Home My Page Projects Code Snippets Project Openings SML/NJ
Summary Activity Forums Tracker Lists Tasks Docs Surveys News SCM Files

SCM Repository

[smlnj] Diff of /sml/trunk/HISTORY
ViewVC logotype

Diff of /sml/trunk/HISTORY

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

revision 977, Wed Nov 14 16:53:16 2001 UTC revision 1641, Wed Oct 6 20:18:55 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: 2004/10/06 15:15:15 CDT
18    Tag: blume-20041006-handler
19    Description:
20    
21    Cleaned up the absyn to reflect the invariant that HANDLE always
22    carries a FNexp as part of the type definition.  This eliminates some
23    superfluous sanity checks at runtime down the road.
24    
25    Some minor cleanup of the btrace code.
26    
27    ----------------------------------------------------------------------
28    Name: Matthias Blume (blume (at) tti - c (dot) org)
29    Date: 2004/10/01 10:20:30 CDT
30    Tag: blume-20041001-slave
31    Description:
32    
33    Added hack to make slave mode work in the presence of the version
34    tool.  (Still, since the master does two passes over the code for
35    CMB.make, the release number gets bumped twice when slaves are
36    attached.  I don't know if this is worth fixing...)
37    
38    ----------------------------------------------------------------------
39    Name: Matthias Blume (blume (at) tti - c (dot) org)
40    Date: 2004/09/30 10:55:00 CDT
41    Tag: blume-20040930-version
42    Description:
43    
44    * Moved the "version" magic into its own little library under
45    src/system/smlnj/internal.  This avoids expensive reconstruction
46    of a stable src/compiler/core.cm.
47    
48    * At the same time, structure CompilerVersion is now known as structure
49    SMLNJVersion.
50    
51    * Arranged for the version tool to NOT kick in when rebuilding the system
52      (makeml -rebuild, fixpt).  Otherwise one would never reach a fixpoint.
53      Also, loading the versiontool does not work when rebuilding the system
54      because CM is not properly initialized at that time.
55    
56    ----------------------------------------------------------------------
57    Name: Matthias Blume (blume (at) tti - c (dot) org)
58    Date: 2004/09/29 14:00:00 CDT
59    Tag: blume-20040929-autoversion
60    Description:
61    
62    Implemented some CM magic to have
63        file src/compiler/TopLevel/main/version.sml
64    generated automagically.
65    The version is taken from two files: config/version and config/release.
66    The first is expected to contain a two-part version number such as 110.49.
67    The second should contain a single number, but it may be missing.
68    
69    If the environment variable VERSIONTOOL_BUMP_RELEASE is defined at the
70    time the version tool is loaded (which is the first time you say CMB.make),
71    then the tool will increment the value stored in config/release every
72    time CMB.make is invoked.
73    
74    The binfile format is now insensitive to anything beyond the first
75    two components of a version number, so bumping the release does not render
76    binfiles incompatible.  Auto-bumping can be used to keep track of versions
77    during development without invalidating existing binfiles.
78    
79    In any case, every CMB.make updates the date information in version.sml.
80    (This is the date that is printed in the banner.)
81    
82    ----------------------------------------------------------------------
83    Name: Matthias Blume (blume (at) tti - c (dot) org)
84    Date: 2004/09/28 10:53:00 CDT
85    Tag: blume-20040928-controls
86    Description:
87    
88    Some cleanup of the controls code.
89    
90    ----------------------------------------------------------------------
91    Name: Matthias Blume (blume (at) tti - c (dot) org)
92    Date: 2004/09/27 22:08:00 CDT
93    Tag: blume-20040927-controls
94    Description:
95    
96    Added two pieces of functionality to the Controls interface:
97    
98    1.   val save'restore: 'a control -> unit -> unit
99    
100       grabs the current value of the control in stage 1 and restores it
101       in stage 2.
102    
103    2.   val set' : 'a control * 'a -> unit -> unit
104    
105       stores the given value into the control in stage 2 (i.e., delayed)
106       but does all error checking in stage 1.
107       (This is for string controls that need to do parse their argument --
108        something that might fail.  In some cases, notably in CM, one
109        already knows the intended argument but wants to delay the actual
110        assignment until a time when error recovery would be more difficult.)
111    
112    Changed the handling of controls in tool arguments to classes "sml" and
113    "lazysml":
114       - use Controls.save'restore as a more robust way of restoring the
115         old value (in particular: without having to re-parse the string)
116       - use controls to handle the "overload" keyword in the init group
117         (I believe this change actually fixes a long-standing obscure bug.)
118    
119    ----------------------------------------------------------------------
120    Name: Matthias Blume (blume (at) tti - c (dot) org)
121    Date: 2004/09/27 17:00:00 CDT
122    Tag: blume-20040927-lazysml
123    Description:
124    
125    Added a new tool class called "lazysml" to CM's tool chest.  The only
126    difference to "sml" is that compilation is done with Control.lazysml
127    set to true.  A source of class "lazysml" is automatically recognized
128    by a file name suffix of ".lml".
129    
130    In addition to the above feature, the original class "sml" now also
131    supports a tool argument "lazy" which has the same effect.  As a
132    result, the following three lines are equivalent:
133    
134        foo.sml : lazysml
135        foo.sml : sml (lazy)
136        foo.sml (lazy)
137    
138    The setting goes into effect both during parsing and during
139    compilation.  The original setting is restored right after parsing and
140    after compilation, respectively.
141    
142    In addition to all the above, there is also a general mechanism to set
143    ANY of the "controls" that are available at the command line via
144    "-C..." on a per-sml-file basis.  The same rules that apply for "lazy"
145    apply as well. (In fact, "lazy" is implemented as a special case of
146    the general mechanism.)
147    
148    The .cm file syntax uses a new keyword tool argument called "with".
149    There are several ways of indicating the desired settings:
150    
151        foo.sml (with:parser.quotations=true)
152        foo.sml (with:(name:parser.quotations value:true))
153        foo.sml (with:(name:name1 value:value1 name:name2 value:value2 ...))
154        foo.sml (with:(name1=value1 name2=value2 ...))
155        foo.sml (with:(name1=value1 name:name2 value:value2 name3=value3 ...))
156    
157        etc.
158    
159    Another possible abbreviation is to leave out the =v or value:v part
160    if the name refers to a boolean control (in which case the value is
161    taken to be true).  Thus, one could get lazy sml also by saying:
162    
163        foo.sml (with:parser.lazy-keyword=true)
164        foo.sml (with:parser.lazy-keyword)
165        foo.sml (with:(name:parser.lazy-keyword value:true))
166        foo.sml (with:(name:parser.lazy-keyword))
167    
168    ----------------------------------------------------------------------
169    Name: Matthias Blume (blume (at) tti - c (dot) org)
170    Date: 2004/09/24 16:22:00 CDT
171    Tag: blume-20040924-ppc-long-branch
172    Description:
173    
174    Turned message about "emiting long form of branch" off by default.
175    Added a control flag to turn it back on when desired.
176    
177    ----------------------------------------------------------------------
178    Name: Matthias Blume (blume (at) tti - c (dot) org)
179    Date: 2004/09/24 16:05:00 CDT
180    Tag: blume-20040924-rounding
181    Description:
182    
183    Applied patch for setting rounding modes under Mac OS X.  Thanks to
184    Melissa O'Neill for providing the code!
185    
186    ----------------------------------------------------------------------
187    Name: Matthias Blume (blume (at) tti - c (dot) org)
188    Date: 2004/09/23 17:30:00 CDT
189    Tag: blume-20040923-envvars
190    Description:
191    
192    1. Changed definition of type ControlRegistry.registry_tree to
193       include control_info (i.e., the name of the controlling environment
194       variable).
195    
196    2. Added command-line flags -e and -E to print the names of environment
197       variables that can be used to control internal settings.
198       (This uses the new API mentioned in 1.)
199    
200    ----------------------------------------------------------------------
201    Name: Matthias Blume (blume (at) tti - c (dot) org)
202    Date: 2004/09/13 16:50:00 CDT
203    Tag: Release_110_49
204    Description:
205    
206    New working version (110.49).  NEW BOOTFILES!
207    
208    ----------------------------------------------------------------------
209    Name: Matthias Blume
210    Date: 2004/09/13 16:20:00 CDT
211    Tag: blume-20040913-config-mlrisc
212    Description:
213    
214    Put target "mlrisc" back into the default list.
215    (There is no harm in having it, and some users have expressed their
216    wish to have "mlrisc" included by default.)
217    
218    ----------------------------------------------------------------------
219    Name: John Reppy
220    Date: 2004/09/13
221    Tag: jhr-20040913-signals
222    Description:
223    
224    Fixed the signal masking code to properly nest mask/unmask operations
225    on a per-signal basis.
226    
227    ----------------------------------------------------------------------
228    Name: Matthias Blume (blume (at) tti - c (dot) org)
229    Date: 2004/09/08 13:20:00 CDT
230    Tag: blume-20040908-heap-magic
231    Description:
232    
233    Bumped the heap macig number to 0x09082004 to account for the changed
234    layout of the ML frame under MacOS X.
235    
236    ----------------------------------------------------------------------
237    Name: Allen Leung (leunga (at) reservoir (dot) com)
238    Date: 2004/09/03 11:26:00 EST
239    Tag: leunga-20040903-cygwin-install
240    Description:
241    
242       Added a patch to _arch-n-opsys to enable the Cygwin runtime.
243    The Cygwin runtime is turned on by setting the environment
244    variable SMLNJ_CYGWIN_RUNTIME to 1.
245    
246    ----------------------------------------------------------------------
247    Name: Matthias Blume (blume (at) tti - c (dot) org)
248    Date: 2004/08/31 17:15:00 CDT
249    Tag: blume-20040831-core
250    Description:
251    
252    Added some exports to src/compiler/core.cm upon request by J. Joemann.
253    
254    ----------------------------------------------------------------------
255    Name: Matthias Blume (blume (at) tti - c (dot) org)
256    Date: 2004/08/30 17:55:00 CDT
257    Tag: blume-20040830-installer
258    Description:
259    
260    Upon request by Johannes Joemann:
261    
262    - improved ML code of installer to fall back to coping when renaming
263      fails (i.e., when source and target are on different file systems);
264      the code compiles but has yet to be tested in anger
265    - removed mlrisc from list of default targets (config/targets)
266    
267    ----------------------------------------------------------------------
268    Name: Matthias Blume (blume (at) tti - c (dot) org)
269    Date: 2004/08/27 17:20:00 CDT
270    Tag: blume-20040827-ptreql
271    Description:
272    
273    Added ptreql primop to structure InlineT (upon request from Larry
274    Paulson).
275    
276    ----------------------------------------------------------------------
277    Name: Allen Leung (leunga (at) reservoir (dot) com)
278    Date: 2004/08/15 21:21:00 EST
279    Tag: leunga-110_48-udgraph
280    Description:
281    
282    Another bug fix from Carl Hauser:
283    
284    diff /net/niflab/smlnj48/src/MLRISC/graphs/udgraph.sml udgraph.sml
285    > 48c48
286    > <              | rmv((e as (k,_))::es,L) = rmv(es,if k = i then es else
287    > e::L)
288    > ---
289    > >              | rmv((e as (k,_))::es,L) = rmv(es,if k = i then L else e::L)
290    > Without this, any deletion of an edge in an undirected graph does severe
291    > violence to the graph.
292    
293    ----------------------------------------------------------------------
294    Name: Allen Leung (leunga (at) reservoir (dot) com)
295    Date: 2004/08/10 23:35:00 EST
296    Tag: leunga-110_48-ppc
297    Description:
298    
299       The IBM/MacOS syntax switch on PPC was incorrectly swapped.
300    
301    ----------------------------------------------------------------------
302    Name: Matthias Blume (blume (at) tti - c (dot) org)
303    Date: 2004/08/10 12:00:00 CDT
304    Tag: Release_110_48
305    Description:
306    
307    New working version (110.48).  NEW BOOTFILES!
308    
309    ----------------------------------------------------------------------
310    Name: Allen Leung (leunga (at) reservoir (dot) com)
311    Date: 2004/08/09 12:21:00 EST
312    Tag: leunga-110_47-dijsktra
313    Description:
314    
315    Bug fix from Carl Hauser:
316    
317    single_source_shortest_paths in dijkstra.sml was observed to get wrong
318    answers (by comparing to single_source_shortest_paths in bellman-ford.sml).
319    
320    The problem is that following the expression A.update(dist,s,Num.zero)
321    it is necessary to update the priority queue using Q.decreaseWeight(Q,s).
322    
323    ----------------------------------------------------------------------
324    Name: Matthias Blume (blume (at) tti - c (dot) org)
325    Date: 2004/08/06 18:10:00 CDT
326    Tag: blume-20040806-cmdline
327    Description:
328    
329    Fiddled with handling of command-line options:
330    
331      * sml now quits after processing the command line
332        if -H, -S, -h<n>, or -s<n> appears as the last
333        command-line argument
334      * a new option -q terminates the session when encountered on
335        the command line; subsequent arguments will be ignored
336      * bug fixes: short (erroneous) arguments are no longer ignored
337        completely
338    
339    ----------------------------------------------------------------------
340    Name: Allen Leung (leunga (at) reservoir (dot) com)
341    Date: 2004/08/04 18:17:00 EST
342    Tag: leunga-110_47-ppc-ibm-asm
343    Description:
344    
345      - Added minimal IBM assembly syntax support for PowerPC.
346    
347      - Cygwin: manually changed the file cygwin.def.  Some exported symbols have
348        been altered in the runtime.  We need an automatic way to keep the file
349        in sync.
350    
351    ----------------------------------------------------------------------
352    Name: Matthias Blume (blume (at) tti - c (dot) org)
353    Date: 2004/08/04 14:00:00 CDT
354    Tag: Release_110_47
355    Description:
356    
357    New working version (110.47).  NEW BOOTFILES!
358    
359    ----------------------------------------------------------------------
360    Name: Matthias Blume (blume (at) tti - c (dot) org)
361    Date: 2004/08/03 14:25:00 CDT
362    Tag: blume-20040803-callingconv
363    Description:
364    
365    Added low-level support for choosing C calling conventions by
366    twiddling the type of rawccall.  (See
367    src/compiler/Semant/types/cproto.sml for details.)
368    
369    ----------------------------------------------------------------------
370    Name: Matthias Blume (blume (at) tti - c (dot) org)
371    Date: 2004/08/02 15:55:00 CDT
372    Tag: blume-20040802-backout
373    Description:
374    
375    Backed out of change to win32-filesys.c.  The earlier patch to
376    get_file_time caused CM to produce files with the wrong time stamp.
377    
378    ----------------------------------------------------------------------
379    Name: Matthias Blume (blume (at) tti - c (dot) org)
380    Date: 2004/08/02 14:45:00 CDT
381    Tag: blume-20040802-nlffi-win32
382    Description:
383    
384    Added NLFFI support for Win32, adapted from a patch provided by David
385    Hansel.  This is currently completely untested.  Also, the issue
386    concerning stdcall vs. ccall is still unresolved.
387    
388    ----------------------------------------------------------------------
389    Name: Matthias Blume (blume (at) tti - c (dot) org)
390    Date: 2004/07/30 17:55:00 CDT
391    Tag: blume-20040730-various
392    Description:
393    
394    Gearing up towards 110.47...
395    
396    - various minor bugfixes to ml-nlffigen
397    - a beginning of a manual for nlffi
398    
399    - eliminated 'export name=value' in config/install.sh as this does
400      not work with certain versions of /bin/sh
401      (Thanks to David King at Motorola for catching this.)
402    
403    - several bugfixes provided or suggested by David Hansel at Reactive Systems:
404      - added a test for tm==NULL to gmtime.c and localtime.c
405      - applied patch for incorrect GetFileTime under win32
406      - toSeconds -> toMilliseconds in Win32/win32-process.sml
407    
408    ----------------------------------------------------------------------
409    Name: Matthias Blume (blume (at) tti - c (dot) org)
410    Date: 2004/07/21 18:20:00 CDT
411    Tag: blume-20040721-nlffigen
412    Description:
413    
414    - Fixed minor issue in ml-nlffigen: Now generate structure T_foo
415      for a typedef to an incomplete type, but leave out the "typ" member.
416      (This is just for consistency.)
417    - Started to produce what is supposed to become better (i.e., comprehensive)
418      documentation of what ml-nlffigen does and produces.
