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/NOTES/HISTORY
ViewVC logotype

Diff of /sml/trunk/NOTES/HISTORY

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 758, Fri Dec 22 04:12:36 2000 UTC revision 1638, Wed Sep 29 19:07:58 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: 2004/09/29 14:00:00 CDT
18    Tag: blume-20040929-autoversion
19    Description:
20    
21    Implemented some CM magic to have
22        file src/compiler/TopLevel/main/version.sml
23    generated automagically.
24    The version is taken from two files: config/version and config/release.
25    The first is expected to contain a two-part version number such as 110.49.
26    The second should contain a single number, but it may be missing.
27    
28    If the environment variable VERSIONTOOL_BUMP_RELEASE is defined at the
29    time the version tool is loaded (which is the first time you say CMB.make),
30    then the tool will increment the value stored in config/release every
31    time CMB.make is invoked.
32    
33    The binfile format is now insensitive to anything beyond the first
34    two components of a version number, so bumping the release does not render
35    binfiles incompatible.  Auto-bumping can be used to keep track of versions
36    during development without invalidating existing binfiles.
37    
38    In any case, every CMB.make updates the date information in version.sml.
39    (This is the date that is printed in the banner.)
40    
41    ----------------------------------------------------------------------
42    Name: Matthias Blume (blume (at) tti - c (dot) org)
43    Date: 2004/09/28 10:53:00 CDT
44    Tag: blume-20040928-controls
45    Description:
46    
47    Some cleanup of the controls code.
48    
49    ----------------------------------------------------------------------
50    Name: Matthias Blume (blume (at) tti - c (dot) org)
51    Date: 2004/09/27 22:08:00 CDT
52    Tag: blume-20040927-controls
53    Description:
54    
55    Added two pieces of functionality to the Controls interface:
56    
57    1.   val save'restore: 'a control -> unit -> unit
58    
59       grabs the current value of the control in stage 1 and restores it
60       in stage 2.
61    
62    2.   val set' : 'a control * 'a -> unit -> unit
63    
64       stores the given value into the control in stage 2 (i.e., delayed)
65       but does all error checking in stage 1.
66       (This is for string controls that need to do parse their argument --
67        something that might fail.  In some cases, notably in CM, one
68        already knows the intended argument but wants to delay the actual
69        assignment until a time when error recovery would be more difficult.)
70    
71    Changed the handling of controls in tool arguments to classes "sml" and
72    "lazysml":
73       - use Controls.save'restore as a more robust way of restoring the
74         old value (in particular: without having to re-parse the string)
75       - use controls to handle the "overload" keyword in the init group
76         (I believe this change actually fixes a long-standing obscure bug.)
77    
78    ----------------------------------------------------------------------
79    Name: Matthias Blume (blume (at) tti - c (dot) org)
80    Date: 2004/09/27 17:00:00 CDT
81    Tag: blume-20040927-lazysml
82    Description:
83    
84    Added a new tool class called "lazysml" to CM's tool chest.  The only
85    difference to "sml" is that compilation is done with Control.lazysml
86    set to true.  A source of class "lazysml" is automatically recognized
87    by a file name suffix of ".lml".
88    
89    In addition to the above feature, the original class "sml" now also
90    supports a tool argument "lazy" which has the same effect.  As a
91    result, the following three lines are equivalent:
92    
93        foo.sml : lazysml
94        foo.sml : sml (lazy)
95        foo.sml (lazy)
96    
97    The setting goes into effect both during parsing and during
98    compilation.  The original setting is restored right after parsing and
99    after compilation, respectively.
100    
101    In addition to all the above, there is also a general mechanism to set
102    ANY of the "controls" that are available at the command line via
103    "-C..." on a per-sml-file basis.  The same rules that apply for "lazy"
104    apply as well. (In fact, "lazy" is implemented as a special case of
105    the general mechanism.)
106    
107    The .cm file syntax uses a new keyword tool argument called "with".
108    There are several ways of indicating the desired settings:
109    
110        foo.sml (with:parser.quotations=true)
111        foo.sml (with:(name:parser.quotations value:true))
112        foo.sml (with:(name:name1 value:value1 name:name2 value:value2 ...))
113        foo.sml (with:(name1=value1 name2=value2 ...))
114        foo.sml (with:(name1=value1 name:name2 value:value2 name3=value3 ...))
115    
116        etc.
117    
118    Another possible abbreviation is to leave out the =v or value:v part
119    if the name refers to a boolean control (in which case the value is
120    taken to be true).  Thus, one could get lazy sml also by saying:
121    
122        foo.sml (with:parser.lazy-keyword=true)
123        foo.sml (with:parser.lazy-keyword)
124        foo.sml (with:(name:parser.lazy-keyword value:true))
125        foo.sml (with:(name:parser.lazy-keyword))
126    
127    ----------------------------------------------------------------------
128    Name: Matthias Blume (blume (at) tti - c (dot) org)
129    Date: 2004/09/24 16:22:00 CDT
130    Tag: blume-20040924-ppc-long-branch
131    Description:
132    
133    Turned message about "emiting long form of branch" off by default.
134    Added a control flag to turn it back on when desired.
135    
136    ----------------------------------------------------------------------
137    Name: Matthias Blume (blume (at) tti - c (dot) org)
138    Date: 2004/09/24 16:05:00 CDT
139    Tag: blume-20040924-rounding
140    Description:
141    
142    Applied patch for setting rounding modes under Mac OS X.  Thanks to
143    Melissa O'Neill for providing the code!
144    
145    ----------------------------------------------------------------------
146    Name: Matthias Blume (blume (at) tti - c (dot) org)
147    Date: 2004/09/23 17:30:00 CDT
148    Tag: blume-20040923-envvars
149    Description:
150    
151    1. Changed definition of type ControlRegistry.registry_tree to
152       include control_info (i.e., the name of the controlling environment
153       variable).
154    
155    2. Added command-line flags -e and -E to print the names of environment
156       variables that can be used to control internal settings.
157       (This uses the new API mentioned in 1.)
158    
159    ----------------------------------------------------------------------
160    Name: Matthias Blume (blume (at) tti - c (dot) org)
161    Date: 2004/09/13 16:50:00 CDT
162    Tag: Release_110_49
163    Description:
164    
165    New working version (110.49).  NEW BOOTFILES!
166    
167    ----------------------------------------------------------------------
168    Name: Matthias Blume
169    Date: 2004/09/13 16:20:00 CDT
170    Tag: blume-20040913-config-mlrisc
171    Description:
172    
173    Put target "mlrisc" back into the default list.
174    (There is no harm in having it, and some users have expressed their
175    wish to have "mlrisc" included by default.)
176    
177    ----------------------------------------------------------------------
178    Name: John Reppy
179    Date: 2004/09/13
180    Tag: jhr-20040913-signals
181    Description:
182    
183    Fixed the signal masking code to properly nest mask/unmask operations
184    on a per-signal basis.
185    
186    ----------------------------------------------------------------------
187    Name: Matthias Blume (blume (at) tti - c (dot) org)
188    Date: 2004/09/08 13:20:00 CDT
189    Tag: blume-20040908-heap-magic
190    Description:
191    
192    Bumped the heap macig number to 0x09082004 to account for the changed
193    layout of the ML frame under MacOS X.
194    
195    ----------------------------------------------------------------------
196    Name: Allen Leung (leunga (at) reservoir (dot) com)
197    Date: 2004/09/03 11:26:00 EST
198    Tag: leunga-20040903-cygwin-install
199    Description:
200    
201       Added a patch to _arch-n-opsys to enable the Cygwin runtime.
202    The Cygwin runtime is turned on by setting the environment
203    variable SMLNJ_CYGWIN_RUNTIME to 1.
204    
205    ----------------------------------------------------------------------
206    Name: Matthias Blume (blume (at) tti - c (dot) org)
207    Date: 2004/08/31 17:15:00 CDT
208    Tag: blume-20040831-core
209    Description:
210    
211    Added some exports to src/compiler/core.cm upon request by J. Joemann.
212    
213    ----------------------------------------------------------------------
214    Name: Matthias Blume (blume (at) tti - c (dot) org)
215    Date: 2004/08/30 17:55:00 CDT
216    Tag: blume-20040830-installer
217    Description:
218    
219    Upon request by Johannes Joemann:
220    
221    - improved ML code of installer to fall back to coping when renaming
222      fails (i.e., when source and target are on different file systems);
223      the code compiles but has yet to be tested in anger
224    - removed mlrisc from list of default targets (config/targets)
225    
226    ----------------------------------------------------------------------
227    Name: Matthias Blume (blume (at) tti - c (dot) org)
228    Date: 2004/08/27 17:20:00 CDT
229    Tag: blume-20040827-ptreql
230    Description:
231    
232    Added ptreql primop to structure InlineT (upon request from Larry
233    Paulson).
234    
235    ----------------------------------------------------------------------
236    Name: Allen Leung (leunga (at) reservoir (dot) com)
237    Date: 2004/08/15 21:21:00 EST
238    Tag: leunga-110_48-udgraph
239    Description:
240    
241    Another bug fix from Carl Hauser:
242    
243    diff /net/niflab/smlnj48/src/MLRISC/graphs/udgraph.sml udgraph.sml
244    > 48c48
245    > <              | rmv((e as (k,_))::es,L) = rmv(es,if k = i then es else
246    > e::L)
247    > ---
248    > >              | rmv((e as (k,_))::es,L) = rmv(es,if k = i then L else e::L)
249    > Without this, any deletion of an edge in an undirected graph does severe
250    > violence to the graph.
251    
252    ----------------------------------------------------------------------
253    Name: Allen Leung (leunga (at) reservoir (dot) com)
254    Date: 2004/08/10 23:35:00 EST
255    Tag: leunga-110_48-ppc
256    Description:
257    
258       The IBM/MacOS syntax switch on PPC was incorrectly swapped.
259    
260    ----------------------------------------------------------------------
261    Name: Matthias Blume (blume (at) tti - c (dot) org)
262    Date: 2004/08/10 12:00:00 CDT
263    Tag: Release_110_48
264    Description:
265    
266    New working version (110.48).  NEW BOOTFILES!
267    
268    ----------------------------------------------------------------------
269    Name: Allen Leung (leunga (at) reservoir (dot) com)
270    Date: 2004/08/09 12:21:00 EST
271    Tag: leunga-110_47-dijsktra
272    Description:
273    
274    Bug fix from Carl Hauser:
275    
276    single_source_shortest_paths in dijkstra.sml was observed to get wrong
277    answers (by comparing to single_source_shortest_paths in bellman-ford.sml).
278    
279    The problem is that following the expression A.update(dist,s,Num.zero)
280    it is necessary to update the priority queue using Q.decreaseWeight(Q,s).
281    
282    ----------------------------------------------------------------------
283    Name: Matthias Blume (blume (at) tti - c (dot) org)
284    Date: 2004/08/06 18:10:00 CDT
285    Tag: blume-20040806-cmdline
286    Description:
287    
288    Fiddled with handling of command-line options:
289    
290      * sml now quits after processing the command line
291        if -H, -S, -h<n>, or -s<n> appears as the last
292        command-line argument
293      * a new option -q terminates the session when encountered on
294        the command line; subsequent arguments will be ignored
295      * bug fixes: short (erroneous) arguments are no longer ignored
296        completely
297    
298    ----------------------------------------------------------------------
299    Name: Allen Leung (leunga (at) reservoir (dot) com)
300    Date: 2004/08/04 18:17:00 EST
301    Tag: leunga-110_47-ppc-ibm-asm
302    Description:
303    
304      - Added minimal IBM assembly syntax support for PowerPC.
305    
306      - Cygwin: manually changed the file cygwin.def.  Some exported symbols have
307        been altered in the runtime.  We need an automatic way to keep the file
308        in sync.
309    
310    ----------------------------------------------------------------------
311    Name: Matthias Blume (blume (at) tti - c (dot) org)
312    Date: 2004/08/04 14:00:00 CDT
313    Tag: Release_110_47
314    Description:
315    
316    New working version (110.47).  NEW BOOTFILES!
317    
318    ----------------------------------------------------------------------
319    Name: Matthias Blume (blume (at) tti - c (dot) org)
320    Date: 2004/08/03 14:25:00 CDT
321    Tag: blume-20040803-callingconv
322    Description:
323    
324    Added low-level support for choosing C calling conventions by
325    twiddling the type of rawccall.  (See
326    src/compiler/Semant/types/cproto.sml for details.)
327    
328    ----------------------------------------------------------------------
329    Name: Matthias Blume (blume (at) tti - c (dot) org)
330    Date: 2004/08/02 15:55:00 CDT
331    Tag: blume-20040802-backout
332    Description:
333    
334    Backed out of change to win32-filesys.c.  The earlier patch to
335    get_file_time caused CM to produce files with the wrong time stamp.
336    
337    ----------------------------------------------------------------------
338    Name: Matthias Blume (blume (at) tti - c (dot) org)
339    Date: 2004/08/02 14:45:00 CDT
340    Tag: blume-20040802-nlffi-win32
341    Description:
342    
343    Added NLFFI support for Win32, adapted from a patch provided by David
344    Hansel.  This is currently completely untested.  Also, the issue
345    concerning stdcall vs. ccall is still unresolved.
346    
347    ----------------------------------------------------------------------
348    Name: Matthias Blume (blume (at) tti - c (dot) org)
349    Date: 2004/07/30 17:55:00 CDT
350    Tag: blume-20040730-various
351    Description:
352    
353    Gearing up towards 110.47...
354    
355    - various minor bugfixes to ml-nlffigen
356    - a beginning of a manual for nlffi
357    
358    - eliminated 'export name=value' in config/install.sh as this does
359      not work with certain versions of /bin/sh
360      (Thanks to David King at Motorola for catching this.)
361    
362    - several bugfixes provided or suggested by David Hansel at Reactive Systems:
363      - added a test for tm==NULL to gmtime.c and localtime.c
364      - applied patch for incorrect GetFileTime under win32
365      - toSeconds -> toMilliseconds in Win32/win32-process.sml
366    
367    ----------------------------------------------------------------------
368    Name: Matthias Blume (blume (at) tti - c (dot) org)
369    Date: 2004/07/21 18:20:00 CDT
370    Tag: blume-20040721-nlffigen
371    Description:
372    
373    - Fixed minor issue in ml-nlffigen: Now generate structure T_foo
374      for a typedef to an incomplete type, but leave out the "typ" member.
375      (This is just for consistency.)
376    - Started to produce what is supposed to become better (i.e., comprehensive)
377      documentation of what ml-nlffigen does and produces.
378    
379    ----------------------------------------------------------------------
380    Name: Matthias Blume (blume (at) tti - c (dot) org)
381    Date: 2004/07/14 16:25:00 CDT
382    Tag: blume-20040714-union
383    Description:
384    
385    Added C_UNION to c-calls/c-types.sml and updated the machinery
386    (ml-nlffigen, cproto.sml) that conveys C function interface
387    information to the code generator.
388    
389    However, the actual architecture-specific implementation of function
390    arguments and results that are C unions is still not implemented.
391    
392    ----------------------------------------------------------------------
393    Name: Allen Leung (leunga (at) reservoir (dot) com)
394    Date: 2004/07/14 14:38:00 EST
395    Tag: leunga-110_46_1-ppc-lwzu
396    Description:
397    
398        Added these instructions to the PowerPC architecture:
399        LBZU(X), LHZU(X), LWZU(X),
400        STWU(X), STFDU, STFSU
401    
402        etc...
403    
404        Note: I haven't added their instruction encoding into the description.
405    
406    ----------------------------------------------------------------------
407    Name: Allen Leung (leunga (at) reservoir (dot) com)
408    Date: 2004/07/13 15:04:00 EST
409    Tag: leunga-110_46_1-ppc-lwarx
410    Description:
411    
412       Added the two instructions LWARX and STWCX to the PowerPC
413    instruction set.
414    
415       A (untested) rewrite of loop-structure.sml.  The old version
416    is completely broken.
417    
418    ----------------------------------------------------------------------
419    Name: Matthias Blume (blume (at) tti - c (dot) org)
420    Date: 2004/07/13 13:50:00 CDT
421    Tag: blume-20040713-nlffi
422    Description:
423    
424    - use paramAlloc to report c-calls with too many arguments
425      (for PPC version where parameter area is pre-allocated)
426    - added ccall_maxargspace to machspec (to implement the above)
427    - made "make" commend in CM's "make" tool configurable
428    - added option (default: on) for passing the name of the SML/NJ's "bin"
429      directory to "make"; the call looks like this:
430    
431         make <options> SMLNJ_BINDIR=<dir> <target>
432    
433      This can be used by the Makefile to, e.g., pick the "right" version
434      of ml-nlffigen.
435    - minor code tweaks
436    
437    ----------------------------------------------------------------------
438    Name: Matthias Blume (blume (at) tti - c (dot) org)
439    Date: 2004/07/12 22:50:00 CDT
440    Tag: blume-110_46_1-macosx-nlffi
441    Description:
442    
443    NLFFI under Mac OS X now working (sort of).  This is largely untested,
444    though.
