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/branches/primop-branch-3/NOTES/HISTORY
ViewVC logotype

Diff of /sml/branches/primop-branch-3/NOTES/HISTORY

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

revision 643, Fri May 12 08:29:15 2000 UTC revision 1689, Sat Nov 13 06:25:04 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: 2004/11/13 00:20:00 CST
18    Tag: blume-20041113-versiontool
19    Description:
20    
21    - Stripped down the versiontool: It now only handles the version number.
22      The date string is generated at bootstrap time (during makeml).
23    
24    - In a previous commit, fixed a minor issue with how polyequal is being
25      translated.  In particular, the code now "looks through" abstractions.
26      This results in slightly fewer polyEqual warnings and hopefully slightly
27      more efficient code.  Important examples for where this matters are
28      the new int64 and word64 types.
29    
30    ----------------------------------------------------------------------
31    Name: Matthias Blume (blume (at) tti - c (dot) org)
32    Date: 2004/11/12 00:30:00 CST
33    Tag: blume-20041112-int64
34    Description:
35    
36    Structure Int64 fully hooked in.  (The implementation is not very
37    efficient, though.)
38    
39    ----------------------------------------------------------------------
40    Name: Matthias Blume (blume (at) tti - c (dot) org)
41    Date: 2004/11/11 17:30:00 CST
42    Tag: blume-20041111-more64
43    Description:
44    
45    All the pieces of Word64 are now there, with the exception of the
46    conversions from and to LargeWord. (Eventually these need to be identities,
47    but for the time being they don't even make sense because LargeWord is
48    32-bit wide.)
49    
50    Also started to add similar support for Int64, but major pieces of that
51    are still missing.
52    
53    ----------------------------------------------------------------------
54    Name: Matthias Blume (blume (at) tti - c (dot) org)
55    Date: 2004/11/11 00:15:00 CST
56    Tag: blume-20041111-word64
57    Description:
58    
59    Structure Word64 is now (almost) complete, word literals and patterns
60    seem to work.  There are a few odd pieces missing.  In particular,
61    I didn't do the {from,to}LargeWord parts because LargeWord is still Word32
62    at the moment.
63    
64    Making Word64 official would mean that LargeWord becomes Word64.  But
65    this requires extreme care because most word-word conversions have to
66    go through LargeWord, so making a mistake means loss of efficiency or
67    worse.  Eventually there will be a solution similar to (but actually
68    simpler than) what I did with IntInf.
69    
70    ----------------------------------------------------------------------
71    Name: Matthias Blume (blume (at) tti - c (dot) org)
72    Date: 2004/11/10 18:12:00 CST
73    Tag: blume-20041110-64bit
74    Description:
75    
76    More 64-bit hacking (but still not even half-way there yet).
77    Also, some assorted improvements to the handling of 8-bit words.
78    
79    ----------------------------------------------------------------------
80    Name: Matthias Blume (blume (at) tti - c (dot) org)
81    Date: 2004/11/09 17:50:00 CST
82    Tag: <none>
83    Description:
84    
85    Started adding some infrastructure for supporting 64-bit int- and
86    word-types.  (Still in its very early stages.)
87    
88    ----------------------------------------------------------------------
89    Name: Matthias Blume (blume (at) tti - c (dot) org)
90    Date: 2004/10/28 10:45:00 CDT
91    Tag: Release_110_50
92    Description:
93    
94    New working version (110.50).  NEW BOOTFILES!
95    
96                 =====================
97    
98    Also:
99    
100      - Changend config/srcarchiveurl from a file just
101        containing the URL string into a file containing
102        shell script code.  The code has access to the $VERSION variable.
103      - Made corresponding changes to config/install.sh and config/unpack.
104      - Default contents of config/srcarchiveurl uses $VERSION and
105        normally does not have to be edited to reflect a version change.
106    
107      (As a result, a version change can be done by just editing
108       config/version, the rest is now automatic.)
109    
110    ----------------------------------------------------------------------
111    Name: Matthias Blume (blume (at) tti - c (dot) org)
112    Date: 2004/10/27 17:50:00 CDT
113    Tag: blume-20041027-btrace-msg
114    Description:
115    
116    BackTrace.monitor now also reports the source of the exception that
117    triggered the trace.
118    
119    ----------------------------------------------------------------------
120    Name: Matthias Blume (blume (at) tti - c (dot) org)
121    Date: 2004/10/27 17:20:00 CDT
122    Tag: blume-20041027-x86-c-calls
123    Description:
124    
125    This is the HISTORY entry for two earlier commits, both concerning
126    the x86 c-calls code in MLRISC:
127    
128      - added a missing LOAD in the code that deals with struct arguments
129      - made sure the caller does not add the wrong number of bytes to the
130        stack pointer after a call of a function returning a struct
131        (the callee already pops the implicit argument which points to
132         the space reserved for the result)
133    
134    ----------------------------------------------------------------------
135    Name: Allen Leung (leunga (at) reservoir (dot) com)
136    Date: 2004/10/24 14:00:00 EST
137    Tag: leunga-20041024-x86-gas-fucomip
138    Description:
139    
140       John discovered a bug in the syntax of fucomip.
141       The opcodes FU?COMIP? have been changed to
142    
143           fu?comip? %st(i), %st
144    
145    ----------------------------------------------------------------------
146    Name: Matthias Blume (blume (at) tti - c (dot) org)
147    Date: 2004/10/20 15:06:00 CDT
148    Tag: blume-20041020-standalone-backtrace
149    Description:
150    
151    Added a mechanism for getting back-trace information from standalone
152    programs.  Here is how it works:
153    
154       1. The part of the program from which you want to get backtrace
155          information (usually the whole program) should be wrapped with
156          BackTrace.monitor.  This is a (unit->'a)->'a function, and your
157          main program could be modified from something like
158    
159             fun main (pgm, args) = ...
160    
161         to
162    
163            fun main (pgm, args) = BackTrace.monitor (fn () => ...)
164    
165       2. To be able to access BackTrace.monitor, you have to add
166          library $smlnj-tdp/plugins.cm to the .cm file that contains your
167          main function.
168    
169       3. Remove all compiled code (i.e., all the .cm/ subdirectories that
170          CM might have created in the past for your project).
171    
172       4. Build the system using this command line:
173    
174             ml-build -Ctdp.instrument=true \$smlnj-tdp/back-trace.cm \
175                      myprog.cm MyProg.main myprog
176    
177          instead of the usual
178    
179             ml-build myprog.cm MyProg.main myprog
180    
181    I changed a library name:
182    
183       $/trace-debug-profile.cm --> $smlnj-tdp/plugins.cm
184    
185    New libraries:
186    
187       $smlnj-tdp/back-trace.cm
188          -- when loaded causes the back-trace plugin to be installed
189       $smlnj-tdp/coverage.cm
190          -- when loaded causes the coverage plugin to be installed
191    
192    ----------------------------------------------------------------------
193    Name: Matthias Blume (blume (at) tti - c (dot) org)
194    Date: 2004/10/18 16:45:00 CDT
195    Tag: blume-20041018-groupowner
196    Description:
197    
198    Added an "obsolete" warning for the "group owner" syntax to CM's parser.
199    
200    Eliminated group owner specs from .cm files throughout the source tree.
201    
202    ----------------------------------------------------------------------
203    Name: Matthias Blume (blume (at) tti - c (dot) org)
204    Date: 2004/10/15 15:45:00 CDT
205    Tag: blume-20041015-coverage
206    Description:
207    
208    * Test coverage tool added!
209    * Further reorganization of tracing-, debugging-, and profiling support:
210    
211        - moved original BTImp -- now called BackTrace -- into a separate
212          library called $/trace-debug-profile.cm
213        - eliminated all mentions of BTrace from SMLofNJ.Internals
214        - only the instrumentation mechanism is now left in the compiler proper
215        - BackTrace module is a plugin which is NOT plugged in by default
216        - Coverage module is another such plugin
217    
218    To get the benefits of any of these plugin modules, the code in
219    question must be compiled with tdp instrumentation turned on.  This
220    can be done by setting SMLofNJ.Internals.TDP.mode to true.
221    (The ref cell is also controlled via the -Ctdp.instrument=... switch.)
222    
223    Plugins are selected at link time.  (Pre-compiled instrumented code
224    can be re-loaded with different plugins in effect.)  When an
225    instrumented module is linked, whatever plugins are at that time
226    enabled will come into effect for that module.
227    
228    To enable the back-trace plugin, load library $/trace-debug-profile.cm
229    and invoke BackTrace.install() (e.g., from the interactive prompt).
230    To enable the coverage plugin, load the same library and invoke
231    Coverage.install().
232    
233    Back-traces are generated automatically on uncaught exceptions and
234    when the code in question explicitly invokes BackTrace.trigger().
235    
236    Coverage (and execution frequency-) information must be queried
237    explicitly by calling Coverage.not_covered and Coverage.hot_spots.
238    
239    ----------------------------------------------------------------------
240    Name: Matthias Blume (blume (at) tti - c (dot) org)
241    Date: 2004/10/14 17:40:00 CDT
242    Tag: blume-20041014-tdp-core
243    Description:
244    
245    Snapshot of a significant overhaul of how the trace/debug/profile support
246    is hooked into the system (specifically: Core and SMLofNJ.Internals).
247    
248    ----------------------------------------------------------------------
249    Name: Matthias Blume (blume (at) tti - c (dot) org)
250    Date: 2004/10/13 16:34:00 CDT
251    Tag: blume-20041013-tdp
252    Description:
253    
254    Some rationalization of names:
255    
256       structure BTrace -> structure TDPInstrument
257       etc.
258    
259    This is is preparation of using the original back-trace
260    instrumentation for other purposes.  "TDP" stands for
261    Trace/Debug/Profile.
262    
263    The control flag controlling whether instrumentation is on or off is now
264    registered under a different name, so instead of running sml as
265    
266     sml -Cinstrument.btrace-mode=true
267    
268    one has to say
269    
270     sml -Ctdp.instrument=true
271    
272    ----------------------------------------------------------------------
273    Name: Matthias Blume (blume (at) tti - c (dot) org)
274    Date: 2004/10/11 16:37:00 CDT
275    Tag: blume-20041011-regions
276    Description:
277    
278    Made some minor modifications to elabcore.sml to have source regions
279    be propagated more tightly -- resulting in better (i.e., smaller)
280    regions being reported in error- and debug messages.
281    
282    ----------------------------------------------------------------------
283    Name: Matthias Blume (blume (at) tti - c (dot) org)
284    Date: 2004/10/08 22:50:00 CDT
285    Tag: blume-20041008-cmkw
286    Description:
287    
288    Fixed handling of keywords in .cm files: After seeing "is" the lexer
289    treats subsequent occurrences of "group", "library", "source", "is",
290    "*", and "-" as ordinary identifiers rather than keywords.
291    
292    Most seriously, this fixes a problem with CM's "shell" tool.  The tool
293    is supposed to accept a tool argument called "source", but this did
294    not work because of the clash with the keyword.
295    
296    ----------------------------------------------------------------------
297    Name: Matthias Blume (blume (at) tti - c (dot) org)
298    Date: 2004/10/07 16:00:00 CDT
299    Tag: blume-20041007-cleanup
300    Description:
301    
302    Assorted cleanup work:
303    
304      - got rid of intstrmap in favor of using the library's
305        hash table implementation
306      - threw out most of the pathnames stuff, as it was not used anyway
307      - simplified tokentable implementation
308      - fixed some minor spelling errors
309    
310    ----------------------------------------------------------------------
311    Name: Matthias Blume (blume (at) tti - c (dot) org)
312    Date: 2004/10/06 15:15:15 CDT
313    Tag: blume-20041006-handler
314    Description:
315    
316    Cleaned up the absyn to reflect the invariant that HANDLE always
317    carries a FNexp as part of the type definition.  This eliminates some
318    superfluous sanity checks at runtime down the road.
319    
320    Some minor cleanup of the btrace code.
321    
322    ----------------------------------------------------------------------
323    Name: Matthias Blume (blume (at) tti - c (dot) org)
324    Date: 2004/10/01 10:20:30 CDT
325    Tag: blume-20041001-slave
326    Description:
327    
328    Added hack to make slave mode work in the presence of the version
329    tool.  (Still, since the master does two passes over the code for
330    CMB.make, the release number gets bumped twice when slaves are
331    attached.  I don't know if this is worth fixing...)
332    
333    ----------------------------------------------------------------------
334    Name: Matthias Blume (blume (at) tti - c (dot) org)
335    Date: 2004/09/30 10:55:00 CDT
336    Tag: blume-20040930-version
337    Description:
338    
339    * Moved the "version" magic into its own little library under
340      src/system/smlnj/internal.  This avoids expensive reconstruction of
341      a stable src/compiler/core.cm.
342    
343    * At the same time, structure CompilerVersion is now known as
344      structure SMLNJVersion.
345    
346    * Arranged for the version tool to NOT kick in when rebuilding the system
347      (makeml -rebuild, fixpt).  Otherwise one would never reach a fixpoint.
348      Also, loading the versiontool does not work when rebuilding the system
349      because CM is not properly initialized at that time.
350    
351    ----------------------------------------------------------------------
352    Name: Matthias Blume (blume (at) tti - c (dot) org)
353    Date: 2004/09/29 14:00:00 CDT
354    Tag: blume-20040929-autoversion
355    Description:
356    
357    Implemented some CM magic to have
358        file src/compiler/TopLevel/main/version.sml
359    generated automagically.
360    The version is taken from two files: config/version and config/release.
361    The first is expected to contain a two-part version number such as 110.49.
362    The second should contain a single number, but it may be missing.
363    
364    If the environment variable VERSIONTOOL_BUMP_RELEASE is defined at the
365    time the version tool is loaded (which is the first time you say CMB.make),
366    then the tool will increment the value stored in config/release every
367    time CMB.make is invoked.
368    
369    The binfile format is now insensitive to anything beyond the first
370    two components of a version number, so bumping the release does not render
371    binfiles incompatible.  Auto-bumping can be used to keep track of versions
372    during development without invalidating existing binfiles.
373    
374    In any case, every CMB.make updates the date information in version.sml.
375    (This is the date that is printed in the banner.)
376    
377    ----------------------------------------------------------------------
378    Name: Matthias Blume (blume (at) tti - c (dot) org)
379    Date: 2004/09/28 10:53:00 CDT
380    Tag: blume-20040928-controls
381    Description:
382    
383    Some cleanup of the controls code.
384    
385    ----------------------------------------------------------------------
386    Name: Matthias Blume (blume (at) tti - c (dot) org)
387    Date: 2004/09/27 22:08:00 CDT
388    Tag: blume-20040927-controls
389    Description:
390    
391    Added two pieces of functionality to the Controls interface:
392    
393    1.   val save'restore: 'a control -> unit -> unit
394    
395       grabs the current value of the control in stage 1 and restores it
396       in stage 2.
397    
398    2.   val set' : 'a control * 'a -> unit -> unit
399    
400       stores the given value into the control in stage 2 (i.e., delayed)
401       but does all error checking in stage 1.
402       (This is for string controls that need to do parse their argument --
403        something that might fail.  In some cases, notably in CM, one
404        already knows the intended argument but wants to delay the actual
405        assignment until a time when error recovery would be more difficult.)
406    
407    Changed the handling of controls in tool arguments to classes "sml" and
408    "lazysml":
409       - use Controls.save'restore as a more robust way of restoring the
410         old value (in particular: without having to re-parse the string)
411       - use controls to handle the "overload" keyword in the init group
412         (I believe this change actually fixes a long-standing obscure bug.)
413    
414    ----------------------------------------------------------------------
415    Name: Matthias Blume (blume (at) tti - c (dot) org)
416    Date: 2004/09/27 17:00:00 CDT
417    Tag: blume-20040927-lazysml
418    Description:
419    
420    Added a new tool class called "lazysml" to CM's tool chest.  The only
421    difference to "sml" is that compilation is done with Control.lazysml
422    set to true.  A source of class "lazysml" is automatically recognized
423    by a file name suffix of ".lml".
424    
425    In addition to the above feature, the original class "sml" now also
426    supports a tool argument "lazy" which has the same effect.  As a
427    result, the following three lines are equivalent:
428    
429        foo.sml : lazysml
430        foo.sml : sml (lazy)
431        foo.sml (lazy)
432    
433    The setting goes into effect both during parsing and during
434    compilation.  The original setting is restored right after parsing and
435    after compilation, respectively.
436    
437    In addition to all the above, there is also a general mechanism to set
438    ANY of the "controls" that are available at the command line via
439    "-C..." on a per-sml-file basis.  The same rules that apply for "lazy"
440    apply as well. (In fact, "lazy" is implemented as a special case of
441    the general mechanism.)