419    
420    ----------------------------------------------------------------------
421    Name: Matthias Blume (blume (at) tti - c (dot) org)
422    Date: 2004/07/14 16:25:00 CDT
423    Tag: blume-20040714-union
424    Description:
425    
426    Added C_UNION to c-calls/c-types.sml and updated the machinery
427    (ml-nlffigen, cproto.sml) that conveys C function interface
428    information to the code generator.
429    
430    However, the actual architecture-specific implementation of function
431    arguments and results that are C unions is still not implemented.
432    
433    ----------------------------------------------------------------------
434    Name: Allen Leung (leunga (at) reservoir (dot) com)
435    Date: 2004/07/14 14:38:00 EST
436    Tag: leunga-110_46_1-ppc-lwzu
437    Description:
438    
439        Added these instructions to the PowerPC architecture:
440        LBZU(X), LHZU(X), LWZU(X),
441        STWU(X), STFDU, STFSU
442    
443        etc...
444    
445        Note: I haven't added their instruction encoding into the description.
446    
447    ----------------------------------------------------------------------
448    Name: Allen Leung (leunga (at) reservoir (dot) com)
449    Date: 2004/07/13 15:04:00 EST
450    Tag: leunga-110_46_1-ppc-lwarx
451    Description:
452    
453       Added the two instructions LWARX and STWCX to the PowerPC
454    instruction set.
455    
456       A (untested) rewrite of loop-structure.sml.  The old version
457    is completely broken.
458    
459    ----------------------------------------------------------------------
460    Name: Matthias Blume (blume (at) tti - c (dot) org)
461    Date: 2004/07/13 13:50:00 CDT
462    Tag: blume-20040713-nlffi
463    Description:
464    
465    - use paramAlloc to report c-calls with too many arguments
466      (for PPC version where parameter area is pre-allocated)
467    - added ccall_maxargspace to machspec (to implement the above)
468    - made "make" commend in CM's "make" tool configurable
469    - added option (default: on) for passing the name of the SML/NJ's "bin"
470      directory to "make"; the call looks like this:
471    
472         make <options> SMLNJ_BINDIR=<dir> <target>
473    
474      This can be used by the Makefile to, e.g., pick the "right" version
475      of ml-nlffigen.
476    - minor code tweaks
477    
478    ----------------------------------------------------------------------
479    Name: Matthias Blume (blume (at) tti - c (dot) org)
480    Date: 2004/07/12 22:50:00 CDT
481    Tag: blume-110_46_1-macosx-nlffi
482    Description:
483    
484    NLFFI under Mac OS X now working (sort of).  This is largely untested,
485    though.
486    
487    Note:  1. You have to make a new, clean build of the runtime system.
488           2. There are new BOOTFILES, you have to use them!
489              (Doing the bootstrap process yourself would be *very* painful!
490               If you absolutely have to do it, build the system under
491               a different architecture and then cross-compile.)
492    
493    Version bumped to 110.46.1 to account for runtime data format changes.
494    
495    ----------------------------------------------------------------------
496    Name: Matthias Blume (blume (at) tti - c (dot) org)
497    Date: 2004/06/18 14:30:00 CDT
498    Tag: blume-20040618-unix
499    Description:
500    
501    Changed the implementation of structure Unix so that the same stream
502    is returned every time one of the {text,bin}{In,Out}streamOf functions
503    is invoked on the same proc.  This is not what the spec currently
504    says -- although IMO it arguably should.  (See discussion below.)
505    
506    ----------------------------------------------------------------------
507    Name: Matthias Blume (blume (at) tti - c (dot) org)
508    Date: 2004/06/17 18:15:00 CDT
509    Tag: Release_110_46
510    Description:
511    
512    New working version (110.46).  NEW BOOTFILES!
513    
514    ----------------------------------------------------------------------
515    Name: Matthias Blume (blume (at) tti - c (dot) org)
516    Date: 2004/06/17 17:20:00 CDT
517    Tag: blume-20040617-timer-unix
518    Description:
519    
520    Changed the interface of structures Timer and Unix to match the most
521    recent Basis spec.
522    
523    In the case of Unix there still seems to be an open/weird issue:
524    
525       The {text,bin}{In,Out}streamOf functions are supposed to create
526       fresh streams whenever they are called -- as opposed to have them
527       return the same stream every time.  This design is supposed to
528       prevent space leaks caused by proc values hanging on to streams.
529    
530       The reap function, on the other hand, is supposed to close the
531       streams.  This cannot be done without having a handle on the
532       stream in proc after all...
533    
534    I took the liberty to implement the following stopgap solution:
535    
536      The proc value hangs on to the most recently created stream(s).
537      Reap closes those.  If either or both of the two streams hadn't
538      been created at all yet, then reap will close the corresponding
539      file descriptors directly.
540    
541    PS: I don't understand the original space leak argument anymore.  If
542    a proc hangs on to the imperative stream, then I/O operations on those
543    will advance the state of the cached stream and avoid the space leak.
544    
545    ----------------------------------------------------------------------
546    Name: Matthias Blume (blume (at) tti - c (dot) org)
547    Date: 2004/05/28 16:45:00 CDT
548    Tag: blume-20040528-basis
549    Description:
550    
551    Added signature PACK_REAL and exported functor PrimIO.
552    
553    ----------------------------------------------------------------------
554    Name: Matthias Blume (blume (at) tti - c (dot) org)
555    Date: 2004/05/25 16:00:00 CDT
556    Tag: blume-20040525-group-owner
557    Description:
558    
559    CM now ignores (but still accepts) the "owner" information in group
560    descriptions.  The owner of a group is its next enclosing
561    library. Each group must have a unique owner.  (There is a virtual
562    "toplevel" library that owns groups which are not nested within a real
563    library.)  Previously, each group had to explicitly declare its owner,
564    and CM would check that such a declaration is correct.  The new scheme
565    is to have CM check that for each group there is precisely one owning
566    library.
567    
568    The advantage of the new scheme is that the programmer no longer needs
569    to maintain the somewhat annoying owner information.  The downside is
570    that CM cannot enforce the ownership rule across multiple runs of
571    CM.make.  Fortunately, enclosing the same group in two different
572    libraries A and B which are not part of the same program does not
573    cause real problems.
574    
575    ----------------------------------------------------------------------
576    Name: Matthias Blume (blume (at) tti - c (dot) org)
577    Date: 2004/05/20 16:00:00 CDT
578    Tag: blume-20040520-win32
579    Description:
580    
581    Made the win32 version work again.  (Strangely, a misplaced comma had
582    slipped into win32-process.c which prevented the runtime from being
583    compiled correctly.)
584    
585    Also, included a minor addition to ml-build.bat analogous to what was
586    done in blume-20040519-ml-build.
587    
588    ----------------------------------------------------------------------
589    Name: Matthias Blume (blume (at) tti - c (dot) org)
590    Date: 2004/05/19 22:10:00 CDT
591    Tag: blume-20040519-ml-build
592    Description:
593    
594    Arranged for ml-build to clean up after itself a little bit better.
595    The script generates a temporary SML source file and compiles it using
596    CM, so CM generates metadata (GUID, SKEL, objectfile) for it.  It now
597    gets rid of those at the end, so they don't accumulate under .cm.
598    
599    This required a minor change to install.sh because the name of the
600    metadata directory (default: .cm) is actually configurable at
601    installation time.
602    
603    ----------------------------------------------------------------------
604    Name: Matthias Blume (blume (at) tti - c (dot) org)
605    Date: 2004/05/18 15:50:00 CDT
606    Tag: blume-20040518-mkreader
607    Description:
608    
609    Added Posix.IO.mk{Bin,Text}{Reader,Writer} by lifting their respective
610    implementations from internal modules PosixBinPrimIO and PosixTextPrimIO.
611    
612    ----------------------------------------------------------------------
613    Name: Matthias Blume (blume (at) tti - c (dot) org)
614    Date: 2004/05/11 14:35:00 CDT
615    Tag: blume-20040511-win32sock
616    Description:
617    
618    Added previously missing support for many socket-related functions
619    under win32.  Thanks to David Hansel <hansel@reactive-systems.com>
620    for the voluminous patch!
621    
622    (I have not tested this patch under win32 yet.)
623    
624    Here is David's e-mail:
625    
626    Hi,
627    
628    Attached to this email you find a diff against sml/nj 110.45
629    that will enable socket support under Windows.
630    
631    To apply the patch (using unix or cygwin)
632    1) gunzip runtime.diff.gz
633    2) "cd" into "src/runtime" in the source tree of a fresh
634       110.45 installation.
635    3) patch -p 1 < [your/path/to]runtime.diff
636    
637    The code compiles fine but has NOT yet been extensively tested.
638    I only ran a few tests for basic socket client functionality
639    (which worked fine).  Especially the functions that use ioctl
640    are not tested at all and might not work (see below).
641    
642    I implemented this since we want to move to a newer version of sml/nj
643    but need socket support in order to use it.  This is the first time I
644    even had a look at the sml/nj source,  so please review my changes
645    before making this part of the distribution!  Here are a few issues
646    that I think might be better for someone to solve who is more
647    familiar with the sml/nj source (and socket programming):
648    
649    - getnetbyaddr.c and getnetbyname.c will raise a "not implemented"
650      exception since I could not figure out what the windows equivalent
651      of these functions is
652    
653    - In sockets-osdep.h there are a some #include statements that are
654      only used in a few files that include sockets-osdep.h
655    
656    - In smlnj-sock-lib.c, function init_fn() calls WSAStartup() but
657      does not process its return value since I don't know how to
658      report an error upwards.
659    
660    - It would probably be good to have a call to WSACleanup() when
661      the library is unloaded (if there is such a possibility).
662      Otherwise I think Windows will take care of this automatically
663      when the process finishes.
664    
665    - I used ioctlsocket() as a replacement for ioctl() but I have
666      no idea if that is actually the proper replacement on Windows.
667    
668    - All these issues are marked in the code by "FIXME" comments.
669    
670    We use sml/nj extensively in our products and are quite happy
671    with it.  I hope this contribution will help you.
672    
673    Keep up the good work!
674    
675    David
676    
677    ----------------------------------------------------------------------
678    Name: Matthias Blume (blume (at) tti - c (dot) org)
679    Date: 2004/05/11 14:20:00 CDT
680    Tag: blume-20040511-installml
681    Description:
682    
683    Fixed two bugs in installml script.  (Thanks to Vesa A. Norrman for
684    the patch.)
685    
686    ----------------------------------------------------------------------
687    Name: Matthias Blume (blume (at) tti - c (dot) org)
688    Date: 2004/05/11 14:05:00 CDT
689    Tag: blume-20040511-nlffi-netbsd
690    Description:
691    
692    Added support for nlffi under netbsd.  (Thanks to Vesa A. Norrman for
693    the patch.)
694    
695    ----------------------------------------------------------------------
696    Name: Matthias Blume (blume (at) tti - c (dot) org)
697    Date: 2004/05/11 12:05:00 CDT
698    Tag: blume-20040511-exports
699    Description:
700    
701    As per request by Adam Chlipala <adam@hcoop.net>, extended various
702    export lists in compiler-related .cm-files.
703    
704    ----------------------------------------------------------------------
705    Name: Matthias Blume (blume (at) tti - c (dot) org)
706    Date: 2004/05/11 11:35:00 CDT
707    Tag: blume-20040511-allsource
708    Description:
709    
710    The installer now honors the "src-smlnj" target again, although its meaning
711    has changed from "all sources required for the compiler" to "all sources
712    the installer knows about".  In other words, if you enable "src-smlnj"
713    in the "targets" file, then the installer will pull in sources for
714    everything.  (Notice that this refers to source code only.  Compiled
715    code is still only installed for modules that were requested explicitly
716    or which are required for other modules that were requested explicitly.)
717    
718    ----------------------------------------------------------------------
719    Name: Matthias Blume (blume (at) tti - c (dot) org)
720    Date: 2004/04/23 17:40:00 CDT
721    Tag: blume-20040423-ieee-scan
722    Description:
723    
724    Fixed IEEEReal.scan (and .fromString) so that if there is an overflow
725    in the exponent calculation we get INF or ZERO (depending on the mantissa
726    and the sign of the exponent).
727    
728    ----------------------------------------------------------------------
729    Name: Matthias Blume (blume (at) tti - c (dot) org)
730    Date: 2004/04/23 10:40:00 CDT
731    Tag: blume-20040423-ml-build
732    Description:
733    
734    The ml-build script now terminates with a non-0 status when something
735    goes wrong.
736    
737    ----------------------------------------------------------------------
738    Name: Matthias Blume (blume (at) tti - c (dot) org)
739    Date: 2004/04/22 16:35:00 CDT
740    Tag: blume-20040422-Option
741    Description:
742    
743    Made exception Option to be the same as exception Option.Option
744    (as it should be).
745    
746    ----------------------------------------------------------------------
747    Name: Allen Leung (leunga (at) reservoir (dot) com)
748    Date: 2004/03/19 14:40:00 EST
749    Tag: leunga-20040319-cygwin-nlffi
750    Description:
751    
752    Fixed the runtime so that ml-nlffi-lib runs on the cygwin version
753    of SML/NJ.  The problem is that
754    
755       lib = dlopen(NULL, ...)
756       f   = dlsym(lib, "malloc");
757    
758    does not work on Windows unless we explicitly export symbols
759    such as 'malloc' during linking.  We fixed this by explicitly
760    exporting the required symbols with the magic gcc incantation:
761    
762        -Wl,--export-all cygwin.def
763    
764    where cygwin.def is a file containing all the symbols that
765    we wish to export.
766    
767    I suspect this is a Windows problem and we'll have to
768    do the same (somehow with windows compilers) when
769    we build the native win32 version with the system
770    calls LoadLibrary/GetProcAddress.
771    
772    ----------------------------------------------------------------------
773    Name: Matthias Blume (blume (at) tti - c (dot) org)
774    Date: 2004/03/04 16:35:00 CST
775    Tag: blume-20040304-intinf-fmt
776    Description:
777    
778    Fixed problem with IntInf.fmt (sign would show up on the right instead
779    of on the left for BIN, OCT, and HEX).
780    
781    ----------------------------------------------------------------------
782    Name: Matthias Blume (blume (at) tti - c (dot) org)
783    Date: 2004/03/04 11:25:00 CST
784    Tag: blume-20040304-symlinks
785    Description:
786    
787    Fixed problem with installer script (unix only) where bin/ml-yacc and
788    friends pointed (via symlinks) to absolute locations instead of just
789    .run-sml.  This was reported by Vesa A Norrman.
790    
791    ----------------------------------------------------------------------
792    Name: Matthias Blume (blume (at) tti - c (dot) org)
793    Date: 2004/02/13 14:50:00 CST
794    Tag: Release_110_45
795    Description:
796    
797    New working version (110.45). New bootfiles.
798    
799    ----------------------------------------------------------------------
800    Name: Matthias Blume (blume (at) tti - c (dot) org)
801    Date: 2004/01/26 15:15:15 CST
802    Tag: blume-20040126-toplevel
803    Description:
804    
805    Improved handling of exceptions at the interactive toplevel.
806    
807    ----------------------------------------------------------------------
808    Name: Matthias Blume (blume (at) tti - c (dot) org)
809    Date: 2004/01/26 11:25:00
810    Tag: blume-20040126-app
811    Description:
812    
813    Type of top-level "app" corrected.
814    Added code for setting vp_limitPtrMask to Win32-specific runtime.
815    
816    ----------------------------------------------------------------------
817    Name: Matthias Blume (blume (at) tti - c (dot) org)
818    Date: 2003/11/18 17:10 CST
819    Tag: blume-20031118-basis-fiddle
820    Description:
821    
822    - changed Timer interface to what might become the spec
823    - POSIX_FLAGS -> BIT_FLAGS according to spec
824    - some other minor discrepancies wrt. spec eliminated
825    
826    ----------------------------------------------------------------------
827    Name: Matthias Blume (blume (at) tti - c (dot) org)
828    Date: 2003/11/06 12:00:00 CST
829    Tag: Release_110_44
830    Description:
831    
832    New working version (110.44). New bootfiles.
833    
834    ----------------------------------------------------------------------
835    Name: Matthias Blume (blume (at) tti - c (dot) org)
836    Date: 2003/11/04 11:50:00 CST
837    Tag: blume-20031104-move-libraries
838    Description:
839    
840    Eliminated the "dont_move_libraries" directive in config/targets.
841    (The mechanism was broken and could not be fixed easily.  Moreover,
842    there does not seem to be any reason not to move all libraries into
843    lib during installation.  I originally implemented this directive as a
844    backward-compatibility feature when I first introduced the new CM.