445    
446    Note:  1. You have to make a new, clean build of the runtime system.
447           2. There are new BOOTFILES, you have to use them!
448              (Doing the bootstrap process yourself would be *very* painful!
449               If you absolutely have to do it, build the system under
450               a different architecture and then cross-compile.)
451    
452    Version bumped to 110.46.1 to account for runtime data format changes.
453    
454    ----------------------------------------------------------------------
455    Name: Matthias Blume (blume (at) tti - c (dot) org)
456    Date: 2004/06/18 14:30:00 CDT
457    Tag: blume-20040618-unix
458    Description:
459    
460    Changed the implementation of structure Unix so that the same stream
461    is returned every time one of the {text,bin}{In,Out}streamOf functions
462    is invoked on the same proc.  This is not what the spec currently
463    says -- although IMO it arguably should.  (See discussion below.)
464    
465    ----------------------------------------------------------------------
466    Name: Matthias Blume (blume (at) tti - c (dot) org)
467    Date: 2004/06/17 18:15:00 CDT
468    Tag: Release_110_46
469    Description:
470    
471    New working version (110.46).  NEW BOOTFILES!
472    
473    ----------------------------------------------------------------------
474    Name: Matthias Blume (blume (at) tti - c (dot) org)
475    Date: 2004/06/17 17:20:00 CDT
476    Tag: blume-20040617-timer-unix
477    Description:
478    
479    Changed the interface of structures Timer and Unix to match the most
480    recent Basis spec.
481    
482    In the case of Unix there still seems to be an open/weird issue:
483    
484       The {text,bin}{In,Out}streamOf functions are supposed to create
485       fresh streams whenever they are called -- as opposed to have them
486       return the same stream every time.  This design is supposed to
487       prevent space leaks caused by proc values hanging on to streams.
488    
489       The reap function, on the other hand, is supposed to close the
490       streams.  This cannot be done without having a handle on the
491       stream in proc after all...
492    
493    I took the liberty to implement the following stopgap solution:
494    
495      The proc value hangs on to the most recently created stream(s).
496      Reap closes those.  If either or both of the two streams hadn't
497      been created at all yet, then reap will close the corresponding
498      file descriptors directly.
499    
500    PS: I don't understand the original space leak argument anymore.  If
501    a proc hangs on to the imperative stream, then I/O operations on those
502    will advance the state of the cached stream and avoid the space leak.
503    
504    ----------------------------------------------------------------------
505    Name: Matthias Blume (blume (at) tti - c (dot) org)
506    Date: 2004/05/28 16:45:00 CDT
507    Tag: blume-20040528-basis
508    Description:
509    
510    Added signature PACK_REAL and exported functor PrimIO.
511    
512    ----------------------------------------------------------------------
513    Name: Matthias Blume (blume (at) tti - c (dot) org)
514    Date: 2004/05/25 16:00:00 CDT
515    Tag: blume-20040525-group-owner
516    Description:
517    
518    CM now ignores (but still accepts) the "owner" information in group
519    descriptions.  The owner of a group is its next enclosing
520    library. Each group must have a unique owner.  (There is a virtual
521    "toplevel" library that owns groups which are not nested within a real
522    library.)  Previously, each group had to explicitly declare its owner,
523    and CM would check that such a declaration is correct.  The new scheme
524    is to have CM check that for each group there is precisely one owning
525    library.
526    
527    The advantage of the new scheme is that the programmer no longer needs
528    to maintain the somewhat annoying owner information.  The downside is
529    that CM cannot enforce the ownership rule across multiple runs of
530    CM.make.  Fortunately, enclosing the same group in two different
531    libraries A and B which are not part of the same program does not
532    cause real problems.
533    
534    ----------------------------------------------------------------------
535    Name: Matthias Blume (blume (at) tti - c (dot) org)
536    Date: 2004/05/20 16:00:00 CDT
537    Tag: blume-20040520-win32
538    Description:
539    
540    Made the win32 version work again.  (Strangely, a misplaced comma had
541    slipped into win32-process.c which prevented the runtime from being
542    compiled correctly.)
543    
544    Also, included a minor addition to ml-build.bat analogous to what was
545    done in blume-20040519-ml-build.
546    
547    ----------------------------------------------------------------------
548    Name: Matthias Blume (blume (at) tti - c (dot) org)
549    Date: 2004/05/19 22:10:00 CDT
550    Tag: blume-20040519-ml-build
551    Description:
552    
553    Arranged for ml-build to clean up after itself a little bit better.
554    The script generates a temporary SML source file and compiles it using
555    CM, so CM generates metadata (GUID, SKEL, objectfile) for it.  It now
556    gets rid of those at the end, so they don't accumulate under .cm.
557    
558    This required a minor change to install.sh because the name of the
559    metadata directory (default: .cm) is actually configurable at
560    installation time.
561    
562    ----------------------------------------------------------------------
563    Name: Matthias Blume (blume (at) tti - c (dot) org)
564    Date: 2004/05/18 15:50:00 CDT
565    Tag: blume-20040518-mkreader
566    Description:
567    
568    Added Posix.IO.mk{Bin,Text}{Reader,Writer} by lifting their respective
569    implementations from internal modules PosixBinPrimIO and PosixTextPrimIO.
570    
571    ----------------------------------------------------------------------
572    Name: Matthias Blume (blume (at) tti - c (dot) org)
573    Date: 2004/05/11 14:35:00 CDT
574    Tag: blume-20040511-win32sock
575    Description:
576    
577    Added previously missing support for many socket-related functions
578    under win32.  Thanks to David Hansel <hansel@reactive-systems.com>
579    for the voluminous patch!
580    
581    (I have not tested this patch under win32 yet.)
582    
583    Here is David's e-mail:
584    
585    Hi,
586    
587    Attached to this email you find a diff against sml/nj 110.45
588    that will enable socket support under Windows.
589    
590    To apply the patch (using unix or cygwin)
591    1) gunzip runtime.diff.gz
592    2) "cd" into "src/runtime" in the source tree of a fresh
593       110.45 installation.
594    3) patch -p 1 < [your/path/to]runtime.diff
595    
596    The code compiles fine but has NOT yet been extensively tested.
597    I only ran a few tests for basic socket client functionality
598    (which worked fine).  Especially the functions that use ioctl
599    are not tested at all and might not work (see below).
600    
601    I implemented this since we want to move to a newer version of sml/nj
602    but need socket support in order to use it.  This is the first time I
603    even had a look at the sml/nj source,  so please review my changes
604    before making this part of the distribution!  Here are a few issues
605    that I think might be better for someone to solve who is more
606    familiar with the sml/nj source (and socket programming):
607    
608    - getnetbyaddr.c and getnetbyname.c will raise a "not implemented"
609      exception since I could not figure out what the windows equivalent
610      of these functions is
611    
612    - In sockets-osdep.h there are a some #include statements that are
613      only used in a few files that include sockets-osdep.h
614    
615    - In smlnj-sock-lib.c, function init_fn() calls WSAStartup() but
616      does not process its return value since I don't know how to
617      report an error upwards.
618    
619    - It would probably be good to have a call to WSACleanup() when
620      the library is unloaded (if there is such a possibility).
621      Otherwise I think Windows will take care of this automatically
622      when the process finishes.
623    
624    - I used ioctlsocket() as a replacement for ioctl() but I have
625      no idea if that is actually the proper replacement on Windows.
626    
627    - All these issues are marked in the code by "FIXME" comments.
628    
629    We use sml/nj extensively in our products and are quite happy
630    with it.  I hope this contribution will help you.
631    
632    Keep up the good work!
633    
634    David
635    
636    ----------------------------------------------------------------------
637    Name: Matthias Blume (blume (at) tti - c (dot) org)
638    Date: 2004/05/11 14:20:00 CDT
639    Tag: blume-20040511-installml
640    Description:
641    
642    Fixed two bugs in installml script.  (Thanks to Vesa A. Norrman for
643    the patch.)
644    
645    ----------------------------------------------------------------------
646    Name: Matthias Blume (blume (at) tti - c (dot) org)
647    Date: 2004/05/11 14:05:00 CDT
648    Tag: blume-20040511-nlffi-netbsd
649    Description:
650    
651    Added support for nlffi under netbsd.  (Thanks to Vesa A. Norrman for
652    the patch.)
653    
654    ----------------------------------------------------------------------
655    Name: Matthias Blume (blume (at) tti - c (dot) org)
656    Date: 2004/05/11 12:05:00 CDT
657    Tag: blume-20040511-exports
658    Description:
659    
660    As per request by Adam Chlipala <adam@hcoop.net>, extended various
661    export lists in compiler-related .cm-files.
662    
663    ----------------------------------------------------------------------
664    Name: Matthias Blume (blume (at) tti - c (dot) org)
665    Date: 2004/05/11 11:35:00 CDT
666    Tag: blume-20040511-allsource
667    Description:
668    
669    The installer now honors the "src-smlnj" target again, although its meaning
670    has changed from "all sources required for the compiler" to "all sources
671    the installer knows about".  In other words, if you enable "src-smlnj"
672    in the "targets" file, then the installer will pull in sources for
673    everything.  (Notice that this refers to source code only.  Compiled
674    code is still only installed for modules that were requested explicitly
675    or which are required for other modules that were requested explicitly.)
676    
677    ----------------------------------------------------------------------
678    Name: Matthias Blume (blume (at) tti - c (dot) org)
679    Date: 2004/04/23 17:40:00 CDT
680    Tag: blume-20040423-ieee-scan
681    Description:
682    
683    Fixed IEEEReal.scan (and .fromString) so that if there is an overflow
684    in the exponent calculation we get INF or ZERO (depending on the mantissa
685    and the sign of the exponent).
686    
687    ----------------------------------------------------------------------
688    Name: Matthias Blume (blume (at) tti - c (dot) org)
689    Date: 2004/04/23 10:40:00 CDT
690    Tag: blume-20040423-ml-build
691    Description:
692    
693    The ml-build script now terminates with a non-0 status when something
694    goes wrong.
695    
696    ----------------------------------------------------------------------
697    Name: Matthias Blume (blume (at) tti - c (dot) org)
698    Date: 2004/04/22 16:35:00 CDT
699    Tag: blume-20040422-Option
700    Description:
701    
702    Made exception Option to be the same as exception Option.Option
703    (as it should be).
704    
705    ----------------------------------------------------------------------
706    Name: Allen Leung (leunga (at) reservoir (dot) com)
707    Date: 2004/03/19 14:40:00 EST
708    Tag: leunga-20040319-cygwin-nlffi
709    Description:
710    
711    Fixed the runtime so that ml-nlffi-lib runs on the cygwin version
712    of SML/NJ.  The problem is that
713    
714       lib = dlopen(NULL, ...)
715       f   = dlsym(lib, "malloc");
716    
717    does not work on Windows unless we explicitly export symbols
718    such as 'malloc' during linking.  We fixed this by explicitly
719    exporting the required symbols with the magic gcc incantation:
720    
721        -Wl,--export-all cygwin.def
722    
723    where cygwin.def is a file containing all the symbols that
724    we wish to export.
725    
726    I suspect this is a Windows problem and we'll have to
727    do the same (somehow with windows compilers) when
728    we build the native win32 version with the system
729    calls LoadLibrary/GetProcAddress.
730    
731    ----------------------------------------------------------------------
732    Name: Matthias Blume (blume (at) tti - c (dot) org)
733    Date: 2004/03/04 16:35:00 CST
734    Tag: blume-20040304-intinf-fmt
735    Description:
736    
737    Fixed problem with IntInf.fmt (sign would show up on the right instead
738    of on the left for BIN, OCT, and HEX).
739    
740    ----------------------------------------------------------------------
741    Name: Matthias Blume (blume (at) tti - c (dot) org)
742    Date: 2004/03/04 11:25:00 CST
743    Tag: blume-20040304-symlinks
744    Description:
745    
746    Fixed problem with installer script (unix only) where bin/ml-yacc and
747    friends pointed (via symlinks) to absolute locations instead of just
748    .run-sml.  This was reported by Vesa A Norrman.
749    
750    ----------------------------------------------------------------------
751    Name: Matthias Blume (blume (at) tti - c (dot) org)
752    Date: 2004/02/13 14:50:00 CST
753    Tag: Release_110_45
754    Description:
755    
756    New working version (110.45). New bootfiles.
757    
758    ----------------------------------------------------------------------
759    Name: Matthias Blume (blume (at) tti - c (dot) org)
760    Date: 2004/01/26 15:15:15 CST
761    Tag: blume-20040126-toplevel
762    Description:
763    
764    Improved handling of exceptions at the interactive toplevel.
765    
766    ----------------------------------------------------------------------
767    Name: Matthias Blume (blume (at) tti - c (dot) org)
768    Date: 2004/01/26 11:25:00
769    Tag: blume-20040126-app
770    Description:
771    
772    Type of top-level "app" corrected.
773    Added code for setting vp_limitPtrMask to Win32-specific runtime.
774    
775    ----------------------------------------------------------------------
776    Name: Matthias Blume (blume (at) tti - c (dot) org)
777    Date: 2003/11/18 17:10 CST
778    Tag: blume-20031118-basis-fiddle
779    Description:
780    
781    - changed Timer interface to what might become the spec
782    - POSIX_FLAGS -> BIT_FLAGS according to spec
783    - some other minor discrepancies wrt. spec eliminated
784    
785    ----------------------------------------------------------------------
786    Name: Matthias Blume (blume (at) tti - c (dot) org)
787    Date: 2003/11/06 12:00:00 CST
788    Tag: Release_110_44
789    Description:
790    
791    New working version (110.44). New bootfiles.
792    
793    ----------------------------------------------------------------------
794    Name: Matthias Blume (blume (at) tti - c (dot) org)
795    Date: 2003/11/04 11:50:00 CST
796    Tag: blume-20031104-move-libraries
797    Description:
798    
799    Eliminated the "dont_move_libraries" directive in config/targets.
800    (The mechanism was broken and could not be fixed easily.  Moreover,
801    there does not seem to be any reason not to move all libraries into
802    lib during installation.  I originally implemented this directive as a
803    backward-compatibility feature when I first introduced the new CM.
804    Now that things have been stable for a long time and going back to the
805    old CM is not an option, there is no reason to keep it around.)
806    
807    ----------------------------------------------------------------------
808    Name: Matthias Blume (blume (at) tti - c (dot) org)
809    Date: 2003/11/03 16:00:00 CST
810    Tag: blume-20031103-installdir
811    Description:
812    
813    Made installer honor INSTALLDIR variable again.  (Thanks to Chris
814    Richards for pointing out the problem and providing the solution.)
815    
816    ----------------------------------------------------------------------
817    Name: Matthias Blume (blume (at) tti - c (dot) org)
818    Date: 2003/10/01 17:05:00 CDT
819    Tag: blume-20031001-lal-mlrisc
820    Description:
821    
822    MLRISC bug fix from Lal.
823    
824    ----------------------------------------------------------------------
825    Name: Matthias Blume (blume (at) tti - c (dot) org)
826    Date: 2003/09/30 16:10:00 CDT
827    Tag: blume-20030930-primio-bat
828    Description:
829    
830    1. Added openVector, nullRd, and nullWr to PRIM_IO.
831    2. Improved .bat files (for Win32 port) to make things work under Win95.
832       (thanks to Aaron S. Hawley for this one)
833    
834    ----------------------------------------------------------------------
835    Name: Matthias Blume (blume (at) tti - c (dot) org)
836    Date: 2003/09/26 16:05:00 CDT
837    Tag: blume-20030926-wrappriv
838    Description:
839    
840    Added missing wrapper for privilege "primitive" in $smlnj/viscomp/core.cm.
841    
842    ----------------------------------------------------------------------
843    Name: Matthias Blume (blume (at) tti - c (dot) org)
844    Date: 2003/09/26 15:00:00 CDT
845    Tag: blume-20030926-110_43_3
846    Description:
847    
848    - additional cleanup
849    - version number bump, NEW BOOTFILES
850    
851    ----------------------------------------------------------------------
852    Name: Matthias Blume (blume (at) tti - c (dot) org)
853    Date: 2003/09/26 12:00:00 CDT
854    Tag: blume-20030926-ppautoload
855    Description:
856    
857    I modified the read-eval-print loop so that the autoloader gets
858    invoked whenever the prettyprinter tries to look up a symbol that
859    is not currently defined in the toplevel environment but which
860    appears in CM's autoload registry.  As a result, we see far fewer of
861    those ?.Foo.Bar.xxx names in the prettyprinter's output.
862    
863    In addition to this I tried to clean up some pieces of the Basis
864    implementation (e.g., Socket, Word8Array) in order to prevent other
865    instances of these ?.Foo.Bar.xxx names from being printed.