442    
443    The .cm file syntax uses a new keyword tool argument called "with".
444    There are several ways of indicating the desired settings:
445    
446        foo.sml (with:parser.quotations=true)
447        foo.sml (with:(name:parser.quotations value:true))
448        foo.sml (with:(name:name1 value:value1 name:name2 value:value2 ...))
449        foo.sml (with:(name1=value1 name2=value2 ...))
450        foo.sml (with:(name1=value1 name:name2 value:value2 name3=value3 ...))
451    
452        etc.
453    
454    Another possible abbreviation is to leave out the =v or value:v part
455    if the name refers to a boolean control (in which case the value is
456    taken to be true).  Thus, one could get lazy sml also by saying:
457    
458        foo.sml (with:parser.lazy-keyword=true)
459        foo.sml (with:parser.lazy-keyword)
460        foo.sml (with:(name:parser.lazy-keyword value:true))
461        foo.sml (with:(name:parser.lazy-keyword))
462    
463    ----------------------------------------------------------------------
464    Name: Matthias Blume (blume (at) tti - c (dot) org)
465    Date: 2004/09/24 16:22:00 CDT
466    Tag: blume-20040924-ppc-long-branch
467    Description:
468    
469    Turned message about "emiting long form of branch" off by default.
470    Added a control flag to turn it back on when desired.
471    
472    ----------------------------------------------------------------------
473    Name: Matthias Blume (blume (at) tti - c (dot) org)
474    Date: 2004/09/24 16:05:00 CDT
475    Tag: blume-20040924-rounding
476    Description:
477    
478    Applied patch for setting rounding modes under Mac OS X.  Thanks to
479    Melissa O'Neill for providing the code!
480    
481    ----------------------------------------------------------------------
482    Name: Matthias Blume (blume (at) tti - c (dot) org)
483    Date: 2004/09/23 17:30:00 CDT
484    Tag: blume-20040923-envvars
485    Description:
486    
487    1. Changed definition of type ControlRegistry.registry_tree to
488       include control_info (i.e., the name of the controlling environment
489       variable).
490    
491    2. Added command-line flags -e and -E to print the names of environment
492       variables that can be used to control internal settings.
493       (This uses the new API mentioned in 1.)
494    
495    ----------------------------------------------------------------------
496    Name: Matthias Blume (blume (at) tti - c (dot) org)
497    Date: 2004/09/13 16:50:00 CDT
498    Tag: Release_110_49
499    Description:
500    
501    New working version (110.49).  NEW BOOTFILES!
502    
503    ----------------------------------------------------------------------
504    Name: Matthias Blume
505    Date: 2004/09/13 16:20:00 CDT
506    Tag: blume-20040913-config-mlrisc
507    Description:
508    
509    Put target "mlrisc" back into the default list.
510    (There is no harm in having it, and some users have expressed their
511    wish to have "mlrisc" included by default.)
512    
513    ----------------------------------------------------------------------
514    Name: John Reppy
515    Date: 2004/09/13
516    Tag: jhr-20040913-signals
517    Description:
518    
519    Fixed the signal masking code to properly nest mask/unmask operations
520    on a per-signal basis.
521    
522    ----------------------------------------------------------------------
523    Name: Matthias Blume (blume (at) tti - c (dot) org)
524    Date: 2004/09/08 13:20:00 CDT
525    Tag: blume-20040908-heap-magic
526    Description:
527    
528    Bumped the heap macig number to 0x09082004 to account for the changed
529    layout of the ML frame under MacOS X.
530    
531    ----------------------------------------------------------------------
532    Name: Allen Leung (leunga (at) reservoir (dot) com)
533    Date: 2004/09/03 11:26:00 EST
534    Tag: leunga-20040903-cygwin-install
535    Description:
536    
537       Added a patch to _arch-n-opsys to enable the Cygwin runtime.
538    The Cygwin runtime is turned on by setting the environment
539    variable SMLNJ_CYGWIN_RUNTIME to 1.
540    
541    ----------------------------------------------------------------------
542    Name: Matthias Blume (blume (at) tti - c (dot) org)
543    Date: 2004/08/31 17:15:00 CDT
544    Tag: blume-20040831-core
545    Description:
546    
547    Added some exports to src/compiler/core.cm upon request by J. Joemann.
548    
549    ----------------------------------------------------------------------
550    Name: Matthias Blume (blume (at) tti - c (dot) org)
551    Date: 2004/08/30 17:55:00 CDT
552    Tag: blume-20040830-installer
553    Description:
554    
555    Upon request by Johannes Joemann:
556    
557    - improved ML code of installer to fall back to coping when renaming
558      fails (i.e., when source and target are on different file systems);
559      the code compiles but has yet to be tested in anger
560    - removed mlrisc from list of default targets (config/targets)
561    
562    ----------------------------------------------------------------------
563    Name: Matthias Blume (blume (at) tti - c (dot) org)
564    Date: 2004/08/27 17:20:00 CDT
565    Tag: blume-20040827-ptreql
566    Description:
567    
568    Added ptreql primop to structure InlineT (upon request from Larry
569    Paulson).
570    
571    ----------------------------------------------------------------------
572    Name: Allen Leung (leunga (at) reservoir (dot) com)
573    Date: 2004/08/15 21:21:00 EST
574    Tag: leunga-110_48-udgraph
575    Description:
576    
577    Another bug fix from Carl Hauser:
578    
579    diff /net/niflab/smlnj48/src/MLRISC/graphs/udgraph.sml udgraph.sml
580    > 48c48
581    > <              | rmv((e as (k,_))::es,L) = rmv(es,if k = i then es else
582    > e::L)
583    > ---
584    > >              | rmv((e as (k,_))::es,L) = rmv(es,if k = i then L else e::L)
585    > Without this, any deletion of an edge in an undirected graph does severe
586    > violence to the graph.
587    
588    ----------------------------------------------------------------------
589    Name: Allen Leung (leunga (at) reservoir (dot) com)
590    Date: 2004/08/10 23:35:00 EST
591    Tag: leunga-110_48-ppc
592    Description:
593    
594       The IBM/MacOS syntax switch on PPC was incorrectly swapped.
595    
596    ----------------------------------------------------------------------
597    Name: Matthias Blume (blume (at) tti - c (dot) org)
598    Date: 2004/08/10 12:00:00 CDT
599    Tag: Release_110_48
600    Description:
601    
602    New working version (110.48).  NEW BOOTFILES!
603    
604    ----------------------------------------------------------------------
605    Name: Allen Leung (leunga (at) reservoir (dot) com)
606    Date: 2004/08/09 12:21:00 EST
607    Tag: leunga-110_47-dijsktra
608    Description:
609    
610    Bug fix from Carl Hauser:
611    
612    single_source_shortest_paths in dijkstra.sml was observed to get wrong
613    answers (by comparing to single_source_shortest_paths in bellman-ford.sml).
614    
615    The problem is that following the expression A.update(dist,s,Num.zero)
616    it is necessary to update the priority queue using Q.decreaseWeight(Q,s).
617    
618    ----------------------------------------------------------------------
619    Name: Matthias Blume (blume (at) tti - c (dot) org)
620    Date: 2004/08/06 18:10:00 CDT
621    Tag: blume-20040806-cmdline
622    Description:
623    
624    Fiddled with handling of command-line options:
625    
626      * sml now quits after processing the command line
627        if -H, -S, -h<n>, or -s<n> appears as the last
628        command-line argument
629      * a new option -q terminates the session when encountered on
630        the command line; subsequent arguments will be ignored
631      * bug fixes: short (erroneous) arguments are no longer ignored
632        completely
633    
634    ----------------------------------------------------------------------
635    Name: Allen Leung (leunga (at) reservoir (dot) com)
636    Date: 2004/08/04 18:17:00 EST
637    Tag: leunga-110_47-ppc-ibm-asm
638    Description:
639    
640      - Added minimal IBM assembly syntax support for PowerPC.
641    
642      - Cygwin: manually changed the file cygwin.def.  Some exported symbols have
643        been altered in the runtime.  We need an automatic way to keep the file
644        in sync.
645    
646    ----------------------------------------------------------------------
647    Name: Matthias Blume (blume (at) tti - c (dot) org)
648    Date: 2004/08/04 14:00:00 CDT
649    Tag: Release_110_47
650    Description:
651    
652    New working version (110.47).  NEW BOOTFILES!
653    
654    ----------------------------------------------------------------------
655    Name: Matthias Blume (blume (at) tti - c (dot) org)
656    Date: 2004/08/03 14:25:00 CDT
657    Tag: blume-20040803-callingconv
658    Description:
659    
660    Added low-level support for choosing C calling conventions by
661    twiddling the type of rawccall.  (See
662    src/compiler/Semant/types/cproto.sml for details.)
663    
664    ----------------------------------------------------------------------
665    Name: Matthias Blume (blume (at) tti - c (dot) org)
666    Date: 2004/08/02 15:55:00 CDT
667    Tag: blume-20040802-backout
668    Description:
669    
670    Backed out of change to win32-filesys.c.  The earlier patch to
671    get_file_time caused CM to produce files with the wrong time stamp.
672    
673    ----------------------------------------------------------------------
674    Name: Matthias Blume (blume (at) tti - c (dot) org)
675    Date: 2004/08/02 14:45:00 CDT
676    Tag: blume-20040802-nlffi-win32
677    Description:
678    
679    Added NLFFI support for Win32, adapted from a patch provided by David
680    Hansel.  This is currently completely untested.  Also, the issue
681    concerning stdcall vs. ccall is still unresolved.
682    
683    ----------------------------------------------------------------------
684    Name: Matthias Blume (blume (at) tti - c (dot) org)
685    Date: 2004/07/30 17:55:00 CDT
686    Tag: blume-20040730-various
687    Description:
688    
689    Gearing up towards 110.47...
690    
691    - various minor bugfixes to ml-nlffigen
692    - a beginning of a manual for nlffi
693    
694    - eliminated 'export name=value' in config/install.sh as this does
695      not work with certain versions of /bin/sh
696      (Thanks to David King at Motorola for catching this.)
697    
698    - several bugfixes provided or suggested by David Hansel at Reactive Systems:
699      - added a test for tm==NULL to gmtime.c and localtime.c
700      - applied patch for incorrect GetFileTime under win32
701      - toSeconds -> toMilliseconds in Win32/win32-process.sml
702    
703    ----------------------------------------------------------------------
704    Name: Matthias Blume (blume (at) tti - c (dot) org)
705    Date: 2004/07/21 18:20:00 CDT
706    Tag: blume-20040721-nlffigen
707    Description:
708    
709    - Fixed minor issue in ml-nlffigen: Now generate structure T_foo
710      for a typedef to an incomplete type, but leave out the "typ" member.
711      (This is just for consistency.)
712    - Started to produce what is supposed to become better (i.e., comprehensive)
713      documentation of what ml-nlffigen does and produces.
714    
715    ----------------------------------------------------------------------
716    Name: Matthias Blume (blume (at) tti - c (dot) org)
717    Date: 2004/07/14 16:25:00 CDT
718    Tag: blume-20040714-union
719    Description:
720    
721    Added C_UNION to c-calls/c-types.sml and updated the machinery
722    (ml-nlffigen, cproto.sml) that conveys C function interface
723    information to the code generator.
724    
725    However, the actual architecture-specific implementation of function
726    arguments and results that are C unions is still not implemented.
727    
728    ----------------------------------------------------------------------
729    Name: Allen Leung (leunga (at) reservoir (dot) com)
730    Date: 2004/07/14 14:38:00 EST
731    Tag: leunga-110_46_1-ppc-lwzu
732    Description:
733    
734        Added these instructions to the PowerPC architecture:
735        LBZU(X), LHZU(X), LWZU(X),
736        STWU(X), STFDU, STFSU
737    
738        etc...
739    
740        Note: I haven't added their instruction encoding into the description.
741    
742    ----------------------------------------------------------------------
743    Name: Allen Leung (leunga (at) reservoir (dot) com)
744    Date: 2004/07/13 15:04:00 EST
745    Tag: leunga-110_46_1-ppc-lwarx
746    Description:
747    
748       Added the two instructions LWARX and STWCX to the PowerPC
749    instruction set.
750    
751       A (untested) rewrite of loop-structure.sml.  The old version
752    is completely broken.
753    
754    ----------------------------------------------------------------------
755    Name: Matthias Blume (blume (at) tti - c (dot) org)
756    Date: 2004/07/13 13:50:00 CDT
757    Tag: blume-20040713-nlffi
758    Description:
759    
760    - use paramAlloc to report c-calls with too many arguments
761      (for PPC version where parameter area is pre-allocated)
762    - added ccall_maxargspace to machspec (to implement the above)
763    - made "make" commend in CM's "make" tool configurable
764    - added option (default: on) for passing the name of the SML/NJ's "bin"
765      directory to "make"; the call looks like this:
766    
767         make <options> SMLNJ_BINDIR=<dir> <target>
768    
769      This can be used by the Makefile to, e.g., pick the "right" version
770      of ml-nlffigen.
771    - minor code tweaks
772    
773    ----------------------------------------------------------------------
774    Name: Matthias Blume (blume (at) tti - c (dot) org)
775    Date: 2004/07/12 22:50:00 CDT
776    Tag: blume-110_46_1-macosx-nlffi
777    Description:
778    
779    NLFFI under Mac OS X now working (sort of).  This is largely untested,
780    though.
781    
782    Note:  1. You have to make a new, clean build of the runtime system.
783           2. There are new BOOTFILES, you have to use them!
784              (Doing the bootstrap process yourself would be *very* painful!
785               If you absolutely have to do it, build the system under
786               a different architecture and then cross-compile.)
787    
788    Version bumped to 110.46.1 to account for runtime data format changes.
789    
790    ----------------------------------------------------------------------
791    Name: Matthias Blume (blume (at) tti - c (dot) org)
792    Date: 2004/06/18 14:30:00 CDT
793    Tag: blume-20040618-unix
794    Description:
795    
796    Changed the implementation of structure Unix so that the same stream
797    is returned every time one of the {text,bin}{In,Out}streamOf functions
798    is invoked on the same proc.  This is not what the spec currently
799    says -- although IMO it arguably should.  (See discussion below.)
800    
801    ----------------------------------------------------------------------
802    Name: Matthias Blume (blume (at) tti - c (dot) org)
803    Date: 2004/06/17 18:15:00 CDT
804    Tag: Release_110_46
805    Description:
806    
807    New working version (110.46).  NEW BOOTFILES!
808    
809    ----------------------------------------------------------------------
810    Name: Matthias Blume (blume (at) tti - c (dot) org)
811    Date: 2004/06/17 17:20:00 CDT
812    Tag: blume-20040617-timer-unix
813    Description:
814    
815    Changed the interface of structures Timer and Unix to match the most
816    recent Basis spec.
817    
818    In the case of Unix there still seems to be an open/weird issue:
819    
820       The {text,bin}{In,Out}streamOf functions are supposed to create
821       fresh streams whenever they are called -- as opposed to have them
822       return the same stream every time.  This design is supposed to
823       prevent space leaks caused by proc values hanging on to streams.
824    
825       The reap function, on the other hand, is supposed to close the
826       streams.  This cannot be done without having a handle on the
827       stream in proc after all...
828    
829    I took the liberty to implement the following stopgap solution:
830    
831      The proc value hangs on to the most recently created stream(s).
832      Reap closes those.  If either or both of the two streams hadn't
833      been created at all yet, then reap will close the corresponding
834      file descriptors directly.
835    
836    PS: I don't understand the original space leak argument anymore.  If
837    a proc hangs on to the imperative stream, then I/O operations on those
838    will advance the state of the cached stream and avoid the space leak.
839    
840    ----------------------------------------------------------------------
841    Name: Matthias Blume (blume (at) tti - c (dot) org)
842    Date: 2004/05/28 16:45:00 CDT
843    Tag: blume-20040528-basis
844    Description:
845    
846    Added signature PACK_REAL and exported functor PrimIO.
847    
848    ----------------------------------------------------------------------
849    Name: Matthias Blume (blume (at) tti - c (dot) org)
850    Date: 2004/05/25 16:00:00 CDT
851    Tag: blume-20040525-group-owner
852    Description:
853    
854    CM now ignores (but still accepts) the "owner" information in group
855    descriptions.  The owner of a group is its next enclosing
856    library. Each group must have a unique owner.  (There is a virtual
857    "toplevel" library that owns groups which are not nested within a real
858    library.)  Previously, each group had to explicitly declare its owner,
859    and CM would check that such a declaration is correct.  The new scheme
860    is to have CM check that for each group there is precisely one owning
861    library.