845    Now that things have been stable for a long time and going back to the
846    old CM is not an option, there is no reason to keep it around.)
847    
848    ----------------------------------------------------------------------
849    Name: Matthias Blume (blume (at) tti - c (dot) org)
850    Date: 2003/11/03 16:00:00 CST
851    Tag: blume-20031103-installdir
852    Description:
853    
854    Made installer honor INSTALLDIR variable again.  (Thanks to Chris
855    Richards for pointing out the problem and providing the solution.)
856    
857    ----------------------------------------------------------------------
858    Name: Matthias Blume (blume (at) tti - c (dot) org)
859    Date: 2003/10/01 17:05:00 CDT
860    Tag: blume-20031001-lal-mlrisc
861    Description:
862    
863    MLRISC bug fix from Lal.
864    
865    ----------------------------------------------------------------------
866    Name: Matthias Blume (blume (at) tti - c (dot) org)
867    Date: 2003/09/30 16:10:00 CDT
868    Tag: blume-20030930-primio-bat
869    Description:
870    
871    1. Added openVector, nullRd, and nullWr to PRIM_IO.
872    2. Improved .bat files (for Win32 port) to make things work under Win95.
873       (thanks to Aaron S. Hawley for this one)
874    
875    ----------------------------------------------------------------------
876    Name: Matthias Blume (blume (at) tti - c (dot) org)
877    Date: 2003/09/26 16:05:00 CDT
878    Tag: blume-20030926-wrappriv
879    Description:
880    
881    Added missing wrapper for privilege "primitive" in $smlnj/viscomp/core.cm.
882    
883    ----------------------------------------------------------------------
884    Name: Matthias Blume (blume (at) tti - c (dot) org)
885    Date: 2003/09/26 15:00:00 CDT
886    Tag: blume-20030926-110_43_3
887    Description:
888    
889    - additional cleanup
890    - version number bump, NEW BOOTFILES
891    
892    ----------------------------------------------------------------------
893    Name: Matthias Blume (blume (at) tti - c (dot) org)
894    Date: 2003/09/26 12:00:00 CDT
895    Tag: blume-20030926-ppautoload
896    Description:
897    
898    I modified the read-eval-print loop so that the autoloader gets
899    invoked whenever the prettyprinter tries to look up a symbol that
900    is not currently defined in the toplevel environment but which
901    appears in CM's autoload registry.  As a result, we see far fewer of
902    those ?.Foo.Bar.xxx names in the prettyprinter's output.
903    
904    In addition to this I tried to clean up some pieces of the Basis
905    implementation (e.g., Socket, Word8Array) in order to prevent other
906    instances of these ?.Foo.Bar.xxx names from being printed.
907    
908    The mechanism that picks names for types still needs some work, though.
909    (Right now it seems that if there is a type A.t which is defined to
910    be B.u, but B is unavailable at toplevel, then A.t gets printed as
911    "?.B.u" although the perhaps more sensible solution would be to use
912    "A.t" in this case.  In other words, the prettyprinter should follow
913    a chain of DEFtycs not farther than there are corresponding toplevel
914    names in the current environment.)
915    
916    ----------------------------------------------------------------------
917    Name: Matthias Blume (blume (at) tti - c (dot) org)
918    Date: 2003/09/24 16:31:00 CDT
919    Tag: blume-20030924-installer
920    Description:
921    
922    Another installer tweak:  All the ML code for the installer is now
923    compiled during CMB.make and put into a little library called
924    $smlnj/installer.cm.  The installation then simply invokes
925    
926       sml -m $smlnj/installer.cm
927    
928    and everything happens automagically.
929    
930    Win32:  ML code senses value of environment variable SMLNJ_HOME.
931    Unix:   ML code senses values of environment variables ROOT, CONFIGDIR,
932            and BINDIR.
933    
934    The new scheme guarantees that the ML code responsible for the installation
935    is in sync with the APIs of the main system.  Also, the installer is
936    somewhat faster because the installer script is precompiled.
937    
938    ----------------------------------------------------------------------
939    Name: Matthias Blume (blume (at) tti - c (dot) org)
940    Date: 2003/09/24 15:35:00 CDT
941    Tag: blume-20030924-synsock
942    Description:
943    
944    Added a signature SYNCHRONOUS_SOCKET to basis.cm.  This is like SOCKET
945    but excludes all non-blocking operations.  Defined SOCKET (in Basis)
946    and CML_SOCKET in terms of SYNCHRONOUS_SOCKET.  Removed superfluous
947    implementations of non-blocking operations from CML's Socket
948    structure.
949    
950    ----------------------------------------------------------------------
951    Name: Matthias Blume (blume (at) tti - c (dot) org)
952    Date: 2003/09/24 15:10:05 CDT
953    Tag: blume-20030924-sockets
954    Description:
955    
956    1. Fixed SOCKET API and implementation to match Basis spec.
957       This required changing the internal representation of sockets to one
958       that remembers (for each socket file descriptor) whether it is currently
959       blocking or non-blocking.  This state is maintained lazily (i.e., a system
960       call is made only if the state actually needs to change).
961    
962    2. OS-specific details of sockets were moved into separate files, thus
963       making it possible to unify the bulk of the socket implementations
964       between Unix and Win32.
965    
966    3. CML's socket API changed accordingly.
967       (Note that we need to remove non-blocking functions from this API
968       since they are redundant in the case of CML!)
969    
970    4. CML's socket implementation now makes use of non-blocking functions
971       provided by Basis, thus removing all OS-dependent code from this part
972       of CML.
973    
974    5. Changed Real64.precision from 52 to 53.  Minor cleanup in Real64 code.
975    
976    ----------------------------------------------------------------------
977    Name: Matthias Blume (blume (at) tti - c (dot) org)
978    Date: 2003/09/22 12:10:00 CDT
979    Tag: blume-20030922-110_43_2
980    Description:
981    
982    Made a new interim version and bootfiles for developer's bootstrapping
983    convenience.
984    
985       110.43.2 -- NEW BOOTFILES
986    
987    ----------------------------------------------------------------------
988    Name: Matthias Blume (blume (at) tti - c (dot) org)
989    Date: 2003/09/19 15:55:00 CDT
990    Tag: blume-20030919-cmdir
991    Description:
992    
993    1. new-install.sh -> install.sh
994    2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
995    3. tweaked installer so that another name instead of .cm can be chosen
996       at install time (by setting the CM_DIR_ARC environment variable
997       during installation); once installation is complete, the name is
998       fixed
999    
1000    ----------------------------------------------------------------------
1001    Name: Matthias Blume (blume (at) tti - c (dot) org)
1002    Date: 2003/09/18 16:00:00 CDT
1003    Tag: blume-20030918-110_43_1
1004    Description:
1005    
1006    Made a new interim version and bootfiles for developer's bootstrapping
1007    convenience.
1008    
1009       110.43.1 -- NEW BOOTFILES
1010    
1011    ----------------------------------------------------------------------
1012    Name: Matthias Blume (blume (at) tti - c (dot) org)
1013    Date: 2003/09/18 15:20:00 CDT
1014    Tag: blume-20030918-misc
1015    Description:
1016    
1017    1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
1018       this seems to be controversial at the moment) and used those in
1019       Posix.ProcEnv.times.
1020    
1021    2. Added Time.{from,to}Nanoseconds to Time.
1022    
1023    3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
1024       For example, fromLargeInt never needs to look at more than 3 "big
1025       digits" to get its 53 bits of precision.
1026    
1027    ----------------------------------------------------------------------
1028    Name: Matthias Blume (blume (at) tti - c (dot) org)
1029    Date: 2003/09/17 16:30:00 CDT
1030    Tag: blume-20030917-real32-slices
1031    Description:
1032    
1033    Added an entry to the primitive environment
1034    (compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
1035    added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
1036    
1037    Removed some of the "magic" constants in real64.sml and replaced them
1038    with code that generates these values from their corresponding
1039    integer counterparts.
1040    
1041    Made all(?) the slice-related changes to the Basis and made everything
1042    compile again...
1043    
1044    ----------------------------------------------------------------------
1045    Name: Matthias Blume (blume (at) tti - c (dot) org)
1046    Date: 2003/09/15 17:45:00 CDT
1047    Tag: blume-20030915-rbase
1048    Description:
1049    
1050    Fixed bug in Real.fromLargeInt.
1051    
1052    ----------------------------------------------------------------------
1053    Name: Matthias Blume (blume (at) tti - c (dot) org)
1054    Date: 2003/09/13 18:11:00 CDT
1055    Tag: blume-20030913-libinstall
1056    Description:
1057    
1058    Minor bugfix in config/libinstall (set anchor with path to
1059    standalone tool after installing it, otherwise libraries that
1060    need ml-lex or ml-yacc won't compile the first time the installer
1061    runs).
1062    
1063    ----------------------------------------------------------------------
1064    Name: Matthias Blume (blume (at) tti - c (dot) org)
1065    Date: 2003/09/12 11:45:00 CDT
1066    Tag: blume-20030912-various
1067    Description:
1068    
1069     - fixed bug in Real.toLargeInt
1070     - fixed bug in Posix.ProcEnv.times
1071     - changed inputLine functions to return an option
1072     - minor installer improvements / bugfixes
1073     - changed default @SMLalloc parameter for x86/celeron to 64k
1074    
1075    ----------------------------------------------------------------------
1076    Name: Matthias Blume (blume (at) tti - c (dot) org)
1077    Date: 2003/09/09 22:00:00 CDT
1078    Tag: Release_110_43
1079    Description:
1080    
1081        New working release 110.43.  New bootfiles.
1082    
1083    ----------------------------------------------------------------------
1084    Name: Matthias Blume (blume (at) tti - c (dot) org)
1085    Date: 2003/09/09 19:20:00 CDT
1086    Tag: blume-20030909-installer
1087    Description:
1088    
1089    Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
1090    Modified config/install.bat to take advantage of it.  Also modified
1091    config/install.sh (and called it config/new-install.sh) to take advantage
1092    of it on Unix systems.  (The SML code is (supposed to be) platform-
1093    independent.)
1094    
1095    The installer can now install everything under Win32
1096    as well as under *nix as long as it compiles.
1097    
1098    Other changes:
1099    
1100      - made CML compile again under Win32
1101      - made eXene compile under Win32 (by providing a fake structure UnixSock
1102        and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
1103      - fixed a bug in nowhere: it assumed that type OS.Process.status is the
1104        same as type int;  under Win32 it isn't
1105      - fixed some slice-related problems in the win32-specific parts of CML
1106      - added a functor argument "sameVol" to os-path-fn.sml in the Basis
1107        (under Win32, the volume name is case-insensitive, and the
1108        OS.Path code compares volume names for equality)
1109    
1110    ----------------------------------------------------------------------
1111    Name: Matthias Blume (blume (at) tti - c (dot) org)
1112    Date: 2003/09/08 11:55:00 CDT
1113    Tag: blume-20030908-fullpath
1114    Description:
1115    
1116    Made Win32 version of OS.FileSys.fullPath return current directory
1117    when given an empty string.  This is what the spec says, and incidentally,
1118    CM depends on it.  (CM otherwise goes into an infinite loop in certain
1119    cases when presented with the name of a non-existing .cm file.)
1120    
1121    ----------------------------------------------------------------------
1122    Name: Matthias Blume (blume (at) tti - c (dot) org)
1123    Date: 2003/09/04 16:30:00 CDT
1124    Tag: blume-20030905-slices-etc
1125    Description:
1126    
1127      1. Changed interface to vectors and arrays in Basis to match
1128         (draft) Basis spec.
1129      2. Added signatures and implementations of slices according to
1130         Basis spec.
1131      3. Edited source code throughout the system to make it compile again
1132         under 1. and 2.  (In some cases code had to be added to have it
1133         match the new signatures.)
1134      4. MLRISC should be backward-compatible: the copies of the originals
1135         of files that needed to change under 3. were retained, the .cm files
1136         check the compiler version number and use old versions when
1137         appropriate.
1138      5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
1139         dirstream -> string option (in accordance with Basis spec).
1140      6. When generating code that counts lines, ml-lex used function
1141         CharVector.foldli, taking advantage of its old interface.
1142         This has been replaced with the corresponding code from
1143         CharVectorSlice.  (html-lex must be re-lexed!)
1144      7. BitArray in smlnj-lib/Util has been extended/modified to match the
1145         new MONO_ARRAY signature.  (Do we need BitArraySlice?)
1146      8. Removed temporary additions (fromInternal, toInternal) from the
1147         (now obsolete) IntInf in smlnj-lib/Util.
1148      9. Cleaned up structure Byte.
1149     10. Added localOffset, scan, and fromString to Date (according to spec).
1150         Cleaned/corrected implementation of Date.
1151         (Still need to check for correctness; implement better canonicalizeDate.)
1152     11. Added "scan" to signature IEEE_REAL.
1153     12. Some improvements to IntInf [in particular: efficiency-hack for
1154         mod and rem when second operand is 2 (for parity checks).]
1155     13. Changed representation of type Time.time, using a single IntInf.int
1156         value counting microseconds.  This considerably simplified the
1157         implementation of structure Time.  We now support negative time
1158         values; scan and fromString handle signs.
1159     14. Functor PrimIO now takes two additional arguments (VectorSlice and
1160         ArraySlice).
1161    
1162    ----------------------------------------------------------------------
1163    Name: Matthias Blume (blume (at) tti - c (dot) org)
1164    Date: 2003/08/28 17:00:00 CDT
1165    Tag: blume-20030828-intinf
1166    Description:
1167    
1168    This is a major update which comes with a version number bump
1169    (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
1170    and an implementation of IntInf in the Basis.
1171    
1172    There are a fairly large number of related changes and updates throughout
1173    the system:
1174    
1175    Basis:
1176            - Implemented IntInf.
1177            - Made LargeInt a projection of IntInf (by filtering through INTEGER).
1178            - Added some missing Real64 operations, most notably Real.toLargeInt.
1179            - Added FixedInt as a synonym for Int32.
1180    
1181    compiler:
1182            * Added support for a built-in intinf type.
1183              - literals
1184              - pattern matching
1185              - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
1186              - overloading on literals and operations
1187    
1188            This required adding a primitive type intinf, some additional
1189            primops, and implementations for several non-trivial intinf
1190            operations in Core.  (The intinf type is completely abstract
1191            to the compiler; all operations get delegated back to the Core.)
1192    
1193            * Intinf equality is handled by polyequal.  However, the compiler
1194            does not print its usual warning in this case (since polyequal
1195            is the right thing to do there).
1196    
1197            * Improved the organization of structure InlineT.
1198    
1199            * A word about conversion primops:
1200            If conversions involving intinf do not cancel out during
1201            CPS contract, then the compiler must insert calls to Core functions.
1202            Since all core access must be resolved already during the FLINT
1203            translate phase, it would be too late a the time of CPS contract
1204            to add new Core calls.  For this reason, conversion primops
1205            for intinf carry two arguments:  1. the numeric argument that
1206            they are supposed to convert, and 2. the Core function that
1207            can help with this conversion if necessary.  If CPS contract
1208            eliminates a primop, then the associated Core function becomes
1209            dead and goes away.  Intinf conversion primops that do not get
1210            eliminated by CPS contract get rewritten into calls of their
1211            core functions by a separate, new phase.
1212    
1213    interactive system:
1214            - Control.Print.intinfDepth controls max length of intinf constants
1215            being printed. (Analogous to Control.Print.stringDepth.)
1216            - Cleanup in printutil and pputil: got rid of unused stuff and
1217            duplicates; replaced some of the code with code that makes better
1218            use of library functionality.
1219    
1220    CM:
1221            Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
1222            are no longer silent.
1223    
1224    CKIT:
1225            Fixed mismatched uses of Int32 and LargeInt.  I always decided
1226            in favor of LargeInt -- which is now the same as IntInf.
1227            CKIT-knowledgable people should check whether this is what's
1228            intended and otherwise change things back to using Int32 or
1229            FixedInt.
1230    
1231    Throughout the code:
1232            Started using IntInf.int literals and built-in operations
1233            (e.g., comparison with 0) where this seems appropriate.