866    
867    The mechanism that picks names for types still needs some work, though.
868    (Right now it seems that if there is a type A.t which is defined to
869    be B.u, but B is unavailable at toplevel, then A.t gets printed as
870    "?.B.u" although the perhaps more sensible solution would be to use
871    "A.t" in this case.  In other words, the prettyprinter should follow
872    a chain of DEFtycs not farther than there are corresponding toplevel
873    names in the current environment.)
874    
875    ----------------------------------------------------------------------
876    Name: Matthias Blume (blume (at) tti - c (dot) org)
877    Date: 2003/09/24 16:31:00 CDT
878    Tag: blume-20030924-installer
879    Description:
880    
881    Another installer tweak:  All the ML code for the installer is now
882    compiled during CMB.make and put into a little library called
883    $smlnj/installer.cm.  The installation then simply invokes
884    
885       sml -m $smlnj/installer.cm
886    
887    and everything happens automagically.
888    
889    Win32:  ML code senses value of environment variable SMLNJ_HOME.
890    Unix:   ML code senses values of environment variables ROOT, CONFIGDIR,
891            and BINDIR.
892    
893    The new scheme guarantees that the ML code responsible for the installation
894    is in sync with the APIs of the main system.  Also, the installer is
895    somewhat faster because the installer script is precompiled.
896    
897    ----------------------------------------------------------------------
898    Name: Matthias Blume (blume (at) tti - c (dot) org)
899    Date: 2003/09/24 15:35:00 CDT
900    Tag: blume-20030924-synsock
901    Description:
902    
903    Added a signature SYNCHRONOUS_SOCKET to basis.cm.  This is like SOCKET
904    but excludes all non-blocking operations.  Defined SOCKET (in Basis)
905    and CML_SOCKET in terms of SYNCHRONOUS_SOCKET.  Removed superfluous
906    implementations of non-blocking operations from CML's Socket
907    structure.
908    
909    ----------------------------------------------------------------------
910    Name: Matthias Blume (blume (at) tti - c (dot) org)
911    Date: 2003/09/24 15:10:05 CDT
912    Tag: blume-20030924-sockets
913    Description:
914    
915    1. Fixed SOCKET API and implementation to match Basis spec.
916       This required changing the internal representation of sockets to one
917       that remembers (for each socket file descriptor) whether it is currently
918       blocking or non-blocking.  This state is maintained lazily (i.e., a system
919       call is made only if the state actually needs to change).
920    
921    2. OS-specific details of sockets were moved into separate files, thus
922       making it possible to unify the bulk of the socket implementations
923       between Unix and Win32.
924    
925    3. CML's socket API changed accordingly.
926       (Note that we need to remove non-blocking functions from this API
927       since they are redundant in the case of CML!)
928    
929    4. CML's socket implementation now makes use of non-blocking functions
930       provided by Basis, thus removing all OS-dependent code from this part
931       of CML.
932    
933    5. Changed Real64.precision from 52 to 53.  Minor cleanup in Real64 code.
934    
935    ----------------------------------------------------------------------
936    Name: Matthias Blume (blume (at) tti - c (dot) org)
937    Date: 2003/09/22 12:10:00 CDT
938    Tag: blume-20030922-110_43_2
939    Description:
940    
941    Made a new interim version and bootfiles for developer's bootstrapping
942    convenience.
943    
944       110.43.2 -- NEW BOOTFILES
945    
946    ----------------------------------------------------------------------
947    Name: Matthias Blume (blume (at) tti - c (dot) org)
948    Date: 2003/09/19 15:55:00 CDT
949    Tag: blume-20030919-cmdir
950    Description:
951    
952    1. new-install.sh -> install.sh
953    2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
954    3. tweaked installer so that another name instead of .cm can be chosen
955       at install time (by setting the CM_DIR_ARC environment variable
956       during installation); once installation is complete, the name is
957       fixed
958    
959    ----------------------------------------------------------------------
960    Name: Matthias Blume (blume (at) tti - c (dot) org)
961    Date: 2003/09/18 16:00:00 CDT
962    Tag: blume-20030918-110_43_1
963    Description:
964    
965    Made a new interim version and bootfiles for developer's bootstrapping
966    convenience.
967    
968       110.43.1 -- NEW BOOTFILES
969    
970    ----------------------------------------------------------------------
971    Name: Matthias Blume (blume (at) tti - c (dot) org)
972    Date: 2003/09/18 15:20:00 CDT
973    Tag: blume-20030918-misc
974    Description:
975    
976    1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
977       this seems to be controversial at the moment) and used those in
978       Posix.ProcEnv.times.
979    
980    2. Added Time.{from,to}Nanoseconds to Time.
981    
982    3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
983       For example, fromLargeInt never needs to look at more than 3 "big
984       digits" to get its 53 bits of precision.
985    
986    ----------------------------------------------------------------------
987    Name: Matthias Blume (blume (at) tti - c (dot) org)
988    Date: 2003/09/17 16:30:00 CDT
989    Tag: blume-20030917-real32-slices
990    Description:
991    
992    Added an entry to the primitive environment
993    (compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
994    added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
995    
996    Removed some of the "magic" constants in real64.sml and replaced them
997    with code that generates these values from their corresponding
998    integer counterparts.
999    
1000    Made all(?) the slice-related changes to the Basis and made everything
1001    compile again...
1002    
1003    ----------------------------------------------------------------------
1004    Name: Matthias Blume (blume (at) tti - c (dot) org)
1005    Date: 2003/09/15 17:45:00 CDT
1006    Tag: blume-20030915-rbase
1007    Description:
1008    
1009    Fixed bug in Real.fromLargeInt.
1010    
1011    ----------------------------------------------------------------------
1012    Name: Matthias Blume (blume (at) tti - c (dot) org)
1013    Date: 2003/09/13 18:11:00 CDT
1014    Tag: blume-20030913-libinstall
1015    Description:
1016    
1017    Minor bugfix in config/libinstall (set anchor with path to
1018    standalone tool after installing it, otherwise libraries that
1019    need ml-lex or ml-yacc won't compile the first time the installer
1020    runs).
1021    
1022    ----------------------------------------------------------------------
1023    Name: Matthias Blume (blume (at) tti - c (dot) org)
1024    Date: 2003/09/12 11:45:00 CDT
1025    Tag: blume-20030912-various
1026    Description:
1027    
1028     - fixed bug in Real.toLargeInt
1029     - fixed bug in Posix.ProcEnv.times
1030     - changed inputLine functions to return an option
1031     - minor installer improvements / bugfixes
1032     - changed default @SMLalloc parameter for x86/celeron to 64k
1033    
1034    ----------------------------------------------------------------------
1035    Name: Matthias Blume (blume (at) tti - c (dot) org)
1036    Date: 2003/09/09 22:00:00 CDT
1037    Tag: Release_110_43
1038    Description:
1039    
1040        New working release 110.43.  New bootfiles.
1041    
1042    ----------------------------------------------------------------------
1043    Name: Matthias Blume (blume (at) tti - c (dot) org)
1044    Date: 2003/09/09 19:20:00 CDT
1045    Tag: blume-20030909-installer
1046    Description:
1047    
1048    Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
1049    Modified config/install.bat to take advantage of it.  Also modified
1050    config/install.sh (and called it config/new-install.sh) to take advantage
1051    of it on Unix systems.  (The SML code is (supposed to be) platform-
1052    independent.)
1053    
1054    The installer can now install everything under Win32
1055    as well as under *nix as long as it compiles.
1056    
1057    Other changes:
1058    
1059      - made CML compile again under Win32
1060      - made eXene compile under Win32 (by providing a fake structure UnixSock
1061        and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
1062      - fixed a bug in nowhere: it assumed that type OS.Process.status is the
1063        same as type int;  under Win32 it isn't
1064      - fixed some slice-related problems in the win32-specific parts of CML
1065      - added a functor argument "sameVol" to os-path-fn.sml in the Basis
1066        (under Win32, the volume name is case-insensitive, and the
1067        OS.Path code compares volume names for equality)
1068    
1069    ----------------------------------------------------------------------
1070    Name: Matthias Blume (blume (at) tti - c (dot) org)
1071    Date: 2003/09/08 11:55:00 CDT
1072    Tag: blume-20030908-fullpath
1073    Description:
1074    
1075    Made Win32 version of OS.FileSys.fullPath return current directory
1076    when given an empty string.  This is what the spec says, and incidentally,
1077    CM depends on it.  (CM otherwise goes into an infinite loop in certain
1078    cases when presented with the name of a non-existing .cm file.)
1079    
1080    ----------------------------------------------------------------------
1081    Name: Matthias Blume (blume (at) tti - c (dot) org)
1082    Date: 2003/09/04 16:30:00 CDT
1083    Tag: blume-20030905-slices-etc
1084    Description:
1085    
1086      1. Changed interface to vectors and arrays in Basis to match
1087         (draft) Basis spec.
1088      2. Added signatures and implementations of slices according to
1089         Basis spec.
1090      3. Edited source code throughout the system to make it compile again
1091         under 1. and 2.  (In some cases code had to be added to have it
1092         match the new signatures.)
1093      4. MLRISC should be backward-compatible: the copies of the originals
1094         of files that needed to change under 3. were retained, the .cm files
1095         check the compiler version number and use old versions when
1096         appropriate.
1097      5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
1098         dirstream -> string option (in accordance with Basis spec).
1099      6. When generating code that counts lines, ml-lex used function
1100         CharVector.foldli, taking advantage of its old interface.
1101         This has been replaced with the corresponding code from
1102         CharVectorSlice.  (html-lex must be re-lexed!)
1103      7. BitArray in smlnj-lib/Util has been extended/modified to match the
1104         new MONO_ARRAY signature.  (Do we need BitArraySlice?)
1105      8. Removed temporary additions (fromInternal, toInternal) from the
1106         (now obsolete) IntInf in smlnj-lib/Util.
1107      9. Cleaned up structure Byte.
1108     10. Added localOffset, scan, and fromString to Date (according to spec).
1109         Cleaned/corrected implementation of Date.
1110         (Still need to check for correctness; implement better canonicalizeDate.)
1111     11. Added "scan" to signature IEEE_REAL.
1112     12. Some improvements to IntInf [in particular: efficiency-hack for
1113         mod and rem when second operand is 2 (for parity checks).]
1114     13. Changed representation of type Time.time, using a single IntInf.int
1115         value counting microseconds.  This considerably simplified the
1116         implementation of structure Time.  We now support negative time
1117         values; scan and fromString handle signs.
1118     14. Functor PrimIO now takes two additional arguments (VectorSlice and
1119         ArraySlice).
1120    
1121    ----------------------------------------------------------------------
1122    Name: Matthias Blume (blume (at) tti - c (dot) org)
1123    Date: 2003/08/28 17:00:00 CDT
1124    Tag: blume-20030828-intinf
1125    Description:
1126    
1127    This is a major update which comes with a version number bump
1128    (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
1129    and an implementation of IntInf in the Basis.
1130    
1131    There are a fairly large number of related changes and updates throughout
1132    the system:
1133    
1134    Basis:
1135            - Implemented IntInf.
1136            - Made LargeInt a projection of IntInf (by filtering through INTEGER).
1137            - Added some missing Real64 operations, most notably Real.toLargeInt.
1138            - Added FixedInt as a synonym for Int32.
1139    
1140    compiler:
1141            * Added support for a built-in intinf type.
1142              - literals
1143              - pattern matching
1144              - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
1145              - overloading on literals and operations
1146    
1147            This required adding a primitive type intinf, some additional
1148            primops, and implementations for several non-trivial intinf
1149            operations in Core.  (The intinf type is completely abstract
1150            to the compiler; all operations get delegated back to the Core.)
1151    
1152            * Intinf equality is handled by polyequal.  However, the compiler
1153            does not print its usual warning in this case (since polyequal
1154            is the right thing to do there).
1155    
1156            * Improved the organization of structure InlineT.
1157    
1158            * A word about conversion primops:
1159            If conversions involving intinf do not cancel out during
1160            CPS contract, then the compiler must insert calls to Core functions.
1161            Since all core access must be resolved already during the FLINT
1162            translate phase, it would be too late a the time of CPS contract
1163            to add new Core calls.  For this reason, conversion primops
1164            for intinf carry two arguments:  1. the numeric argument that
1165            they are supposed to convert, and 2. the Core function that
1166            can help with this conversion if necessary.  If CPS contract
1167            eliminates a primop, then the associated Core function becomes
1168            dead and goes away.  Intinf conversion primops that do not get
1169            eliminated by CPS contract get rewritten into calls of their
1170            core functions by a separate, new phase.
1171    
1172    interactive system:
1173            - Control.Print.intinfDepth controls max length of intinf constants
1174            being printed. (Analogous to Control.Print.stringDepth.)
1175            - Cleanup in printutil and pputil: got rid of unused stuff and
1176            duplicates; replaced some of the code with code that makes better
1177            use of library functionality.
1178    
1179    CM:
1180            Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
1181            are no longer silent.
1182    
1183    CKIT:
1184            Fixed mismatched uses of Int32 and LargeInt.  I always decided
1185            in favor of LargeInt -- which is now the same as IntInf.
1186            CKIT-knowledgable people should check whether this is what's
1187            intended and otherwise change things back to using Int32 or
1188            FixedInt.
1189    
1190    Throughout the code:
1191            Started using IntInf.int literals and built-in operations
1192            (e.g., comparison with 0) where this seems appropriate.
1193    
1194    
1195    ----------------------------------------------------------------------
1196    Name: Dave MacQueen (dbm@cs.uchicago.edu)
1197    Date: 2003/08/13 11:36:00 CDT
1198    Tag: dbm-20030813-mcz-merge1
1199    Description:
1200    
1201    Merging changes from the mcz-branch development branch into trunk.
1202    These changes involve replacement of the emulated old prettyprinter
1203    interface with direct use of the SML/NJ Lib PP library, and fixing
1204    of a couple of bugs (895, 1186) relating to error messages.  A new
1205    prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
1206    has been added.
1207    
1208    ----------------------------------------------------------------------
1209    Name: Matthias Blume (blume (at) tti - c (dot) org)
1210    Date: 2003/08/11 15:45:00 CDT
1211    Tag: blume-20030811-windows
1212    Description:
1213    
1214    Version number bumped to 110.42.9.  NEW BOOTFILES!!!
1215    
1216        http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
1217    
1218    This patch restores SML/NJ's ability to run under win32.  There are a
1219    number of changes, including fixes for several bugs that had gone
1220    unnoticed until now:
1221    
1222      - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
1223        trying to run the win32 version from within cygwin.)
1224    
1225      - There are a number of simple .bat scripts that substitute for their
1226        corresponding Unix shell-scripts.  (See below.)
1227    
1228      - The internals of ml-build have been modified slightly.  The main
1229        difference is that instead of calling ".link-sml" (or link-sml.bat)
1230        using OS.Process.system, the ML process delegates this task back
1231        to the script.  Otherwise problems arise in mixed environments such
1232        as Cygwin where scripts look and work like Unix scripts, but
1233        where OS.Process.system cannot run them.
1234    
1235      - In CM, the srcpath pickler used native pathname syntax -- which
1236        is incorrect in the case of cross-compilation.  The new pickle format
1237        is independent of platform-specific naming conventions.
1238    
1239      - Path configuration files (such as lib/pathconfig) can now choose
1240        between native and standard syntax.  Placing a line of the form
1241    
1242            standard!
1243    
1244        into the file causes all subsequent paths to be interpreted using
1245        CM standard pathname syntax (= Unix conventions); a line
1246    
1247            native!
1248    
1249        switches back to native style.  This was needed so that
1250        path config files can be written portably, see src/system/pathconfig.
1251    
1252      - Runtime system:
1253    
1254          - win32-filesys.c:  get_file_time and set_file_time now
1255            access modification time, not creation time.
1256    
1257          - I/O code made aware of new array representation.
1258    
1259          - Bug fixes in X86.prim.masm.
1260    
1261       - src/system/makeml made aware of win32. (For use under cygwin
1262            and other Unix-environments for windows.)
1263    
1264       - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
1265         which caused BinIO.inputAll to fail consistently.
1266    
1267    .bat scripts:
1268    
1269       Windows .bat scripts assume that SMLNJ_HOME is defined.
1270    
1271       - sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
1272         applications (sml, ml-yacc, ml-lex).
1273       - ml-build.bat: analogous to ml-build.
1274       - config\install.bat:  Analogous to config/install.sh.  This requires
1275         that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
1276         (nmake etc. must be on the path, and vcvars32 must have been run.)
1277         Moreover, sources for ml-lex and ml-yacc need to exist under src,
1278         and the bootfile hierarchy must have been unpacked under
1279         sml.boot.x86-win32.
1280         The script is very primitive and does a poor job at error checking.
1281         It only installs the base system, ml-lex, and ml-yacc.  No other
1282         libraries are being installed (i.e., you get only those that
1283         are part of the compiler.)
1284       - link-sml.bat:  analogous to .link-sml, but not currently used
1285    
1286    Unrelated bug fixes:
1287    
1288      - ml-nlffigen now exports structures ST_* corresponding to incomplete
1289        types.