862    
863    The advantage of the new scheme is that the programmer no longer needs
864    to maintain the somewhat annoying owner information.  The downside is
865    that CM cannot enforce the ownership rule across multiple runs of
866    CM.make.  Fortunately, enclosing the same group in two different
867    libraries A and B which are not part of the same program does not
868    cause real problems.
869    
870    ----------------------------------------------------------------------
871    Name: Matthias Blume (blume (at) tti - c (dot) org)
872    Date: 2004/05/20 16:00:00 CDT
873    Tag: blume-20040520-win32
874    Description:
875    
876    Made the win32 version work again.  (Strangely, a misplaced comma had
877    slipped into win32-process.c which prevented the runtime from being
878    compiled correctly.)
879    
880    Also, included a minor addition to ml-build.bat analogous to what was
881    done in blume-20040519-ml-build.
882    
883    ----------------------------------------------------------------------
884    Name: Matthias Blume (blume (at) tti - c (dot) org)
885    Date: 2004/05/19 22:10:00 CDT
886    Tag: blume-20040519-ml-build
887    Description:
888    
889    Arranged for ml-build to clean up after itself a little bit better.
890    The script generates a temporary SML source file and compiles it using
891    CM, so CM generates metadata (GUID, SKEL, objectfile) for it.  It now
892    gets rid of those at the end, so they don't accumulate under .cm.
893    
894    This required a minor change to install.sh because the name of the
895    metadata directory (default: .cm) is actually configurable at
896    installation time.
897    
898    ----------------------------------------------------------------------
899    Name: Matthias Blume (blume (at) tti - c (dot) org)
900    Date: 2004/05/18 15:50:00 CDT
901    Tag: blume-20040518-mkreader
902    Description:
903    
904    Added Posix.IO.mk{Bin,Text}{Reader,Writer} by lifting their respective
905    implementations from internal modules PosixBinPrimIO and PosixTextPrimIO.
906    
907    ----------------------------------------------------------------------
908    Name: Matthias Blume (blume (at) tti - c (dot) org)
909    Date: 2004/05/11 14:35:00 CDT
910    Tag: blume-20040511-win32sock
911    Description:
912    
913    Added previously missing support for many socket-related functions
914    under win32.  Thanks to David Hansel <hansel@reactive-systems.com>
915    for the voluminous patch!
916    
917    (I have not tested this patch under win32 yet.)
918    
919    Here is David's e-mail:
920    
921    Hi,
922    
923    Attached to this email you find a diff against sml/nj 110.45
924    that will enable socket support under Windows.
925    
926    To apply the patch (using unix or cygwin)
927    1) gunzip runtime.diff.gz
928    2) "cd" into "src/runtime" in the source tree of a fresh
929       110.45 installation.
930    3) patch -p 1 < [your/path/to]runtime.diff
931    
932    The code compiles fine but has NOT yet been extensively tested.
933    I only ran a few tests for basic socket client functionality
934    (which worked fine).  Especially the functions that use ioctl
935    are not tested at all and might not work (see below).
936    
937    I implemented this since we want to move to a newer version of sml/nj
938    but need socket support in order to use it.  This is the first time I
939    even had a look at the sml/nj source,  so please review my changes
940    before making this part of the distribution!  Here are a few issues
941    that I think might be better for someone to solve who is more
942    familiar with the sml/nj source (and socket programming):
943    
944    - getnetbyaddr.c and getnetbyname.c will raise a "not implemented"
945      exception since I could not figure out what the windows equivalent
946      of these functions is
947    
948    - In sockets-osdep.h there are a some #include statements that are
949      only used in a few files that include sockets-osdep.h
950    
951    - In smlnj-sock-lib.c, function init_fn() calls WSAStartup() but
952      does not process its return value since I don't know how to
953      report an error upwards.
954    
955    - It would probably be good to have a call to WSACleanup() when
956      the library is unloaded (if there is such a possibility).
957      Otherwise I think Windows will take care of this automatically
958      when the process finishes.
959    
960    - I used ioctlsocket() as a replacement for ioctl() but I have
961      no idea if that is actually the proper replacement on Windows.
962    
963    - All these issues are marked in the code by "FIXME" comments.
964    
965    We use sml/nj extensively in our products and are quite happy
966    with it.  I hope this contribution will help you.
967    
968    Keep up the good work!
969    
970    David
971    
972    ----------------------------------------------------------------------
973    Name: Matthias Blume (blume (at) tti - c (dot) org)
974    Date: 2004/05/11 14:20:00 CDT
975    Tag: blume-20040511-installml
976    Description:
977    
978    Fixed two bugs in installml script.  (Thanks to Vesa A. Norrman for
979    the patch.)
980    
981    ----------------------------------------------------------------------
982    Name: Matthias Blume (blume (at) tti - c (dot) org)
983    Date: 2004/05/11 14:05:00 CDT
984    Tag: blume-20040511-nlffi-netbsd
985    Description:
986    
987    Added support for nlffi under netbsd.  (Thanks to Vesa A. Norrman for
988    the patch.)
989    
990    ----------------------------------------------------------------------
991    Name: Matthias Blume (blume (at) tti - c (dot) org)
992    Date: 2004/05/11 12:05:00 CDT
993    Tag: blume-20040511-exports
994    Description:
995    
996    As per request by Adam Chlipala <adam@hcoop.net>, extended various
997    export lists in compiler-related .cm-files.
998    
999    ----------------------------------------------------------------------
1000    Name: Matthias Blume (blume (at) tti - c (dot) org)
1001    Date: 2004/05/11 11:35:00 CDT
1002    Tag: blume-20040511-allsource
1003    Description:
1004    
1005    The installer now honors the "src-smlnj" target again, although its meaning
1006    has changed from "all sources required for the compiler" to "all sources
1007    the installer knows about".  In other words, if you enable "src-smlnj"
1008    in the "targets" file, then the installer will pull in sources for
1009    everything.  (Notice that this refers to source code only.  Compiled
1010    code is still only installed for modules that were requested explicitly
1011    or which are required for other modules that were requested explicitly.)
1012    
1013    ----------------------------------------------------------------------
1014    Name: Matthias Blume (blume (at) tti - c (dot) org)
1015    Date: 2004/04/23 17:40:00 CDT
1016    Tag: blume-20040423-ieee-scan
1017    Description:
1018    
1019    Fixed IEEEReal.scan (and .fromString) so that if there is an overflow
1020    in the exponent calculation we get INF or ZERO (depending on the mantissa
1021    and the sign of the exponent).
1022    
1023    ----------------------------------------------------------------------
1024    Name: Matthias Blume (blume (at) tti - c (dot) org)
1025    Date: 2004/04/23 10:40:00 CDT
1026    Tag: blume-20040423-ml-build
1027    Description:
1028    
1029    The ml-build script now terminates with a non-0 status when something
1030    goes wrong.
1031    
1032    ----------------------------------------------------------------------
1033    Name: Matthias Blume (blume (at) tti - c (dot) org)
1034    Date: 2004/04/22 16:35:00 CDT
1035    Tag: blume-20040422-Option
1036    Description:
1037    
1038    Made exception Option to be the same as exception Option.Option
1039    (as it should be).
1040    
1041    ----------------------------------------------------------------------
1042    Name: Allen Leung (leunga (at) reservoir (dot) com)
1043    Date: 2004/03/19 14:40:00 EST
1044    Tag: leunga-20040319-cygwin-nlffi
1045    Description:
1046    
1047    Fixed the runtime so that ml-nlffi-lib runs on the cygwin version
1048    of SML/NJ.  The problem is that
1049    
1050       lib = dlopen(NULL, ...)
1051       f   = dlsym(lib, "malloc");
1052    
1053    does not work on Windows unless we explicitly export symbols
1054    such as 'malloc' during linking.  We fixed this by explicitly
1055    exporting the required symbols with the magic gcc incantation:
1056    
1057        -Wl,--export-all cygwin.def
1058    
1059    where cygwin.def is a file containing all the symbols that
1060    we wish to export.
1061    
1062    I suspect this is a Windows problem and we'll have to
1063    do the same (somehow with windows compilers) when
1064    we build the native win32 version with the system
1065    calls LoadLibrary/GetProcAddress.
1066    
1067    ----------------------------------------------------------------------
1068    Name: Matthias Blume (blume (at) tti - c (dot) org)
1069    Date: 2004/03/04 16:35:00 CST
1070    Tag: blume-20040304-intinf-fmt
1071    Description:
1072    
1073    Fixed problem with IntInf.fmt (sign would show up on the right instead
1074    of on the left for BIN, OCT, and HEX).
1075    
1076    ----------------------------------------------------------------------
1077    Name: Matthias Blume (blume (at) tti - c (dot) org)
1078    Date: 2004/03/04 11:25:00 CST
1079    Tag: blume-20040304-symlinks
1080    Description:
1081    
1082    Fixed problem with installer script (unix only) where bin/ml-yacc and
1083    friends pointed (via symlinks) to absolute locations instead of just
1084    .run-sml.  This was reported by Vesa A Norrman.
1085    
1086    ----------------------------------------------------------------------
1087    Name: Matthias Blume (blume (at) tti - c (dot) org)
1088    Date: 2004/02/13 14:50:00 CST
1089    Tag: Release_110_45
1090    Description:
1091    
1092    New working version (110.45). New bootfiles.
1093    
1094    ----------------------------------------------------------------------
1095    Name: Matthias Blume (blume (at) tti - c (dot) org)
1096    Date: 2004/01/26 15:15:15 CST
1097    Tag: blume-20040126-toplevel
1098    Description:
1099    
1100    Improved handling of exceptions at the interactive toplevel.
1101    
1102    ----------------------------------------------------------------------
1103    Name: Matthias Blume (blume (at) tti - c (dot) org)
1104    Date: 2004/01/26 11:25:00
1105    Tag: blume-20040126-app
1106    Description:
1107    
1108    Type of top-level "app" corrected.
1109    Added code for setting vp_limitPtrMask to Win32-specific runtime.
1110    
1111    ----------------------------------------------------------------------
1112    Name: Matthias Blume (blume (at) tti - c (dot) org)
1113    Date: 2003/11/18 17:10 CST
1114    Tag: blume-20031118-basis-fiddle
1115    Description:
1116    
1117    - changed Timer interface to what might become the spec
1118    - POSIX_FLAGS -> BIT_FLAGS according to spec
1119    - some other minor discrepancies wrt. spec eliminated
1120    
1121    ----------------------------------------------------------------------
1122    Name: Matthias Blume (blume (at) tti - c (dot) org)
1123    Date: 2003/11/06 12:00:00 CST
1124    Tag: Release_110_44
1125    Description:
1126    
1127    New working version (110.44). New bootfiles.
1128    
1129    ----------------------------------------------------------------------
1130    Name: Matthias Blume (blume (at) tti - c (dot) org)
1131    Date: 2003/11/04 11:50:00 CST
1132    Tag: blume-20031104-move-libraries
1133    Description:
1134    
1135    Eliminated the "dont_move_libraries" directive in config/targets.
1136    (The mechanism was broken and could not be fixed easily.  Moreover,
1137    there does not seem to be any reason not to move all libraries into
1138    lib during installation.  I originally implemented this directive as a
1139    backward-compatibility feature when I first introduced the new CM.
1140    Now that things have been stable for a long time and going back to the
1141    old CM is not an option, there is no reason to keep it around.)
1142    
1143    ----------------------------------------------------------------------
1144    Name: Matthias Blume (blume (at) tti - c (dot) org)
1145    Date: 2003/11/03 16:00:00 CST
1146    Tag: blume-20031103-installdir
1147    Description:
1148    
1149    Made installer honor INSTALLDIR variable again.  (Thanks to Chris
1150    Richards for pointing out the problem and providing the solution.)
1151    
1152    ----------------------------------------------------------------------
1153    Name: Matthias Blume (blume (at) tti - c (dot) org)
1154    Date: 2003/10/01 17:05:00 CDT
1155    Tag: blume-20031001-lal-mlrisc
1156    Description:
1157    
1158    MLRISC bug fix from Lal.
1159    
1160    ----------------------------------------------------------------------
1161    Name: Matthias Blume (blume (at) tti - c (dot) org)
1162    Date: 2003/09/30 16:10:00 CDT
1163    Tag: blume-20030930-primio-bat
1164    Description:
1165    
1166    1. Added openVector, nullRd, and nullWr to PRIM_IO.
1167    2. Improved .bat files (for Win32 port) to make things work under Win95.
1168       (thanks to Aaron S. Hawley for this one)
1169    
1170    ----------------------------------------------------------------------
1171    Name: Matthias Blume (blume (at) tti - c (dot) org)
1172    Date: 2003/09/26 16:05:00 CDT
1173    Tag: blume-20030926-wrappriv
1174    Description:
1175    
1176    Added missing wrapper for privilege "primitive" in $smlnj/viscomp/core.cm.
1177    
1178    ----------------------------------------------------------------------
1179    Name: Matthias Blume (blume (at) tti - c (dot) org)
1180    Date: 2003/09/26 15:00:00 CDT
1181    Tag: blume-20030926-110_43_3
1182    Description:
1183    
1184    - additional cleanup
1185    - version number bump, NEW BOOTFILES
1186    
1187    ----------------------------------------------------------------------
1188    Name: Matthias Blume (blume (at) tti - c (dot) org)
1189    Date: 2003/09/26 12:00:00 CDT
1190    Tag: blume-20030926-ppautoload
1191    Description:
1192    
1193    I modified the read-eval-print loop so that the autoloader gets
1194    invoked whenever the prettyprinter tries to look up a symbol that
1195    is not currently defined in the toplevel environment but which
1196    appears in CM's autoload registry.  As a result, we see far fewer of
1197    those ?.Foo.Bar.xxx names in the prettyprinter's output.
1198    
1199    In addition to this I tried to clean up some pieces of the Basis
1200    implementation (e.g., Socket, Word8Array) in order to prevent other
1201    instances of these ?.Foo.Bar.xxx names from being printed.
1202    
1203    The mechanism that picks names for types still needs some work, though.
1204    (Right now it seems that if there is a type A.t which is defined to
1205    be B.u, but B is unavailable at toplevel, then A.t gets printed as
1206    "?.B.u" although the perhaps more sensible solution would be to use
1207    "A.t" in this case.  In other words, the prettyprinter should follow
1208    a chain of DEFtycs not farther than there are corresponding toplevel
1209    names in the current environment.)
1210    
1211    ----------------------------------------------------------------------
1212    Name: Matthias Blume (blume (at) tti - c (dot) org)
1213    Date: 2003/09/24 16:31:00 CDT
1214    Tag: blume-20030924-installer
1215    Description:
1216    
1217    Another installer tweak:  All the ML code for the installer is now
1218    compiled during CMB.make and put into a little library called
1219    $smlnj/installer.cm.  The installation then simply invokes
1220    
1221       sml -m $smlnj/installer.cm
1222    
1223    and everything happens automagically.
1224    
1225    Win32:  ML code senses value of environment variable SMLNJ_HOME.
1226    Unix:   ML code senses values of environment variables ROOT, CONFIGDIR,
1227            and BINDIR.
1228    
1229    The new scheme guarantees that the ML code responsible for the installation
1230    is in sync with the APIs of the main system.  Also, the installer is
1231    somewhat faster because the installer script is precompiled.
1232    
1233    ----------------------------------------------------------------------
1234    Name: Matthias Blume (blume (at) tti - c (dot) org)
1235    Date: 2003/09/24 15:35:00 CDT
1236    Tag: blume-20030924-synsock
1237    Description:
1238    
1239    Added a signature SYNCHRONOUS_SOCKET to basis.cm.  This is like SOCKET
1240    but excludes all non-blocking operations.  Defined SOCKET (in Basis)
1241    and CML_SOCKET in terms of SYNCHRONOUS_SOCKET.  Removed superfluous
1242    implementations of non-blocking operations from CML's Socket
1243    structure.
1244    
1245    ----------------------------------------------------------------------
1246    Name: Matthias Blume (blume (at) tti - c (dot) org)
1247    Date: 2003/09/24 15:10:05 CDT
1248    Tag: blume-20030924-sockets
1249    Description:
1250    
1251    1. Fixed SOCKET API and implementation to match Basis spec.
1252       This required changing the internal representation of sockets to one
1253       that remembers (for each socket file descriptor) whether it is currently
1254       blocking or non-blocking.  This state is maintained lazily (i.e., a system
1255       call is made only if the state actually needs to change).
1256    
1257    2. OS-specific details of sockets were moved into separate files, thus
1258       making it possible to unify the bulk of the socket implementations
1259       between Unix and Win32.
1260    
1261    3. CML's socket API changed accordingly.
1262       (Note that we need to remove non-blocking functions from this API
1263       since they are redundant in the case of CML!)