1234    
1235    
1236    ----------------------------------------------------------------------
1237    Name: Dave MacQueen (dbm@cs.uchicago.edu)
1238    Date: 2003/08/13 11:36:00 CDT
1239    Tag: dbm-20030813-mcz-merge1
1240    Description:
1241    
1242    Merging changes from the mcz-branch development branch into trunk.
1243    These changes involve replacement of the emulated old prettyprinter
1244    interface with direct use of the SML/NJ Lib PP library, and fixing
1245    of a couple of bugs (895, 1186) relating to error messages.  A new
1246    prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
1247    has been added.
1248    
1249    ----------------------------------------------------------------------
1250    Name: Matthias Blume (blume (at) tti - c (dot) org)
1251    Date: 2003/08/11 15:45:00 CDT
1252    Tag: blume-20030811-windows
1253    Description:
1254    
1255    Version number bumped to 110.42.9.  NEW BOOTFILES!!!
1256    
1257        http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
1258    
1259    This patch restores SML/NJ's ability to run under win32.  There are a
1260    number of changes, including fixes for several bugs that had gone
1261    unnoticed until now:
1262    
1263      - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
1264        trying to run the win32 version from within cygwin.)
1265    
1266      - There are a number of simple .bat scripts that substitute for their
1267        corresponding Unix shell-scripts.  (See below.)
1268    
1269      - The internals of ml-build have been modified slightly.  The main
1270        difference is that instead of calling ".link-sml" (or link-sml.bat)
1271        using OS.Process.system, the ML process delegates this task back
1272        to the script.  Otherwise problems arise in mixed environments such
1273        as Cygwin where scripts look and work like Unix scripts, but
1274        where OS.Process.system cannot run them.
1275    
1276      - In CM, the srcpath pickler used native pathname syntax -- which
1277        is incorrect in the case of cross-compilation.  The new pickle format
1278        is independent of platform-specific naming conventions.
1279    
1280      - Path configuration files (such as lib/pathconfig) can now choose
1281        between native and standard syntax.  Placing a line of the form
1282    
1283            standard!
1284    
1285        into the file causes all subsequent paths to be interpreted using
1286        CM standard pathname syntax (= Unix conventions); a line
1287    
1288            native!
1289    
1290        switches back to native style.  This was needed so that
1291        path config files can be written portably, see src/system/pathconfig.
1292    
1293      - Runtime system:
1294    
1295          - win32-filesys.c:  get_file_time and set_file_time now
1296            access modification time, not creation time.
1297    
1298          - I/O code made aware of new array representation.
1299    
1300          - Bug fixes in X86.prim.masm.
1301    
1302       - src/system/makeml made aware of win32. (For use under cygwin
1303            and other Unix-environments for windows.)
1304    
1305       - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
1306         which caused BinIO.inputAll to fail consistently.
1307    
1308    .bat scripts:
1309    
1310       Windows .bat scripts assume that SMLNJ_HOME is defined.
1311    
1312       - sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
1313         applications (sml, ml-yacc, ml-lex).
1314       - ml-build.bat: analogous to ml-build.
1315       - config\install.bat:  Analogous to config/install.sh.  This requires
1316         that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
1317         (nmake etc. must be on the path, and vcvars32 must have been run.)
1318         Moreover, sources for ml-lex and ml-yacc need to exist under src,
1319         and the bootfile hierarchy must have been unpacked under
1320         sml.boot.x86-win32.
1321         The script is very primitive and does a poor job at error checking.
1322         It only installs the base system, ml-lex, and ml-yacc.  No other
1323         libraries are being installed (i.e., you get only those that
1324         are part of the compiler.)
1325       - link-sml.bat:  analogous to .link-sml, but not currently used
1326    
1327    Unrelated bug fixes:
1328    
1329      - ml-nlffigen now exports structures ST_* corresponding to incomplete
1330        types.
1331      - Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
1332        otherwise.)
1333    
1334    ----------------------------------------------------------------------
1335    Name: Dave MacQueen (macqueen@cs.uchicago.edu)
1336    Date: 2003/06/17
1337    Tag: macqueen-20030617-bug895
1338    Description:
1339    
1340    Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
1341    Tag will be used for new development branch (mcz-branch) for
1342    use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
1343    
1344    ----------------------------------------------------------------------
1345    Name: Matthias Blume (blume (at) tti - c (dot) org)
1346    Date: 2003/05/27 16:55:00 CDT
1347    Tag: blume-20030527-polyeq
1348    Description:
1349    
1350    Tried to eliminated most cases of polymorphic equality.
1351    
1352    ----------------------------------------------------------------------
1353    Name: Matthias Blume (blume (at) tti - c (dot) org)
1354    Date: 2003/05/21 17:45:00 CDT
1355    Tag: blume-20030517-complete
1356    Description:
1357    
1358    Two changes:
1359    
1360      1. Added a flag for controlling whether non-exhaustive bindings will
1361         be treated as errors (default is false).
1362      2. Cleaned up the *entire* source tree so that CMB.make goes through
1363         without a single non-exhaustive match- or bind warning.
1364    
1365    ----------------------------------------------------------------------
1366    Name: Matthias Blume (blume (at) tti - c (dot) org)
1367    Date: 2003/05/17 10:20:00 CDT
1368    Tag: blume-20030517-absyn
1369    Description:
1370    
1371    1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
1372    
1373       This mainly affects the quality of error messages.  However, some
1374       of the code is now more straightforward than before. (Treatment of
1375       the above four constructs in translate.sml is much simpler than
1376       the "macro-expansion" that was going on before.  Plus, the mach-
1377       compiler no longer gets invoked just to be able to compile an
1378       if-expression.)
1379    
1380    2. The ErrorMsg.Error exception is now caught and absorbed by the
1381       interactive loop.
1382    
1383    ----------------------------------------------------------------------
1384    Name: Allen Leung
1385    Date: 2003/05/16 13:05:00 CDT
1386    Tag: leunga-20030516-cygwin-runtime
1387    Description:
1388    
1389        Ported the runtime system to cygwin, which uses the unix
1390    x86-unix bin files.  Missing/buggy features:
1391    
1392       o getnetbyname, getnetbyaddr: these functions seem to be missing in
1393         the Cygwin library.
1394       o Ctrl-C handling may be flaky.
1395       o Windows system calls and Windows I/O are not supported.
1396    
1397       A new set of binfiles is located at:
1398    
1399         http://www.dorsai.org/~leunga/boot.x86-unix.tgz
1400    
1401       This is only needed for bootstrapping the cygwin version of smlnj.
1402       Other x86 versions can use the existing binfiles.
1403    
1404    ----------------------------------------------------------------------
1405    Name: Matthias Blume
1406    Date: 2003/04/08 15:42:00 CDT
1407    Tag: blume-20030408-listpair
1408    Description:
1409    
1410    1. Added a target 'mlrisc' to installer.
1411    
1412    2. Added missing elements to structure ListPair.
1413    
1414    ----------------------------------------------------------------------
1415    Name: Allen Leung
1416    Date: 2003/01/07 10:40:00 EST
1417    Tag: leunga-20030107-int-rem
1418    Description:
1419    
1420    Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
1421    The arguments to the SUBL instruction were swapped.
1422    
1423    ----------------------------------------------------------------------
1424    Name: Matthias Blume
1425    Date: 2002/12/12 16:25:00 EST
1426    Tag: blume-20021212-risc-ra
1427    Description:
1428    
1429    Fixed a serious bug in the rewrite code for FP spilling/reloading that
1430    sent the RA into an infinite loop when floating point registers get
1431    spilled.  (Because of this bug, e.g., nucleic stopped compiling between
1432    110.37 and 110.38.)
1433    There was another set of potential problems related to the handling of
1434    MLRISC annotations (but those did not yet cause real problems, apparently).
1435    
1436    ----------------------------------------------------------------------
1437    Name: Matthias Blume
1438    Date: 2002/12/06 22:40:00 EST
1439    Tag: blume-20021206-cm-fileid
1440    Description:
1441    
1442    Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
1443    This fixes the problem of CM getting confused by files that suddenly
1444    change their identity (e.g., by getting unlinked and recreated by some
1445    text editor such as vi).  There might be a better/cheaper/cleaner way
1446    of doing this, but for now this will have to do.
1447    
1448    ----------------------------------------------------------------------
1449    Name: Matthias Blume
1450    Date: 2002/10/28 09:50:00 EST
1451    Tag: blume-20021028-typecheck
1452    Description:
1453    
1454    Exported structure Typecheck from $smlnj/viscomp/core.cm.
1455    
1456    ----------------------------------------------------------------------
1457    Name: Matthias Blume
1458    Date: 2002/10/17 09:10:00 EDT
1459    Tag: Release_110_42
1460    Description:
1461    
1462    In good old tradition, there has been a slight hiccup so that we have
1463    to patch 110.42 after the fact.   The old release tag has been replaced
1464    (see below).
1465    
1466    The change solves a problem with two competing approaches the
1467    configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
1468    each other's way.
1469    
1470    This change only affects the runtime system code and the installer script.
1471    (No new bootfiles.)
1472    
1473    ----------------------------------------------------------------------
1474    Name: Matthias Blume
1475    Date: 2002/10/16 12:00:00 EDT
1476    Tag: Release_110_42_removed
1477    Description:
1478    
1479    New working release.  New bootfiles.
1480    
1481    ----------------------------------------------------------------------
1482    Name: Matthias Blume
1483    Date: 2002/10/10 13:10:00 EDT
1484    Tag: blume-20021010-ppc-divs
1485    Description:
1486    
1487    The mltree operator DIVS must be implemented with an overflow check on
1488    the PPC because the hardware indicates divide-by-zero using "overflow" as
1489    well.
1490    
1491    ----------------------------------------------------------------------
1492    Name: Matthias Blume
1493    Date: 2002/07/23 11:45:00
1494    Tag: blume-20020723-smlnj-home
1495    Description:
1496    
1497    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
1498    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
1499    CM_PATHCONFIG is also set), the path configuration file is assumed
1500    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
1501    the entire tree to some other place and everything will "just work".
1502    
1503    (Companion commands such as ml-build and ml-makedepend also sense this
1504    variable.)
1505    
1506    ----------------------------------------------------------------------
1507    Name: Matthias Blume
1508    Date: 2002/07/12 21:19:00 EDT
1509    Tag: blume-20020712-liveness
1510    Description:
1511    
1512    Exported two useful "step" functions from liveness module (MLRISC).
1513    
1514    ----------------------------------------------------------------------
1515    Name: Matthias Blume
1516    Date: 2002/07/05 16:00 EDT
1517    Tag: Release_110_41
1518    Description:
1519    
1520    New working release.  New bootfiles.
1521    
1522    ----------------------------------------------------------------------
1523    Name: Matthias Blume
1524    Date: 2002/07/05 10:25:00 EDT
1525    Tag: blume-20020705-btimp
1526    Description:
1527    
1528    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
1529    other clients can set up backtracing support.
1530    
1531    ----------------------------------------------------------------------
1532    Name: Matthias Blume
1533    Date: 2002/06/25 17:23:00 EDT
1534    Tag: blume-20020625-fpmax
1535    Description:
1536    
1537    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
1538    case.  (The sense of the isNaN test was reversed -- which made min and
1539    max always return their first argument.)
1540    
1541    ----------------------------------------------------------------------
1542    Name: Matthias Blume
1543    Date: 2002/06/11
1544    Tag: blume-20020611-unixpath
1545    Description:
1546    
1547    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
1548    
1549    ----------------------------------------------------------------------
1550    Name: Matthias Blume
1551    Date: 2002/06/10 16:35:00 EDT
1552    Tag: blume-20020610-ieeereal
1553    Description:
1554    
1555    I back-ported my implementation of IEEEReal.fromString from the
1556    idlbasis-devel branch so that we can test it.
1557    
1558    Another small change is that ppDec tries to give more information
1559    than just "<sig>" in the case of functors.  However, this code is
1560    broken in some mysterious way if the functor's body's signature
1561    has not been declared by ascription but gets inferred from the
1562    implementation.  This needs fixing...
1563    
1564    ----------------------------------------------------------------------
1565    Name: Matthias Blume
1566    Date: 2002/05/31
1567    Tag: blume-20020531-btrace-mode
1568    Description:
1569    
1570    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
1571    the wayside when I switched over to using Controls everywhere.)
1572    
1573    ----------------------------------------------------------------------
1574    Name: Lal George
1575    Date: 2002/05/23 12:21:40 EDT
1576    Tag: george-20020523-visual-labels
1577    Description:
1578    
1579    Labels are now displayed in the graphical output to make
1580    the fall-through and target blocks obvious.
1581    
1582    ----------------------------------------------------------------------
1583    Name: Matthias Blume
1584    Date: 2002/05/22 11:03:00 EDT
1585    Tag: blume-20020522-shrink
1586    Description:
1587    
1588    John tweaked yesterday's fix for 1131 to handle an out-of-memory
1589    situation that comes up when allocating huge arrays.
1590    
1591    ----------------------------------------------------------------------
1592    Name: Matthias Blume
1593    Date: 2002/05/21 16:00:00 EDT
1594    Tag: Release_110_40
1595    Description:
1596    
1597    New working release (110.40).  New bootfiles.
1598    
1599    [Also: John Reppy fixed GC bug 1131.]
1600    
1601    ----------------------------------------------------------------------
1602    Name: Matthias Blume
1603    Date: 2002/05/21 12:35:00 EDT
1604    Tag: blume-20020521-cmdoc
1605    Description:
1606    
1607    CM documentation update.
1608    
1609    ----------------------------------------------------------------------
1610    Name: Matthias Blume
1611    Date: 2002/05/21 10:55:00 EDT
1612    Tag: blume-20020521-misc
1613    Description:
1614    
1615    - John tweaked runtime to be silent on heap export (except when
1616      GC messages are on).
1617    - I added a few more things (cross-compiling versions of CMB) to
1618      config/preloads (as suggestions).
1619    
1620    ----------------------------------------------------------------------
1621    Name: Matthias Blume
1622    Date: 2002/05/20 22:25:00 EDT
1623    Tag: blume-20020520-controls
1624    Description:
1625    
1626    - Added ControlUtil structure to control-lib.cm.
1627    - Use it throughout.
1628    - Used Controls facility to define MLRISC controls (as opposed to
1629      registering MLRISC control ref cells with Controls after the fact)
1630    - Fixed messed-up controls priorities.
1631    
1632    * Removed again all the stuff from config/preloads that one wouldn't
1633      be able to preload at the time the initial heap image is built.
1634      (Many libraries, e.g., CML, do not exist yet at this time.  The
1635       only libraries that can be preloaded via config/preloads are those
1636       that come bundled with the bootfiles.)
1637    
1638    ----------------------------------------------------------------------
1639    Name: Matthias Blume
1640    Date: 2002/05/20 10:59:00 EDT
1641    Tag: blume-20020520-preloads
1642    Description:
1643    
1644    Added a lot of commented-out suggestions for things to be included
1645    in config/preloads.
1646    
1647    ----------------------------------------------------------------------
1648    Name: Allen Leung
1649    Date: 2002/05/18 14:20:00 EDT
1650    Tag: leunga-20020518-mdl
1651    Description:
1652    
1653    o Made the mdl tool stuff compile and run again.
1654    o I've disabled all the stuff that depends on RTL specifications; they
1655      are all badly broken anyway.
1656    
1657    ----------------------------------------------------------------------
1658    Name: Matthias Blume
1659    Date: 2002/05/17 16:49:00 EDT
1660    Tag: blume-20020517-controls
1661    Description:
1662    
1663    0. John Reppy made several modifications to the SML/NJ library.
1664       In particular, there is a shiny new controls-lib.cm.
1665    
1666    1. Pushed new controls interface through compiler so that everything
1667       compiles again.
1668    
1669    2. Added FormatComb and FORMAT_COMB to the CML version of the
1670       SML/NJ library (so that CML compiles again).
1671    
1672    3. Modified init scripts because XXX_DEFAULT environment variables
1673       are no longer with us.  (Boot-time initialization is now done
1674       using the same environment variables that are also used for
1675       startup-time initialization of controls.)
1676    
1677    ----------------------------------------------------------------------
1678    Name: Lal George
1679    Date: 2002/05/15 09:20:10 EDT
1680    Tag: george-20020515-pseudo-op-decls
1681    Description:
1682    
1683            All pseudo-ops emitted before the first segment declaration
1684    such as TEXT, DATA, and BSS directives are assumed to be global
1685    declarations and are emitted first in the assembly file. This is
1686    useful in a number of situations where one has pseudo-ops that are not
1687    specific to any segment, and also works around the constraint that one
1688    cannot have client pseudo-ops in the TEXT segment.