1290      - Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
1291        otherwise.)
1292    
1293    ----------------------------------------------------------------------
1294    Name: Dave MacQueen (macqueen@cs.uchicago.edu)
1295    Date: 2003/06/17
1296    Tag: macqueen-20030617-bug895
1297    Description:
1298    
1299    Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
1300    Tag will be used for new development branch (mcz-branch) for
1301    use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
1302    
1303    ----------------------------------------------------------------------
1304    Name: Matthias Blume (blume (at) tti - c (dot) org)
1305    Date: 2003/05/27 16:55:00 CDT
1306    Tag: blume-20030527-polyeq
1307    Description:
1308    
1309    Tried to eliminated most cases of polymorphic equality.
1310    
1311    ----------------------------------------------------------------------
1312    Name: Matthias Blume (blume (at) tti - c (dot) org)
1313    Date: 2003/05/21 17:45:00 CDT
1314    Tag: blume-20030517-complete
1315    Description:
1316    
1317    Two changes:
1318    
1319      1. Added a flag for controlling whether non-exhaustive bindings will
1320         be treated as errors (default is false).
1321      2. Cleaned up the *entire* source tree so that CMB.make goes through
1322         without a single non-exhaustive match- or bind warning.
1323    
1324    ----------------------------------------------------------------------
1325    Name: Matthias Blume (blume (at) tti - c (dot) org)
1326    Date: 2003/05/17 10:20:00 CDT
1327    Tag: blume-20030517-absyn
1328    Description:
1329    
1330    1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
1331    
1332       This mainly affects the quality of error messages.  However, some
1333       of the code is now more straightforward than before. (Treatment of
1334       the above four constructs in translate.sml is much simpler than
1335       the "macro-expansion" that was going on before.  Plus, the mach-
1336       compiler no longer gets invoked just to be able to compile an
1337       if-expression.)
1338    
1339    2. The ErrorMsg.Error exception is now caught and absorbed by the
1340       interactive loop.
1341    
1342    ----------------------------------------------------------------------
1343    Name: Allen Leung
1344    Date: 2003/05/16 13:05:00 CDT
1345    Tag: leunga-20030516-cygwin-runtime
1346    Description:
1347    
1348        Ported the runtime system to cygwin, which uses the unix
1349    x86-unix bin files.  Missing/buggy features:
1350    
1351       o getnetbyname, getnetbyaddr: these functions seem to be missing in
1352         the Cygwin library.
1353       o Ctrl-C handling may be flaky.
1354       o Windows system calls and Windows I/O are not supported.
1355    
1356       A new set of binfiles is located at:
1357    
1358         http://www.dorsai.org/~leunga/boot.x86-unix.tgz
1359    
1360       This is only needed for bootstrapping the cygwin version of smlnj.
1361       Other x86 versions can use the existing binfiles.
1362    
1363    ----------------------------------------------------------------------
1364    Name: Matthias Blume
1365    Date: 2003/04/08 15:42:00 CDT
1366    Tag: blume-20030408-listpair
1367    Description:
1368    
1369    1. Added a target 'mlrisc' to installer.
1370    
1371    2. Added missing elements to structure ListPair.
1372    
1373    ----------------------------------------------------------------------
1374    Name: Allen Leung
1375    Date: 2003/01/07 10:40:00 EST
1376    Tag: leunga-20030107-int-rem
1377    Description:
1378    
1379    Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
1380    The arguments to the SUBL instruction were swapped.
1381    
1382    ----------------------------------------------------------------------
1383    Name: Matthias Blume
1384    Date: 2002/12/12 16:25:00 EST
1385    Tag: blume-20021212-risc-ra
1386    Description:
1387    
1388    Fixed a serious bug in the rewrite code for FP spilling/reloading that
1389    sent the RA into an infinite loop when floating point registers get
1390    spilled.  (Because of this bug, e.g., nucleic stopped compiling between
1391    110.37 and 110.38.)
1392    There was another set of potential problems related to the handling of
1393    MLRISC annotations (but those did not yet cause real problems, apparently).
1394    
1395    ----------------------------------------------------------------------
1396    Name: Matthias Blume
1397    Date: 2002/12/06 22:40:00 EST
1398    Tag: blume-20021206-cm-fileid
1399    Description:
1400    
1401    Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
1402    This fixes the problem of CM getting confused by files that suddenly
1403    change their identity (e.g., by getting unlinked and recreated by some
1404    text editor such as vi).  There might be a better/cheaper/cleaner way
1405    of doing this, but for now this will have to do.
1406    
1407    ----------------------------------------------------------------------
1408    Name: Matthias Blume
1409    Date: 2002/10/28 09:50:00 EST
1410    Tag: blume-20021028-typecheck
1411    Description:
1412    
1413    Exported structure Typecheck from $smlnj/viscomp/core.cm.
1414    
1415    ----------------------------------------------------------------------
1416    Name: Matthias Blume
1417    Date: 2002/10/17 09:10:00 EDT
1418    Tag: Release_110_42
1419    Description:
1420    
1421    In good old tradition, there has been a slight hiccup so that we have
1422    to patch 110.42 after the fact.   The old release tag has been replaced
1423    (see below).
1424    
1425    The change solves a problem with two competing approaches the
1426    configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
1427    each other's way.
1428    
1429    This change only affects the runtime system code and the installer script.
1430    (No new bootfiles.)
1431    
1432    ----------------------------------------------------------------------
1433    Name: Matthias Blume
1434    Date: 2002/10/16 12:00:00 EDT
1435    Tag: Release_110_42_removed
1436    Description:
1437    
1438    New working release.  New bootfiles.
1439    
1440    ----------------------------------------------------------------------
1441    Name: Matthias Blume
1442    Date: 2002/10/10 13:10:00 EDT
1443    Tag: blume-20021010-ppc-divs
1444    Description:
1445    
1446    The mltree operator DIVS must be implemented with an overflow check on
1447    the PPC because the hardware indicates divide-by-zero using "overflow" as
1448    well.
1449    
1450    ----------------------------------------------------------------------
1451    Name: Matthias Blume
1452    Date: 2002/07/23 11:45:00
1453    Tag: blume-20020723-smlnj-home
1454    Description:
1455    
1456    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
1457    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
1458    CM_PATHCONFIG is also set), the path configuration file is assumed
1459    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
1460    the entire tree to some other place and everything will "just work".
1461    
1462    (Companion commands such as ml-build and ml-makedepend also sense this
1463    variable.)
1464    
1465    ----------------------------------------------------------------------
1466    Name: Matthias Blume
1467    Date: 2002/07/12 21:19:00 EDT
1468    Tag: blume-20020712-liveness
1469    Description:
1470    
1471    Exported two useful "step" functions from liveness module (MLRISC).
1472    
1473    ----------------------------------------------------------------------
1474    Name: Matthias Blume
1475    Date: 2002/07/05 16:00 EDT
1476    Tag: Release_110_41
1477    Description:
1478    
1479    New working release.  New bootfiles.
1480    
1481    ----------------------------------------------------------------------
1482    Name: Matthias Blume
1483    Date: 2002/07/05 10:25:00 EDT
1484    Tag: blume-20020705-btimp
1485    Description:
1486    
1487    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
1488    other clients can set up backtracing support.
1489    
1490    ----------------------------------------------------------------------
1491    Name: Matthias Blume
1492    Date: 2002/06/25 17:23:00 EDT
1493    Tag: blume-20020625-fpmax
1494    Description:
1495    
1496    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
1497    case.  (The sense of the isNaN test was reversed -- which made min and
1498    max always return their first argument.)
1499    
1500    ----------------------------------------------------------------------
1501    Name: Matthias Blume
1502    Date: 2002/06/11
1503    Tag: blume-20020611-unixpath
1504    Description:
1505    
1506    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
1507    
1508    ----------------------------------------------------------------------
1509    Name: Matthias Blume
1510    Date: 2002/06/10 16:35:00 EDT
1511    Tag: blume-20020610-ieeereal
1512    Description:
1513    
1514    I back-ported my implementation of IEEEReal.fromString from the
1515    idlbasis-devel branch so that we can test it.
1516    
1517    Another small change is that ppDec tries to give more information
1518    than just "<sig>" in the case of functors.  However, this code is
1519    broken in some mysterious way if the functor's body's signature
1520    has not been declared by ascription but gets inferred from the
1521    implementation.  This needs fixing...
1522    
1523    ----------------------------------------------------------------------
1524    Name: Matthias Blume
1525    Date: 2002/05/31
1526    Tag: blume-20020531-btrace-mode
1527    Description:
1528    
1529    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
1530    the wayside when I switched over to using Controls everywhere.)
1531    
1532    ----------------------------------------------------------------------
1533    Name: Lal George
1534    Date: 2002/05/23 12:21:40 EDT
1535    Tag: george-20020523-visual-labels
1536    Description:
1537    
1538    Labels are now displayed in the graphical output to make
1539    the fall-through and target blocks obvious.
1540    
1541    ----------------------------------------------------------------------
1542    Name: Matthias Blume
1543    Date: 2002/05/22 11:03:00 EDT
1544    Tag: blume-20020522-shrink
1545    Description:
1546    
1547    John tweaked yesterday's fix for 1131 to handle an out-of-memory
1548    situation that comes up when allocating huge arrays.
1549    
1550    ----------------------------------------------------------------------
1551    Name: Matthias Blume
1552    Date: 2002/05/21 16:00:00 EDT
1553    Tag: Release_110_40
1554    Description:
1555    
1556    New working release (110.40).  New bootfiles.
1557    
1558    [Also: John Reppy fixed GC bug 1131.]
1559    
1560    ----------------------------------------------------------------------
1561    Name: Matthias Blume
1562    Date: 2002/05/21 12:35:00 EDT
1563    Tag: blume-20020521-cmdoc
1564    Description:
1565    
1566    CM documentation update.
1567    
1568    ----------------------------------------------------------------------
1569    Name: Matthias Blume
1570    Date: 2002/05/21 10:55:00 EDT
1571    Tag: blume-20020521-misc
1572    Description:
1573    
1574    - John tweaked runtime to be silent on heap export (except when
1575      GC messages are on).
1576    - I added a few more things (cross-compiling versions of CMB) to
1577      config/preloads (as suggestions).
1578    
1579    ----------------------------------------------------------------------
1580    Name: Matthias Blume
1581    Date: 2002/05/20 22:25:00 EDT
1582    Tag: blume-20020520-controls
1583    Description:
1584    
1585    - Added ControlUtil structure to control-lib.cm.
1586    - Use it throughout.
1587    - Used Controls facility to define MLRISC controls (as opposed to
1588      registering MLRISC control ref cells with Controls after the fact)
1589    - Fixed messed-up controls priorities.
1590    
1591    * Removed again all the stuff from config/preloads that one wouldn't
1592      be able to preload at the time the initial heap image is built.
1593      (Many libraries, e.g., CML, do not exist yet at this time.  The
1594       only libraries that can be preloaded via config/preloads are those
1595       that come bundled with the bootfiles.)
1596    
1597    ----------------------------------------------------------------------
1598    Name: Matthias Blume
1599    Date: 2002/05/20 10:59:00 EDT
1600    Tag: blume-20020520-preloads
1601    Description:
1602    
1603    Added a lot of commented-out suggestions for things to be included
1604    in config/preloads.
1605    
1606    ----------------------------------------------------------------------
1607    Name: Allen Leung
1608    Date: 2002/05/18 14:20:00 EDT
1609    Tag: leunga-20020518-mdl
1610    Description:
1611    
1612    o Made the mdl tool stuff compile and run again.
1613    o I've disabled all the stuff that depends on RTL specifications; they
1614      are all badly broken anyway.
1615    
1616    ----------------------------------------------------------------------
1617    Name: Matthias Blume
1618    Date: 2002/05/17 16:49:00 EDT
1619    Tag: blume-20020517-controls
1620    Description:
1621    
1622    0. John Reppy made several modifications to the SML/NJ library.
1623       In particular, there is a shiny new controls-lib.cm.
1624    
1625    1. Pushed new controls interface through compiler so that everything
1626       compiles again.
1627    
1628    2. Added FormatComb and FORMAT_COMB to the CML version of the
1629       SML/NJ library (so that CML compiles again).
1630    
1631    3. Modified init scripts because XXX_DEFAULT environment variables
1632       are no longer with us.  (Boot-time initialization is now done
1633       using the same environment variables that are also used for
1634       startup-time initialization of controls.)
1635    
1636    ----------------------------------------------------------------------
1637    Name: Lal George
1638    Date: 2002/05/15 09:20:10 EDT
1639    Tag: george-20020515-pseudo-op-decls
1640    Description:
1641    
1642            All pseudo-ops emitted before the first segment declaration
1643    such as TEXT, DATA, and BSS directives are assumed to be global
1644    declarations and are emitted first in the assembly file. This is
1645    useful in a number of situations where one has pseudo-ops that are not
1646    specific to any segment, and also works around the constraint that one
1647    cannot have client pseudo-ops in the TEXT segment.
1648    
1649            Because no segment is associated with these declarations it is
1650    an error to allocate any space or objects before the first segment
1651    directive and an exception will be raised. However, we cannot make
1652    this check for client pseudo-ops.
1653    
1654            These top level declarations are a field in the CFG graph_info.
1655    In theory you can continue to add to this field after the CFG has been
1656    built -- provided you know what you are doing;-)
1657    
1658    ----------------------------------------------------------------------
1659    Name: Matthias Blume
1660    Date: 2002/05/13 16:40:00 EDT
1661    Tag: blume-20020513-pp-etc
1662    Description:
1663    
1664    A few minor bugfixes:
1665    
1666      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
1667        (Bogus printouts for redefined bindings still occur.  Compiler
1668        bug should no longer occur now.  We need to redo the prettyprinter
1669        from scratch.)
1670    
1671      - CM pathname printer now also adds escape sequences for ( and )
1672    
1673      - commend and docu fixes for ml-nlffi
1674    
1675    ----------------------------------------------------------------------
1676    Name: Matthias Blume
1677    Date: 2002/05/10 16:40:00 EDT
1678    Tag: blume-20020510-erg-textio
1679    Description:
1680    
1681    Applied the following bugfix provided by Emden Gansner:
1682    
1683        Output is corrupted when outputSubstr is used rather than output.
1684    
1685        The problem occurs when a substring
1686    
1687            ss = (s, dataStart, dataLen)
1688    
1689        where dataStart > 0, fills a stream buffer with avail bytes left.
1690        avail bytes of s, starting at index dataStart, are copied into the
1691        buffer, the buffer is flushed, and then the remaining dataLen-avail
1692        bytes of ss are copied into the beginning of the buffer. Instead of
1693        starting this copy at index dataStart+avail in s, the current code
1694        starts the copy at index avail.
1695    
1696        Fix:
1697        In text-io-fn.sml, change line 695 from
1698             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
1699        to
1700             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
1701    
1702    ----------------------------------------------------------------------
1703    Name: Matthias Blume
1704    Date: 2002/04/12 13:55:00 EDT
1705    Tag: blume-20020412-assyntax
1706    Description:
1707    
1708    1. Grabbed newer assyntax.h from the XFree86 project.
1709    2. Fiddled with how to compile X86.prim.asm without warnings.
1710    3. (Very) Minor cleanup in CM.
1711    
1712    ----------------------------------------------------------------------
1713    Name: Matthias Blume
1714    Date: 2002/04/01 (no joke!) 17:07:00 EST
1715    Tag: blume-20020401-x86div
1716    Description:
1717    
1718    Added full support for div/mod/rem/quot on the x86, using the machine
1719    instruction's two results (without clumsily recomputing the remainder)
1720    directly where appropriate.
1721    
1722    Some more extensive power-of-two support was added to the x86 instruction
1723    selector (avoiding expensive divs, mods, and muls where they can be
1724    replaced with cheaper shifts and masks).  However, this sort of thing
1725    ought to be done earlier, e.g., within the CPS optimizer so that
1726    all architectures benefit from it.
1727    
1728    The compiler compiles to a fixed point, but changes might be somewhat
1729    fragile nevertheless.  Please, report any strange things that you might
1730    see wrt. div/mod/quot/rem...
1731    
1732    ----------------------------------------------------------------------
1733    Name: Matthias Blume
1734    Date: 2002/03/29 17:22:00
1735    Tag: blume-20020329-div
1736    Description:
1737    
1738    Fixed my broken div/mod logic.  Unfortunately, this means that the
1739    inline code for div/mod now has one more comparison than before.
1740    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
1741    The problem was with quotient = 0, because that alone does not tell
1742    us which way the rounding went.  One then has to look at whether
1743    remainder and divisor have the same sign...  :(
1744    
1745    Anyway, I replaced the bootfiles with fresh ones...
1746    
1747    ----------------------------------------------------------------------
1748    Name: Matthias Blume
1749    Date: 2002/03/29 14:10:00 EST
1750    Tag: blume-20020329-inlprims
1751    Description:
1752    
1753    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
1754    
1755    Primops have changed. This means that the bin/boot-file formats have
1756    changed as well.