1264    
1265    4. CML's socket implementation now makes use of non-blocking functions
1266       provided by Basis, thus removing all OS-dependent code from this part
1267       of CML.
1268    
1269    5. Changed Real64.precision from 52 to 53.  Minor cleanup in Real64 code.
1270    
1271    ----------------------------------------------------------------------
1272    Name: Matthias Blume (blume (at) tti - c (dot) org)
1273    Date: 2003/09/22 12:10:00 CDT
1274    Tag: blume-20030922-110_43_2
1275    Description:
1276    
1277    Made a new interim version and bootfiles for developer's bootstrapping
1278    convenience.
1279    
1280       110.43.2 -- NEW BOOTFILES
1281    
1282    ----------------------------------------------------------------------
1283    Name: Matthias Blume (blume (at) tti - c (dot) org)
1284    Date: 2003/09/19 15:55:00 CDT
1285    Tag: blume-20030919-cmdir
1286    Description:
1287    
1288    1. new-install.sh -> install.sh
1289    2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
1290    3. tweaked installer so that another name instead of .cm can be chosen
1291       at install time (by setting the CM_DIR_ARC environment variable
1292       during installation); once installation is complete, the name is
1293       fixed
1294    
1295    ----------------------------------------------------------------------
1296    Name: Matthias Blume (blume (at) tti - c (dot) org)
1297    Date: 2003/09/18 16:00:00 CDT
1298    Tag: blume-20030918-110_43_1
1299    Description:
1300    
1301    Made a new interim version and bootfiles for developer's bootstrapping
1302    convenience.
1303    
1304       110.43.1 -- NEW BOOTFILES
1305    
1306    ----------------------------------------------------------------------
1307    Name: Matthias Blume (blume (at) tti - c (dot) org)
1308    Date: 2003/09/18 15:20:00 CDT
1309    Tag: blume-20030918-misc
1310    Description:
1311    
1312    1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
1313       this seems to be controversial at the moment) and used those in
1314       Posix.ProcEnv.times.
1315    
1316    2. Added Time.{from,to}Nanoseconds to Time.
1317    
1318    3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
1319       For example, fromLargeInt never needs to look at more than 3 "big
1320       digits" to get its 53 bits of precision.
1321    
1322    ----------------------------------------------------------------------
1323    Name: Matthias Blume (blume (at) tti - c (dot) org)
1324    Date: 2003/09/17 16:30:00 CDT
1325    Tag: blume-20030917-real32-slices
1326    Description:
1327    
1328    Added an entry to the primitive environment
1329    (compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
1330    added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
1331    
1332    Removed some of the "magic" constants in real64.sml and replaced them
1333    with code that generates these values from their corresponding
1334    integer counterparts.
1335    
1336    Made all(?) the slice-related changes to the Basis and made everything
1337    compile again...
1338    
1339    ----------------------------------------------------------------------
1340    Name: Matthias Blume (blume (at) tti - c (dot) org)
1341    Date: 2003/09/15 17:45:00 CDT
1342    Tag: blume-20030915-rbase
1343    Description:
1344    
1345    Fixed bug in Real.fromLargeInt.
1346    
1347    ----------------------------------------------------------------------
1348    Name: Matthias Blume (blume (at) tti - c (dot) org)
1349    Date: 2003/09/13 18:11:00 CDT
1350    Tag: blume-20030913-libinstall
1351    Description:
1352    
1353    Minor bugfix in config/libinstall (set anchor with path to
1354    standalone tool after installing it, otherwise libraries that
1355    need ml-lex or ml-yacc won't compile the first time the installer
1356    runs).
1357    
1358    ----------------------------------------------------------------------
1359    Name: Matthias Blume (blume (at) tti - c (dot) org)
1360    Date: 2003/09/12 11:45:00 CDT
1361    Tag: blume-20030912-various
1362    Description:
1363    
1364     - fixed bug in Real.toLargeInt
1365     - fixed bug in Posix.ProcEnv.times
1366     - changed inputLine functions to return an option
1367     - minor installer improvements / bugfixes
1368     - changed default @SMLalloc parameter for x86/celeron to 64k
1369    
1370    ----------------------------------------------------------------------
1371    Name: Matthias Blume (blume (at) tti - c (dot) org)
1372    Date: 2003/09/09 22:00:00 CDT
1373    Tag: Release_110_43
1374    Description:
1375    
1376        New working release 110.43.  New bootfiles.
1377    
1378    ----------------------------------------------------------------------
1379    Name: Matthias Blume (blume (at) tti - c (dot) org)
1380    Date: 2003/09/09 19:20:00 CDT
1381    Tag: blume-20030909-installer
1382    Description:
1383    
1384    Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
1385    Modified config/install.bat to take advantage of it.  Also modified
1386    config/install.sh (and called it config/new-install.sh) to take advantage
1387    of it on Unix systems.  (The SML code is (supposed to be) platform-
1388    independent.)
1389    
1390    The installer can now install everything under Win32
1391    as well as under *nix as long as it compiles.
1392    
1393    Other changes:
1394    
1395      - made CML compile again under Win32
1396      - made eXene compile under Win32 (by providing a fake structure UnixSock
1397        and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
1398      - fixed a bug in nowhere: it assumed that type OS.Process.status is the
1399        same as type int;  under Win32 it isn't
1400      - fixed some slice-related problems in the win32-specific parts of CML
1401      - added a functor argument "sameVol" to os-path-fn.sml in the Basis
1402        (under Win32, the volume name is case-insensitive, and the
1403        OS.Path code compares volume names for equality)
1404    
1405    ----------------------------------------------------------------------
1406    Name: Matthias Blume (blume (at) tti - c (dot) org)
1407    Date: 2003/09/08 11:55:00 CDT
1408    Tag: blume-20030908-fullpath
1409    Description:
1410    
1411    Made Win32 version of OS.FileSys.fullPath return current directory
1412    when given an empty string.  This is what the spec says, and incidentally,
1413    CM depends on it.  (CM otherwise goes into an infinite loop in certain
1414    cases when presented with the name of a non-existing .cm file.)
1415    
1416    ----------------------------------------------------------------------
1417    Name: Matthias Blume (blume (at) tti - c (dot) org)
1418    Date: 2003/09/04 16:30:00 CDT
1419    Tag: blume-20030905-slices-etc
1420    Description:
1421    
1422      1. Changed interface to vectors and arrays in Basis to match
1423         (draft) Basis spec.
1424      2. Added signatures and implementations of slices according to
1425         Basis spec.
1426      3. Edited source code throughout the system to make it compile again
1427         under 1. and 2.  (In some cases code had to be added to have it
1428         match the new signatures.)
1429      4. MLRISC should be backward-compatible: the copies of the originals
1430         of files that needed to change under 3. were retained, the .cm files
1431         check the compiler version number and use old versions when
1432         appropriate.
1433      5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
1434         dirstream -> string option (in accordance with Basis spec).
1435      6. When generating code that counts lines, ml-lex used function
1436         CharVector.foldli, taking advantage of its old interface.
1437         This has been replaced with the corresponding code from
1438         CharVectorSlice.  (html-lex must be re-lexed!)
1439      7. BitArray in smlnj-lib/Util has been extended/modified to match the
1440         new MONO_ARRAY signature.  (Do we need BitArraySlice?)
1441      8. Removed temporary additions (fromInternal, toInternal) from the
1442         (now obsolete) IntInf in smlnj-lib/Util.
1443      9. Cleaned up structure Byte.
1444     10. Added localOffset, scan, and fromString to Date (according to spec).
1445         Cleaned/corrected implementation of Date.
1446         (Still need to check for correctness; implement better canonicalizeDate.)
1447     11. Added "scan" to signature IEEE_REAL.
1448     12. Some improvements to IntInf [in particular: efficiency-hack for
1449         mod and rem when second operand is 2 (for parity checks).]
1450     13. Changed representation of type Time.time, using a single IntInf.int
1451         value counting microseconds.  This considerably simplified the
1452         implementation of structure Time.  We now support negative time
1453         values; scan and fromString handle signs.
1454     14. Functor PrimIO now takes two additional arguments (VectorSlice and
1455         ArraySlice).
1456    
1457    ----------------------------------------------------------------------
1458    Name: Matthias Blume (blume (at) tti - c (dot) org)
1459    Date: 2003/08/28 17:00:00 CDT
1460    Tag: blume-20030828-intinf
1461    Description:
1462    
1463    This is a major update which comes with a version number bump
1464    (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
1465    and an implementation of IntInf in the Basis.
1466    
1467    There are a fairly large number of related changes and updates throughout
1468    the system:
1469    
1470    Basis:
1471            - Implemented IntInf.
1472            - Made LargeInt a projection of IntInf (by filtering through INTEGER).
1473            - Added some missing Real64 operations, most notably Real.toLargeInt.
1474            - Added FixedInt as a synonym for Int32.
1475    
1476    compiler:
1477            * Added support for a built-in intinf type.
1478              - literals
1479              - pattern matching
1480              - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
1481              - overloading on literals and operations
1482    
1483            This required adding a primitive type intinf, some additional
1484            primops, and implementations for several non-trivial intinf
1485            operations in Core.  (The intinf type is completely abstract
1486            to the compiler; all operations get delegated back to the Core.)
1487    
1488            * Intinf equality is handled by polyequal.  However, the compiler
1489            does not print its usual warning in this case (since polyequal
1490            is the right thing to do there).
1491    
1492            * Improved the organization of structure InlineT.
1493    
1494            * A word about conversion primops:
1495            If conversions involving intinf do not cancel out during
1496            CPS contract, then the compiler must insert calls to Core functions.
1497            Since all core access must be resolved already during the FLINT
1498            translate phase, it would be too late a the time of CPS contract
1499            to add new Core calls.  For this reason, conversion primops
1500            for intinf carry two arguments:  1. the numeric argument that
1501            they are supposed to convert, and 2. the Core function that
1502            can help with this conversion if necessary.  If CPS contract
1503            eliminates a primop, then the associated Core function becomes
1504            dead and goes away.  Intinf conversion primops that do not get
1505            eliminated by CPS contract get rewritten into calls of their
1506            core functions by a separate, new phase.
1507    
1508    interactive system:
1509            - Control.Print.intinfDepth controls max length of intinf constants
1510            being printed. (Analogous to Control.Print.stringDepth.)
1511            - Cleanup in printutil and pputil: got rid of unused stuff and
1512            duplicates; replaced some of the code with code that makes better
1513            use of library functionality.
1514    
1515    CM:
1516            Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
1517            are no longer silent.
1518    
1519    CKIT:
1520            Fixed mismatched uses of Int32 and LargeInt.  I always decided
1521            in favor of LargeInt -- which is now the same as IntInf.
1522            CKIT-knowledgable people should check whether this is what's
1523            intended and otherwise change things back to using Int32 or
1524            FixedInt.
1525    
1526    Throughout the code:
1527            Started using IntInf.int literals and built-in operations
1528            (e.g., comparison with 0) where this seems appropriate.
1529    
1530    
1531    ----------------------------------------------------------------------
1532    Name: Dave MacQueen (dbm@cs.uchicago.edu)
1533    Date: 2003/08/13 11:36:00 CDT
1534    Tag: dbm-20030813-mcz-merge1
1535    Description:
1536    
1537    Merging changes from the mcz-branch development branch into trunk.
1538    These changes involve replacement of the emulated old prettyprinter
1539    interface with direct use of the SML/NJ Lib PP library, and fixing
1540    of a couple of bugs (895, 1186) relating to error messages.  A new
1541    prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
1542    has been added.
1543    
1544    ----------------------------------------------------------------------
1545    Name: Matthias Blume (blume (at) tti - c (dot) org)
1546    Date: 2003/08/11 15:45:00 CDT
1547    Tag: blume-20030811-windows
1548    Description:
1549    
1550    Version number bumped to 110.42.9.  NEW BOOTFILES!!!
1551    
1552        http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
1553    
1554    This patch restores SML/NJ's ability to run under win32.  There are a
1555    number of changes, including fixes for several bugs that had gone
1556    unnoticed until now:
1557    
1558      - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
1559        trying to run the win32 version from within cygwin.)
1560    
1561      - There are a number of simple .bat scripts that substitute for their
1562        corresponding Unix shell-scripts.  (See below.)
1563    
1564      - The internals of ml-build have been modified slightly.  The main
1565        difference is that instead of calling ".link-sml" (or link-sml.bat)
1566        using OS.Process.system, the ML process delegates this task back
1567        to the script.  Otherwise problems arise in mixed environments such
1568        as Cygwin where scripts look and work like Unix scripts, but
1569        where OS.Process.system cannot run them.
1570    
1571      - In CM, the srcpath pickler used native pathname syntax -- which
1572        is incorrect in the case of cross-compilation.  The new pickle format
1573        is independent of platform-specific naming conventions.
1574    
1575      - Path configuration files (such as lib/pathconfig) can now choose
1576        between native and standard syntax.  Placing a line of the form
1577    
1578            standard!
1579    
1580        into the file causes all subsequent paths to be interpreted using
1581        CM standard pathname syntax (= Unix conventions); a line
1582    
1583            native!
1584    
1585        switches back to native style.  This was needed so that
1586        path config files can be written portably, see src/system/pathconfig.
1587    
1588      - Runtime system:
1589    
1590          - win32-filesys.c:  get_file_time and set_file_time now
1591            access modification time, not creation time.
1592    
1593          - I/O code made aware of new array representation.
1594    
1595          - Bug fixes in X86.prim.masm.
1596    
1597       - src/system/makeml made aware of win32. (For use under cygwin
1598            and other Unix-environments for windows.)
1599    
1600       - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
1601         which caused BinIO.inputAll to fail consistently.
1602    
1603    .bat scripts:
1604    
1605       Windows .bat scripts assume that SMLNJ_HOME is defined.
1606    
1607       - sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
1608         applications (sml, ml-yacc, ml-lex).
1609       - ml-build.bat: analogous to ml-build.
1610       - config\install.bat:  Analogous to config/install.sh.  This requires
1611         that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
1612         (nmake etc. must be on the path, and vcvars32 must have been run.)
1613         Moreover, sources for ml-lex and ml-yacc need to exist under src,
1614         and the bootfile hierarchy must have been unpacked under
1615         sml.boot.x86-win32.
1616         The script is very primitive and does a poor job at error checking.
1617         It only installs the base system, ml-lex, and ml-yacc.  No other
1618         libraries are being installed (i.e., you get only those that
1619         are part of the compiler.)
1620       - link-sml.bat:  analogous to .link-sml, but not currently used
1621    
1622    Unrelated bug fixes:
1623    
1624      - ml-nlffigen now exports structures ST_* corresponding to incomplete
1625        types.
1626      - Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
1627        otherwise.)
1628    
1629    ----------------------------------------------------------------------
1630    Name: Dave MacQueen (macqueen@cs.uchicago.edu)
1631    Date: 2003/06/17
1632    Tag: macqueen-20030617-bug895
1633    Description:
1634    
1635    Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
1636    Tag will be used for new development branch (mcz-branch) for
1637    use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
1638    
1639    ----------------------------------------------------------------------
1640    Name: Matthias Blume (blume (at) tti - c (dot) org)
1641    Date: 2003/05/27 16:55:00 CDT
1642    Tag: blume-20030527-polyeq
1643    Description:
1644    
1645    Tried to eliminated most cases of polymorphic equality.
1646    
1647    ----------------------------------------------------------------------
1648    Name: Matthias Blume (blume (at) tti - c (dot) org)
1649    Date: 2003/05/21 17:45:00 CDT
1650    Tag: blume-20030517-complete
1651    Description:
1652    
1653    Two changes:
1654    
1655      1. Added a flag for controlling whether non-exhaustive bindings will
1656         be treated as errors (default is false).
1657      2. Cleaned up the *entire* source tree so that CMB.make goes through
1658         without a single non-exhaustive match- or bind warning.
1659    
1660    ----------------------------------------------------------------------
1661    Name: Matthias Blume (blume (at) tti - c (dot) org)
1662    Date: 2003/05/17 10:20:00 CDT
1663    Tag: blume-20030517-absyn
1664    Description:
1665    
1666    1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
1667    
1668       This mainly affects the quality of error messages.  However, some
1669       of the code is now more straightforward than before. (Treatment of
1670       the above four constructs in translate.sml is much simpler than
1671       the "macro-expansion" that was going on before.  Plus, the mach-
1672       compiler no longer gets invoked just to be able to compile an
1673       if-expression.)
1674    
1675    2. The ErrorMsg.Error exception is now caught and absorbed by the
1676       interactive loop.