1689    
1690            Because no segment is associated with these declarations it is
1691    an error to allocate any space or objects before the first segment
1692    directive and an exception will be raised. However, we cannot make
1693    this check for client pseudo-ops.
1694    
1695            These top level declarations are a field in the CFG graph_info.
1696    In theory you can continue to add to this field after the CFG has been
1697    built -- provided you know what you are doing;-)
1698    
1699    ----------------------------------------------------------------------
1700    Name: Matthias Blume
1701    Date: 2002/05/13 16:40:00 EDT
1702    Tag: blume-20020513-pp-etc
1703    Description:
1704    
1705    A few minor bugfixes:
1706    
1707      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
1708        (Bogus printouts for redefined bindings still occur.  Compiler
1709        bug should no longer occur now.  We need to redo the prettyprinter
1710        from scratch.)
1711    
1712      - CM pathname printer now also adds escape sequences for ( and )
1713    
1714      - commend and docu fixes for ml-nlffi
1715    
1716    ----------------------------------------------------------------------
1717    Name: Matthias Blume
1718    Date: 2002/05/10 16:40:00 EDT
1719    Tag: blume-20020510-erg-textio
1720    Description:
1721    
1722    Applied the following bugfix provided by Emden Gansner:
1723    
1724        Output is corrupted when outputSubstr is used rather than output.
1725    
1726        The problem occurs when a substring
1727    
1728            ss = (s, dataStart, dataLen)
1729    
1730        where dataStart > 0, fills a stream buffer with avail bytes left.
1731        avail bytes of s, starting at index dataStart, are copied into the
1732        buffer, the buffer is flushed, and then the remaining dataLen-avail
1733        bytes of ss are copied into the beginning of the buffer. Instead of
1734        starting this copy at index dataStart+avail in s, the current code
1735        starts the copy at index avail.
1736    
1737        Fix:
1738        In text-io-fn.sml, change line 695 from
1739             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
1740        to
1741             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
1742    
1743    ----------------------------------------------------------------------
1744    Name: Matthias Blume
1745    Date: 2002/04/12 13:55:00 EDT
1746    Tag: blume-20020412-assyntax
1747    Description:
1748    
1749    1. Grabbed newer assyntax.h from the XFree86 project.
1750    2. Fiddled with how to compile X86.prim.asm without warnings.
1751    3. (Very) Minor cleanup in CM.
1752    
1753    ----------------------------------------------------------------------
1754    Name: Matthias Blume
1755    Date: 2002/04/01 (no joke!) 17:07:00 EST
1756    Tag: blume-20020401-x86div
1757    Description:
1758    
1759    Added full support for div/mod/rem/quot on the x86, using the machine
1760    instruction's two results (without clumsily recomputing the remainder)
1761    directly where appropriate.
1762    
1763    Some more extensive power-of-two support was added to the x86 instruction
1764    selector (avoiding expensive divs, mods, and muls where they can be
1765    replaced with cheaper shifts and masks).  However, this sort of thing
1766    ought to be done earlier, e.g., within the CPS optimizer so that
1767    all architectures benefit from it.
1768    
1769    The compiler compiles to a fixed point, but changes might be somewhat
1770    fragile nevertheless.  Please, report any strange things that you might
1771    see wrt. div/mod/quot/rem...
1772    
1773    ----------------------------------------------------------------------
1774    Name: Matthias Blume
1775    Date: 2002/03/29 17:22:00
1776    Tag: blume-20020329-div
1777    Description:
1778    
1779    Fixed my broken div/mod logic.  Unfortunately, this means that the
1780    inline code for div/mod now has one more comparison than before.
1781    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
1782    The problem was with quotient = 0, because that alone does not tell
1783    us which way the rounding went.  One then has to look at whether
1784    remainder and divisor have the same sign...  :(
1785    
1786    Anyway, I replaced the bootfiles with fresh ones...
1787    
1788    ----------------------------------------------------------------------
1789    Name: Matthias Blume
1790    Date: 2002/03/29 14:10:00 EST
1791    Tag: blume-20020329-inlprims
1792    Description:
1793    
1794    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
1795    
1796    Primops have changed. This means that the bin/boot-file formats have
1797    changed as well.
1798    
1799    To make sure that there is no confusion, I made a new version.
1800    
1801    
1802    CHANGES:
1803    
1804    * removed REMT from mltree (remainder should never overflow).
1805    
1806    * added primops to deal with divisions of all flavors to the frontend
1807    
1808    * handled these primops all the way through so they map to their respective
1809      MLRISC support
1810    
1811    * used these primops in the implementation of Int, Int32, Word, Word32
1812    
1813    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
1814    
1815    * parameterized INLMIN, INLMAX, and INLABS by a numkind
1816    
1817    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
1818      floating point
1819    
1820    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
1821      and Real (but Real.abs maps to a separate floating-point-only primop)
1822    
1823    
1824    TODO items:
1825    
1826    * Hacked Alpha32 instruction selection, disabling the selection of REMx
1827      instructions because the machine instruction encoder cannot handle
1828      them.  (Hppa, PPC, and Sparc instruction selection did not handle
1829      REM in the first place, and REM is supported by the x86 machine coder.)
1830    
1831    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
1832      selection phase.  (The two can be streamlined because the hardware
1833      delivers both quotient and remainder at the same time anyway.)
1834    
1835    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
1836      (Currently the behavior is inconsistent both across architectures and
1837      wrt. the draft Basis spec.)
1838    
1839    * Word8 should eventually be handled natively, too.
1840    
1841    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
1842      as if there currently is no execution path that could trigger it in
1843      SML/NJ.  (The assumptions underlying functions arith and promotable do not
1844      hold for things like multiplication and division.)
1845    
1846    ----------------------------------------------------------------------
1847    Name: Matthias Blume
1848    Date: 2002/03/27 16:27:00 EST
1849    Tag: blume-20020327-mlrisc-divisions
1850    Description:
1851    
1852    Added support for all four division operations (ML's div, mod, quot,
1853    and rem) to MLRISC.  In the course of doing so, I also rationalized
1854    the naming (no more annoying switch-around of DIV and QUOT), by
1855    parameterizing the operation by div_rounding_mode (which can be either
1856    DIV_TO_ZERO or DIV_TO_NEGINF).
1857    
1858    The generic MLTreeGen functor takes care of compiling all four
1859    operations down to only round-to-zero div.
1860    
1861    Missing pieces:
1862    
1863      * Doing something smarter than relying on MLTreeGen on architectures
1864        like, e.g., the x86 where hardware division delivers both quotient and
1865        remainder at the same time.  With this, the implementation of the
1866        round-to-neginf operations could be further streamlined.
1867    
1868      * Remove inlining support for div/mod/rem from the frontend and replace it
1869        with primops that get carried through to the backend.  Do this for all
1870        int and word types.
1871    
1872    ----------------------------------------------------------------------
1873    Name: Matthias Blume
1874    Date: 2002/03/25 17:25:00 EST
1875    Tag: blume-20020325-divmod
1876    Description:
1877    
1878    I improved (hopefully without breaking them) the implementation of Int.div,
1879    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
1880    advantage of the following observations:
1881    
1882      Let  q = x quot y      r = x rem y
1883           d = x div  y      m = x mod y
1884    
1885    where "quot" is the round-to-zero version of integer division that
1886    hardware usually provides.  Then we have:
1887    
1888         r = x - q * y        where neither the * nor the - will overflow
1889         d = if q >= 0 orelse x = q * y then q else q - 1
1890                              where neither the * nor the - will overflow
1891         m = if q >= 0 orelse r = 0 then r else r + y
1892                              where the + will not overflow
1893    
1894    This results in substantial simplification of the generated code.
1895    The following table shows the number of CFG nodes and edges generated
1896    for
1897            fun f (x, y) = x OPER y
1898            (* with OPER \in div, mod, quot, rem *)
1899    
1900    
1901        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
1902        --------------------------------------------------------
1903         div |         24 |         39 |         12 |         16
1904         mod |         41 |         71 |         12 |         16
1905        quot |          8 |         10 |          8 |         10
1906         rem |         10 |         14 |          8 |         10
1907    
1908    
1909    ----------------------------------------------------------------------
1910    Name: Matthias Blume
1911    Date: 2002/03/25 22:06:00 EST
1912    Tag: blume-20020325-cprotobug
1913    Description:
1914    
1915    Fixed a bug in cproto (c prototype decoder).
1916    
1917    ----------------------------------------------------------------------
1918    Name: Matthias Blume
1919    Date: 2002/03/25 16:00:00 EST
1920    Tag: blume-20020325-raw-primops
1921    Description:
1922    
1923    I did some cleanup to Allen's new primop code and
1924    replaced yesterday's bootfiles with new ones.
1925    (But they are stored in the same place.)
1926    
1927    ----------------------------------------------------------------------
1928    Name: Matthias Blume
1929    Date: 2002/03/24 22:40:00 EST
1930    Tag: blume-20020324-bootfiles
1931    Description:
1932    
1933    Made the bootfiles that Allen asked for.
1934    
1935    ----------------------------------------------------------------------
1936    Name: Allen Leung
1937    Date: 2002/03/23 15:50:00 EST
1938    Tag: leunga-20020323-flint-cps-rcc-primops
1939    Description:
1940    
1941      1. Changes to FLINT primops:
1942    
1943        (* make a call to a C-function;
1944         * The primop carries C function prototype information and specifies
1945         * which of its (ML-) arguments are floating point. C prototype
1946         * information is for use by the backend, ML information is for
1947         * use by the CPS converter. *)
1948      | RAW_CCALL of { c_proto: CTypes.c_proto,
1949                       ml_args: ccall_type list,
1950                       ml_res_opt: ccall_type option,
1951                       reentrant : bool
1952                     } option
1953       (* Allocate uninitialized storage on the heap.
1954        * The record is meant to hold short-lived C objects, i.e., they
1955        * are not ML pointers.  With the tag, the representation is
1956        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
1957        *)
1958      | RAW_RECORD of {tag:bool,sz:int}
1959      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
1960    
1961      2.  These CPS primops are now overloaded:
1962    
1963           rawload of {kind:numkind}
1964           rawstore of {kind:numkind}
1965    
1966          The one argument form is:
1967    
1968             rawload {kind} address
1969    
1970          The two argument form is:
1971    
1972             rawload {kind} [ml object, byte-offset]
1973    
1974      3. RAW_CCALL/RCC now takes two extra arguments:
1975    
1976         a. The first is whether the C call is reentrant, i.e., whether
1977            ML state should be saved and restored.
1978         b. The second argument is a string argument specifying the name of
1979            library and the C function.
1980    
1981         These things are currently not handled in the code generator, yet.
1982    
1983      4. In CProto,
1984    
1985         An encoding type of "bool" means "ml object" and is mapped into
1986         C prototype of PTR.  Note that "bool" is different than "string",
1987         even though "string" is also mapped into PTR, because "bool"
1988         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
1989    
1990      5. Pickler/unpicker
1991    
1992         Changed to handle RAW_RECORD and newest RAW_CCALL
1993    
1994      6. MLRiscGen,
1995    
1996         1. Changed to handle the new rawload/rawstore/rawrecord operators.
1997         2. Code for handling C Calls has been moved to a new module CPSCCalls,
1998            in the file CodeGen/cpscompile/cps-c-calls.sml
1999    
2000      7. Added the conditional move operator
2001    
2002             condmove of branch
2003    
2004         to cps.  Generation of this is still buggy so it is currently
2005         disabled.
2006    
2007    ----------------------------------------------------------------------
2008    Name: Lal George
2009    Date: 2002/03/22 14:18:25 EST
2010    Tag: george-20020322-cps-branch-prob
2011    Description:
2012    
2013    Implemented the Ball-Larus branch prediction-heuristics, and
2014    incorporated graphical viewers for control flow graphs.
2015    
2016    Ball-Larus Heuristics:
2017    ---------------------
2018    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
2019    
2020    By design it uses the Dempster-Shafer theory for combining
2021    probabilities.  For example, in the function:
2022    
2023        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
2024    
2025    the ball-larus heuristics predicts that the n=0 is unlikely
2026    (OH-heuristic), and the 'then' branch is unlikely because of the
2027    RH-heuristic -- giving the 'then' branch an even lower combined
2028    probability using the Dempster-Shafer theory.
2029    
2030    Finally, John Reppy's loop analysis in MLRISC, further lowers the
2031    probability of the 'then' branch because of the loop in the else
2032    branch.
2033    
2034    
2035    Graphical Viewing:
2036    ------------------
2037    I merely plugged in Allen's graphical viewers into the compiler. The
2038    additional code is not much. At the top level, saying:
2039    
2040            Control.MLRISC.getFlag "cfg-graphical-view" := true;
2041    
2042    will display the graphical view of the control flow graph just before
2043    back-patching.  daVinci must be in your path for this to work. If
2044    daVinci is not available, then the default viewer can be changed
2045    using:
2046    
2047            Control.MLRISC.getString "viewer"
2048    
2049    which can be set to "dot" or "vcg" for the corresponding viewers. Of
2050    course, these viewers must be in your path.
2051    
2052    The above will display the compilation unit at the level of clusters,
2053    many of which are small, boring, and un-interesting. Also setting:
2054    
2055            Control.MLRISC.getInt "cfg-graphical-view_size"
2056    
2057    will display clusters that are larger than the value set by the above.
2058    
2059    
2060    ----------------------------------------------------------------------
2061    Name: Matthias Blume
2062    Date: 2002/03/21 22:20:00 EST
2063    Tag: blume-20020321-kmp-bugfix
2064    Description:
2065    
2066    Changed the interface to the KMP routine in PreString and fixed
2067    a minor bug in one place where it was used.
2068    
2069    ----------------------------------------------------------------------
2070    Name: Allen Leung
2071    Date: 2002/03/21 20:30:00 EST
2072    Tag: leunga-20020321-cfg
2073    Description:
2074    
2075      Fixed a potential problem in cfg edge splitting.
2076    
2077    ----------------------------------------------------------------------
2078    Name: Allen Leung
2079    Date: 2002/03/21 17:15:00 EST
2080    Tag: leunga-20020321-x86-fp-cfg
2081    Description:
2082    
2083      1. Recoded the buggy parts of x86-fp.
2084    
2085         a. All the block reordering code has been removed.
2086            We now depend on the block placement phases to do this work.
2087    
2088         b. Critical edge splitting code has been simplified and moved into the
2089            CFG modules, as where they belong.
2090    
2091         Both of these were quite buggy and complex.  The code is now much, much
2092         simpler.
2093    
2094      2. X86 backend.
2095    
2096         a. Added instructions for 64-bit support.  Instruction selection for
2097            64-bit has not been committed, however, since that
2098            requires changes to MLTREE which haven't been approved by
2099            Lal and John.
2100    
2101         b. Added support for FUCOMI and FUCOMIP when generating code for
2102            PentiumPro and above.  We only generate these instructions in
2103            the fast-fp mode.
2104    
2105         c. Added cases for JP and JNP in X86FreqProps.
2106    
2107      3. CFG
2108    
2109         CFG now has a bunch of methods for edge splitting and merging.
2110    
2111      4. Machine description.
2112    
2113         John's simplification of MLTREE_BASIS.fcond broke a few machine
2114         description things:
2115    
2116         rtl-build.{sig,sml} and hppa.mdl fixed.
2117    
2118         NOTE: the machine description stuff in the repository is still broken.
2119               Again, I can't put my fixes in because that involves
2120               changes to MLTREE.
2121    
2122    ----------------------------------------------------------------------
2123    Name: Matthias Blume
2124    Date: 2002/03/20 15:55:00 EST
2125    Tag: blume-20020320-kmp
2126    Description:
2127    
2128    Implemented Knuth-Morris-Pratt string matching in PreString and used
2129    it for String.isSubstring, Substring.isSubstring, and
2130    Substring.position.
2131    
2132    (Might need some stress-testing.  Simple examples worked fine.)
2133    
2134    ----------------------------------------------------------------------
2135    Name: Matthias Blume
2136    Date: 2002/03/19 16:37:00 EST
2137    Tag: blume-20020319-witnesses
2138    Description:
2139    
2140    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
2141    
2142    This implements a generic mechanism for changing constness qualifiers
2143    anywhere within big C types without resorting to outright "casts".
2144    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
2145    modify the constness at the outermost level.)
2146    The implementation of "convert" is based on the idea of "witness"
2147    values -- values that are not used by the operation but whose types
2148    "testify" to their applicability.  On the implementation side, "convert"
2149    is simply a projection (returning its second curried argument).  With
2150    cross-module inlining, it should not result in any machine code being
2151    generated.