1757    
1758    To make sure that there is no confusion, I made a new version.
1759    
1760    
1761    CHANGES:
1762    
1763    * removed REMT from mltree (remainder should never overflow).
1764    
1765    * added primops to deal with divisions of all flavors to the frontend
1766    
1767    * handled these primops all the way through so they map to their respective
1768      MLRISC support
1769    
1770    * used these primops in the implementation of Int, Int32, Word, Word32
1771    
1772    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
1773    
1774    * parameterized INLMIN, INLMAX, and INLABS by a numkind
1775    
1776    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
1777      floating point
1778    
1779    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
1780      and Real (but Real.abs maps to a separate floating-point-only primop)
1781    
1782    
1783    TODO items:
1784    
1785    * Hacked Alpha32 instruction selection, disabling the selection of REMx
1786      instructions because the machine instruction encoder cannot handle
1787      them.  (Hppa, PPC, and Sparc instruction selection did not handle
1788      REM in the first place, and REM is supported by the x86 machine coder.)
1789    
1790    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
1791      selection phase.  (The two can be streamlined because the hardware
1792      delivers both quotient and remainder at the same time anyway.)
1793    
1794    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
1795      (Currently the behavior is inconsistent both across architectures and
1796      wrt. the draft Basis spec.)
1797    
1798    * Word8 should eventually be handled natively, too.
1799    
1800    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
1801      as if there currently is no execution path that could trigger it in
1802      SML/NJ.  (The assumptions underlying functions arith and promotable do not
1803      hold for things like multiplication and division.)
1804    
1805    ----------------------------------------------------------------------
1806    Name: Matthias Blume
1807    Date: 2002/03/27 16:27:00 EST
1808    Tag: blume-20020327-mlrisc-divisions
1809    Description:
1810    
1811    Added support for all four division operations (ML's div, mod, quot,
1812    and rem) to MLRISC.  In the course of doing so, I also rationalized
1813    the naming (no more annoying switch-around of DIV and QUOT), by
1814    parameterizing the operation by div_rounding_mode (which can be either
1815    DIV_TO_ZERO or DIV_TO_NEGINF).
1816    
1817    The generic MLTreeGen functor takes care of compiling all four
1818    operations down to only round-to-zero div.
1819    
1820    Missing pieces:
1821    
1822      * Doing something smarter than relying on MLTreeGen on architectures
1823        like, e.g., the x86 where hardware division delivers both quotient and
1824        remainder at the same time.  With this, the implementation of the
1825        round-to-neginf operations could be further streamlined.
1826    
1827      * Remove inlining support for div/mod/rem from the frontend and replace it
1828        with primops that get carried through to the backend.  Do this for all
1829        int and word types.
1830    
1831    ----------------------------------------------------------------------
1832    Name: Matthias Blume
1833    Date: 2002/03/25 17:25:00 EST
1834    Tag: blume-20020325-divmod
1835    Description:
1836    
1837    I improved (hopefully without breaking them) the implementation of Int.div,
1838    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
1839    advantage of the following observations:
1840    
1841      Let  q = x quot y      r = x rem y
1842           d = x div  y      m = x mod y
1843    
1844    where "quot" is the round-to-zero version of integer division that
1845    hardware usually provides.  Then we have:
1846    
1847         r = x - q * y        where neither the * nor the - will overflow
1848         d = if q >= 0 orelse x = q * y then q else q - 1
1849                              where neither the * nor the - will overflow
1850         m = if q >= 0 orelse r = 0 then r else r + y
1851                              where the + will not overflow
1852    
1853    This results in substantial simplification of the generated code.
1854    The following table shows the number of CFG nodes and edges generated
1855    for
1856            fun f (x, y) = x OPER y
1857            (* with OPER \in div, mod, quot, rem *)
1858    
1859    
1860        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
1861        --------------------------------------------------------
1862         div |         24 |         39 |         12 |         16
1863         mod |         41 |         71 |         12 |         16
1864        quot |          8 |         10 |          8 |         10
1865         rem |         10 |         14 |          8 |         10
1866    
1867    
1868    ----------------------------------------------------------------------
1869    Name: Matthias Blume
1870    Date: 2002/03/25 22:06:00 EST
1871    Tag: blume-20020325-cprotobug
1872    Description:
1873    
1874    Fixed a bug in cproto (c prototype decoder).
1875    
1876    ----------------------------------------------------------------------
1877    Name: Matthias Blume
1878    Date: 2002/03/25 16:00:00 EST
1879    Tag: blume-20020325-raw-primops
1880    Description:
1881    
1882    I did some cleanup to Allen's new primop code and
1883    replaced yesterday's bootfiles with new ones.
1884    (But they are stored in the same place.)
1885    
1886    ----------------------------------------------------------------------
1887    Name: Matthias Blume
1888    Date: 2002/03/24 22:40:00 EST
1889    Tag: blume-20020324-bootfiles
1890    Description:
1891    
1892    Made the bootfiles that Allen asked for.
1893    
1894    ----------------------------------------------------------------------
1895    Name: Allen Leung
1896    Date: 2002/03/23 15:50:00 EST
1897    Tag: leunga-20020323-flint-cps-rcc-primops
1898    Description:
1899    
1900      1. Changes to FLINT primops:
1901    
1902        (* make a call to a C-function;
1903         * The primop carries C function prototype information and specifies
1904         * which of its (ML-) arguments are floating point. C prototype
1905         * information is for use by the backend, ML information is for
1906         * use by the CPS converter. *)
1907      | RAW_CCALL of { c_proto: CTypes.c_proto,
1908                       ml_args: ccall_type list,
1909                       ml_res_opt: ccall_type option,
1910                       reentrant : bool
1911                     } option
1912       (* Allocate uninitialized storage on the heap.
1913        * The record is meant to hold short-lived C objects, i.e., they
1914        * are not ML pointers.  With the tag, the representation is
1915        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
1916        *)
1917      | RAW_RECORD of {tag:bool,sz:int}
1918      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
1919    
1920      2.  These CPS primops are now overloaded:
1921    
1922           rawload of {kind:numkind}
1923           rawstore of {kind:numkind}
1924    
1925          The one argument form is:
1926    
1927             rawload {kind} address
1928    
1929          The two argument form is:
1930    
1931             rawload {kind} [ml object, byte-offset]
1932    
1933      3. RAW_CCALL/RCC now takes two extra arguments:
1934    
1935         a. The first is whether the C call is reentrant, i.e., whether
1936            ML state should be saved and restored.
1937         b. The second argument is a string argument specifying the name of
1938            library and the C function.
1939    
1940         These things are currently not handled in the code generator, yet.
1941    
1942      4. In CProto,
1943    
1944         An encoding type of "bool" means "ml object" and is mapped into
1945         C prototype of PTR.  Note that "bool" is different than "string",
1946         even though "string" is also mapped into PTR, because "bool"
1947         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
1948    
1949      5. Pickler/unpicker
1950    
1951         Changed to handle RAW_RECORD and newest RAW_CCALL
1952    
1953      6. MLRiscGen,
1954    
1955         1. Changed to handle the new rawload/rawstore/rawrecord operators.
1956         2. Code for handling C Calls has been moved to a new module CPSCCalls,
1957            in the file CodeGen/cpscompile/cps-c-calls.sml
1958    
1959      7. Added the conditional move operator
1960    
1961             condmove of branch
1962    
1963         to cps.  Generation of this is still buggy so it is currently
1964         disabled.
1965    
1966    ----------------------------------------------------------------------
1967    Name: Lal George
1968    Date: 2002/03/22 14:18:25 EST
1969    Tag: george-20020322-cps-branch-prob
1970    Description:
1971    
1972    Implemented the Ball-Larus branch prediction-heuristics, and
1973    incorporated graphical viewers for control flow graphs.
1974    
1975    Ball-Larus Heuristics:
1976    ---------------------
1977    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
1978    
1979    By design it uses the Dempster-Shafer theory for combining
1980    probabilities.  For example, in the function:
1981    
1982        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
1983    
1984    the ball-larus heuristics predicts that the n=0 is unlikely
1985    (OH-heuristic), and the 'then' branch is unlikely because of the
1986    RH-heuristic -- giving the 'then' branch an even lower combined
1987    probability using the Dempster-Shafer theory.
1988    
1989    Finally, John Reppy's loop analysis in MLRISC, further lowers the
1990    probability of the 'then' branch because of the loop in the else
1991    branch.
1992    
1993    
1994    Graphical Viewing:
1995    ------------------
1996    I merely plugged in Allen's graphical viewers into the compiler. The
1997    additional code is not much. At the top level, saying:
1998    
1999            Control.MLRISC.getFlag "cfg-graphical-view" := true;
2000    
2001    will display the graphical view of the control flow graph just before
2002    back-patching.  daVinci must be in your path for this to work. If
2003    daVinci is not available, then the default viewer can be changed
2004    using:
2005    
2006            Control.MLRISC.getString "viewer"
2007    
2008    which can be set to "dot" or "vcg" for the corresponding viewers. Of
2009    course, these viewers must be in your path.
2010    
2011    The above will display the compilation unit at the level of clusters,
2012    many of which are small, boring, and un-interesting. Also setting:
2013    
2014            Control.MLRISC.getInt "cfg-graphical-view_size"
2015    
2016    will display clusters that are larger than the value set by the above.
2017    
2018    
2019    ----------------------------------------------------------------------
2020    Name: Matthias Blume
2021    Date: 2002/03/21 22:20:00 EST
2022    Tag: blume-20020321-kmp-bugfix
2023    Description:
2024    
2025    Changed the interface to the KMP routine in PreString and fixed
2026    a minor bug in one place where it was used.
2027    
2028    ----------------------------------------------------------------------
2029    Name: Allen Leung
2030    Date: 2002/03/21 20:30:00 EST
2031    Tag: leunga-20020321-cfg
2032    Description:
2033    
2034      Fixed a potential problem in cfg edge splitting.
2035    
2036    ----------------------------------------------------------------------
2037    Name: Allen Leung
2038    Date: 2002/03/21 17:15:00 EST
2039    Tag: leunga-20020321-x86-fp-cfg
2040    Description:
2041    
2042      1. Recoded the buggy parts of x86-fp.
2043    
2044         a. All the block reordering code has been removed.
2045            We now depend on the block placement phases to do this work.
2046    
2047         b. Critical edge splitting code has been simplified and moved into the
2048            CFG modules, as where they belong.
2049    
2050         Both of these were quite buggy and complex.  The code is now much, much
2051         simpler.
2052    
2053      2. X86 backend.
2054    
2055         a. Added instructions for 64-bit support.  Instruction selection for
2056            64-bit has not been committed, however, since that
2057            requires changes to MLTREE which haven't been approved by
2058            Lal and John.
2059    
2060         b. Added support for FUCOMI and FUCOMIP when generating code for
2061            PentiumPro and above.  We only generate these instructions in
2062            the fast-fp mode.
2063    
2064         c. Added cases for JP and JNP in X86FreqProps.
2065    
2066      3. CFG
2067    
2068         CFG now has a bunch of methods for edge splitting and merging.
2069    
2070      4. Machine description.
2071    
2072         John's simplification of MLTREE_BASIS.fcond broke a few machine
2073         description things:
2074    
2075         rtl-build.{sig,sml} and hppa.mdl fixed.
2076    
2077         NOTE: the machine description stuff in the repository is still broken.
2078               Again, I can't put my fixes in because that involves
2079               changes to MLTREE.
2080    
2081    ----------------------------------------------------------------------
2082    Name: Matthias Blume
2083    Date: 2002/03/20 15:55:00 EST
2084    Tag: blume-20020320-kmp
2085    Description:
2086    
2087    Implemented Knuth-Morris-Pratt string matching in PreString and used
2088    it for String.isSubstring, Substring.isSubstring, and
2089    Substring.position.
2090    
2091    (Might need some stress-testing.  Simple examples worked fine.)
2092    
2093    ----------------------------------------------------------------------
2094    Name: Matthias Blume
2095    Date: 2002/03/19 16:37:00 EST
2096    Tag: blume-20020319-witnesses
2097    Description:
2098    
2099    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
2100    
2101    This implements a generic mechanism for changing constness qualifiers
2102    anywhere within big C types without resorting to outright "casts".
2103    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
2104    modify the constness at the outermost level.)
2105    The implementation of "convert" is based on the idea of "witness"
2106    values -- values that are not used by the operation but whose types
2107    "testify" to their applicability.  On the implementation side, "convert"
2108    is simply a projection (returning its second curried argument).  With
2109    cross-module inlining, it should not result in any machine code being
2110    generated.
2111    
2112    ----------------------------------------------------------------------
2113    Name: Matthias Blume
2114    Date: 2002/03/15 16:40:00 EST
2115    Tag: blume-20020315-basis
2116    Description:
2117    
2118    Provided (preliminary?) implementations for
2119    
2120      {String,Substring}.{concatWith,isSuffix,isSubstring}
2121    
2122    and
2123    
2124      Substring.full
2125    
2126    Those are in the Basis spec but they were missing in SML/NJ.
2127    
2128    ----------------------------------------------------------------------
2129    Name: Matthias Blume
2130    Date: 2002/03/14 21:30:00 EST
2131    Tag: blume-20020314-controls
2132    Description:
2133    
2134    Controls:
2135    ---------
2136    
2137    1. Factored out the recently-added Controls : CONTROLS stuff and put
2138       it into its own library $/controls-lib.cm.  The source tree for
2139       this is under src/smlnj-lib/Controls.
2140    
2141    2. Changed the names of types and functions in this interface, so they
2142       make a bit more "sense":
2143    
2144          module -> registry
2145          'a registry -> 'a group
2146    
2147    3. The interface now deals in ref cells only.  The getter/setter interface
2148       is (mostly) gone.
2149    
2150    4. Added a function that lets one register an already-existing ref cell.
2151    
2152    5. Made the corresponding modifications to the rest of the code so that
2153       everything compiles again.
2154    
2155    6. Changed the implementation of Controls.MLRISC back to something closer
2156       to the original.  In particular, this module (and therefore MLRISC)
2157       does not depend on Controls.  There now is some link-time code in
2158       int-sys.sml that registers the MLRISC controls with the Controls
2159       module.
2160    
2161    CM:
2162    ---
2163    
2164      * One can now specify the lambda-split aggressiveness in init.cmi.
2165    
2166    ----------------------------------------------------------------------
2167    Name: Allen Leung
2168    Date: 2002/03/13 17:30:00 EST
2169    Tag: leunga-20020313-x86-fp-unary
2170    Description:
2171    
2172    Bug fix for:
2173    
2174    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
2175    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
2176    > - fun f(x,(y,z)) = Real.~ y;
2177    > [autoloading]
2178    > [autoloading done]
2179    >       fchsl   (%eax), 184(%esp)
2180    > Error: MLRisc bug: X86MCEmitter.emitInstr
2181    >
2182    > uncaught exception Error
2183    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
2184    
2185    The problem was that the code generator did not generate any fp registers
2186    in this case, and the ra didn't know that it needed to run the X86FP phase to
2187    translate the pseudo fp instruction.   This only happened with unary fp
2188    operators in certain situations.
2189    
2190    ----------------------------------------------------------------------
2191    Name: Matthias Blume
2192    Date: 2002/03/13 14:00:00 EST
2193    Tag: blume-20020313-overload-etc
2194    Description:
2195    
2196    1. Added _overload as a synonym for overload for backward compatibility.
2197       (Control.overloadKW must be true for either version to be accepted.)
2198    
2199    2. Fixed bug in install script that caused more things to be installed
2200       than what was requested in config/targets.
2201    
2202    3. Made CM aware of the (_)overload construct so that autoloading
2203       works.
2204    
2205    ----------------------------------------------------------------------
2206    Name: Matthias Blume
2207    Date: 2002/03/12 22:03:00 EST
2208    Tag: blume-20020312-url
2209    Description:
2210    
2211    Forgot to update BOOT and srcarchiveurl.
2212    
2213    ----------------------------------------------------------------------
2214    Name: Matthias Blume
2215    Date: 2002/03/12 17:30:00 EST
2216    Tag: blume-20020312-version110392
2217    Description:
2218    
2219    Yet another version number bump (because of small changes to the
2220    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
2221    
2222    Changes:
2223    
2224      The new pid generation scheme described a few weeks ago was overly
2225      complicated.  I implemented a new mechanism that is simpler and
2226      provides a bit more "stability":  Once CM has seen a compilation
2227      unit, it keeps its identity constant (as long as you do not delete
2228      those crucial CM/GUID/* files).  This means that when you change
2229      an interface, compile, then go back to the old interface, and
2230      compile again, you arrive at the original pid.
2231    
2232      There now also is a mechanism that instructs CM to use the plain
2233      environment hash as a module's pid (effectively making its GUID
2234      the empty string).  For this, "noguid" must be specified as an
2235      option to the .sml file in question within its .cm file.