1677    
1678    ----------------------------------------------------------------------
1679    Name: Allen Leung
1680    Date: 2003/05/16 13:05:00 CDT
1681    Tag: leunga-20030516-cygwin-runtime
1682    Description:
1683    
1684        Ported the runtime system to cygwin, which uses the unix
1685    x86-unix bin files.  Missing/buggy features:
1686    
1687       o getnetbyname, getnetbyaddr: these functions seem to be missing in
1688         the Cygwin library.
1689       o Ctrl-C handling may be flaky.
1690       o Windows system calls and Windows I/O are not supported.
1691    
1692       A new set of binfiles is located at:
1693    
1694         http://www.dorsai.org/~leunga/boot.x86-unix.tgz
1695    
1696       This is only needed for bootstrapping the cygwin version of smlnj.
1697       Other x86 versions can use the existing binfiles.
1698    
1699    ----------------------------------------------------------------------
1700    Name: Matthias Blume
1701    Date: 2003/04/08 15:42:00 CDT
1702    Tag: blume-20030408-listpair
1703    Description:
1704    
1705    1. Added a target 'mlrisc' to installer.
1706    
1707    2. Added missing elements to structure ListPair.
1708    
1709    ----------------------------------------------------------------------
1710    Name: Allen Leung
1711    Date: 2003/01/07 10:40:00 EST
1712    Tag: leunga-20030107-int-rem
1713    Description:
1714    
1715    Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
1716    The arguments to the SUBL instruction were swapped.
1717    
1718    ----------------------------------------------------------------------
1719    Name: Matthias Blume
1720    Date: 2002/12/12 16:25:00 EST
1721    Tag: blume-20021212-risc-ra
1722    Description:
1723    
1724    Fixed a serious bug in the rewrite code for FP spilling/reloading that
1725    sent the RA into an infinite loop when floating point registers get
1726    spilled.  (Because of this bug, e.g., nucleic stopped compiling between
1727    110.37 and 110.38.)
1728    There was another set of potential problems related to the handling of
1729    MLRISC annotations (but those did not yet cause real problems, apparently).
1730    
1731    ----------------------------------------------------------------------
1732    Name: Matthias Blume
1733    Date: 2002/12/06 22:40:00 EST
1734    Tag: blume-20021206-cm-fileid
1735    Description:
1736    
1737    Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
1738    This fixes the problem of CM getting confused by files that suddenly
1739    change their identity (e.g., by getting unlinked and recreated by some
1740    text editor such as vi).  There might be a better/cheaper/cleaner way
1741    of doing this, but for now this will have to do.
1742    
1743    ----------------------------------------------------------------------
1744    Name: Matthias Blume
1745    Date: 2002/10/28 09:50:00 EST
1746    Tag: blume-20021028-typecheck
1747    Description:
1748    
1749    Exported structure Typecheck from $smlnj/viscomp/core.cm.
1750    
1751    ----------------------------------------------------------------------
1752    Name: Matthias Blume
1753    Date: 2002/10/17 09:10:00 EDT
1754    Tag: Release_110_42
1755    Description:
1756    
1757    In good old tradition, there has been a slight hiccup so that we have
1758    to patch 110.42 after the fact.   The old release tag has been replaced
1759    (see below).
1760    
1761    The change solves a problem with two competing approaches the
1762    configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
1763    each other's way.
1764    
1765    This change only affects the runtime system code and the installer script.
1766    (No new bootfiles.)
1767    
1768    ----------------------------------------------------------------------
1769    Name: Matthias Blume
1770    Date: 2002/10/16 12:00:00 EDT
1771    Tag: Release_110_42_removed
1772    Description:
1773    
1774    New working release.  New bootfiles.
1775    
1776    ----------------------------------------------------------------------
1777    Name: Matthias Blume
1778    Date: 2002/10/10 13:10:00 EDT
1779    Tag: blume-20021010-ppc-divs
1780    Description:
1781    
1782    The mltree operator DIVS must be implemented with an overflow check on
1783    the PPC because the hardware indicates divide-by-zero using "overflow" as
1784    well.
1785    
1786    ----------------------------------------------------------------------
1787    Name: Matthias Blume
1788    Date: 2002/07/23 11:45:00
1789    Tag: blume-20020723-smlnj-home
1790    Description:
1791    
1792    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
1793    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
1794    CM_PATHCONFIG is also set), the path configuration file is assumed
1795    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
1796    the entire tree to some other place and everything will "just work".
1797    
1798    (Companion commands such as ml-build and ml-makedepend also sense this
1799    variable.)
1800    
1801    ----------------------------------------------------------------------
1802    Name: Matthias Blume
1803    Date: 2002/07/12 21:19:00 EDT
1804    Tag: blume-20020712-liveness
1805    Description:
1806    
1807    Exported two useful "step" functions from liveness module (MLRISC).
1808    
1809    ----------------------------------------------------------------------
1810    Name: Matthias Blume
1811    Date: 2002/07/05 16:00 EDT
1812    Tag: Release_110_41
1813    Description:
1814    
1815    New working release.  New bootfiles.
1816    
1817    ----------------------------------------------------------------------
1818    Name: Matthias Blume
1819    Date: 2002/07/05 10:25:00 EDT
1820    Tag: blume-20020705-btimp
1821    Description:
1822    
1823    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
1824    other clients can set up backtracing support.
1825    
1826    ----------------------------------------------------------------------
1827    Name: Matthias Blume
1828    Date: 2002/06/25 17:23:00 EDT
1829    Tag: blume-20020625-fpmax
1830    Description:
1831    
1832    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
1833    case.  (The sense of the isNaN test was reversed -- which made min and
1834    max always return their first argument.)
1835    
1836    ----------------------------------------------------------------------
1837    Name: Matthias Blume
1838    Date: 2002/06/11
1839    Tag: blume-20020611-unixpath
1840    Description:
1841    
1842    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
1843    
1844    ----------------------------------------------------------------------
1845    Name: Matthias Blume
1846    Date: 2002/06/10 16:35:00 EDT
1847    Tag: blume-20020610-ieeereal
1848    Description:
1849    
1850    I back-ported my implementation of IEEEReal.fromString from the
1851    idlbasis-devel branch so that we can test it.
1852    
1853    Another small change is that ppDec tries to give more information
1854    than just "<sig>" in the case of functors.  However, this code is
1855    broken in some mysterious way if the functor's body's signature
1856    has not been declared by ascription but gets inferred from the
1857    implementation.  This needs fixing...
1858    
1859    ----------------------------------------------------------------------
1860    Name: Matthias Blume
1861    Date: 2002/05/31
1862    Tag: blume-20020531-btrace-mode
1863    Description:
1864    
1865    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
1866    the wayside when I switched over to using Controls everywhere.)
1867    
1868    ----------------------------------------------------------------------
1869    Name: Lal George
1870    Date: 2002/05/23 12:21:40 EDT
1871    Tag: george-20020523-visual-labels
1872    Description:
1873    
1874    Labels are now displayed in the graphical output to make
1875    the fall-through and target blocks obvious.
1876    
1877    ----------------------------------------------------------------------
1878    Name: Matthias Blume
1879    Date: 2002/05/22 11:03:00 EDT
1880    Tag: blume-20020522-shrink
1881    Description:
1882    
1883    John tweaked yesterday's fix for 1131 to handle an out-of-memory
1884    situation that comes up when allocating huge arrays.
1885    
1886    ----------------------------------------------------------------------
1887    Name: Matthias Blume
1888    Date: 2002/05/21 16:00:00 EDT
1889    Tag: Release_110_40
1890    Description:
1891    
1892    New working release (110.40).  New bootfiles.
1893    
1894    [Also: John Reppy fixed GC bug 1131.]
1895    
1896    ----------------------------------------------------------------------
1897    Name: Matthias Blume
1898    Date: 2002/05/21 12:35:00 EDT
1899    Tag: blume-20020521-cmdoc
1900    Description:
1901    
1902    CM documentation update.
1903    
1904    ----------------------------------------------------------------------
1905    Name: Matthias Blume
1906    Date: 2002/05/21 10:55:00 EDT
1907    Tag: blume-20020521-misc
1908    Description:
1909    
1910    - John tweaked runtime to be silent on heap export (except when
1911      GC messages are on).
1912    - I added a few more things (cross-compiling versions of CMB) to
1913      config/preloads (as suggestions).
1914    
1915    ----------------------------------------------------------------------
1916    Name: Matthias Blume
1917    Date: 2002/05/20 22:25:00 EDT
1918    Tag: blume-20020520-controls
1919    Description:
1920    
1921    - Added ControlUtil structure to control-lib.cm.
1922    - Use it throughout.
1923    - Used Controls facility to define MLRISC controls (as opposed to
1924      registering MLRISC control ref cells with Controls after the fact)
1925    - Fixed messed-up controls priorities.
1926    
1927    * Removed again all the stuff from config/preloads that one wouldn't
1928      be able to preload at the time the initial heap image is built.
1929      (Many libraries, e.g., CML, do not exist yet at this time.  The
1930       only libraries that can be preloaded via config/preloads are those
1931       that come bundled with the bootfiles.)
1932    
1933    ----------------------------------------------------------------------
1934    Name: Matthias Blume
1935    Date: 2002/05/20 10:59:00 EDT
1936    Tag: blume-20020520-preloads
1937    Description:
1938    
1939    Added a lot of commented-out suggestions for things to be included
1940    in config/preloads.
1941    
1942    ----------------------------------------------------------------------
1943    Name: Allen Leung
1944    Date: 2002/05/18 14:20:00 EDT
1945    Tag: leunga-20020518-mdl
1946    Description:
1947    
1948    o Made the mdl tool stuff compile and run again.
1949    o I've disabled all the stuff that depends on RTL specifications; they
1950      are all badly broken anyway.
1951    
1952    ----------------------------------------------------------------------
1953    Name: Matthias Blume
1954    Date: 2002/05/17 16:49:00 EDT
1955    Tag: blume-20020517-controls
1956    Description:
1957    
1958    0. John Reppy made several modifications to the SML/NJ library.
1959       In particular, there is a shiny new controls-lib.cm.
1960    
1961    1. Pushed new controls interface through compiler so that everything
1962       compiles again.
1963    
1964    2. Added FormatComb and FORMAT_COMB to the CML version of the
1965       SML/NJ library (so that CML compiles again).
1966    
1967    3. Modified init scripts because XXX_DEFAULT environment variables
1968       are no longer with us.  (Boot-time initialization is now done
1969       using the same environment variables that are also used for
1970       startup-time initialization of controls.)
1971    
1972    ----------------------------------------------------------------------
1973    Name: Lal George
1974    Date: 2002/05/15 09:20:10 EDT
1975    Tag: george-20020515-pseudo-op-decls
1976    Description:
1977    
1978            All pseudo-ops emitted before the first segment declaration
1979    such as TEXT, DATA, and BSS directives are assumed to be global
1980    declarations and are emitted first in the assembly file. This is
1981    useful in a number of situations where one has pseudo-ops that are not
1982    specific to any segment, and also works around the constraint that one
1983    cannot have client pseudo-ops in the TEXT segment.
1984    
1985            Because no segment is associated with these declarations it is
1986    an error to allocate any space or objects before the first segment
1987    directive and an exception will be raised. However, we cannot make
1988    this check for client pseudo-ops.
1989    
1990            These top level declarations are a field in the CFG graph_info.
1991    In theory you can continue to add to this field after the CFG has been
1992    built -- provided you know what you are doing;-)
1993    
1994    ----------------------------------------------------------------------
1995    Name: Matthias Blume
1996    Date: 2002/05/13 16:40:00 EDT
1997    Tag: blume-20020513-pp-etc
1998    Description:
1999    
2000    A few minor bugfixes:
2001    
2002      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
2003        (Bogus printouts for redefined bindings still occur.  Compiler
2004        bug should no longer occur now.  We need to redo the prettyprinter
2005        from scratch.)
2006    
2007      - CM pathname printer now also adds escape sequences for ( and )
2008    
2009      - commend and docu fixes for ml-nlffi
2010    
2011    ----------------------------------------------------------------------
2012    Name: Matthias Blume
2013    Date: 2002/05/10 16:40:00 EDT
2014    Tag: blume-20020510-erg-textio
2015    Description:
2016    
2017    Applied the following bugfix provided by Emden Gansner:
2018    
2019        Output is corrupted when outputSubstr is used rather than output.
2020    
2021        The problem occurs when a substring
2022    
2023            ss = (s, dataStart, dataLen)
2024    
2025        where dataStart > 0, fills a stream buffer with avail bytes left.
2026        avail bytes of s, starting at index dataStart, are copied into the
2027        buffer, the buffer is flushed, and then the remaining dataLen-avail
2028        bytes of ss are copied into the beginning of the buffer. Instead of
2029        starting this copy at index dataStart+avail in s, the current code
2030        starts the copy at index avail.
2031    
2032        Fix:
2033        In text-io-fn.sml, change line 695 from
2034             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
2035        to
2036             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
2037    
2038    ----------------------------------------------------------------------
2039    Name: Matthias Blume
2040    Date: 2002/04/12 13:55:00 EDT
2041    Tag: blume-20020412-assyntax
2042    Description:
2043    
2044    1. Grabbed newer assyntax.h from the XFree86 project.
2045    2. Fiddled with how to compile X86.prim.asm without warnings.
2046    3. (Very) Minor cleanup in CM.
2047    
2048    ----------------------------------------------------------------------
2049    Name: Matthias Blume
2050    Date: 2002/04/01 (no joke!) 17:07:00 EST
2051    Tag: blume-20020401-x86div
2052    Description:
2053    
2054    Added full support for div/mod/rem/quot on the x86, using the machine
2055    instruction's two results (without clumsily recomputing the remainder)
2056    directly where appropriate.
2057    
2058    Some more extensive power-of-two support was added to the x86 instruction
2059    selector (avoiding expensive divs, mods, and muls where they can be
2060    replaced with cheaper shifts and masks).  However, this sort of thing
2061    ought to be done earlier, e.g., within the CPS optimizer so that
2062    all architectures benefit from it.
2063    
2064    The compiler compiles to a fixed point, but changes might be somewhat
2065    fragile nevertheless.  Please, report any strange things that you might
2066    see wrt. div/mod/quot/rem...
2067    
2068    ----------------------------------------------------------------------
2069    Name: Matthias Blume
2070    Date: 2002/03/29 17:22:00
2071    Tag: blume-20020329-div
2072    Description:
2073    
2074    Fixed my broken div/mod logic.  Unfortunately, this means that the
2075    inline code for div/mod now has one more comparison than before.
2076    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
2077    The problem was with quotient = 0, because that alone does not tell
2078    us which way the rounding went.  One then has to look at whether
2079    remainder and divisor have the same sign...  :(
2080    
2081    Anyway, I replaced the bootfiles with fresh ones...
2082    
2083    ----------------------------------------------------------------------
2084    Name: Matthias Blume
2085    Date: 2002/03/29 14:10:00 EST
2086    Tag: blume-20020329-inlprims
2087    Description:
2088    
2089    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
2090    
2091    Primops have changed. This means that the bin/boot-file formats have
2092    changed as well.
2093    
2094    To make sure that there is no confusion, I made a new version.
2095    
2096    
2097    CHANGES:
2098    
2099    * removed REMT from mltree (remainder should never overflow).
2100    
2101    * added primops to deal with divisions of all flavors to the frontend
2102    
2103    * handled these primops all the way through so they map to their respective
2104      MLRISC support
2105    
2106    * used these primops in the implementation of Int, Int32, Word, Word32
2107    
2108    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
2109    
2110    * parameterized INLMIN, INLMAX, and INLABS by a numkind
2111    
2112    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
2113      floating point
2114    
2115    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
2116      and Real (but Real.abs maps to a separate floating-point-only primop)
2117    
2118    
2119    TODO items:
2120    
2121    * Hacked Alpha32 instruction selection, disabling the selection of REMx
2122      instructions because the machine instruction encoder cannot handle
2123      them.  (Hppa, PPC, and Sparc instruction selection did not handle
2124      REM in the first place, and REM is supported by the x86 machine coder.)
2125    
2126    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
2127      selection phase.  (The two can be streamlined because the hardware
2128      delivers both quotient and remainder at the same time anyway.)
2129    
2130    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
2131      (Currently the behavior is inconsistent both across architectures and
2132      wrt. the draft Basis spec.)
2133    
2134    * Word8 should eventually be handled natively, too.
2135    
2136    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
2137      as if there currently is no execution path that could trigger it in
2138      SML/NJ.  (The assumptions underlying functions arith and promotable do not
2139      hold for things like multiplication and division.)