2152    
2153    ----------------------------------------------------------------------
2154    Name: Matthias Blume
2155    Date: 2002/03/15 16:40:00 EST
2156    Tag: blume-20020315-basis
2157    Description:
2158    
2159    Provided (preliminary?) implementations for
2160    
2161      {String,Substring}.{concatWith,isSuffix,isSubstring}
2162    
2163    and
2164    
2165      Substring.full
2166    
2167    Those are in the Basis spec but they were missing in SML/NJ.
2168    
2169    ----------------------------------------------------------------------
2170    Name: Matthias Blume
2171    Date: 2002/03/14 21:30:00 EST
2172    Tag: blume-20020314-controls
2173    Description:
2174    
2175    Controls:
2176    ---------
2177    
2178    1. Factored out the recently-added Controls : CONTROLS stuff and put
2179       it into its own library $/controls-lib.cm.  The source tree for
2180       this is under src/smlnj-lib/Controls.
2181    
2182    2. Changed the names of types and functions in this interface, so they
2183       make a bit more "sense":
2184    
2185          module -> registry
2186          'a registry -> 'a group
2187    
2188    3. The interface now deals in ref cells only.  The getter/setter interface
2189       is (mostly) gone.
2190    
2191    4. Added a function that lets one register an already-existing ref cell.
2192    
2193    5. Made the corresponding modifications to the rest of the code so that
2194       everything compiles again.
2195    
2196    6. Changed the implementation of Controls.MLRISC back to something closer
2197       to the original.  In particular, this module (and therefore MLRISC)
2198       does not depend on Controls.  There now is some link-time code in
2199       int-sys.sml that registers the MLRISC controls with the Controls
2200       module.
2201    
2202    CM:
2203    ---
2204    
2205      * One can now specify the lambda-split aggressiveness in init.cmi.
2206    
2207    ----------------------------------------------------------------------
2208    Name: Allen Leung
2209    Date: 2002/03/13 17:30:00 EST
2210    Tag: leunga-20020313-x86-fp-unary
2211    Description:
2212    
2213    Bug fix for:
2214    
2215    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
2216    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
2217    > - fun f(x,(y,z)) = Real.~ y;
2218    > [autoloading]
2219    > [autoloading done]
2220    >       fchsl   (%eax), 184(%esp)
2221    > Error: MLRisc bug: X86MCEmitter.emitInstr
2222    >
2223    > uncaught exception Error
2224    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
2225    
2226    The problem was that the code generator did not generate any fp registers
2227    in this case, and the ra didn't know that it needed to run the X86FP phase to
2228    translate the pseudo fp instruction.   This only happened with unary fp
2229    operators in certain situations.
2230    
2231    ----------------------------------------------------------------------
2232    Name: Matthias Blume
2233    Date: 2002/03/13 14:00:00 EST
2234    Tag: blume-20020313-overload-etc
2235    Description:
2236    
2237    1. Added _overload as a synonym for overload for backward compatibility.
2238       (Control.overloadKW must be true for either version to be accepted.)
2239    
2240    2. Fixed bug in install script that caused more things to be installed
2241       than what was requested in config/targets.
2242    
2243    3. Made CM aware of the (_)overload construct so that autoloading
2244       works.
2245    
2246    ----------------------------------------------------------------------
2247    Name: Matthias Blume
2248    Date: 2002/03/12 22:03:00 EST
2249    Tag: blume-20020312-url
2250    Description:
2251    
2252    Forgot to update BOOT and srcarchiveurl.
2253    
2254    ----------------------------------------------------------------------
2255    Name: Matthias Blume
2256    Date: 2002/03/12 17:30:00 EST
2257    Tag: blume-20020312-version110392
2258    Description:
2259    
2260    Yet another version number bump (because of small changes to the
2261    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
2262    
2263    Changes:
2264    
2265      The new pid generation scheme described a few weeks ago was overly
2266      complicated.  I implemented a new mechanism that is simpler and
2267      provides a bit more "stability":  Once CM has seen a compilation
2268      unit, it keeps its identity constant (as long as you do not delete
2269      those crucial CM/GUID/* files).  This means that when you change
2270      an interface, compile, then go back to the old interface, and
2271      compile again, you arrive at the original pid.
2272    
2273      There now also is a mechanism that instructs CM to use the plain
2274      environment hash as a module's pid (effectively making its GUID
2275      the empty string).  For this, "noguid" must be specified as an
2276      option to the .sml file in question within its .cm file.
2277      This is most useful for code that is being generated by tools such
2278      as ml-nlffigen (because during development programmers tend to
2279      erase the tool's entire output directory tree including CM's cached
2280      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
2281      revert to the old, broken behavior of SML/NJ, but in specific cases
2282      where there is no danger of interface confusion, its use is ok
2283      (I think).
2284    
2285      ml-nlffigen by default generates "noguid" annotations.  They can be
2286      turned off by specifying -guid in its command line.
2287    
2288    ----------------------------------------------------------------------
2289    Name: Lal George
2290    Date: 2002/03/12 12 14:42:36 EST
2291    Tag: george-20020312-frequency-computation
2292    Description:
2293    
2294    Integrated jump chaining and static block frequency into the
2295    compiler. More details and numbers later.
2296    
2297    ----------------------------------------------------------------------
2298    Name: Lal George
2299    Date: 2002/03/11 11 22:38:53 EST
2300    Tag: george-20020311-jump-chain-elim
2301    Description:
2302    
2303    Tested the jump chain elimination on all architectures (except the
2304    hppa).  This is on by default right now and is profitable for the
2305    alpha and x86, however, it may not be profitable for the sparc and ppc
2306    when compiling the compiler.
2307    
2308    The gc test will typically jump to a label at the end of the cluster,
2309    where there is another jump to an external cluster containing the actual
2310    code to invoke gc. This is to allow factoring of common gc invocation
2311    sequences. That is to say, we generate:
2312    
2313            f:
2314               testgc
2315               ja   L1      % jump if above to L1
2316    
2317            L1:
2318               jmp L2
2319    
2320    
2321    After jump chain elimination the 'ja L1' instructions is converted to
2322    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
2323    up being implemented in their long form (if L2 is far away) using:
2324    
2325            jbe     L3      % jump if below or equal to L3
2326            jmp     L2
2327         L3:
2328            ...
2329    
2330    
2331    For large compilation units L2  may be far away.
2332    
2333    
2334    ----------------------------------------------------------------------
2335    Name: Matthias Blume
2336    Date: 2002/03/11 13:30:00 EST
2337    Tag: blume-20020311-mltreeeval
2338    Description:
2339    
2340    A functor parameter was missing.
2341    
2342    ----------------------------------------------------------------------
2343    Name: Allen Leung
2344    Date: 2002/03/11 10:30:00 EST
2345    Tag: leunga-20020311-runtime-string0
2346    Description:
2347    
2348       The representation of the empty string now points to a
2349    legal null terminated C string instead of unit.  It is now possible
2350    to convert an ML string into C string with InlineT.CharVector.getData.
2351    This compiles into one single machine instruction.
2352    
2353    ----------------------------------------------------------------------
2354    Name: Allen Leung
2355    Date: 2002/03/10 23:55:00 EST
2356    Tag: leunga-20020310-x86-call
2357    Description:
2358    
2359       Added machine generation for CALL instruction (relative displacement mode)
2360    
2361    ----------------------------------------------------------------------
2362    Name: Matthias Blume
2363    Date: 2002/03/08 16:05:00
2364    Tag: blume-20020308-entrypoints
2365    Description:
2366    
2367    Version number bumped to 110.39.1.  NEW BOOTFILES!
2368    
2369    Entrypoints: non-zero offset into a code object where execution should begin.
2370    
2371    - Added the notion of an entrypoint to CodeObj.
2372    - Added reading/writing of entrypoint info to Binfile.
2373    - Made runtime system bootloader aware of entrypoints.
2374    - Use the address of the label of the first function given to mlriscGen
2375      as the entrypoint.  This address is currently always 0, but it will
2376      not be 0 once we turn on block placement.
2377    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
2378      with entry points) from mlriscGen.
2379    
2380    ----------------------------------------------------------------------
2381    Name: Allen Leung
2382    Date: 2002/03/07 20:45:00 EST
2383    Tag: leunga-20020307-x86-cmov
2384    Description:
2385    
2386       Bug fixes for CMOVcc on x86.
2387    
2388       1. Added machine code generation for CMOVcc
2389       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
2390       3. CMOVcc cannot have an immediate operand as argument.
2391    
2392    ----------------------------------------------------------------------
2393    Name: Matthias Blume
2394    Date: 2002/03/07 16:15:00 EST
2395    Tag: blume-20020307-controls
2396    Description:
2397    
2398    This is a very large but mostly boring patch which makes (almost)
2399    every tuneable compiler knob (i.e., pretty much everything under
2400    Control.* plus a few other things) configurable via both the command
2401    line and environment variables in the style CM did its configuration
2402    until now.
2403    
2404    Try starting sml with '-h' (or, if you are brave, '-H')
2405    
2406    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
2407    implements the underlying generic mechanism.
2408    
2409    The interface to some of the existing such facilities has changed somewhat.
2410    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
2411    (The getFoo interface is still there for backward-compatibility, but its
2412    use is deprecated.)
2413    
2414    The ml-build script passes -Cxxx=yyy command-line arguments through so
2415    that one can now twiddle the compiler settings when using this "batch"
2416    compiler.
2417    
2418    TODO items:
2419    
2420    We should go through and throw out all controls that are no longer
2421    connected to anything.  Moreover, we should go through and provide
2422    meaningful (and correct!) documentation strings for those controls
2423    that still are connected.
2424    
2425    Currently, multiple calls to Controls.new are accepted (only the first
2426    has any effect).  Eventually we should make sure that every control
2427    is being made (via Controls.new) exactly once.  Future access can then
2428    be done using Controls.acc.
2429    
2430    Finally, it would probably be a good idea to use the getter-setter
2431    interface to controls rather than ref cells.  For the time being, both
2432    styles are provided by the Controls module, but getter-setter pairs are
2433    better if thread-safety is of any concern because they can be wrapped.
2434    
2435    *****************************************
2436    
2437    One bug fix: The function blockPlacement in three of the MLRISC
2438    backpatch files used to be hard-wired to one of two possibilities at
2439    link time (according to the value of the placementFlag).  But (I
2440    think) it should rather sense the flag every time.
2441    
2442    *****************************************
2443    
2444    Other assorted changes (by other people who did not supply a HISTORY entry):
2445    
2446    1. the cross-module inliner now works much better (Monnier)
2447    2. representation of weights, frequencies, and probabilities in MLRISC
2448       changed in preparation of using those for weighted block placement
2449       (Reppy, George)
2450    
2451    ----------------------------------------------------------------------
2452    Name: Lal George
2453    Date: 2002/03/07 14:44:24 EST 2002
2454    Tag: george-20020307-weighted-block-placement
2455    
2456    Tested the weighted block placement optimization on all architectures
2457    (except the hppa) using AMPL to generate the block and edge frequencies.
2458    Changes were required in the machine properties to correctly
2459    categorize trap instructions. There is an MLRISC flag
2460    "weighted-block-placement" that can be used to enable weighted block
2461    placement, but this will be ineffective without block/edge
2462    frequencies (coming soon).
2463    
2464    
2465    ----------------------------------------------------------------------
2466    Name: Lal George
2467    Date: 2002/03/05 17:24:48 EST
2468    Tag: george-20020305-linkage-cluster
2469    
2470    In order to support the block placement optimization, a new cluster
2471    is generated as the very first cluster (called the linkage cluster).
2472    It contains a single jump to the 'real' entry point for the compilation
2473    unit. Block placement has no effect on the linkage cluster itself, but
2474    all the other clusters  have full freedom in the manner in which they
2475    reorder blocks or functions.
2476    
2477    On the x86 the typical linkage code that is generated is:
2478       ----------------------
2479            .align 2
2480       L0:
2481            addl    $L1-L0, 72(%esp)
2482            jmp     L1
2483    
2484    
2485            .align  2
2486       L1:
2487       ----------------------
2488    
2489    72(%esp) is the memory location for the stdlink register. This
2490    must contain the address of the CPS function being called. In the
2491    above example, it contains the address of  L0; before
2492    calling L1 (the real entry point for the compilation unit), it
2493    must contain the address for L1, and hence
2494    
2495            addl $L1-L0, 72(%esp)
2496    
2497    I have tested this on all architectures except the hppa.The increase
2498    in code size is of course negligible
2499    
2500    ----------------------------------------------------------------------
2501    Name: Allen Leung
2502    Date: 2002/03/03 13:20:00 EST
2503    Tag: leunga-20020303-mlrisc-tools
2504    
2505      Added #[ ... ] expressions to mlrisc tools
2506    
2507    ----------------------------------------------------------------------
2508    Name: Matthias Blume
2509    Date: 2002/02/27 12:29:00 EST
2510    Tag: blume-20020227-cdebug
2511    Description:
2512    
2513    - made types in structure C and C_Debug to be equal
2514    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
2515    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
2516    
2517    ----------------------------------------------------------------------
2518    Name: Matthias Blume
2519    Date: 2002/02/26 12:00:00 EST
2520    Tag: blume-20020226-ffi
2521    Description:
2522    
2523    1. Fixed a minor bug in CM's "noweb" tool:
2524       If numbering is turned off, then truly don't number (i.e., do not
2525       supply the -L option to noweb).  The previous behavior was to supply
2526       -L'' -- which caused noweb to use the "default" line numbering scheme.
2527       Thanks to Chris Richards for pointing this out (and supplying the fix).
2528    
2529    2. Once again, I reworked some aspects of the FFI:
2530    
2531       A. The incomplete/complete type business:
2532    
2533       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
2534         gone!
2535       - ML types representing an incomplete type are now *equal* to
2536         ML types representing their corresponding complete types (just like
2537         in C).  This is still safe because ml-nlffigen will not generate
2538         RTTI for incomplete types, nor will it generate functions that
2539         require access to such RTTI.   But when ML code generated from both
2540         incomplete and complete versions of the C type meet, the ML types
2541         are trivially interoperable.
2542    
2543         NOTE:  These changes restore the full generality of the translation
2544         (which was previously lost when I eliminated functorization)!
2545    
2546       B. Enum types:
2547    
2548       - Structure C now has a type constructor "enum" that is similar to
2549         how the "su" constructor works.  However, "enum" is not a phantom
2550         type because each "T enum" has values (and is isomorphic to
2551         MLRep.Signed.int).
2552       - There are generic access operations for enum objects (using
2553         MLRep.Signed.int).
2554       - ml-nlffigen will generate a structure E_foo for each "enum foo".
2555         * The structure contains the definition of type "mlrep" (the ML-side
2556         representation type of the enum).  Normally, mlrep is the same
2557         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
2558         then mlrep will be defined as a datatype -- thus facilitating
2559         pattern matching on mlrep values.
2560         ("-ec" will be suppressed if there are duplicate values in an
2561          enumeration.)
2562         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
2563         will be generated for each C enum constant xxx.
2564         * Conversion functions m2i and i2m convert between mlrep and
2565         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
2566         * Coversion functions c and ml convert between mlrep and "tag enum".
2567         * Access functions (get/set) fetch and store mlrep values.
2568       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
2569         enumerations are merged into one single enumeration represented by
2570         structure E_'.
2571    
2572    ----------------------------------------------------------------------
2573    Name: Allen Leung
2574    Date: 2002/02/25 04:45:00 EST
2575    Tag: leunga-20020225-cps-spill
2576    
2577    This is a new implementation of the CPS spill phase.
2578    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
2579    In case of problems, replace it with the old file spill.sml
2580    
2581    The current compiler runs into some serious performance problems when
2582    constructing a large record.  This can happen when we try to compile a
2583    structure with many items.  Even a very simple structure like the following
2584    makes the compiler slow down.
2585    
2586        structure Foo = struct
2587           val x_1 = 0w1 : Word32.int
2588           val x_2 = 0w2 : Word32.int
2589           val x_3 = 0w3 : Word32.int
2590           ...
