explanation on our web page.
3424    
3425    ----------------------------------------------------------------------
3426    Name: Matthias Blume
3427    Date: 2001/07/19 15:00:00 EDT
3428    Tag: blume-20010719-libreorg
3429    Description:
3430    
3431    This update puts together a fairly extensive but straightforward change
3432    to the way the libraries that implement the interactive system are
3433    organized:
3434    
3435       The biggest change is the elimination of structure Compiler.  As a
3436       replacement for this structure, there is now a CM library
3437       (known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
3438       that exports all the substructures of the original structure Compiler
3439       directly.  So instead of saying Compiler.Foo.bar one now simply
3440       says Foo.bar.  (The CM libraries actually export a collection of
3441       structures that is richer than the collection of substructures of
3442       structure Compiler.)
3443    
3444       To make the transition smooth, there is a separate library called
3445       $smlnj/compiler/compiler.cm which puts together and exports the
3446       original structure Compiler (or at least something very close to it).
3447    
3448       There are five members of the original structure Compiler
3449       that are not exported directly but which instead became members
3450       of a new structure Backend (described by signature BACKEND).  These are:
3451       structure Profile (: PROFILE), structure Compile (: COMPILE), structure
3452       Interact (: INTERACT), structure Machine (: MACHINE), and val
3453       architecture (: string).
3454    
3455       Structure Compiler.Version has become structure CompilerVersion.
3456    
3457       Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
3458       by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
3459       or x86, respectively.
3460       Each of these exports the same frontend structures that
3461       $smlnj/compiler.cm exports.  But they do not have a structure Backend
3462       and instead export some structure <Arch>Backend where <Arch> is Alpha32,
3463       Hppa, PPC, Sparc, or X86, respectively.
3464    
3465       Library $smlnj/compiler/all.cm exports the union of the exports of
3466       $smlnj/compiler/<arch>.cm
3467    
3468       There are no structures <Arch>Compiler anymore, use
3469       $smlnj/compiler/<arch>.cm instead.
3470    
3471       Library host-compiler-0.cm is gone.  Instead, the internal library
3472       that instantiates CM is now called cm0.cm.  Selection of the host
3473       compiler (backend) is no longer done here but. (Responsibility for it
3474       now lies with $smlnj/compiler/current.cm.  This seems to be more
3475       logical.)
3476    
3477       Many individual files have been moved or renamed.  Some files have
3478       been split into multiple files, and some "dead" files have been deleted.
3479    
3480    Aside from these changes to library organization, there are also changes
3481    to the way the code itself is organized:
3482    
3483       Structure Binfile has been re-implemented in such a way that it no
3484       longer needs any knowledge of the compiler.  It exclusively deals
3485       with the details of binfile layout.  It no longer invokes the
3486       compiler (for the purpose of creating new prospective binfile
3487       content), and it no longer has any knowledge of how to interpret
3488       pickles.
3489    
3490       Structure Compile (: COMPILE) has been stripped down to the bare
3491       essentials of compilation.  It no longer deals with linking/execution.
3492       The interface has been cleaned up considerably.
3493    
3494       Utility routines for dealing with linking and execution have been
3495       moved into their own substructures.
3496    
3497       (The ultimate goal of these changes is to provide a light-weight
3498       binfile loader/linker (at least for, e.g., stable libraries) that
3499       does not require CM or the compiler to be present.)
3500    
3501    CM documentation has been updated to reflect the changes to library
3502    organization.
3503    
3504    ----------------------------------------------------------------------
3505    Name: Matthias Blume
3506    Date: 2001/07/10 17:30:00 EDT
3507    Tag: Release_110_34
3508    Description:
3509    
3510    Minor tweak to 110.34 (re-tagged):
3511    
3512      - README.html file added to CVS repository
3513      - runtime compiles properly under FreeBSD 3.X and 4.X
3514    
3515    ----------------------------------------------------------------------
3516    Name: Matthias Blume
3517    Date: 2001/07/10 17:30:00 EDT
3518    Tag: Release_110_34
3519    Description:
3520    
3521    New version number (110.34). New bootfiles.
3522    
3523    ----------------------------------------------------------------------
3524    Name: Matthias Blume
3525    Date: 2001/07/09 16:00:00 EDT
3526    Tag: blume-20010709-more-varargs
3527    Description:
3528    
3529    I changed the handling of varargs in ml-nlffigen again:
3530    The ellipsis ... will now simply be ignored (with an accompanying warning).
3531    
3532    The immediate effect is that you can actually call a varargs function
3533    from ML -- but you can't actually supply any arguments beyond the ones
3534    specified explicitly.  (For example, you can call printf with its format
3535    string, but you cannot pass additional arguments.)
3536    
3537    This behavior is only marginally more useful than the one before, but
3538    it has the advantage that a function or, more importantly, a function
3539    type never gets dropped on the floor, thus avoiding follow-up problems with
3540    other types that refer to the offending one.
3541    
3542    ----------------------------------------------------------------------
3543  Name: Matthias Blume  Name: Matthias Blume
3544  Date: 2001/07/09 11:25:00 EDT  Date: 2001/07/09 11:25:00 EDT
3545  Tag: blume-20010709-varargs  Tag: blume-20010709-varargs
# Line 2551  Line 6078 
6078       elaborator). There were a lot of changes during my "linkpath" trials       elaborator). There were a lot of changes during my "linkpath" trials
6079       that could have been reverted to their original state but weren't.       that could have been reverted to their original state but weren't.
6080       Please, don't be too harsh on me for messing with this code a bit more       Please, don't be too harsh on me for messing with this code a bit more
6081       than what was strictly necessary...  (I _did_ resist the tempation       than what was strictly necessary...  (I _did_ resist the temptation
6082       of doing any "global reformatting" to avoid an untimely death at       of doing any "global reformatting" to avoid an untimely death at
6083       Dave's hands. :)       Dave's hands. :)
6084    

Legend:
Removed from v.873  
changed lines
  Added in v.1496

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