2140    
2141    ----------------------------------------------------------------------
2142    Name: Matthias Blume
2143    Date: 2002/03/27 16:27:00 EST
2144    Tag: blume-20020327-mlrisc-divisions
2145    Description:
2146    
2147    Added support for all four division operations (ML's div, mod, quot,
2148    and rem) to MLRISC.  In the course of doing so, I also rationalized
2149    the naming (no more annoying switch-around of DIV and QUOT), by
2150    parameterizing the operation by div_rounding_mode (which can be either
2151    DIV_TO_ZERO or DIV_TO_NEGINF).
2152    
2153    The generic MLTreeGen functor takes care of compiling all four
2154    operations down to only round-to-zero div.
2155    
2156    Missing pieces:
2157    
2158      * Doing something smarter than relying on MLTreeGen on architectures
2159        like, e.g., the x86 where hardware division delivers both quotient and
2160        remainder at the same time.  With this, the implementation of the
2161        round-to-neginf operations could be further streamlined.
2162    
2163      * Remove inlining support for div/mod/rem from the frontend and replace it
2164        with primops that get carried through to the backend.  Do this for all
2165        int and word types.
2166    
2167    ----------------------------------------------------------------------
2168    Name: Matthias Blume
2169    Date: 2002/03/25 17:25:00 EST
2170    Tag: blume-20020325-divmod
2171    Description:
2172    
2173    I improved (hopefully without breaking them) the implementation of Int.div,
2174    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
2175    advantage of the following observations:
2176    
2177      Let  q = x quot y      r = x rem y
2178           d = x div  y      m = x mod y
2179    
2180    where "quot" is the round-to-zero version of integer division that
2181    hardware usually provides.  Then we have:
2182    
2183         r = x - q * y        where neither the * nor the - will overflow
2184         d = if q >= 0 orelse x = q * y then q else q - 1
2185                              where neither the * nor the - will overflow
2186         m = if q >= 0 orelse r = 0 then r else r + y
2187                              where the + will not overflow
2188    
2189    This results in substantial simplification of the generated code.
2190    The following table shows the number of CFG nodes and edges generated
2191    for
2192            fun f (x, y) = x OPER y
2193            (* with OPER \in div, mod, quot, rem *)
2194    
2195    
2196        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
2197        --------------------------------------------------------
2198         div |         24 |         39 |         12 |         16
2199         mod |         41 |         71 |         12 |         16
2200        quot |          8 |         10 |          8 |         10
2201         rem |         10 |         14 |          8 |         10
2202    
2203    
2204    ----------------------------------------------------------------------
2205    Name: Matthias Blume
2206    Date: 2002/03/25 22:06:00 EST
2207    Tag: blume-20020325-cprotobug
2208    Description:
2209    
2210    Fixed a bug in cproto (c prototype decoder).
2211    
2212    ----------------------------------------------------------------------
2213    Name: Matthias Blume
2214    Date: 2002/03/25 16:00:00 EST
2215    Tag: blume-20020325-raw-primops
2216    Description:
2217    
2218    I did some cleanup to Allen's new primop code and
2219    replaced yesterday's bootfiles with new ones.
2220    (But they are stored in the same place.)
2221    
2222    ----------------------------------------------------------------------
2223    Name: Matthias Blume
2224    Date: 2002/03/24 22:40:00 EST
2225    Tag: blume-20020324-bootfiles
2226    Description:
2227    
2228    Made the bootfiles that Allen asked for.
2229    
2230    ----------------------------------------------------------------------
2231    Name: Allen Leung
2232    Date: 2002/03/23 15:50:00 EST
2233    Tag: leunga-20020323-flint-cps-rcc-primops
2234    Description:
2235    
2236      1. Changes to FLINT primops:
2237    
2238        (* make a call to a C-function;
2239         * The primop carries C function prototype information and specifies
2240         * which of its (ML-) arguments are floating point. C prototype
2241         * information is for use by the backend, ML information is for
2242         * use by the CPS converter. *)
2243      | RAW_CCALL of { c_proto: CTypes.c_proto,
2244                       ml_args: ccall_type list,
2245                       ml_res_opt: ccall_type option,
2246                       reentrant : bool
2247                     } option
2248       (* Allocate uninitialized storage on the heap.
2249        * The record is meant to hold short-lived C objects, i.e., they
2250        * are not ML pointers.  With the tag, the representation is
2251        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
2252        *)
2253      | RAW_RECORD of {tag:bool,sz:int}
2254      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
2255    
2256      2.  These CPS primops are now overloaded:
2257    
2258           rawload of {kind:numkind}
2259           rawstore of {kind:numkind}
2260    
2261          The one argument form is:
2262    
2263             rawload {kind} address
2264    
2265          The two argument form is:
2266    
2267             rawload {kind} [ml object, byte-offset]
2268    
2269      3. RAW_CCALL/RCC now takes two extra arguments:
2270    
2271         a. The first is whether the C call is reentrant, i.e., whether
2272            ML state should be saved and restored.
2273         b. The second argument is a string argument specifying the name of
2274            library and the C function.
2275    
2276         These things are currently not handled in the code generator, yet.
2277    
2278      4. In CProto,
2279    
2280         An encoding type of "bool" means "ml object" and is mapped into
2281         C prototype of PTR.  Note that "bool" is different than "string",
2282         even though "string" is also mapped into PTR, because "bool"
2283         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
2284    
2285      5. Pickler/unpicker
2286    
2287         Changed to handle RAW_RECORD and newest RAW_CCALL
2288    
2289      6. MLRiscGen,
2290    
2291         1. Changed to handle the new rawload/rawstore/rawrecord operators.
2292         2. Code for handling C Calls has been moved to a new module CPSCCalls,
2293            in the file CodeGen/cpscompile/cps-c-calls.sml
2294    
2295      7. Added the conditional move operator
2296    
2297             condmove of branch
2298    
2299         to cps.  Generation of this is still buggy so it is currently
2300         disabled.
2301    
2302    ----------------------------------------------------------------------
2303    Name: Lal George
2304    Date: 2002/03/22 14:18:25 EST
2305    Tag: george-20020322-cps-branch-prob
2306    Description:
2307    
2308    Implemented the Ball-Larus branch prediction-heuristics, and
2309    incorporated graphical viewers for control flow graphs.
2310    
2311    Ball-Larus Heuristics:
2312    ---------------------
2313    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
2314    
2315    By design it uses the Dempster-Shafer theory for combining
2316    probabilities.  For example, in the function:
2317    
2318        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
2319    
2320    the ball-larus heuristics predicts that the n=0 is unlikely
2321    (OH-heuristic), and the 'then' branch is unlikely because of the
2322    RH-heuristic -- giving the 'then' branch an even lower combined
2323    probability using the Dempster-Shafer theory.
2324    
2325    Finally, John Reppy's loop analysis in MLRISC, further lowers the
2326    probability of the 'then' branch because of the loop in the else
2327    branch.
2328    
2329    
2330    Graphical Viewing:
2331    ------------------
2332    I merely plugged in Allen's graphical viewers into the compiler. The
2333    additional code is not much. At the top level, saying:
2334    
2335            Control.MLRISC.getFlag "cfg-graphical-view" := true;
2336    
2337    will display the graphical view of the control flow graph just before
2338    back-patching.  daVinci must be in your path for this to work. If
2339    daVinci is not available, then the default viewer can be changed
2340    using:
2341    
2342            Control.MLRISC.getString "viewer"
2343    
2344    which can be set to "dot" or "vcg" for the corresponding viewers. Of
2345    course, these viewers must be in your path.
2346    
2347    The above will display the compilation unit at the level of clusters,
2348    many of which are small, boring, and un-interesting. Also setting:
2349    
2350            Control.MLRISC.getInt "cfg-graphical-view_size"
2351    
2352    will display clusters that are larger than the value set by the above.
2353    
2354    
2355    ----------------------------------------------------------------------
2356    Name: Matthias Blume
2357    Date: 2002/03/21 22:20:00 EST
2358    Tag: blume-20020321-kmp-bugfix
2359    Description:
2360    
2361    Changed the interface to the KMP routine in PreString and fixed
2362    a minor bug in one place where it was used.
2363    
2364    ----------------------------------------------------------------------
2365    Name: Allen Leung
2366    Date: 2002/03/21 20:30:00 EST
2367    Tag: leunga-20020321-cfg
2368    Description:
2369    
2370      Fixed a potential problem in cfg edge splitting.
2371    
2372    ----------------------------------------------------------------------
2373    Name: Allen Leung
2374    Date: 2002/03/21 17:15:00 EST
2375    Tag: leunga-20020321-x86-fp-cfg
2376    Description:
2377    
2378      1. Recoded the buggy parts of x86-fp.
2379    
2380         a. All the block reordering code has been removed.
2381            We now depend on the block placement phases to do this work.
2382    
2383         b. Critical edge splitting code has been simplified and moved into the
2384            CFG modules, as where they belong.
2385    
2386         Both of these were quite buggy and complex.  The code is now much, much
2387         simpler.
2388    
2389      2. X86 backend.
2390    
2391         a. Added instructions for 64-bit support.  Instruction selection for
2392            64-bit has not been committed, however, since that
2393            requires changes to MLTREE which haven't been approved by
2394            Lal and John.
2395    
2396         b. Added support for FUCOMI and FUCOMIP when generating code for
2397            PentiumPro and above.  We only generate these instructions in
2398            the fast-fp mode.
2399    
2400         c. Added cases for JP and JNP in X86FreqProps.
2401    
2402      3. CFG
2403    
2404         CFG now has a bunch of methods for edge splitting and merging.
2405    
2406      4. Machine description.
2407    
2408         John's simplification of MLTREE_BASIS.fcond broke a few machine
2409         description things:
2410    
2411         rtl-build.{sig,sml} and hppa.mdl fixed.
2412    
2413         NOTE: the machine description stuff in the repository is still broken.
2414               Again, I can't put my fixes in because that involves
2415               changes to MLTREE.
2416    
2417    ----------------------------------------------------------------------
2418    Name: Matthias Blume
2419    Date: 2002/03/20 15:55:00 EST
2420    Tag: blume-20020320-kmp
2421    Description:
2422    
2423    Implemented Knuth-Morris-Pratt string matching in PreString and used
2424    it for String.isSubstring, Substring.isSubstring, and
2425    Substring.position.
2426    
2427    (Might need some stress-testing.  Simple examples worked fine.)
2428    
2429    ----------------------------------------------------------------------
2430    Name: Matthias Blume
2431    Date: 2002/03/19 16:37:00 EST
2432    Tag: blume-20020319-witnesses
2433    Description:
2434    
2435    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
2436    
2437    This implements a generic mechanism for changing constness qualifiers
2438    anywhere within big C types without resorting to outright "casts".
2439    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
2440    modify the constness at the outermost level.)
2441    The implementation of "convert" is based on the idea of "witness"
2442    values -- values that are not used by the operation but whose types
2443    "testify" to their applicability.  On the implementation side, "convert"
2444    is simply a projection (returning its second curried argument).  With
2445    cross-module inlining, it should not result in any machine code being
2446    generated.
2447    
2448    ----------------------------------------------------------------------
2449    Name: Matthias Blume
2450    Date: 2002/03/15 16:40:00 EST
2451    Tag: blume-20020315-basis
2452    Description:
2453    
2454    Provided (preliminary?) implementations for
2455    
2456      {String,Substring}.{concatWith,isSuffix,isSubstring}
2457    
2458    and
2459    
2460      Substring.full
2461    
2462    Those are in the Basis spec but they were missing in SML/NJ.
2463    
2464    ----------------------------------------------------------------------
2465    Name: Matthias Blume
2466    Date: 2002/03/14 21:30:00 EST
2467    Tag: blume-20020314-controls
2468    Description:
2469    
2470    Controls:
2471    ---------
2472    
2473    1. Factored out the recently-added Controls : CONTROLS stuff and put
2474       it into its own library $/controls-lib.cm.  The source tree for
2475       this is under src/smlnj-lib/Controls.
2476    
2477    2. Changed the names of types and functions in this interface, so they
2478       make a bit more "sense":
2479    
2480          module -> registry
2481          'a registry -> 'a group
2482    
2483    3. The interface now deals in ref cells only.  The getter/setter interface
2484       is (mostly) gone.
2485    
2486    4. Added a function that lets one register an already-existing ref cell.
2487    
2488    5. Made the corresponding modifications to the rest of the code so that
2489       everything compiles again.
2490    
2491    6. Changed the implementation of Controls.MLRISC back to something closer
2492       to the original.  In particular, this module (and therefore MLRISC)
2493       does not depend on Controls.  There now is some link-time code in
2494       int-sys.sml that registers the MLRISC controls with the Controls
2495       module.
2496    
2497    CM:
2498    ---
2499    
2500      * One can now specify the lambda-split aggressiveness in init.cmi.
2501    
2502    ----------------------------------------------------------------------
2503    Name: Allen Leung
2504    Date: 2002/03/13 17:30:00 EST
2505    Tag: leunga-20020313-x86-fp-unary
2506    Description:
2507    
2508    Bug fix for:
2509    
2510    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
2511    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
2512    > - fun f(x,(y,z)) = Real.~ y;
2513    > [autoloading]
2514    > [autoloading done]
2515    >       fchsl   (%eax), 184(%esp)
2516    > Error: MLRisc bug: X86MCEmitter.emitInstr
2517    >
2518    > uncaught exception Error
2519    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
2520    
2521    The problem was that the code generator did not generate any fp registers
2522    in this case, and the ra didn't know that it needed to run the X86FP phase to
2523    translate the pseudo fp instruction.   This only happened with unary fp
2524    operators in certain situations.
2525    
2526    ----------------------------------------------------------------------
2527    Name: Matthias Blume
2528    Date: 2002/03/13 14:00:00 EST
2529    Tag: blume-20020313-overload-etc
2530    Description:
2531    
2532    1. Added _overload as a synonym for overload for backward compatibility.
2533       (Control.overloadKW must be true for either version to be accepted.)
2534    
2535    2. Fixed bug in install script that caused more things to be installed
2536       than what was requested in config/targets.
2537    
2538    3. Made CM aware of the (_)overload construct so that autoloading
2539       works.
2540    
2541    ----------------------------------------------------------------------
2542    Name: Matthias Blume
2543    Date: 2002/03/12 22:03:00 EST
2544    Tag: blume-20020312-url
2545    Description:
2546    
2547    Forgot to update BOOT and srcarchiveurl.
2548    
2549    ----------------------------------------------------------------------
2550    Name: Matthias Blume
2551    Date: 2002/03/12 17:30:00 EST
2552    Tag: blume-20020312-version110392
2553    Description:
2554    
2555    Yet another version number bump (because of small changes to the
2556    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
2557    
2558    Changes:
2559    
2560      The new pid generation scheme described a few weeks ago was overly
2561      complicated.  I implemented a new mechanism that is simpler and
2562      provides a bit more "stability":  Once CM has seen a compilation
2563      unit, it keeps its identity constant (as long as you do not delete
2564      those crucial CM/GUID/* files).  This means that when you change
2565      an interface, compile, then go back to the old interface, and
2566      compile again, you arrive at the original pid.
2567    
2568      There now also is a mechanism that instructs CM to use the plain
2569      environment hash as a module's pid (effectively making its GUID
2570      the empty string).  For this, "noguid" must be specified as an
2571      option to the .sml file in question within its .cm file.
2572      This is most useful for code that is being generated by tools such
2573      as ml-nlffigen (because during development programmers tend to
2574      erase the tool's entire output directory tree including CM's cached
2575      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
2576      revert to the old, broken behavior of SML/NJ, but in specific cases
2577      where there is no danger of interface confusion, its use is ok
2578      (I think).
2579    
2580      ml-nlffigen by default generates "noguid" annotations.  They can be
2581      turned off by specifying -guid in its command line.
2582    
2583    ----------------------------------------------------------------------
2584    Name: Lal George
2585    Date: 2002/03/12 12 14:42:36 EST
2586    Tag: george-20020312-frequency-computation
2587    Description:
2588    
2589    Integrated jump chaining and static block frequency into the
2590    compiler. More details and numbers later.
2591    
2592    ----------------------------------------------------------------------
2593    Name: Lal George
2594    Date: 2002/03/11 11 22:38:53 EST
2595    Tag: george-20020311-jump-chain-elim
2596    Description:
2597    
2598    Tested the jump chain elimination on all architectures (except the
2599    hppa).  This is on by default right now and is profitable for the
2600    alpha and x86, however, it may not be profitable for the sparc and ppc
2601    when compiling the compiler.