2236      This is most useful for code that is being generated by tools such
2237      as ml-nlffigen (because during development programmers tend to
2238      erase the tool's entire output directory tree including CM's cached
2239      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
2240      revert to the old, broken behavior of SML/NJ, but in specific cases
2241      where there is no danger of interface confusion, its use is ok
2242      (I think).
2243    
2244      ml-nlffigen by default generates "noguid" annotations.  They can be
2245      turned off by specifying -guid in its command line.
2246    
2247    ----------------------------------------------------------------------
2248    Name: Lal George
2249    Date: 2002/03/12 12 14:42:36 EST
2250    Tag: george-20020312-frequency-computation
2251    Description:
2252    
2253    Integrated jump chaining and static block frequency into the
2254    compiler. More details and numbers later.
2255    
2256    ----------------------------------------------------------------------
2257    Name: Lal George
2258    Date: 2002/03/11 11 22:38:53 EST
2259    Tag: george-20020311-jump-chain-elim
2260    Description:
2261    
2262    Tested the jump chain elimination on all architectures (except the
2263    hppa).  This is on by default right now and is profitable for the
2264    alpha and x86, however, it may not be profitable for the sparc and ppc
2265    when compiling the compiler.
2266    
2267    The gc test will typically jump to a label at the end of the cluster,
2268    where there is another jump to an external cluster containing the actual
2269    code to invoke gc. This is to allow factoring of common gc invocation
2270    sequences. That is to say, we generate:
2271    
2272            f:
2273               testgc
2274               ja   L1      % jump if above to L1
2275    
2276            L1:
2277               jmp L2
2278    
2279    
2280    After jump chain elimination the 'ja L1' instructions is converted to
2281    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
2282    up being implemented in their long form (if L2 is far away) using:
2283    
2284            jbe     L3      % jump if below or equal to L3
2285            jmp     L2
2286         L3:
2287            ...
2288    
2289    
2290    For large compilation units L2  may be far away.
2291    
2292    
2293    ----------------------------------------------------------------------
2294    Name: Matthias Blume
2295    Date: 2002/03/11 13:30:00 EST
2296    Tag: blume-20020311-mltreeeval
2297    Description:
2298    
2299    A functor parameter was missing.
2300    
2301    ----------------------------------------------------------------------
2302    Name: Allen Leung
2303    Date: 2002/03/11 10:30:00 EST
2304    Tag: leunga-20020311-runtime-string0
2305    Description:
2306    
2307       The representation of the empty string now points to a
2308    legal null terminated C string instead of unit.  It is now possible
2309    to convert an ML string into C string with InlineT.CharVector.getData.
2310    This compiles into one single machine instruction.
2311    
2312    ----------------------------------------------------------------------
2313    Name: Allen Leung
2314    Date: 2002/03/10 23:55:00 EST
2315    Tag: leunga-20020310-x86-call
2316    Description:
2317    
2318       Added machine generation for CALL instruction (relative displacement mode)
2319    
2320    ----------------------------------------------------------------------
2321    Name: Matthias Blume
2322    Date: 2002/03/08 16:05:00
2323    Tag: blume-20020308-entrypoints
2324    Description:
2325    
2326    Version number bumped to 110.39.1.  NEW BOOTFILES!
2327    
2328    Entrypoints: non-zero offset into a code object where execution should begin.
2329    
2330    - Added the notion of an entrypoint to CodeObj.
2331    - Added reading/writing of entrypoint info to Binfile.
2332    - Made runtime system bootloader aware of entrypoints.
2333    - Use the address of the label of the first function given to mlriscGen
2334      as the entrypoint.  This address is currently always 0, but it will
2335      not be 0 once we turn on block placement.
2336    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
2337      with entry points) from mlriscGen.
2338    
2339    ----------------------------------------------------------------------
2340    Name: Allen Leung
2341    Date: 2002/03/07 20:45:00 EST
2342    Tag: leunga-20020307-x86-cmov
2343    Description:
2344    
2345       Bug fixes for CMOVcc on x86.
2346    
2347       1. Added machine code generation for CMOVcc
2348       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
2349       3. CMOVcc cannot have an immediate operand as argument.
2350    
2351    ----------------------------------------------------------------------
2352    Name: Matthias Blume
2353    Date: 2002/03/07 16:15:00 EST
2354    Tag: blume-20020307-controls
2355    Description:
2356    
2357    This is a very large but mostly boring patch which makes (almost)
2358    every tuneable compiler knob (i.e., pretty much everything under
2359    Control.* plus a few other things) configurable via both the command
2360    line and environment variables in the style CM did its configuration
2361    until now.
2362    
2363    Try starting sml with '-h' (or, if you are brave, '-H')
2364    
2365    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
2366    implements the underlying generic mechanism.
2367    
2368    The interface to some of the existing such facilities has changed somewhat.
2369    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
2370    (The getFoo interface is still there for backward-compatibility, but its
2371    use is deprecated.)
2372    
2373    The ml-build script passes -Cxxx=yyy command-line arguments through so
2374    that one can now twiddle the compiler settings when using this "batch"
2375    compiler.
2376    
2377    TODO items:
2378    
2379    We should go through and throw out all controls that are no longer
2380    connected to anything.  Moreover, we should go through and provide
2381    meaningful (and correct!) documentation strings for those controls
2382    that still are connected.
2383    
2384    Currently, multiple calls to Controls.new are accepted (only the first
2385    has any effect).  Eventually we should make sure that every control
2386    is being made (via Controls.new) exactly once.  Future access can then
2387    be done using Controls.acc.
2388    
2389    Finally, it would probably be a good idea to use the getter-setter
2390    interface to controls rather than ref cells.  For the time being, both
2391    styles are provided by the Controls module, but getter-setter pairs are
2392    better if thread-safety is of any concern because they can be wrapped.
2393    
2394    *****************************************
2395    
2396    One bug fix: The function blockPlacement in three of the MLRISC
2397    backpatch files used to be hard-wired to one of two possibilities at
2398    link time (according to the value of the placementFlag).  But (I
2399    think) it should rather sense the flag every time.
2400    
2401    *****************************************
2402    
2403    Other assorted changes (by other people who did not supply a HISTORY entry):
2404    
2405    1. the cross-module inliner now works much better (Monnier)
2406    2. representation of weights, frequencies, and probabilities in MLRISC
2407       changed in preparation of using those for weighted block placement
2408       (Reppy, George)
2409    
2410    ----------------------------------------------------------------------
2411    Name: Lal George
2412    Date: 2002/03/07 14:44:24 EST 2002
2413    Tag: george-20020307-weighted-block-placement
2414    
2415    Tested the weighted block placement optimization on all architectures
2416    (except the hppa) using AMPL to generate the block and edge frequencies.
2417    Changes were required in the machine properties to correctly
2418    categorize trap instructions. There is an MLRISC flag
2419    "weighted-block-placement" that can be used to enable weighted block
2420    placement, but this will be ineffective without block/edge
2421    frequencies (coming soon).
2422    
2423    
2424    ----------------------------------------------------------------------
2425    Name: Lal George
2426    Date: 2002/03/05 17:24:48 EST
2427    Tag: george-20020305-linkage-cluster
2428    
2429    In order to support the block placement optimization, a new cluster
2430    is generated as the very first cluster (called the linkage cluster).
2431    It contains a single jump to the 'real' entry point for the compilation
2432    unit. Block placement has no effect on the linkage cluster itself, but
2433    all the other clusters  have full freedom in the manner in which they
2434    reorder blocks or functions.
2435    
2436    On the x86 the typical linkage code that is generated is:
2437       ----------------------
2438            .align 2
2439       L0:
2440            addl    $L1-L0, 72(%esp)
2441            jmp     L1
2442    
2443    
2444            .align  2
2445       L1:
2446       ----------------------
2447    
2448    72(%esp) is the memory location for the stdlink register. This
2449    must contain the address of the CPS function being called. In the
2450    above example, it contains the address of  L0; before
2451    calling L1 (the real entry point for the compilation unit), it
2452    must contain the address for L1, and hence
2453    
2454            addl $L1-L0, 72(%esp)
2455    
2456    I have tested this on all architectures except the hppa.The increase
2457    in code size is of course negligible
2458    
2459    ----------------------------------------------------------------------
2460    Name: Allen Leung
2461    Date: 2002/03/03 13:20:00 EST
2462    Tag: leunga-20020303-mlrisc-tools
2463    
2464      Added #[ ... ] expressions to mlrisc tools
2465    
2466    ----------------------------------------------------------------------
2467    Name: Matthias Blume
2468    Date: 2002/02/27 12:29:00 EST
2469    Tag: blume-20020227-cdebug
2470    Description:
2471    
2472    - made types in structure C and C_Debug to be equal
2473    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
2474    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
2475    
2476    ----------------------------------------------------------------------
2477    Name: Matthias Blume
2478    Date: 2002/02/26 12:00:00 EST
2479    Tag: blume-20020226-ffi
2480    Description:
2481    
2482    1. Fixed a minor bug in CM's "noweb" tool:
2483       If numbering is turned off, then truly don't number (i.e., do not
2484       supply the -L option to noweb).  The previous behavior was to supply
2485       -L'' -- which caused noweb to use the "default" line numbering scheme.
2486       Thanks to Chris Richards for pointing this out (and supplying the fix).
2487    
2488    2. Once again, I reworked some aspects of the FFI:
2489    
2490       A. The incomplete/complete type business:
2491    
2492       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
2493         gone!
2494       - ML types representing an incomplete type are now *equal* to
2495         ML types representing their corresponding complete types (just like
2496         in C).  This is still safe because ml-nlffigen will not generate
2497         RTTI for incomplete types, nor will it generate functions that
2498         require access to such RTTI.   But when ML code generated from both
2499         incomplete and complete versions of the C type meet, the ML types
2500         are trivially interoperable.
2501    
2502         NOTE:  These changes restore the full generality of the translation
2503         (which was previously lost when I eliminated functorization)!
2504    
2505       B. Enum types:
2506    
2507       - Structure C now has a type constructor "enum" that is similar to
2508         how the "su" constructor works.  However, "enum" is not a phantom
2509         type because each "T enum" has values (and is isomorphic to
2510         MLRep.Signed.int).
2511       - There are generic access operations for enum objects (using
2512         MLRep.Signed.int).
2513       - ml-nlffigen will generate a structure E_foo for each "enum foo".
2514         * The structure contains the definition of type "mlrep" (the ML-side
2515         representation type of the enum).  Normally, mlrep is the same
2516         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
2517         then mlrep will be defined as a datatype -- thus facilitating
2518         pattern matching on mlrep values.
2519         ("-ec" will be suppressed if there are duplicate values in an
2520          enumeration.)
2521         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
2522         will be generated for each C enum constant xxx.
2523         * Conversion functions m2i and i2m convert between mlrep and
2524         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
2525         * Coversion functions c and ml convert between mlrep and "tag enum".
2526         * Access functions (get/set) fetch and store mlrep values.
2527       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
2528         enumerations are merged into one single enumeration represented by
2529         structure E_'.
2530    
2531    ----------------------------------------------------------------------
2532    Name: Allen Leung
2533    Date: 2002/02/25 04:45:00 EST
2534    Tag: leunga-20020225-cps-spill
2535    
2536    This is a new implementation of the CPS spill phase.
2537    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
2538    In case of problems, replace it with the old file spill.sml
2539    
2540    The current compiler runs into some serious performance problems when
2541    constructing a large record.  This can happen when we try to compile a
2542    structure with many items.  Even a very simple structure like the following
2543    makes the compiler slow down.
2544    
2545        structure Foo = struct
2546           val x_1 = 0w1 : Word32.int
2547           val x_2 = 0w2 : Word32.int
2548           val x_3 = 0w3 : Word32.int
2549           ...
2550           val x_N = 0wN : Word32.int
2551        end
2552    
2553    The following table shows the compile time, from N=1000 to N=4000,
2554    with the old compiler:
2555    
2556    N
2557    1000   CPS 100 spill                           0.04u  0.00s  0.00g
2558           MLRISC ra                               0.06u  0.00s  0.05g
2559              (spills = 0 reloads = 0)
2560           TOTAL                                   0.63u  0.07s  0.21g
2561    
2562    1100   CPS 100 spill                           8.25u  0.32s  0.64g
2563           MLRISC ra                               5.68u  0.59s  3.93g
2564              (spills = 0 reloads = 0)
2565           TOTAL                                   14.71u  0.99s  4.81g
2566    
2567    1500   CPS 100 spill                           58.55u  2.34s  1.74g
2568           MLRISC ra                               5.54u  0.65s  3.91g
2569              (spills = 543 reloads = 1082)
2570           TOTAL                                   65.40u  3.13s  6.00g
2571    
2572    2000   CPS 100 spill                           126.69u  4.84s  3.08g
2573           MLRISC ra                               0.80u  0.10s  0.55g
2574              (spills = 42 reloads = 84)
2575           TOTAL                                   129.42u  5.10s  4.13g
2576    
2577    3000   CPS 100 spill                           675.59u  19.03s  11.64g
2578           MLRISC ra                               2.69u  0.27s  1.38g
2579              (spills = 62 reloads = 124)
2580           TOTAL                                   682.48u  19.61s  13.99g
2581    
2582    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
2583           MLRISC ra                               4.96u  0.27s  2.72g
2584              (spills = 85 reloads = 170)
2585           TOTAL                                   2375.26u  57.21s  48.00g
2586    
2587    As you can see the old cps spill module suffers from some serious
2588    performance problem.  But since I cannot decipher the old code fully,
2589    instead of patching the problems up, I'm reimplementing it
2590    with a different algorithm.  The new code is more modular,
2591    smaller when compiled, and substantially faster
2592    (O(n log n) time and O(n) space).  Timing of the new spill module:
2593    
2594    4000  CPS 100 spill                           0.02u  0.00s  0.00g
2595          MLRISC ra                               0.25u  0.02s  0.15g
2596             (spills=1 reloads=3)
2597          TOTAL                                   7.74u  0.34s  1.62g
2598    
2599    Implementation details:
2600    
2601    As far as I can tell, the purpose of the CPS spill module is to make sure the
2602    number of live variables at any program point (the bandwidth)
2603    does not exceed a certain limit, which is determined by the
2604    size of the spill area.
2605    
2606    When the bandwidth is too large, we decrease the register pressure by
2607    packing live variables into spill records.  How we achieve this is
2608    completely different than what we did in the old code.
2609    
2610    First, there is something about the MLRiscGen code generator
2611    that we should be aware of:
2612    
2613    o MLRiscGen performs code motion!
2614    
2615       In particular, it will move floating point computations and
2616       address computations involving only the heap pointer to
2617       their use sites (if there is only a single use).
2618       What this means is that if we have a CPS record construction
2619       statement
2620    
2621           RECORD(k,vl,w,e)
2622    
2623       we should never count the new record address w as live if w
2624       has only one use (which is often the case).
2625    
2626       We should do something similar to floating point, but the transformation
2627       there is much more complex, so I won't deal with that.
2628    
2629    Secondly, there are now two new cps primops at our disposal:
2630    
2631     1. rawrecord of record_kind option
2632        This pure operator allocates some uninitialized storage from the heap.
2633        There are two forms:
2634    
2635         rawrecord NONE [INT n]  allocates a tagless record of length n
2636         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
2637                                     and initializes the tag.
2638    
2639     2. rawupdate of cty
2640          rawupdate cty (v,i,x)
2641          Assigns to x to the ith component of record v.
2642          The storelist is not updated.
2643    
2644    We use these new primops for both spilling and increment record construction.
2645    
2646     1. Spilling.
2647    
2648        This is implemented with a linear scan algorithm (but generalized
2649        to trees).  The algorithm will create a single spill record at the
2650        beginning of the cps function and use rawupdate to spill to it,
2651        and SELECT or SELp to reload from it.  So both spills and reloads
2652        are fine-grain operations.  In contrast, in the old algorithm
2653        "spills" have to be bundled together in records.
2654    
2655        Ideally, we should sink the spill record construction to where
2656        it is needed.  We can even split the spill record into multiple ones
2657        at the places where they are needed.  But CPS is not a good
2658        representation for global code motion, so I'll keep it simple and
2659        am not attempting this.
2660    
2661     2. Incremental record construction (aka record splitting).
2662    
2663        Long records with many component values which are simulatenously live
2664        (recall that single use record addresses are not considered to
2665         be live) are constructed with rawrecord and rawupdate.
2666        We allocate space on the heap with rawrecord first, then gradually
2667        fill it in with rawupdate.  This is the technique suggested to me
2668        by Matthias.
2669    
2670        Some restrictions on when this is applicable:
2671        1. It is not a VECTOR record.  The code generator currently does not handle
2672           this case. VECTOR record uses double indirection like arrays.
2673        2. All the record component values are defined in the same "basic block"
2674           as the record constructor.  This is to prevent speculative
2675           record construction.