2591           val x_N = 0wN : Word32.int
2592        end
2593    
2594    The following table shows the compile time, from N=1000 to N=4000,
2595    with the old compiler:
2596    
2597    N
2598    1000   CPS 100 spill                           0.04u  0.00s  0.00g
2599           MLRISC ra                               0.06u  0.00s  0.05g
2600              (spills = 0 reloads = 0)
2601           TOTAL                                   0.63u  0.07s  0.21g
2602    
2603    1100   CPS 100 spill                           8.25u  0.32s  0.64g
2604           MLRISC ra                               5.68u  0.59s  3.93g
2605              (spills = 0 reloads = 0)
2606           TOTAL                                   14.71u  0.99s  4.81g
2607    
2608    1500   CPS 100 spill                           58.55u  2.34s  1.74g
2609           MLRISC ra                               5.54u  0.65s  3.91g
2610              (spills = 543 reloads = 1082)
2611           TOTAL                                   65.40u  3.13s  6.00g
2612    
2613    2000   CPS 100 spill                           126.69u  4.84s  3.08g
2614           MLRISC ra                               0.80u  0.10s  0.55g
2615              (spills = 42 reloads = 84)
2616           TOTAL                                   129.42u  5.10s  4.13g
2617    
2618    3000   CPS 100 spill                           675.59u  19.03s  11.64g
2619           MLRISC ra                               2.69u  0.27s  1.38g
2620              (spills = 62 reloads = 124)
2621           TOTAL                                   682.48u  19.61s  13.99g
2622    
2623    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
2624           MLRISC ra                               4.96u  0.27s  2.72g
2625              (spills = 85 reloads = 170)
2626           TOTAL                                   2375.26u  57.21s  48.00g
2627    
2628    As you can see the old cps spill module suffers from some serious
2629    performance problem.  But since I cannot decipher the old code fully,
2630    instead of patching the problems up, I'm reimplementing it
2631    with a different algorithm.  The new code is more modular,
2632    smaller when compiled, and substantially faster
2633    (O(n log n) time and O(n) space).  Timing of the new spill module:
2634    
2635    4000  CPS 100 spill                           0.02u  0.00s  0.00g
2636          MLRISC ra                               0.25u  0.02s  0.15g
2637             (spills=1 reloads=3)
2638          TOTAL                                   7.74u  0.34s  1.62g
2639    
2640    Implementation details:
2641    
2642    As far as I can tell, the purpose of the CPS spill module is to make sure the
2643    number of live variables at any program point (the bandwidth)
2644    does not exceed a certain limit, which is determined by the
2645    size of the spill area.
2646    
2647    When the bandwidth is too large, we decrease the register pressure by
2648    packing live variables into spill records.  How we achieve this is
2649    completely different than what we did in the old code.
2650    
2651    First, there is something about the MLRiscGen code generator
2652    that we should be aware of:
2653    
2654    o MLRiscGen performs code motion!
2655    
2656       In particular, it will move floating point computations and
2657       address computations involving only the heap pointer to
2658       their use sites (if there is only a single use).
2659       What this means is that if we have a CPS record construction
2660       statement
2661    
2662           RECORD(k,vl,w,e)
2663    
2664       we should never count the new record address w as live if w
2665       has only one use (which is often the case).
2666    
2667       We should do something similar to floating point, but the transformation
2668       there is much more complex, so I won't deal with that.
2669    
2670    Secondly, there are now two new cps primops at our disposal:
2671    
2672     1. rawrecord of record_kind option
2673        This pure operator allocates some uninitialized storage from the heap.
2674        There are two forms:
2675    
2676         rawrecord NONE [INT n]  allocates a tagless record of length n
2677         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
2678                                     and initializes the tag.
2679    
2680     2. rawupdate of cty
2681          rawupdate cty (v,i,x)
2682          Assigns to x to the ith component of record v.
2683          The storelist is not updated.
2684    
2685    We use these new primops for both spilling and increment record construction.
2686    
2687     1. Spilling.
2688    
2689        This is implemented with a linear scan algorithm (but generalized
2690        to trees).  The algorithm will create a single spill record at the
2691        beginning of the cps function and use rawupdate to spill to it,
2692        and SELECT or SELp to reload from it.  So both spills and reloads
2693        are fine-grain operations.  In contrast, in the old algorithm
2694        "spills" have to be bundled together in records.
2695    
2696        Ideally, we should sink the spill record construction to where
2697        it is needed.  We can even split the spill record into multiple ones
2698        at the places where they are needed.  But CPS is not a good
2699        representation for global code motion, so I'll keep it simple and
2700        am not attempting this.
2701    
2702     2. Incremental record construction (aka record splitting).
2703    
2704        Long records with many component values which are simulatenously live
2705        (recall that single use record addresses are not considered to
2706         be live) are constructed with rawrecord and rawupdate.
2707        We allocate space on the heap with rawrecord first, then gradually
2708        fill it in with rawupdate.  This is the technique suggested to me
2709        by Matthias.
2710    
2711        Some restrictions on when this is applicable:
2712        1. It is not a VECTOR record.  The code generator currently does not handle
2713           this case. VECTOR record uses double indirection like arrays.
2714        2. All the record component values are defined in the same "basic block"
2715           as the record constructor.  This is to prevent speculative
2716           record construction.
2717    
2718    ----------------------------------------------------------------------
2719    Name: Allen Leung
2720    Date: 2002/02/22 01:02:00 EST
2721    Tag: leunga-20020222-mlrisc-tools
2722    
2723    Minor bug fixes in the parser and rewriter
2724    
2725    ----------------------------------------------------------------------
2726    Name: Allen Leung
2727    Date: 2002/02/21 20:20:00 EST
2728    Tag: leunga-20020221-peephole
2729    
2730    Regenerated the peephole files.  Some contained typos in the specification
2731    and some didn't compile because of pretty printing bugs in the old version
2732    of 'nowhere'.
2733    
2734    ----------------------------------------------------------------------
2735    Name: Allen Leung
2736    Date: 2002/02/19 20:20:00 EST
2737    Tag: leunga-20020219-mlrisc-tools
2738    Description:
2739    
2740       Minor bug fixes to the mlrisc-tools library:
2741    
2742       1.  Fixed up parsing colon suffixed keywords
2743       2.  Added the ability to shut the error messages up
2744       3.  Reimplemented the pretty printer and fixed up/improved
2745           the pretty printing of handle and -> types.
2746       4.  Fixed up generation of literal symbols in the nowhere tool.
2747       5.  Added some SML keywords to to sml.sty
2748    
2749    ----------------------------------------------------------------------
2750    Name: Matthias Blume
2751    Date: 2002/02/19 16:20:00 EST
2752    Tag: blume-20020219-cmffi
2753    Description:
2754    
2755    A wild mix of changes, some minor, some major:
2756    
2757    * All C FFI-related libraries are now anchored under $c:
2758        $/c.cm      --> $c/c.cm
2759        $/c-int.cm  --> $c/internals/c-int.cm
2760        $/memory.cm --> $c/memory/memory.cm
2761    
2762    * "make" tool (in CM) now treats its argument pathname slightly
2763      differently:
2764        1. If the native expansion is an absolute name, then before invoking
2765           the "make" command on it, CM will apply OS.Path.mkRelative
2766           (with relativeTo = OS.FileSys.getDir()) to it.
2767        2. The argument will be passed through to subsequent phases of CM
2768           processing without "going native".  In particular, if the argument
2769           was an anchored path, then "make" will not lose track of that anchor.
2770    
2771    * Compiler backends now "know" their respective C calling conventions
2772      instead of having to be told about it by ml-nlffigen.  This relieves
2773      ml-nlffigen from one of its burdens.
2774    
2775    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
2776    
2777    * Export C_DEBUG and C_Debug from $c/c.cm.
2778    
2779    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
2780      subtyping relationship between incomplete pointers and their complete
2781      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
2782      with the convention of instantiating 'o with ('t, 'c) obj whenever
2783      the pointer target type is complete.  In the incomplete case, 'o
2784      will be instantiated with some "'c iobj" -- a type obtained by
2785      using one of the functors PointerToIncompleteType or PointerToCompleteType.
2786    
2787      Operations that work on both incomplete and complete pointer types are
2788      typed as taking an 'o ptr while operations that require the target to
2789      be known are typed as taking some ('t, 'c) obj ptr.
2790    
2791      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
2792      where void is an eqtype without any values.  This makes it possible
2793      to work on voidptr values using functions meant to operate on light
2794      incomplete pointers.
2795    
2796    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
2797      been vastly simplified.
2798    
2799    ----------------------------------------------------------------------
2800    Name: Matthias Blume
2801    Date: 2002/02/19 10:48:00 EST
2802    Tag: blume-20020219-pqfix
2803    Description:
2804    
2805    Applied Chris Okasaki's bug fix for priority queues.
2806    
2807    ----------------------------------------------------------------------
2808    Name: Matthias Blume
2809    Date: 2002/02/15 17:05:00
2810    Tag: Release_110_39
2811    Description:
2812    
2813    Last-minute retagging is becoming a tradition... :-(
2814    
2815    This is the working release 110.39.
2816    
2817    ----------------------------------------------------------------------
2818    Name: Matthias Blume
2819    Date: 2002/02/15 16:00:00 EST
2820    Tag: Release_110_39-orig
2821    Description:
2822    
2823    Working release 110.39.  New bootfiles.
2824    
2825    (Update: There was a small bug in the installer so it wouldn't work
2826    with all shells.  So I retagged. -Matthias)
2827    
2828    ----------------------------------------------------------------------
2829    Name: Matthias Blume
2830    Date: 2002/02/15 14:17:00 EST
2831    Tag: blume-20020215-showbindings
2832    Description:
2833    
2834    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
2835    the latter can be useful for exploring what bindings are available at
2836    the interactive prompt.  (The first function returns only the list
2837    of symbols that are really bound, the second prints those but also the
2838    ones that CM's autoloading mechanism knows about.)
2839    
2840    ----------------------------------------------------------------------
2841    Name: Matthias Blume
2842    Date: 2002/02/15 12:08:00 EST
2843    Tag: blume-20020215-iptrs
2844    Description:
2845    
2846    Two improvements to ml-nlffigen:
2847    
2848      1. Write files only if they do not exist or if their current contents
2849         do not coincide with what's being written.  (That is, avoid messing
2850         with the time stamps unless absolutely necessary.)
2851    
2852      2. Implement a "repository" mechanism for generated files related
2853         to "incomplete pointer types".   See the README file for details.
2854    
2855    ----------------------------------------------------------------------
2856    Name: Matthias Blume
2857    Date: 2002/02/14 11:50:00 EST
2858    Tag: blume-20020214-quote
2859    Description:
2860    
2861    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
2862    because of the new and improved tag generation scheme.  (Thanks to Allen
2863    Leung for pointing it out.)
2864    
2865    ----------------------------------------------------------------------
2866    Name: Lal George
2867    Date: 2002/02/14 09:55:27 EST 2002
2868    Tag: george-20020214-isabelle-bug
2869    Description:
2870    
2871    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
2872    of Isabelle on the x86.
2873    
2874    From Allen:
2875    -----------
2876     I've found the problem:
2877    
2878         in ra-core.sml, I use the counter "blocked" to keep track of the
2879         true number of elements in the freeze queue.  When the counter goes
2880         to zero, I skip examining the queue.  But I've messed up the
2881         bookkeeping in combine():
2882    
2883             else ();
2884             case !ucol of
2885               PSEUDO => (if !cntv > 0 then
2886                     (if !cntu > 0 then blocked := !blocked - 1 else ();
2887                                        ^^^^^^^^^^^^^^^^^^^^^^^
2888                      moveu := mergeMoveList(!movev, !moveu)
2889                     )
2890                  else ();
2891    
2892         combine() is called to coalesce two nodes u and v.
2893         I think I was thinking that if the move counts of u and v are both
2894         greater than zero then after they are coalesced then one node is
2895         removed from the freeze queue.  Apparently I was thinking that
2896         both u and v are of low degree, but that's clearly not necessarily true.
2897    
2898    
2899    02/12/2002:
2900        Here's the patch.  HOL now compiles.
2901    
2902        I don't know how this impact on performance (compile
2903        time or runtime).  This bug caused the RA (especially on the x86)
2904        to go thru the potential spill phase when there are still nodes on the
2905        freeze queue.
2906    
2907    
2908    
2909    
2910    ----------------------------------------------------------------------
2911    Name: Matthias Blume
2912    Date: 2002/02/13 22:40:00 EST
2913    Tag: blume-20020213-fptr-rtti
2914    Description:
2915    
2916    Fixed a bug in ml-nlffigen that was introduced with one of the previous
2917    updates.
2918    
2919    ----------------------------------------------------------------------
2920    Name: Matthias Blume
2921    Date: 2002/02/13 16:41:00 EST
2922    Tag: blume-20020213-cmlpq
2923    Description:
2924    
2925    Added new priority queue export symbols (which have just been added to
2926    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
2927    not compile and the installer would choke.)
2928    
2929    ----------------------------------------------------------------------
2930    Name: Matthias Blume
2931    Date: 2002/02/13 16:15:00 EST
2932    Tag: blume-20020213-various
2933    Description:
2934    
2935    1. More tweaks to ml-nlffigen:
2936    
2937       - better internal datastructures (resulting in slight speedup)
2938       - "-match" option requires exact match
2939       - "localized" gensym counters (untagged structs/unions nested within
2940         other structs/unions or within typedefs get a fresh counter; their
2941         tag will be prefixed by a concatenation of their parents' tags)
2942       - bug fixes (related to calculation of transitive closure of types
2943         to be included in the output)
2944    
2945    2. Minor Basis updates:
2946    
2947       - added implementations for List.collate and Option.app
2948    
2949    ----------------------------------------------------------------------
2950    Name: Matthias Blume
2951    Date: 2002/02/11 15:55:00 EST
2952    Tag: blume-20020211-gensym
2953    Description:
2954    
2955    Added a "-gensym" option to command line of ml-nlffigen.  This can be
2956    used to specify a "stem" -- a string that is inserted in all "gensym'd"
2957    names (ML structure names that correspond to unnamed C structs, unions,
2958    and enums), so that separate runs of ml-nlffigen do not clash.
2959    
2960    ----------------------------------------------------------------------
2961    Name: Matthias Blume
2962    Date: 2002/02/11 12:05:00 EST
2963    Tag: blume-20020211-gensml
2964    Description:
2965    
2966    A quick fix for a problem with GenSML (in the pgraph-util library):
2967    Make generation of toplevel "local" optional.  (Strictly speaking,
2968    signature definitions within "local" are not legal SML.)
2969    
2970    Other than that: updates to INSTALL and cm/TODO.
2971    
2972    ----------------------------------------------------------------------
2973    Name: Matthias Blume
2974    Date: 2002/02/08 15:00:00 EST
2975    Tag: blume-20020208-uniquepid
2976    Description:
2977    
2978    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
2979    
2980    1. The installer (config/install.sh) has gotten smarter:
2981    
2982         - Configuration options are a bit easier to specify now
2983           (in config/targets).
2984         - Bug in recognizing .tar.bz2 files fixed.
2985         - Installer automatically resolves dependencies between
2986           configuration options (e.g., if you ask for eXene, you will
2987           also get cml -- regardless whether you asked for it or not).
2988         - Installer can run in "quieter mode" by setting the environment
2989           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
2990           "completely silent", though.
2991         - Build HashCons library as part of smlnj-lib.
2992    
2993    2. A new scheme for assigning persistent identifiers to compilation
2994       units (and, by extension, to types etc.) has been put into place.
2995       This fixes a long-standing bug where types and even dynamic values
2996       can get internally confused, thereby compromising type safety
2997       (abstraction) and dynamic correctness.  See
2998    
2999         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
3000    
3001       for an example of how things could go wrong until now.
3002    
3003       The downside of the new scheme is that pids are not quite as
3004       persistent as they used to be: CM will generate a fresh pid
3005       for every compilation unit that it thinks it sees for the first
3006       time.  That means that if you compile starting from a clean, fresh
3007       source tree at two different times, you end up with different
3008       binaries.
3009    
3010       Cutoff recompilation, however, has not been compromised because
3011       CM keeps pid information in special caches between runs.
3012    
3013    ----------------------------------------------------------------------
3014    Name: Lal George
3015    Date: 2002/02/07 15:34:13 EST 2002
3016    Tag: <none>
3017    Description:
3018    
3019    Compilers that generate assembly code may produce  global labels
3020    whose value is resolved at link time. The various peephole optimization
3021    modules did not take this in account.
3022    
3023    TODO. The Labels.addrOf function should really return an option
3024    type so that clients are forced to deal with this issue, rather
3025    than an exception being raised.
3026    
3027    ----------------------------------------------------------------------
3028    Name: Lal George
3029    Date: 2002/02/06 13:55:02 EST
3030    Tag: george-20020206-ra-breakup
3031    Description:
3032    
3033    1. A bug fix from Allen.