2602    
2603    The gc test will typically jump to a label at the end of the cluster,
2604    where there is another jump to an external cluster containing the actual
2605    code to invoke gc. This is to allow factoring of common gc invocation
2606    sequences. That is to say, we generate:
2607    
2608            f:
2609               testgc
2610               ja   L1      % jump if above to L1
2611    
2612            L1:
2613               jmp L2
2614    
2615    
2616    After jump chain elimination the 'ja L1' instructions is converted to
2617    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
2618    up being implemented in their long form (if L2 is far away) using:
2619    
2620            jbe     L3      % jump if below or equal to L3
2621            jmp     L2
2622         L3:
2623            ...
2624    
2625    
2626    For large compilation units L2  may be far away.
2627    
2628    
2629    ----------------------------------------------------------------------
2630    Name: Matthias Blume
2631    Date: 2002/03/11 13:30:00 EST
2632    Tag: blume-20020311-mltreeeval
2633    Description:
2634    
2635    A functor parameter was missing.
2636    
2637    ----------------------------------------------------------------------
2638    Name: Allen Leung
2639    Date: 2002/03/11 10:30:00 EST
2640    Tag: leunga-20020311-runtime-string0
2641    Description:
2642    
2643       The representation of the empty string now points to a
2644    legal null terminated C string instead of unit.  It is now possible
2645    to convert an ML string into C string with InlineT.CharVector.getData.
2646    This compiles into one single machine instruction.
2647    
2648    ----------------------------------------------------------------------
2649    Name: Allen Leung
2650    Date: 2002/03/10 23:55:00 EST
2651    Tag: leunga-20020310-x86-call
2652    Description:
2653    
2654       Added machine generation for CALL instruction (relative displacement mode)
2655    
2656    ----------------------------------------------------------------------
2657    Name: Matthias Blume
2658    Date: 2002/03/08 16:05:00
2659    Tag: blume-20020308-entrypoints
2660    Description:
2661    
2662    Version number bumped to 110.39.1.  NEW BOOTFILES!
2663    
2664    Entrypoints: non-zero offset into a code object where execution should begin.
2665    
2666    - Added the notion of an entrypoint to CodeObj.
2667    - Added reading/writing of entrypoint info to Binfile.
2668    - Made runtime system bootloader aware of entrypoints.
2669    - Use the address of the label of the first function given to mlriscGen
2670      as the entrypoint.  This address is currently always 0, but it will
2671      not be 0 once we turn on block placement.
2672    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
2673      with entry points) from mlriscGen.
2674    
2675    ----------------------------------------------------------------------
2676    Name: Allen Leung
2677    Date: 2002/03/07 20:45:00 EST
2678    Tag: leunga-20020307-x86-cmov
2679    Description:
2680    
2681       Bug fixes for CMOVcc on x86.
2682    
2683       1. Added machine code generation for CMOVcc
2684       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
2685       3. CMOVcc cannot have an immediate operand as argument.
2686    
2687    ----------------------------------------------------------------------
2688    Name: Matthias Blume
2689    Date: 2002/03/07 16:15:00 EST
2690    Tag: blume-20020307-controls
2691    Description:
2692    
2693    This is a very large but mostly boring patch which makes (almost)
2694    every tuneable compiler knob (i.e., pretty much everything under
2695    Control.* plus a few other things) configurable via both the command
2696    line and environment variables in the style CM did its configuration
2697    until now.
2698    
2699    Try starting sml with '-h' (or, if you are brave, '-H')
2700    
2701    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
2702    implements the underlying generic mechanism.
2703    
2704    The interface to some of the existing such facilities has changed somewhat.
2705    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
2706    (The getFoo interface is still there for backward-compatibility, but its
2707    use is deprecated.)
2708    
2709    The ml-build script passes -Cxxx=yyy command-line arguments through so
2710    that one can now twiddle the compiler settings when using this "batch"
2711    compiler.
2712    
2713    TODO items:
2714    
2715    We should go through and throw out all controls that are no longer
2716    connected to anything.  Moreover, we should go through and provide
2717    meaningful (and correct!) documentation strings for those controls
2718    that still are connected.
2719    
2720    Currently, multiple calls to Controls.new are accepted (only the first
2721    has any effect).  Eventually we should make sure that every control
2722    is being made (via Controls.new) exactly once.  Future access can then
2723    be done using Controls.acc.
2724    
2725    Finally, it would probably be a good idea to use the getter-setter
2726    interface to controls rather than ref cells.  For the time being, both
2727    styles are provided by the Controls module, but getter-setter pairs are
2728    better if thread-safety is of any concern because they can be wrapped.
2729    
2730    *****************************************
2731    
2732    One bug fix: The function blockPlacement in three of the MLRISC
2733    backpatch files used to be hard-wired to one of two possibilities at
2734    link time (according to the value of the placementFlag).  But (I
2735    think) it should rather sense the flag every time.
2736    
2737    *****************************************
2738    
2739    Other assorted changes (by other people who did not supply a HISTORY entry):
2740    
2741    1. the cross-module inliner now works much better (Monnier)
2742    2. representation of weights, frequencies, and probabilities in MLRISC
2743       changed in preparation of using those for weighted block placement
2744       (Reppy, George)
2745    
2746    ----------------------------------------------------------------------
2747    Name: Lal George
2748    Date: 2002/03/07 14:44:24 EST 2002
2749    Tag: george-20020307-weighted-block-placement
2750    
2751    Tested the weighted block placement optimization on all architectures
2752    (except the hppa) using AMPL to generate the block and edge frequencies.
2753    Changes were required in the machine properties to correctly
2754    categorize trap instructions. There is an MLRISC flag
2755    "weighted-block-placement" that can be used to enable weighted block
2756    placement, but this will be ineffective without block/edge
2757    frequencies (coming soon).
2758    
2759    
2760    ----------------------------------------------------------------------
2761    Name: Lal George
2762    Date: 2002/03/05 17:24:48 EST
2763    Tag: george-20020305-linkage-cluster
2764    
2765    In order to support the block placement optimization, a new cluster
2766    is generated as the very first cluster (called the linkage cluster).
2767    It contains a single jump to the 'real' entry point for the compilation
2768    unit. Block placement has no effect on the linkage cluster itself, but
2769    all the other clusters  have full freedom in the manner in which they
2770    reorder blocks or functions.
2771    
2772    On the x86 the typical linkage code that is generated is:
2773       ----------------------
2774            .align 2
2775       L0:
2776            addl    $L1-L0, 72(%esp)
2777            jmp     L1
2778    
2779    
2780            .align  2
2781       L1:
2782       ----------------------
2783    
2784    72(%esp) is the memory location for the stdlink register. This
2785    must contain the address of the CPS function being called. In the
2786    above example, it contains the address of  L0; before
2787    calling L1 (the real entry point for the compilation unit), it
2788    must contain the address for L1, and hence
2789    
2790            addl $L1-L0, 72(%esp)
2791    
2792    I have tested this on all architectures except the hppa.The increase
2793    in code size is of course negligible
2794    
2795    ----------------------------------------------------------------------
2796    Name: Allen Leung
2797    Date: 2002/03/03 13:20:00 EST
2798    Tag: leunga-20020303-mlrisc-tools
2799    
2800      Added #[ ... ] expressions to mlrisc tools
2801    
2802    ----------------------------------------------------------------------
2803    Name: Matthias Blume
2804    Date: 2002/02/27 12:29:00 EST
2805    Tag: blume-20020227-cdebug
2806    Description:
2807    
2808    - made types in structure C and C_Debug to be equal
2809    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
2810    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
2811    
2812    ----------------------------------------------------------------------
2813    Name: Matthias Blume
2814    Date: 2002/02/26 12:00:00 EST
2815    Tag: blume-20020226-ffi
2816    Description:
2817    
2818    1. Fixed a minor bug in CM's "noweb" tool:
2819       If numbering is turned off, then truly don't number (i.e., do not
2820       supply the -L option to noweb).  The previous behavior was to supply
2821       -L'' -- which caused noweb to use the "default" line numbering scheme.
2822       Thanks to Chris Richards for pointing this out (and supplying the fix).
2823    
2824    2. Once again, I reworked some aspects of the FFI:
2825    
2826       A. The incomplete/complete type business:
2827    
2828       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
2829         gone!
2830       - ML types representing an incomplete type are now *equal* to
2831         ML types representing their corresponding complete types (just like
2832         in C).  This is still safe because ml-nlffigen will not generate
2833         RTTI for incomplete types, nor will it generate functions that
2834         require access to such RTTI.   But when ML code generated from both
2835         incomplete and complete versions of the C type meet, the ML types
2836         are trivially interoperable.
2837    
2838         NOTE:  These changes restore the full generality of the translation
2839         (which was previously lost when I eliminated functorization)!
2840    
2841       B. Enum types:
2842    
2843       - Structure C now has a type constructor "enum" that is similar to
2844         how the "su" constructor works.  However, "enum" is not a phantom
2845         type because each "T enum" has values (and is isomorphic to
2846         MLRep.Signed.int).
2847       - There are generic access operations for enum objects (using
2848         MLRep.Signed.int).
2849       - ml-nlffigen will generate a structure E_foo for each "enum foo".
2850         * The structure contains the definition of type "mlrep" (the ML-side
2851         representation type of the enum).  Normally, mlrep is the same
2852         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
2853         then mlrep will be defined as a datatype -- thus facilitating
2854         pattern matching on mlrep values.
2855         ("-ec" will be suppressed if there are duplicate values in an
2856          enumeration.)
2857         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
2858         will be generated for each C enum constant xxx.
2859         * Conversion functions m2i and i2m convert between mlrep and
2860         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
2861         * Coversion functions c and ml convert between mlrep and "tag enum".
2862         * Access functions (get/set) fetch and store mlrep values.
2863       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
2864         enumerations are merged into one single enumeration represented by
2865         structure E_'.
2866    
2867    ----------------------------------------------------------------------
2868    Name: Allen Leung
2869    Date: 2002/02/25 04:45:00 EST
2870    Tag: leunga-20020225-cps-spill
2871    
2872    This is a new implementation of the CPS spill phase.
2873    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
2874    In case of problems, replace it with the old file spill.sml
2875    
2876    The current compiler runs into some serious performance problems when
2877    constructing a large record.  This can happen when we try to compile a
2878    structure with many items.  Even a very simple structure like the following
2879    makes the compiler slow down.
2880    
2881        structure Foo = struct
2882           val x_1 = 0w1 : Word32.int
2883           val x_2 = 0w2 : Word32.int
2884           val x_3 = 0w3 : Word32.int
2885           ...
2886           val x_N = 0wN : Word32.int
2887        end
2888    
2889    The following table shows the compile time, from N=1000 to N=4000,
2890    with the old compiler:
2891    
2892    N
2893    1000   CPS 100 spill                           0.04u  0.00s  0.00g
2894           MLRISC ra                               0.06u  0.00s  0.05g
2895              (spills = 0 reloads = 0)
2896           TOTAL                                   0.63u  0.07s  0.21g
2897    
2898    1100   CPS 100 spill                           8.25u  0.32s  0.64g
2899           MLRISC ra                               5.68u  0.59s  3.93g
2900              (spills = 0 reloads = 0)
2901           TOTAL                                   14.71u  0.99s  4.81g
2902    
2903    1500   CPS 100 spill                           58.55u  2.34s  1.74g
2904           MLRISC ra                               5.54u  0.65s  3.91g
2905              (spills = 543 reloads = 1082)
2906           TOTAL                                   65.40u  3.13s  6.00g
2907    
2908    2000   CPS 100 spill                           126.69u  4.84s  3.08g
2909           MLRISC ra                               0.80u  0.10s  0.55g
2910              (spills = 42 reloads = 84)
2911           TOTAL                                   129.42u  5.10s  4.13g
2912    
2913    3000   CPS 100 spill                           675.59u  19.03s  11.64g
2914           MLRISC ra                               2.69u  0.27s  1.38g
2915              (spills = 62 reloads = 124)
2916           TOTAL                                   682.48u  19.61s  13.99g
2917    
2918    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
2919           MLRISC ra                               4.96u  0.27s  2.72g
2920              (spills = 85 reloads = 170)
2921           TOTAL                                   2375.26u  57.21s  48.00g
2922    
2923    As you can see the old cps spill module suffers from some serious
2924    performance problem.  But since I cannot decipher the old code fully,
2925    instead of patching the problems up, I'm reimplementing it
2926    with a different algorithm.  The new code is more modular,
2927    smaller when compiled, and substantially faster
2928    (O(n log n) time and O(n) space).  Timing of the new spill module:
2929    
2930    4000  CPS 100 spill                           0.02u  0.00s  0.00g
2931          MLRISC ra                               0.25u  0.02s  0.15g
2932             (spills=1 reloads=3)
2933          TOTAL                                   7.74u  0.34s  1.62g
2934    
2935    Implementation details:
2936    
2937    As far as I can tell, the purpose of the CPS spill module is to make sure the
2938    number of live variables at any program point (the bandwidth)
2939    does not exceed a certain limit, which is determined by the
2940    size of the spill area.
2941    
2942    When the bandwidth is too large, we decrease the register pressure by
2943    packing live variables into spill records.  How we achieve this is
2944    completely different than what we did in the old code.
2945    
2946    First, there is something about the MLRiscGen code generator
2947    that we should be aware of:
2948    
2949    o MLRiscGen performs code motion!
2950    
2951       In particular, it will move floating point computations and
2952       address computations involving only the heap pointer to
2953       their use sites (if there is only a single use).
2954       What this means is that if we have a CPS record construction
2955       statement
2956    
2957           RECORD(k,vl,w,e)
2958    
2959       we should never count the new record address w as live if w
2960       has only one use (which is often the case).
2961    
2962       We should do something similar to floating point, but the transformation
2963       there is much more complex, so I won't deal with that.
2964    
2965    Secondly, there are now two new cps primops at our disposal:
2966    
2967     1. rawrecord of record_kind option
2968        This pure operator allocates some uninitialized storage from the heap.
2969        There are two forms:
2970    
2971         rawrecord NONE [INT n]  allocates a tagless record of length n
2972         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
2973                                     and initializes the tag.
2974    
2975     2. rawupdate of cty
2976          rawupdate cty (v,i,x)
2977          Assigns to x to the ith component of record v.
2978          The storelist is not updated.
2979    
2980    We use these new primops for both spilling and increment record construction.
2981    
2982     1. Spilling.
2983    
2984        This is implemented with a linear scan algorithm (but generalized
2985        to trees).  The algorithm will create a single spill record at the
2986        beginning of the cps function and use rawupdate to spill to it,
2987        and SELECT or SELp to reload from it.  So both spills and reloads
2988        are fine-grain operations.  In contrast, in the old algorithm
2989        "spills" have to be bundled together in records.
2990    
2991        Ideally, we should sink the spill record construction to where
2992        it is needed.  We can even split the spill record into multiple ones
2993        at the places where they are needed.  But CPS is not a good
2994        representation for global code motion, so I'll keep it simple and
2995        am not attempting this.
2996    
2997     2. Incremental record construction (aka record splitting).
2998    
2999        Long records with many component values which are simulatenously live
3000        (recall that single use record addresses are not considered to
3001         be live) are constructed with rawrecord and rawupdate.
3002        We allocate space on the heap with rawrecord first, then gradually
3003        fill it in with rawupdate.  This is the technique suggested to me
3004        by Matthias.
3005    
3006        Some restrictions on when this is applicable:
3007        1. It is not a VECTOR record.  The code generator currently does not handle
3008           this case. VECTOR record uses double indirection like arrays.
3009        2. All the record component values are defined in the same "basic block"
3010           as the record constructor.  This is to prevent speculative
3011           record construction.
3012    
3013    ----------------------------------------------------------------------
3014    Name: Allen Leung
3015    Date: 2002/02/22 01:02:00 EST
3016    Tag: leunga-20020222-mlrisc-tools
3017    
3018    Minor bug fixes in the parser and rewriter
3019    
3020    ----------------------------------------------------------------------
3021    Name: Allen Leung
3022    Date: 2002/02/21 20:20:00 EST
3023    Tag: leunga-20020221-peephole
3024    
3025    Regenerated the peephole files.  Some contained typos in the specification
3026    and some didn't compile because of pretty printing bugs in the old version
3027    of 'nowhere'.
3028    
3029    ----------------------------------------------------------------------
3030    Name: Allen Leung
3031    Date: 2002/02/19 20:20:00 EST
3032    Tag: leunga-20020219-mlrisc-tools
3033    Description:
3034    
3035       Minor bug fixes to the mlrisc-tools library:
3036    
3037       1.  Fixed up parsing colon suffixed keywords
3038       2.  Added the ability to shut the error messages up
3039       3.  Reimplemented the pretty printer and fixed up/improved
3040           the pretty printing of handle and -> types.
3041       4.  Fixed up generation of literal symbols in the nowhere tool.