2676    
2677    ----------------------------------------------------------------------
2678    Name: Allen Leung
2679    Date: 2002/02/22 01:02:00 EST
2680    Tag: leunga-20020222-mlrisc-tools
2681    
2682    Minor bug fixes in the parser and rewriter
2683    
2684    ----------------------------------------------------------------------
2685    Name: Allen Leung
2686    Date: 2002/02/21 20:20:00 EST
2687    Tag: leunga-20020221-peephole
2688    
2689    Regenerated the peephole files.  Some contained typos in the specification
2690    and some didn't compile because of pretty printing bugs in the old version
2691    of 'nowhere'.
2692    
2693    ----------------------------------------------------------------------
2694    Name: Allen Leung
2695    Date: 2002/02/19 20:20:00 EST
2696    Tag: leunga-20020219-mlrisc-tools
2697    Description:
2698    
2699       Minor bug fixes to the mlrisc-tools library:
2700    
2701       1.  Fixed up parsing colon suffixed keywords
2702       2.  Added the ability to shut the error messages up
2703       3.  Reimplemented the pretty printer and fixed up/improved
2704           the pretty printing of handle and -> types.
2705       4.  Fixed up generation of literal symbols in the nowhere tool.
2706       5.  Added some SML keywords to to sml.sty
2707    
2708    ----------------------------------------------------------------------
2709    Name: Matthias Blume
2710    Date: 2002/02/19 16:20:00 EST
2711    Tag: blume-20020219-cmffi
2712    Description:
2713    
2714    A wild mix of changes, some minor, some major:
2715    
2716    * All C FFI-related libraries are now anchored under $c:
2717        $/c.cm      --> $c/c.cm
2718        $/c-int.cm  --> $c/internals/c-int.cm
2719        $/memory.cm --> $c/memory/memory.cm
2720    
2721    * "make" tool (in CM) now treats its argument pathname slightly
2722      differently:
2723        1. If the native expansion is an absolute name, then before invoking
2724           the "make" command on it, CM will apply OS.Path.mkRelative
2725           (with relativeTo = OS.FileSys.getDir()) to it.
2726        2. The argument will be passed through to subsequent phases of CM
2727           processing without "going native".  In particular, if the argument
2728           was an anchored path, then "make" will not lose track of that anchor.
2729    
2730    * Compiler backends now "know" their respective C calling conventions
2731      instead of having to be told about it by ml-nlffigen.  This relieves
2732      ml-nlffigen from one of its burdens.
2733    
2734    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
2735    
2736    * Export C_DEBUG and C_Debug from $c/c.cm.
2737    
2738    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
2739      subtyping relationship between incomplete pointers and their complete
2740      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
2741      with the convention of instantiating 'o with ('t, 'c) obj whenever
2742      the pointer target type is complete.  In the incomplete case, 'o
2743      will be instantiated with some "'c iobj" -- a type obtained by
2744      using one of the functors PointerToIncompleteType or PointerToCompleteType.
2745    
2746      Operations that work on both incomplete and complete pointer types are
2747      typed as taking an 'o ptr while operations that require the target to
2748      be known are typed as taking some ('t, 'c) obj ptr.
2749    
2750      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
2751      where void is an eqtype without any values.  This makes it possible
2752      to work on voidptr values using functions meant to operate on light
2753      incomplete pointers.
2754    
2755    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
2756      been vastly simplified.
2757    
2758    ----------------------------------------------------------------------
2759    Name: Matthias Blume
2760    Date: 2002/02/19 10:48:00 EST
2761    Tag: blume-20020219-pqfix
2762    Description:
2763    
2764    Applied Chris Okasaki's bug fix for priority queues.
2765    
2766    ----------------------------------------------------------------------
2767    Name: Matthias Blume
2768    Date: 2002/02/15 17:05:00
2769    Tag: Release_110_39
2770    Description:
2771    
2772    Last-minute retagging is becoming a tradition... :-(
2773    
2774    This is the working release 110.39.
2775    
2776    ----------------------------------------------------------------------
2777    Name: Matthias Blume
2778    Date: 2002/02/15 16:00:00 EST
2779    Tag: Release_110_39-orig
2780    Description:
2781    
2782    Working release 110.39.  New bootfiles.
2783    
2784    (Update: There was a small bug in the installer so it wouldn't work
2785    with all shells.  So I retagged. -Matthias)
2786    
2787    ----------------------------------------------------------------------
2788    Name: Matthias Blume
2789    Date: 2002/02/15 14:17:00 EST
2790    Tag: blume-20020215-showbindings
2791    Description:
2792    
2793    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
2794    the latter can be useful for exploring what bindings are available at
2795    the interactive prompt.  (The first function returns only the list
2796    of symbols that are really bound, the second prints those but also the
2797    ones that CM's autoloading mechanism knows about.)
2798    
2799    ----------------------------------------------------------------------
2800    Name: Matthias Blume
2801    Date: 2002/02/15 12:08:00 EST
2802    Tag: blume-20020215-iptrs
2803    Description:
2804    
2805    Two improvements to ml-nlffigen:
2806    
2807      1. Write files only if they do not exist or if their current contents
2808         do not coincide with what's being written.  (That is, avoid messing
2809         with the time stamps unless absolutely necessary.)
2810    
2811      2. Implement a "repository" mechanism for generated files related
2812         to "incomplete pointer types".   See the README file for details.
2813    
2814    ----------------------------------------------------------------------
2815    Name: Matthias Blume
2816    Date: 2002/02/14 11:50:00 EST
2817    Tag: blume-20020214-quote
2818    Description:
2819    
2820    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
2821    because of the new and improved tag generation scheme.  (Thanks to Allen
2822    Leung for pointing it out.)
2823    
2824    ----------------------------------------------------------------------
2825    Name: Lal George
2826    Date: 2002/02/14 09:55:27 EST 2002
2827    Tag: george-20020214-isabelle-bug
2828    Description:
2829    
2830    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
2831    of Isabelle on the x86.
2832    
2833    From Allen:
2834    -----------
2835     I've found the problem:
2836    
2837         in ra-core.sml, I use the counter "blocked" to keep track of the
2838         true number of elements in the freeze queue.  When the counter goes
2839         to zero, I skip examining the queue.  But I've messed up the
2840         bookkeeping in combine():
2841    
2842             else ();
2843             case !ucol of
2844               PSEUDO => (if !cntv > 0 then
2845                     (if !cntu > 0 then blocked := !blocked - 1 else ();
2846                                        ^^^^^^^^^^^^^^^^^^^^^^^
2847                      moveu := mergeMoveList(!movev, !moveu)
2848                     )
2849                  else ();
2850    
2851         combine() is called to coalesce two nodes u and v.
2852         I think I was thinking that if the move counts of u and v are both
2853         greater than zero then after they are coalesced then one node is
2854         removed from the freeze queue.  Apparently I was thinking that
2855         both u and v are of low degree, but that's clearly not necessarily true.
2856    
2857    
2858    02/12/2002:
2859        Here's the patch.  HOL now compiles.
2860    
2861        I don't know how this impact on performance (compile
2862        time or runtime).  This bug caused the RA (especially on the x86)
2863        to go thru the potential spill phase when there are still nodes on the
2864        freeze queue.
2865    
2866    
2867    
2868    
2869    ----------------------------------------------------------------------
2870    Name: Matthias Blume
2871    Date: 2002/02/13 22:40:00 EST
2872    Tag: blume-20020213-fptr-rtti
2873    Description:
2874    
2875    Fixed a bug in ml-nlffigen that was introduced with one of the previous
2876    updates.
2877    
2878    ----------------------------------------------------------------------
2879    Name: Matthias Blume
2880    Date: 2002/02/13 16:41:00 EST
2881    Tag: blume-20020213-cmlpq
2882    Description:
2883    
2884    Added new priority queue export symbols (which have just been added to
2885    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
2886    not compile and the installer would choke.)
2887    
2888    ----------------------------------------------------------------------
2889    Name: Matthias Blume
2890    Date: 2002/02/13 16:15:00 EST
2891    Tag: blume-20020213-various
2892    Description:
2893    
2894    1. More tweaks to ml-nlffigen:
2895    
2896       - better internal datastructures (resulting in slight speedup)
2897       - "-match" option requires exact match
2898       - "localized" gensym counters (untagged structs/unions nested within
2899         other structs/unions or within typedefs get a fresh counter; their
2900         tag will be prefixed by a concatenation of their parents' tags)
2901       - bug fixes (related to calculation of transitive closure of types
2902         to be included in the output)
2903    
2904    2. Minor Basis updates:
2905    
2906       - added implementations for List.collate and Option.app
2907    
2908    ----------------------------------------------------------------------
2909    Name: Matthias Blume
2910    Date: 2002/02/11 15:55:00 EST
2911    Tag: blume-20020211-gensym
2912    Description:
2913    
2914    Added a "-gensym" option to command line of ml-nlffigen.  This can be
2915    used to specify a "stem" -- a string that is inserted in all "gensym'd"
2916    names (ML structure names that correspond to unnamed C structs, unions,
2917    and enums), so that separate runs of ml-nlffigen do not clash.
2918    
2919    ----------------------------------------------------------------------
2920    Name: Matthias Blume
2921    Date: 2002/02/11 12:05:00 EST
2922    Tag: blume-20020211-gensml
2923    Description:
2924    
2925    A quick fix for a problem with GenSML (in the pgraph-util library):
2926    Make generation of toplevel "local" optional.  (Strictly speaking,
2927    signature definitions within "local" are not legal SML.)
2928    
2929    Other than that: updates to INSTALL and cm/TODO.
2930    
2931    ----------------------------------------------------------------------
2932    Name: Matthias Blume
2933    Date: 2002/02/08 15:00:00 EST
2934    Tag: blume-20020208-uniquepid
2935    Description:
2936    
2937    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
2938    
2939    1. The installer (config/install.sh) has gotten smarter:
2940    
2941         - Configuration options are a bit easier to specify now
2942           (in config/targets).
2943         - Bug in recognizing .tar.bz2 files fixed.
2944         - Installer automatically resolves dependencies between
2945           configuration options (e.g., if you ask for eXene, you will
2946           also get cml -- regardless whether you asked for it or not).
2947         - Installer can run in "quieter mode" by setting the environment
2948           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
2949           "completely silent", though.
2950         - Build HashCons library as part of smlnj-lib.
2951    
2952    2. A new scheme for assigning persistent identifiers to compilation
2953       units (and, by extension, to types etc.) has been put into place.
2954       This fixes a long-standing bug where types and even dynamic values
2955       can get internally confused, thereby compromising type safety
2956       (abstraction) and dynamic correctness.  See
2957    
2958         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
2959    
2960       for an example of how things could go wrong until now.
2961    
2962       The downside of the new scheme is that pids are not quite as
2963       persistent as they used to be: CM will generate a fresh pid
2964       for every compilation unit that it thinks it sees for the first
2965       time.  That means that if you compile starting from a clean, fresh
2966       source tree at two different times, you end up with different
2967       binaries.
2968    
2969       Cutoff recompilation, however, has not been compromised because
2970       CM keeps pid information in special caches between runs.
2971    
2972    ----------------------------------------------------------------------
2973    Name: Lal George
2974    Date: 2002/02/07 15:34:13 EST 2002
2975    Tag: <none>
2976    Description:
2977    
2978    Compilers that generate assembly code may produce  global labels
2979    whose value is resolved at link time. The various peephole optimization
2980    modules did not take this in account.
2981    
2982    TODO. The Labels.addrOf function should really return an option
2983    type so that clients are forced to deal with this issue, rather
2984    than an exception being raised.
2985    
2986    ----------------------------------------------------------------------
2987    Name: Lal George
2988    Date: 2002/02/06 13:55:02 EST
2989    Tag: george-20020206-ra-breakup
2990    Description:
2991    
2992    1. A bug fix from Allen.
2993    
2994        A typo causes extra fstp %st(0)'s to be generated at compensation
2995        edges, which might cause stack underflow traps at runtime.  This
2996        occurs in fft where there are extraneous fstps right before the 'into'
2997        trap instruction (in this case they are harmless since none of the
2998        integers overflow.)
2999    
3000    2. Pulled out various utility modules that were embedded in the modules
3001       of the register allocator. I need these modules for other purposes, but
3002       they are not complete enough to put into a library (just yet).
3003    ----------------------------------------------------------------------
3004    Name: Matthias Blume
3005    Date: 2002/01/31 16:05:00 EST
3006    Tag: blume-20020131-sparc-ccalls
3007    Description:
3008    
3009    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
3010       of extra stack space by mistake.  Fixed.
3011    
3012    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
3013    
3014    ----------------------------------------------------------------------
3015    Name: Allen Leung
3016    Date: 2002/01/30
3017    Tag: leunga-20020130-nowhere-bug-fix
3018    Description:
3019    
3020       MLRISC bug fixes:
3021       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
3022       2. MachineInt.fromString was negating its value.
3023    
3024    ----------------------------------------------------------------------
3025    Name: Matthias Blume
3026    Date: 2002/01/29
3027    Tag: blume-20020129-INSTALL
3028    Description:
3029    
3030    - Added somewhat detailed installation instructions (file INSTALL).
3031    - Fixed curl-detection bug in config/install.sh.
3032    - It is now possible to select the URL getter using the URLGETTER
3033      environment variable:
3034    
3035          not set / "unknown"      --> automatic detection (script tries wget,
3036                                       curl, and lynx)
3037          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
3038                                       how to properly invoke them)
3039          other                    --> use $URLGETTER directly, it must take
3040                                       precisely two command-line arguments
3041                                       (source URL and destination file name)
3042    
3043    ----------------------------------------------------------------------
3044    Name: Matthias Blume
3045    Date: 2002/01/28
3046    Tag: blume-20020128-sparc-ccalls
3047    Description:
3048    
3049    - Fixed problem with calculation of "used" registers in sparc-c-calls.
3050    - Make use of the allocParam argument in sparc-c-calls.
3051    
3052    ----------------------------------------------------------------------
3053    Name: Matthias Blume
3054    Date: 2002/01/28
3055    Tag: blume-20020128-allocParam
3056    Description:
3057    
3058    John Reppy:  Changes c-calls API to accept client-callback for
3059    allocating extra stack space.
3060    me: Corresponding changes to mlriscGen (using a dummy argument that
3061        does not change the current behavior).
3062    
3063    ----------------------------------------------------------------------
3064    Name: Matthias Blume
3065    Date: 2002/01/28 12:00:00
3066    Tag: Release_110_38
3067    Description:
3068    
3069    This time for real!!!
3070    
3071    ----------------------------------------------------------------------
3072    Name: Matthias Blume
3073    Date: 2002/01/28 10:56:00 EST
3074    Tag: blume-20020128-retraction
3075    Description:
3076    
3077    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
3078       with blume-Release_110_38-retracted.)
3079    
3080    1. Fixed a problem with incorrect rounding modes in real64.sml.
3081       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
3082    
3083    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
3084       and enums fixed.  The naming of corresponding ML identifiers should
3085       now be consistent again.
3086    
3087    ----------------------------------------------------------------------
3088    Name: Allen Leung
3089    Date: 2002/01/27
3090    Tag: leunga-20020127-nowhere
3091    Description:
3092    
3093       Added a target called nowhere in the configuration scripts.
3094       Enabling this will build the MLRISC 'nowhere' tool (for translating
3095       programs with where-clauses into legal SML code) during installation.
3096    
3097    ----------------------------------------------------------------------
3098    Name: Matthias Blume
3099    Date: 2002/01/25 21:27:00 EST
3100    Tag: blume-Release_110_38-retracted
3101    Description:
3102    
3103    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
3104    
3105    README will be added later.
3106    
3107    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
3108    (Release_110_38) removed.  Reason: Last-minute bug fixes.
3109    
3110    ----------------------------------------------------------------------
3111    Name: Matthias Blume
3112    Date: 2002/01/25
3113    Tag: blume-20020125-ffi
3114    Description:
3115    
3116    A large number of tweaks and improvements to ml-nlffi-lib and
3117    ml-nlffigen:
3118    
3119       - ML represenation types have been streamlined
3120       - getter and setter functions work with concrete values, not abstract
3121         ones where possible
3122       - ml-nlffigen command line more flexible (see README file there)
3123       - some bugs have been fixed (hopefully)
3124    
3125    ----------------------------------------------------------------------
3126    Name: Lal George
3127    Date: 2002/01/24
3128    Tag: george-20020124-risc-ra-interface
3129    Description:
3130    
3131       There is a dramatic simplification in the interface to the
3132       register allocator for RISC architectures as a result of making
3133       parallel copy instructions explicit.
3134    
3135    ----------------------------------------------------------------------
3136    Name: Matthias Blume
3137    Date: 2002/01/22
3138    Tag: blume-20020122-x86-ccalls
3139    Description:
3140    
3141    Bug fix for c-calls on x86 (having to do with how char- and
3142    short-arguments are being handled).