3034    
3035        A typo causes extra fstp %st(0)'s to be generated at compensation
3036        edges, which might cause stack underflow traps at runtime.  This
3037        occurs in fft where there are extraneous fstps right before the 'into'
3038        trap instruction (in this case they are harmless since none of the
3039        integers overflow.)
3040    
3041    2. Pulled out various utility modules that were embedded in the modules
3042       of the register allocator. I need these modules for other purposes, but
3043       they are not complete enough to put into a library (just yet).
3044    ----------------------------------------------------------------------
3045    Name: Matthias Blume
3046    Date: 2002/01/31 16:05:00 EST
3047    Tag: blume-20020131-sparc-ccalls
3048    Description:
3049    
3050    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
3051       of extra stack space by mistake.  Fixed.
3052    
3053    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
3054    
3055    ----------------------------------------------------------------------
3056    Name: Allen Leung
3057    Date: 2002/01/30
3058    Tag: leunga-20020130-nowhere-bug-fix
3059    Description:
3060    
3061       MLRISC bug fixes:
3062       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
3063       2. MachineInt.fromString was negating its value.
3064    
3065    ----------------------------------------------------------------------
3066    Name: Matthias Blume
3067    Date: 2002/01/29
3068    Tag: blume-20020129-INSTALL
3069    Description:
3070    
3071    - Added somewhat detailed installation instructions (file INSTALL).
3072    - Fixed curl-detection bug in config/install.sh.
3073    - It is now possible to select the URL getter using the URLGETTER
3074      environment variable:
3075    
3076          not set / "unknown"      --> automatic detection (script tries wget,
3077                                       curl, and lynx)
3078          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
3079                                       how to properly invoke them)
3080          other                    --> use $URLGETTER directly, it must take
3081                                       precisely two command-line arguments
3082                                       (source URL and destination file name)
3083    
3084    ----------------------------------------------------------------------
3085    Name: Matthias Blume
3086    Date: 2002/01/28
3087    Tag: blume-20020128-sparc-ccalls
3088    Description:
3089    
3090    - Fixed problem with calculation of "used" registers in sparc-c-calls.
3091    - Make use of the allocParam argument in sparc-c-calls.
3092    
3093    ----------------------------------------------------------------------
3094    Name: Matthias Blume
3095    Date: 2002/01/28
3096    Tag: blume-20020128-allocParam
3097    Description:
3098    
3099    John Reppy:  Changes c-calls API to accept client-callback for
3100    allocating extra stack space.
3101    me: Corresponding changes to mlriscGen (using a dummy argument that
3102        does not change the current behavior).
3103    
3104    ----------------------------------------------------------------------
3105    Name: Matthias Blume
3106    Date: 2002/01/28 12:00:00
3107    Tag: Release_110_38
3108    Description:
3109    
3110    This time for real!!!
3111    
3112    ----------------------------------------------------------------------
3113    Name: Matthias Blume
3114    Date: 2002/01/28 10:56:00 EST
3115    Tag: blume-20020128-retraction
3116    Description:
3117    
3118    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
3119       with blume-Release_110_38-retracted.)
3120    
3121    1. Fixed a problem with incorrect rounding modes in real64.sml.
3122       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
3123    
3124    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
3125       and enums fixed.  The naming of corresponding ML identifiers should
3126       now be consistent again.
3127    
3128    ----------------------------------------------------------------------
3129    Name: Allen Leung
3130    Date: 2002/01/27
3131    Tag: leunga-20020127-nowhere
3132    Description:
3133    
3134       Added a target called nowhere in the configuration scripts.
3135       Enabling this will build the MLRISC 'nowhere' tool (for translating
3136       programs with where-clauses into legal SML code) during installation.
3137    
3138    ----------------------------------------------------------------------
3139    Name: Matthias Blume
3140    Date: 2002/01/25 21:27:00 EST
3141    Tag: blume-Release_110_38-retracted
3142    Description:
3143    
3144    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
3145    
3146    README will be added later.
3147    
3148    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
3149    (Release_110_38) removed.  Reason: Last-minute bug fixes.
3150    
3151    ----------------------------------------------------------------------
3152    Name: Matthias Blume
3153    Date: 2002/01/25
3154    Tag: blume-20020125-ffi
3155    Description:
3156    
3157    A large number of tweaks and improvements to ml-nlffi-lib and
3158    ml-nlffigen:
3159    
3160       - ML represenation types have been streamlined
3161       - getter and setter functions work with concrete values, not abstract
3162         ones where possible
3163       - ml-nlffigen command line more flexible (see README file there)
3164       - some bugs have been fixed (hopefully)
3165    
3166    ----------------------------------------------------------------------
3167    Name: Lal George
3168    Date: 2002/01/24
3169    Tag: george-20020124-risc-ra-interface
3170    Description:
3171    
3172       There is a dramatic simplification in the interface to the
3173       register allocator for RISC architectures as a result of making
3174       parallel copy instructions explicit.
3175    
3176    ----------------------------------------------------------------------
3177    Name: Matthias Blume
3178    Date: 2002/01/22
3179    Tag: blume-20020122-x86-ccalls
3180    Description:
3181    
3182    Bug fix for c-calls on x86 (having to do with how char- and
3183    short-arguments are being handled).
3184    
3185    ----------------------------------------------------------------------
3186    Name: Matthias Blume
3187    Date: 2002/01/21
3188    Tag: blume-20020121-ff
3189    Description:
3190    
3191    Another day of fiddling with the FFI...
3192    
3193    1. Bug fix/workaround:  CKIT does not complain about negative array
3194       dimensions, so ml-nlffigen has to guard itself against this possibility.
3195       (Otherwise a negative dimension would send it into an infinite loop.)
3196    
3197    2. Some of the abstract types (light objects, light pointers, most "base"
3198       types) in structure C are now eqtypes.
3199    
3200    3. Added constructors and test functions for NULL function pointers.
3201    
3202    ----------------------------------------------------------------------
3203    Name: Matthias Blume
3204    Date: 2002/01/18
3205    Tag: blume-20020118-ready-for-new-release
3206    Description:
3207    
3208    Made config/srcarchiveurl point to a new place.  (Will provide boot
3209    files shortly.)
3210    
3211    Maybe we christen this to be 110.38?
3212    
3213    ----------------------------------------------------------------------
3214    Name: Matthias Blume
3215    Date: 2002/01/18
3216    Tag: blume-20020118-more-ffifiddle
3217    Description:
3218    
3219    Today's FFI fiddling:
3220    
3221      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
3222        C.Set.* that use concrete (MLRep.*) arguments and results instead
3223        of abstract ones.
3224    
3225      - Provided word-style bit operations etc. for "int" representation
3226        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
3227        and Long.
3228    
3229    ----------------------------------------------------------------------
3230    Name: Matthias Blume
3231    Date: 2002/01/18
3232    Tag: blume-20020118-use-x86-fp
3233    Description:
3234    
3235    Now that x86-fast-fp seems to be working, I turned it back on again
3236    by default.  (Seems to work fine now, even with the FFI.)
3237    
3238    Other than that, I added some documentation about the FFI to
3239    src/ml-nlffigen/README and updated the FFI test examples in
3240    src/ml-nlffi-lib/Tests/*.
3241    
3242    ----------------------------------------------------------------------
3243    Name: Allen Leung
3244    Date: 2002/01/17
3245    Tag: leunga-20020117-x86-fast-fp-call
3246    Description:
3247    
3248       1. Fixed a problem with handling return fp values when x86's fast fp
3249          mode is turned on.
3250    
3251       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
3252          of %f32.
3253    
3254       3. Added a constructor INT32lit to the ast of MLRISC tools.
3255    
3256    ----------------------------------------------------------------------
3257    Name: Matthias Blume
3258    Date: 2002/01/16
3259    Tag: blume-20020116-ffifiddle
3260    Description:
3261    
3262    More fiddling with the FFI interface:
3263    
3264     - Make constness 'c instead of rw wherever possible.  This eliminates
3265       the need for certain explicit coercions.  (However, due to ML's
3266       value polymorphism, there will still be many cases where explicit
3267       coercions are necessary.  Phantom types are not the whole answer
3268       to modeling a subtyping relationship in ML.)
3269    
3270     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
3271    
3272     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
3273       clumsy workaround for varargs problem.)
3274    
3275    ----------------------------------------------------------------------
3276    Name: Lal George
3277    Date: 2002/01/15
3278    Tag: <none>
3279    Description:
3280    
3281    1. Since COPY instructions are no longer native to the architecture,
3282       a generic functor can be used to implement the expandCopies function.
3283    
3284    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
3285       TEXT segment.
3286    
3287    ----------------------------------------------------------------------
3288    Name: Matthias Blume
3289    Date: 2002/01/15
3290    Tag: blume-20020115-ffiupdates
3291    Description:
3292    
3293    1. Fix for bug resulting in single-precision float values being returned
3294       incorrectly from FFI calls.
3295    
3296    2. Small modifications to C FFI API:
3297    
3298        - memory-allocation routines return straight objects (no options)
3299          and raise an exception in out-of-memory situations
3300        - unsafe extensions to cast between function pointers and pointers
3301          from/to ints
3302        - added structure C_Debug as an alternative to structure C where
3303          pointer-dereferencing (|*| and |*!) always check for null-pointers
3304        - added open_lib' to DynLinkage;  open_lib' works like open_lib
3305          but also takes a (possibly empty) list of existing library handles
3306          that the current library depends on
3307    
3308    ----------------------------------------------------------------------
3309    Name: Matthias Blume
3310    Date: 2002/01/10
3311    Tag: blume-20020110-newffigen
3312    Description:
3313    
3314    1. Updates to portable graph code.
3315    
3316    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
3317       more scalable now so that even huge interfaces such as the one
3318       for GTK compile in finite time and space. :-)
3319       See src/ml-nlffigen/README for details on what's new.
3320    
3321    ----------------------------------------------------------------------
3322    Name: Lal George
3323    Date: 2001/01/09 14:31:35 EST 2002
3324    Tag: george-20011206-rm-native-copy
3325    Description:
3326    
3327            Removed the native COPY and FCOPY instructions
3328            from all the architectures and replaced it with the
3329            explicit COPY instruction from the previous commit.
3330    
3331            It is now possible to simplify many of the optimizations
3332            modules that manipulate copies. This has not been
3333            done in this change.
3334    
3335    ----------------------------------------------------------------------
3336    Name: Lal George
3337    Date: 2001/12/06 16:50:13 EST 2001
3338    Tag: george-20011206-mlrisc-instruction
3339    Description:
3340    
3341    Changed the representation of instructions from being fully abstract
3342    to being partially concrete. That is to say:
3343    
3344      from
3345            type instruction
3346    
3347      to
3348            type instr                              (* machine instruction *)
3349    
3350            datatype instruction =
3351                LIVE of {regs: C.cellset, spilled: C.cellset}
3352              | KILL of {regs: C.cellset, spilled: C.cellset}
3353              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
3354              | ANNOTATION of {i: instruction, a: Annotations.annotation}
3355              | INSTR of instr
3356    
3357    This makes the handling of certain special instructions that appear on
3358    all architectures easier and uniform.
3359    
3360    LIVE and KILL say that a list of registers are live or killed at the
3361    program point where they appear. No spill code is generated when an
3362    element of the 'regs' field is spilled, but the register is moved to
3363    the 'spilled' (which is present, more for debugging than anything else).
3364    
3365    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
3366    We used to generate:
3367    
3368            DEFFREG f1
3369            f1 := f2 + f3
3370            trapb
3371    
3372    but now generate:
3373    
3374            f1 := f2 + f3
3375            trapb
3376            LIVE {regs=[f1,f2,f3], spilled=[]}
3377    
3378    Furthermore, the DEFFREG (hack) required that all floating point instruction
3379    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
3380    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
3381    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
3382    this architecture).
3383    
3384    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
3385    all the architectures. This will result in further simplification of the
3386    register allocator that must be aware of them for coalescing purposes, and
3387    will also simplify certain aspects of the machine description that provides
3388    callbacks related to parallel copies.
3389    
3390    ANNOTATION should be obvious, and now INSTR represents the honest to God
3391    machine instruction set!
3392    
3393    The <arch>/instructions/<arch>Instr.sml files define certain utility
3394    functions for making porting easier -- essentially converting upper case
3395    to lower case. All machine instructions (of type instr) are in upper case,
3396    and the lower case form generates an MLRISC instruction. For example on
3397    the alpha we have:
3398    
3399      datatype instr =
3400         LDA of {r:cell, b:cell, d:operand}
3401       | ...
3402    
3403      val lda : {r:cell, b:cell, d:operand} -> instruction
3404        ...
3405    
3406    where lda is just (INSTR o LDA), etc.
3407    
3408    ----------------------------------------------------------------------
3409    Name: Matthias Blume
3410    Date: 2001/11/22 21:40:00 EST
3411    Tag: Release_110_37
3412    Description:
3413    
3414    Release 110.37.  This time for real.
3415    
3416    ----------------------------------------------------------------------
3417    Name: Matthias Blume
3418    Date: 2001/11/21 16:35:00 EST
3419    Tag: blume-20011121-foot-in-mouth
3420    Description:
3421    
3422    Removed the "Release_110_37" tag because of a serious bug.
3423    This will be re-tagged once the bug is fixed.
3424    
3425    ----------------------------------------------------------------------
3426    Name: Matthias Blume
3427    Date: 2001/11/21 16:14:00 EST
3428    Tag: blume-20011121-forgottenfile
3429    Description:
3430    
3431    Forgot to add a file.  (Just a .tex-file -- part of
3432    the CM manual source.)
3433    
3434    ----------------------------------------------------------------------
3435    Name: Matthias Blume
3436    Date: 2001/11/21 16:10:00 EST
3437    Tag: blume-20011121-invalid_110_37
3438    Description:
3439    
3440    Note: I removed the original tag "Release_110_37" from this commit
3441          because we found a serious bug in all non-x86 backends.
3442          - Matthias
3443    
3444    1. Modifications to the SML/NJ code generator and to the runtime system
3445       so that code object name strings are directly inserted into code
3446       objects at code generation time.  The only business the runtime system
3447       has with this is now to read the name strings on occasions.
3448       (The encoding of the name string has also changed somewhat.)
3449    
3450    2. CM now implements a simple "set calculus" for specifying export lists.
3451       In particular, it is now possible to refer to the export lists of
3452       other libraries/groups/sources and form unions as well as differences.
3453       See the latest CM manual for details.
3454    
3455    3. An separate notion of "proxy" libraries has again be eliminated from
3456       CM's model.  (Proxy libraries are now simply a special case of using
3457       the export list calculus.)
3458    
3459    4. Some of the existing libraries now take advantage of the new set
3460       calculus.
3461       (Notice that not all libraries have been converted because some
3462       of the existing .cm-files are supposed to be backward compatible
3463       with 110.0.x.)
3464    
3465    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
3466       "General.exnMessage"!  The former relies on a certain hook to be
3467       initialized, and that often does not happen in the stand-alone case.)
3468    
3469    ----------------------------------------------------------------------
3470    Name: Lal George
3471    Date: 2001/11/21  13:56:18 EST
3472    Tag: george-2001121-pseudo-ops
3473    Description:
3474    
3475      Implemented a complete redesign of MLRISC pseudo-ops. Now there
3476      ought to never be any question of incompatabilities with
3477      pseudo-op syntax expected by host assemblers.
3478    
3479      For now, only modules supporting GAS syntax are implemented
3480      but more should follow, such as MASM, and vendor assembler
3481      syntax, e.g. IBM as, Sun as, etc.
3482    
3483    ----------------------------------------------------------------------
3484  Name: Matthias Blume  Name: Matthias Blume
3485  Date: 2001/11/14 11:52:00 EST  Date: 2001/11/14 11:52:00 EST
3486  Tag: blume-20011114-srcname  Tag: blume-20011114-srcname
# Line 194  Line 3662 
3662    
3663  ----------------------------------------------------------------------  ----------------------------------------------------------------------
3664  Name: Matthias Blume  Name: Matthias Blume
 >>>>>>> 1.169  
3665  Date: 2001/09/18 15:35:00 EDT  Date: 2001/09/18 15:35:00 EDT
3666  Tag: blume-20010918-readme11036  Tag: blume-20010918-readme11036
3667  Description:  Description:

Legend:
Removed from v.977  
changed lines
  Added in v.1641

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