3143    
3144    ----------------------------------------------------------------------
3145    Name: Matthias Blume
3146    Date: 2002/01/21
3147    Tag: blume-20020121-ff
3148    Description:
3149    
3150    Another day of fiddling with the FFI...
3151    
3152    1. Bug fix/workaround:  CKIT does not complain about negative array
3153       dimensions, so ml-nlffigen has to guard itself against this possibility.
3154       (Otherwise a negative dimension would send it into an infinite loop.)
3155    
3156    2. Some of the abstract types (light objects, light pointers, most "base"
3157       types) in structure C are now eqtypes.
3158    
3159    3. Added constructors and test functions for NULL function pointers.
3160    
3161    ----------------------------------------------------------------------
3162    Name: Matthias Blume
3163    Date: 2002/01/18
3164    Tag: blume-20020118-ready-for-new-release
3165    Description:
3166    
3167    Made config/srcarchiveurl point to a new place.  (Will provide boot
3168    files shortly.)
3169    
3170    Maybe we christen this to be 110.38?
3171    
3172    ----------------------------------------------------------------------
3173    Name: Matthias Blume
3174    Date: 2002/01/18
3175    Tag: blume-20020118-more-ffifiddle
3176    Description:
3177    
3178    Today's FFI fiddling:
3179    
3180      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
3181        C.Set.* that use concrete (MLRep.*) arguments and results instead
3182        of abstract ones.
3183    
3184      - Provided word-style bit operations etc. for "int" representation
3185        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
3186        and Long.
3187    
3188    ----------------------------------------------------------------------
3189    Name: Matthias Blume
3190    Date: 2002/01/18
3191    Tag: blume-20020118-use-x86-fp
3192    Description:
3193    
3194    Now that x86-fast-fp seems to be working, I turned it back on again
3195    by default.  (Seems to work fine now, even with the FFI.)
3196    
3197    Other than that, I added some documentation about the FFI to
3198    src/ml-nlffigen/README and updated the FFI test examples in
3199    src/ml-nlffi-lib/Tests/*.
3200    
3201    ----------------------------------------------------------------------
3202    Name: Allen Leung
3203    Date: 2002/01/17
3204    Tag: leunga-20020117-x86-fast-fp-call
3205    Description:
3206    
3207       1. Fixed a problem with handling return fp values when x86's fast fp
3208          mode is turned on.
3209    
3210       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
3211          of %f32.
3212    
3213       3. Added a constructor INT32lit to the ast of MLRISC tools.
3214    
3215    ----------------------------------------------------------------------
3216    Name: Matthias Blume
3217    Date: 2002/01/16
3218    Tag: blume-20020116-ffifiddle
3219    Description:
3220    
3221    More fiddling with the FFI interface:
3222    
3223     - Make constness 'c instead of rw wherever possible.  This eliminates
3224       the need for certain explicit coercions.  (However, due to ML's
3225       value polymorphism, there will still be many cases where explicit
3226       coercions are necessary.  Phantom types are not the whole answer
3227       to modeling a subtyping relationship in ML.)
3228    
3229     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
3230    
3231     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
3232       clumsy workaround for varargs problem.)
3233    
3234    ----------------------------------------------------------------------
3235    Name: Lal George
3236    Date: 2002/01/15
3237    Tag: <none>
3238    Description:
3239    
3240    1. Since COPY instructions are no longer native to the architecture,
3241       a generic functor can be used to implement the expandCopies function.
3242    
3243    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
3244       TEXT segment.
3245    
3246    ----------------------------------------------------------------------
3247    Name: Matthias Blume
3248    Date: 2002/01/15
3249    Tag: blume-20020115-ffiupdates
3250    Description:
3251    
3252    1. Fix for bug resulting in single-precision float values being returned
3253       incorrectly from FFI calls.
3254    
3255    2. Small modifications to C FFI API:
3256    
3257        - memory-allocation routines return straight objects (no options)
3258          and raise an exception in out-of-memory situations
3259        - unsafe extensions to cast between function pointers and pointers
3260          from/to ints
3261        - added structure C_Debug as an alternative to structure C where
3262          pointer-dereferencing (|*| and |*!) always check for null-pointers
3263        - added open_lib' to DynLinkage;  open_lib' works like open_lib
3264          but also takes a (possibly empty) list of existing library handles
3265          that the current library depends on
3266    
3267    ----------------------------------------------------------------------
3268    Name: Matthias Blume
3269    Date: 2002/01/10
3270    Tag: blume-20020110-newffigen
3271    Description:
3272    
3273    1. Updates to portable graph code.
3274    
3275    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
3276       more scalable now so that even huge interfaces such as the one
3277       for GTK compile in finite time and space. :-)
3278       See src/ml-nlffigen/README for details on what's new.
3279    
3280    ----------------------------------------------------------------------
3281    Name: Lal George
3282    Date: 2001/01/09 14:31:35 EST 2002
3283    Tag: george-20011206-rm-native-copy
3284    Description:
3285    
3286            Removed the native COPY and FCOPY instructions
3287            from all the architectures and replaced it with the
3288            explicit COPY instruction from the previous commit.
3289    
3290            It is now possible to simplify many of the optimizations
3291            modules that manipulate copies. This has not been
3292            done in this change.
3293    
3294    ----------------------------------------------------------------------
3295    Name: Lal George
3296    Date: 2001/12/06 16:50:13 EST 2001
3297    Tag: george-20011206-mlrisc-instruction
3298    Description:
3299    
3300    Changed the representation of instructions from being fully abstract
3301    to being partially concrete. That is to say:
3302    
3303      from
3304            type instruction
3305    
3306      to
3307            type instr                              (* machine instruction *)
3308    
3309            datatype instruction =
3310                LIVE of {regs: C.cellset, spilled: C.cellset}
3311              | KILL of {regs: C.cellset, spilled: C.cellset}
3312              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
3313              | ANNOTATION of {i: instruction, a: Annotations.annotation}
3314              | INSTR of instr
3315    
3316    This makes the handling of certain special instructions that appear on
3317    all architectures easier and uniform.
3318    
3319    LIVE and KILL say that a list of registers are live or killed at the
3320    program point where they appear. No spill code is generated when an
3321    element of the 'regs' field is spilled, but the register is moved to
3322    the 'spilled' (which is present, more for debugging than anything else).
3323    
3324    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
3325    We used to generate:
3326    
3327            DEFFREG f1
3328            f1 := f2 + f3
3329            trapb
3330    
3331    but now generate:
3332    
3333            f1 := f2 + f3
3334            trapb
3335            LIVE {regs=[f1,f2,f3], spilled=[]}
3336    
3337    Furthermore, the DEFFREG (hack) required that all floating point instruction
3338    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
3339    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
3340    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
3341    this architecture).
3342    
3343    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
3344    all the architectures. This will result in further simplification of the
3345    register allocator that must be aware of them for coalescing purposes, and
3346    will also simplify certain aspects of the machine description that provides
3347    callbacks related to parallel copies.
3348    
3349    ANNOTATION should be obvious, and now INSTR represents the honest to God
3350    machine instruction set!
3351    
3352    The <arch>/instructions/<arch>Instr.sml files define certain utility
3353    functions for making porting easier -- essentially converting upper case
3354    to lower case. All machine instructions (of type instr) are in upper case,
3355    and the lower case form generates an MLRISC instruction. For example on
3356    the alpha we have:
3357    
3358      datatype instr =
3359         LDA of {r:cell, b:cell, d:operand}
3360       | ...
3361    
3362      val lda : {r:cell, b:cell, d:operand} -> instruction
3363        ...
3364    
3365    where lda is just (INSTR o LDA), etc.
3366    
3367    ----------------------------------------------------------------------
3368    Name: Matthias Blume
3369    Date: 2001/11/22 21:40:00 EST
3370    Tag: Release_110_37
3371    Description:
3372    
3373    Release 110.37.  This time for real.
3374    
3375    ----------------------------------------------------------------------
3376    Name: Matthias Blume
3377    Date: 2001/11/21 16:35:00 EST
3378    Tag: blume-20011121-foot-in-mouth
3379    Description:
3380    
3381    Removed the "Release_110_37" tag because of a serious bug.
3382    This will be re-tagged once the bug is fixed.
3383    
3384    ----------------------------------------------------------------------
3385    Name: Matthias Blume
3386    Date: 2001/11/21 16:14:00 EST
3387    Tag: blume-20011121-forgottenfile
3388    Description:
3389    
3390    Forgot to add a file.  (Just a .tex-file -- part of
3391    the CM manual source.)
3392    
3393    ----------------------------------------------------------------------
3394    Name: Matthias Blume
3395    Date: 2001/11/21 16:10:00 EST
3396    Tag: blume-20011121-invalid_110_37
3397    Description:
3398    
3399    Note: I removed the original tag "Release_110_37" from this commit
3400          because we found a serious bug in all non-x86 backends.
3401          - Matthias
3402    
3403    1. Modifications to the SML/NJ code generator and to the runtime system
3404       so that code object name strings are directly inserted into code
3405       objects at code generation time.  The only business the runtime system
3406       has with this is now to read the name strings on occasions.
3407       (The encoding of the name string has also changed somewhat.)
3408    
3409    2. CM now implements a simple "set calculus" for specifying export lists.
3410       In particular, it is now possible to refer to the export lists of
3411       other libraries/groups/sources and form unions as well as differences.
3412       See the latest CM manual for details.
3413    
3414    3. An separate notion of "proxy" libraries has again be eliminated from
3415       CM's model.  (Proxy libraries are now simply a special case of using
3416       the export list calculus.)
3417    
3418    4. Some of the existing libraries now take advantage of the new set
3419       calculus.
3420       (Notice that not all libraries have been converted because some
3421       of the existing .cm-files are supposed to be backward compatible
3422       with 110.0.x.)
3423    
3424    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
3425       "General.exnMessage"!  The former relies on a certain hook to be
3426       initialized, and that often does not happen in the stand-alone case.)
3427    
3428    ----------------------------------------------------------------------
3429    Name: Lal George
3430    Date: 2001/11/21  13:56:18 EST
3431    Tag: george-2001121-pseudo-ops
3432    Description:
3433    
3434      Implemented a complete redesign of MLRISC pseudo-ops. Now there
3435      ought to never be any question of incompatabilities with
3436      pseudo-op syntax expected by host assemblers.
3437    
3438      For now, only modules supporting GAS syntax are implemented
3439      but more should follow, such as MASM, and vendor assembler
3440      syntax, e.g. IBM as, Sun as, etc.
3441    
3442    ----------------------------------------------------------------------
3443    Name: Matthias Blume
3444    Date: 2001/11/14 11:52:00 EST
3445    Tag: blume-20011114-srcname
3446    Description:
3447    
3448    1. Routed the name of the current source file to mlriscgen where it
3449       should be directly emitted into the code object.  (This last part
3450       is yet to be done.)
3451    
3452    2. Some cleanup of the pgraph code to make it match the proposal that
3453       I put out the other day.  (The proposal notwithstanding, things are
3454       still in flux here.)
3455    
3456    ----------------------------------------------------------------------
3457    Name: Lal George
3458    Date: 2001/11/14 09:44:04 EST
3459    Tag:
3460    Description:
3461    
3462      Fix for a backpatching bug reported by Allen.
3463    
3464      Because the boundary between short and long span-dependent
3465      instructions is +/- 128, there are an astounding number of
3466      span-dependent instructions whose size is over estimated.
3467    
3468      Allen came up with the idea of letting the size of span
3469      dependent instructions be non-monotonic, for a maxIter
3470      number of times, after which the size must be monotonically
3471      increasing.
3472    
3473      This table shows the number of span-dependent instructions
3474      whose size was over-estimated as a function of maxIter, for the
3475      file Parse/parse/ml.grm.sml:
3476    
3477         maxIter            # of instructions:
3478            10                      687
3479            20                      438
3480            30                      198
3481            40                        0
3482    
3483      In compiling the compiler, there is no significant difference in
3484      compilation speed between maxIter=10 and maxIter=40. Actually,
3485      my measurements showed that maxIter=40 was a tad faster than
3486      maxIter=10! Also 96% of the  files in the compiler reach a fix
3487      point within 13 iterations, so fixing maxIter at 40, while high,
3488      is okay.
3489    
3490    ----------------------------------------------------------------------
3491    Name: Matthias Blume
3492    Date: 2001/10/31 15:25:00 EST
3493    Tag: blume-20011031-pgraph
3494    Description:
3495    
3496    CKIT:
3497    * Changed the "Function" constructor of type Ast.ctype to carry optional
3498      argument identifiers.
3499    * Changed the return type of TypeUtil.getFunction accordingly.
3500    * Type equality ignores the argument names.
3501    * TypeUtil.composite tries to preserve argument names but gives up quickly
3502      if there is a mismatch.
3503    
3504    installation script:
3505    * attempts to use "curl" if available (unless "wget" is available as well)
3506    
3507    CM:
3508    * has an experimental implementation of "portable graphs" which I will
3509      soon propose as an implementation-independent library format
3510    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
3511    
3512    NLFFI-LIB:
3513    * some cleanup (all cosmetic)
3514    
3515    NLFFIGEN:
3516    * temporarily disabled the mechanism that suppresses ML output for
3517      C definitions whose identifiers start with an underscore character
3518    * generate val bindings for enum constants
3519    * user can request that only one style (light or heavy) is being used;
3520      default is to use both (command-line arguments: -heavy and -light)
3521    * fixed bug in handling of function types involving incomplete pointers
3522    * generate ML entry points that take record arguments (i.e., using
3523      named arguments) for C functions that have a prototype with named
3524      arguments
3525      (see changes to CKIT)
3526    
3527    ----------------------------------------------------------------------
3528    Name: Allen Leung
3529    Date: 2001/10/27 20:34:00 EDT
3530    Tag: leunga-20011027-x86-fast-fp-call
3531    Description:
3532    
3533       Fixed the bug described in blume-20010920-slowfp.
3534    
3535       The fix involves
3536          1. generating FCOPYs in FSTP in ia32-svid
3537          2. marking a CALL with the appropriate annotation
3538    
3539    ----------------------------------------------------------------------
3540    Name: Matthias Blume
3541    Date: 2001/10/16 11:32:00 EDT
3542    Tag: blume-20011016-netbsd
3543    Description:
3544    
3545    Underscore patch from Chris Richards (fixing problem with compiling
3546    runtime system under recent NetBSD).
3547    
3548    ----------------------------------------------------------------------
3549    Name: Allen Leung
3550    Date: 2001/10/12 17:18:32 EDT 2001
3551    Tag: leung-20011012-x86-printflowgraph
3552    Description:
3553    
3554    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
3555    
3556    ----------------------------------------------------------------------
3557    Name: Lal George
3558    Date: 2001/10/11 23:51:34 EDT
3559    Tag: george-20011011-too-many-instrs
3560    Description:
3561    
3562    The representation of a program point never expected to see more
3563    than 65536 instructions in a basic block!
3564    
3565    ----------------------------------------------------------------------
3566    Name: Lal George
3567    Date: 2001/10/09 09:41:37 EDT
3568    Tag: george-20011008-mlrisc-labels
3569    Description:
3570    
3571    Changed the machine description files to support printing of
3572    local and global labels in assembly code, based on host assembler
3573    conventions.
3574    
3575    ----------------------------------------------------------------------
3576    Name: Matthias Blume
3577    Date: 2001/09/25 15:25:00 EDT
3578    Tag: blume-20010925-exninfo
3579    Description:
3580    
3581    I provided a non-hook implementation of exnName (at the toplevel) and
3582    made the "dummy" implementation of exnMessage (at the toplevel) more
3583    useful: if nothing gets "hooked in", then at least you are going to
3584    see the exception name and a message indicating why you don't see more.
3585    
3586    [For the time being, programs that need exnMessage and want to use
3587    ml-build should either use General.exnMessage (strongly recommended) or
3588    refer to structure General at some other point so that CM sees a
3589    static dependency.]
3590    
3591    [Similar remarks go for "print" and "use":  If you want to use their
3592    functionality in stand-alone programs generated by ml-build, then use
3593    TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
3594    
3595    ----------------------------------------------------------------------
3596    Name: Matthias Blume
3597    Date: 2001/09/20 17:28:00 EDT
3598    Tag: blume-20010920-slowfp
3599    Description:
3600    
3601    Allen says that x86-fast-fp is not safe yet, so I turned it off again...
3602    
3603    ----------------------------------------------------------------------
3604    Name: Matthias Blume
3605    Date: 2001/09/20 17:20:00 EDT
3606    Tag: blume-20010920-canonicalpaths
3607    Description:
3608    
3609    0. Updated the BOOT file (something that I forgot to do earlier).
3610    
3611    1. Small internal change to CM so that it avoids "/../" in filenames
3612       as much as possible (but only where it is safe).
3613    
3614    2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
3615       that arguments that contain delimiters are passed through correctly.
3616       This change also means that all "special" arguments of the form
3617       @SMLxxx... must come first.
3618    
3619    3. Changed install script to put relative anchor names for tool commands