3042       5.  Added some SML keywords to to sml.sty
3043    
3044    ----------------------------------------------------------------------
3045    Name: Matthias Blume
3046    Date: 2002/02/19 16:20:00 EST
3047    Tag: blume-20020219-cmffi
3048    Description:
3049    
3050    A wild mix of changes, some minor, some major:
3051    
3052    * All C FFI-related libraries are now anchored under $c:
3053        $/c.cm      --> $c/c.cm
3054        $/c-int.cm  --> $c/internals/c-int.cm
3055        $/memory.cm --> $c/memory/memory.cm
3056    
3057    * "make" tool (in CM) now treats its argument pathname slightly
3058      differently:
3059        1. If the native expansion is an absolute name, then before invoking
3060           the "make" command on it, CM will apply OS.Path.mkRelative
3061           (with relativeTo = OS.FileSys.getDir()) to it.
3062        2. The argument will be passed through to subsequent phases of CM
3063           processing without "going native".  In particular, if the argument
3064           was an anchored path, then "make" will not lose track of that anchor.
3065    
3066    * Compiler backends now "know" their respective C calling conventions
3067      instead of having to be told about it by ml-nlffigen.  This relieves
3068      ml-nlffigen from one of its burdens.
3069    
3070    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
3071    
3072    * Export C_DEBUG and C_Debug from $c/c.cm.
3073    
3074    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
3075      subtyping relationship between incomplete pointers and their complete
3076      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
3077      with the convention of instantiating 'o with ('t, 'c) obj whenever
3078      the pointer target type is complete.  In the incomplete case, 'o
3079      will be instantiated with some "'c iobj" -- a type obtained by
3080      using one of the functors PointerToIncompleteType or PointerToCompleteType.
3081    
3082      Operations that work on both incomplete and complete pointer types are
3083      typed as taking an 'o ptr while operations that require the target to
3084      be known are typed as taking some ('t, 'c) obj ptr.
3085    
3086      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
3087      where void is an eqtype without any values.  This makes it possible
3088      to work on voidptr values using functions meant to operate on light
3089      incomplete pointers.
3090    
3091    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
3092      been vastly simplified.
3093    
3094    ----------------------------------------------------------------------
3095    Name: Matthias Blume
3096    Date: 2002/02/19 10:48:00 EST
3097    Tag: blume-20020219-pqfix
3098    Description:
3099    
3100    Applied Chris Okasaki's bug fix for priority queues.
3101    
3102    ----------------------------------------------------------------------
3103    Name: Matthias Blume
3104    Date: 2002/02/15 17:05:00
3105    Tag: Release_110_39
3106    Description:
3107    
3108    Last-minute retagging is becoming a tradition... :-(
3109    
3110    This is the working release 110.39.
3111    
3112    ----------------------------------------------------------------------
3113    Name: Matthias Blume
3114    Date: 2002/02/15 16:00:00 EST
3115    Tag: Release_110_39-orig
3116    Description:
3117    
3118    Working release 110.39.  New bootfiles.
3119    
3120    (Update: There was a small bug in the installer so it wouldn't work
3121    with all shells.  So I retagged. -Matthias)
3122    
3123    ----------------------------------------------------------------------
3124    Name: Matthias Blume
3125    Date: 2002/02/15 14:17:00 EST
3126    Tag: blume-20020215-showbindings
3127    Description:
3128    
3129    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
3130    the latter can be useful for exploring what bindings are available at
3131    the interactive prompt.  (The first function returns only the list
3132    of symbols that are really bound, the second prints those but also the
3133    ones that CM's autoloading mechanism knows about.)
3134    
3135    ----------------------------------------------------------------------
3136    Name: Matthias Blume
3137    Date: 2002/02/15 12:08:00 EST
3138    Tag: blume-20020215-iptrs
3139    Description:
3140    
3141    Two improvements to ml-nlffigen:
3142    
3143      1. Write files only if they do not exist or if their current contents
3144         do not coincide with what's being written.  (That is, avoid messing
3145         with the time stamps unless absolutely necessary.)
3146    
3147      2. Implement a "repository" mechanism for generated files related
3148         to "incomplete pointer types".   See the README file for details.
3149    
3150    ----------------------------------------------------------------------
3151    Name: Matthias Blume
3152    Date: 2002/02/14 11:50:00 EST
3153    Tag: blume-20020214-quote
3154    Description:
3155    
3156    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
3157    because of the new and improved tag generation scheme.  (Thanks to Allen
3158    Leung for pointing it out.)
3159    
3160    ----------------------------------------------------------------------
3161    Name: Lal George
3162    Date: 2002/02/14 09:55:27 EST 2002
3163    Tag: george-20020214-isabelle-bug
3164    Description:
3165    
3166    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
3167    of Isabelle on the x86.
3168    
3169    From Allen:
3170    -----------
3171     I've found the problem:
3172    
3173         in ra-core.sml, I use the counter "blocked" to keep track of the
3174         true number of elements in the freeze queue.  When the counter goes
3175         to zero, I skip examining the queue.  But I've messed up the
3176         bookkeeping in combine():
3177    
3178             else ();
3179             case !ucol of
3180               PSEUDO => (if !cntv > 0 then
3181                     (if !cntu > 0 then blocked := !blocked - 1 else ();
3182                                        ^^^^^^^^^^^^^^^^^^^^^^^
3183                      moveu := mergeMoveList(!movev, !moveu)
3184                     )
3185                  else ();
3186    
3187         combine() is called to coalesce two nodes u and v.
3188         I think I was thinking that if the move counts of u and v are both
3189         greater than zero then after they are coalesced then one node is
3190         removed from the freeze queue.  Apparently I was thinking that
3191         both u and v are of low degree, but that's clearly not necessarily true.
3192    
3193    
3194    02/12/2002:
3195        Here's the patch.  HOL now compiles.
3196    
3197        I don't know how this impact on performance (compile
3198        time or runtime).  This bug caused the RA (especially on the x86)
3199        to go thru the potential spill phase when there are still nodes on the
3200        freeze queue.
3201    
3202    
3203    
3204    
3205    ----------------------------------------------------------------------
3206    Name: Matthias Blume
3207    Date: 2002/02/13 22:40:00 EST
3208    Tag: blume-20020213-fptr-rtti
3209    Description:
3210    
3211    Fixed a bug in ml-nlffigen that was introduced with one of the previous
3212    updates.
3213    
3214    ----------------------------------------------------------------------
3215    Name: Matthias Blume
3216    Date: 2002/02/13 16:41:00 EST
3217    Tag: blume-20020213-cmlpq
3218    Description:
3219    
3220    Added new priority queue export symbols (which have just been added to
3221    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
3222    not compile and the installer would choke.)
3223    
3224    ----------------------------------------------------------------------
3225    Name: Matthias Blume
3226    Date: 2002/02/13 16:15:00 EST
3227    Tag: blume-20020213-various
3228    Description:
3229    
3230    1. More tweaks to ml-nlffigen:
3231    
3232       - better internal datastructures (resulting in slight speedup)
3233       - "-match" option requires exact match
3234       - "localized" gensym counters (untagged structs/unions nested within
3235         other structs/unions or within typedefs get a fresh counter; their
3236         tag will be prefixed by a concatenation of their parents' tags)
3237       - bug fixes (related to calculation of transitive closure of types
3238         to be included in the output)
3239    
3240    2. Minor Basis updates:
3241    
3242       - added implementations for List.collate and Option.app
3243    
3244    ----------------------------------------------------------------------
3245    Name: Matthias Blume
3246    Date: 2002/02/11 15:55:00 EST
3247    Tag: blume-20020211-gensym
3248    Description:
3249    
3250    Added a "-gensym" option to command line of ml-nlffigen.  This can be
3251    used to specify a "stem" -- a string that is inserted in all "gensym'd"
3252    names (ML structure names that correspond to unnamed C structs, unions,
3253    and enums), so that separate runs of ml-nlffigen do not clash.
3254    
3255    ----------------------------------------------------------------------
3256    Name: Matthias Blume
3257    Date: 2002/02/11 12:05:00 EST
3258    Tag: blume-20020211-gensml
3259    Description:
3260    
3261    A quick fix for a problem with GenSML (in the pgraph-util library):
3262    Make generation of toplevel "local" optional.  (Strictly speaking,
3263    signature definitions within "local" are not legal SML.)
3264    
3265    Other than that: updates to INSTALL and cm/TODO.
3266    
3267    ----------------------------------------------------------------------
3268    Name: Matthias Blume
3269    Date: 2002/02/08 15:00:00 EST
3270    Tag: blume-20020208-uniquepid
3271    Description:
3272    
3273    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
3274    
3275    1. The installer (config/install.sh) has gotten smarter:
3276    
3277         - Configuration options are a bit easier to specify now
3278           (in config/targets).
3279         - Bug in recognizing .tar.bz2 files fixed.
3280         - Installer automatically resolves dependencies between
3281           configuration options (e.g., if you ask for eXene, you will
3282           also get cml -- regardless whether you asked for it or not).
3283         - Installer can run in "quieter mode" by setting the environment
3284           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
3285           "completely silent", though.
3286         - Build HashCons library as part of smlnj-lib.
3287    
3288    2. A new scheme for assigning persistent identifiers to compilation
3289       units (and, by extension, to types etc.) has been put into place.
3290       This fixes a long-standing bug where types and even dynamic values
3291       can get internally confused, thereby compromising type safety
3292       (abstraction) and dynamic correctness.  See
3293    
3294         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
3295    
3296       for an example of how things could go wrong until now.
3297    
3298       The downside of the new scheme is that pids are not quite as
3299       persistent as they used to be: CM will generate a fresh pid
3300       for every compilation unit that it thinks it sees for the first
3301       time.  That means that if you compile starting from a clean, fresh
3302       source tree at two different times, you end up with different
3303       binaries.
3304    
3305       Cutoff recompilation, however, has not been compromised because
3306       CM keeps pid information in special caches between runs.
3307    
3308    ----------------------------------------------------------------------
3309    Name: Lal George
3310    Date: 2002/02/07 15:34:13 EST 2002
3311    Tag: <none>
3312    Description:
3313    
3314    Compilers that generate assembly code may produce  global labels
3315    whose value is resolved at link time. The various peephole optimization
3316    modules did not take this in account.
3317    
3318    TODO. The Labels.addrOf function should really return an option
3319    type so that clients are forced to deal with this issue, rather
3320    than an exception being raised.
3321    
3322    ----------------------------------------------------------------------
3323    Name: Lal George
3324    Date: 2002/02/06 13:55:02 EST
3325    Tag: george-20020206-ra-breakup
3326    Description:
3327    
3328    1. A bug fix from Allen.
3329    
3330        A typo causes extra fstp %st(0)'s to be generated at compensation
3331        edges, which might cause stack underflow traps at runtime.  This
3332        occurs in fft where there are extraneous fstps right before the 'into'
3333        trap instruction (in this case they are harmless since none of the
3334        integers overflow.)
3335    
3336    2. Pulled out various utility modules that were embedded in the modules
3337       of the register allocator. I need these modules for other purposes, but
3338       they are not complete enough to put into a library (just yet).
3339    ----------------------------------------------------------------------
3340    Name: Matthias Blume
3341    Date: 2002/01/31 16:05:00 EST
3342    Tag: blume-20020131-sparc-ccalls
3343    Description:
3344    
3345    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
3346       of extra stack space by mistake.  Fixed.
3347    
3348    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
3349    
3350    ----------------------------------------------------------------------
3351    Name: Allen Leung
3352    Date: 2002/01/30
3353    Tag: leunga-20020130-nowhere-bug-fix
3354    Description:
3355    
3356       MLRISC bug fixes:
3357       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
3358       2. MachineInt.fromString was negating its value.
3359    
3360    ----------------------------------------------------------------------
3361    Name: Matthias Blume
3362    Date: 2002/01/29
3363    Tag: blume-20020129-INSTALL
3364    Description:
3365    
3366    - Added somewhat detailed installation instructions (file INSTALL).
3367    - Fixed curl-detection bug in config/install.sh.
3368    - It is now possible to select the URL getter using the URLGETTER
3369      environment variable:
3370    
3371          not set / "unknown"      --> automatic detection (script tries wget,
3372                                       curl, and lynx)
3373          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
3374                                       how to properly invoke them)
3375          other                    --> use $URLGETTER directly, it must take
3376                                       precisely two command-line arguments
3377                                       (source URL and destination file name)
3378    
3379    ----------------------------------------------------------------------
3380    Name: Matthias Blume
3381    Date: 2002/01/28
3382    Tag: blume-20020128-sparc-ccalls
3383    Description:
3384    
3385    - Fixed problem with calculation of "used" registers in sparc-c-calls.
3386    - Make use of the allocParam argument in sparc-c-calls.
3387    
3388    ----------------------------------------------------------------------
3389    Name: Matthias Blume
3390    Date: 2002/01/28
3391    Tag: blume-20020128-allocParam
3392    Description:
3393    
3394    John Reppy:  Changes c-calls API to accept client-callback for
3395    allocating extra stack space.
3396    me: Corresponding changes to mlriscGen (using a dummy argument that
3397        does not change the current behavior).
3398    
3399    ----------------------------------------------------------------------
3400    Name: Matthias Blume
3401    Date: 2002/01/28 12:00:00
3402    Tag: Release_110_38
3403    Description:
3404    
3405    This time for real!!!
3406    
3407    ----------------------------------------------------------------------
3408    Name: Matthias Blume
3409    Date: 2002/01/28 10:56:00 EST
3410    Tag: blume-20020128-retraction
3411    Description:
3412    
3413    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
3414       with blume-Release_110_38-retracted.)
3415    
3416    1. Fixed a problem with incorrect rounding modes in real64.sml.
3417       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
3418    
3419    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
3420       and enums fixed.  The naming of corresponding ML identifiers should
3421       now be consistent again.
3422    
3423    ----------------------------------------------------------------------
3424    Name: Allen Leung
3425    Date: 2002/01/27
3426    Tag: leunga-20020127-nowhere
3427    Description:
3428    
3429       Added a target called nowhere in the configuration scripts.
3430       Enabling this will build the MLRISC 'nowhere' tool (for translating
3431       programs with where-clauses into legal SML code) during installation.
3432    
3433    ----------------------------------------------------------------------
3434    Name: Matthias Blume
3435    Date: 2002/01/25 21:27:00 EST
3436    Tag: blume-Release_110_38-retracted
3437    Description:
3438    
3439    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
3440    
3441    README will be added later.
3442    
3443    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
3444    (Release_110_38) removed.  Reason: Last-minute bug fixes.
3445    
3446    ----------------------------------------------------------------------
3447    Name: Matthias Blume
3448    Date: 2002/01/25
3449    Tag: blume-20020125-ffi
3450    Description:
3451    
3452    A large number of tweaks and improvements to ml-nlffi-lib and
3453    ml-nlffigen:
3454    
3455       - ML represenation types have been streamlined
3456       - getter and setter functions work with concrete values, not abstract
3457         ones where possible
3458       - ml-nlffigen command line more flexible (see README file there)
3459       - some bugs have been fixed (hopefully)
3460    
3461    ----------------------------------------------------------------------
3462    Name: Lal George
3463    Date: 2002/01/24
3464    Tag: george-20020124-risc-ra-interface
3465    Description:
3466    
3467       There is a dramatic simplification in the interface to the
3468       register allocator for RISC architectures as a result of making
3469       parallel copy instructions explicit.
3470    
3471    ----------------------------------------------------------------------
3472    Name: Matthias Blume
3473    Date: 2002/01/22
3474    Tag: blume-20020122-x86-ccalls
3475    Description:
3476    
3477    Bug fix for c-calls on x86 (having to do with how char- and
3478    short-arguments are being handled).
3479    
3480    ----------------------------------------------------------------------
3481    Name: Matthias Blume
3482    Date: 2002/01/21
3483    Tag: blume-20020121-ff
3484    Description:
3485    
3486    Another day of fiddling with the FFI...
3487    
3488    1. Bug fix/workaround:  CKIT does not complain about negative array
3489       dimensions, so ml-nlffigen has to guard itself against this possibility.
3490       (Otherwise a negative dimension would send it into an infinite loop.)
3491    
3492    2. Some of the abstract types (light objects, light pointers, most "base"
3493       types) in structure C are now eqtypes.
3494    
3495    3. Added constructors and test functions for NULL function pointers.
3496    
3497    ----------------------------------------------------------------------
3498    Name: Matthias Blume
3499    Date: 2002/01/18
3500    Tag: blume-20020118-ready-for-new-release
3501    Description:
3502    
3503    Made config/srcarchiveurl point to a new place.  (Will provide boot
3504    files shortly.)
3505    
3506    Maybe we christen this to be 110.38?
3507    
3508    ----------------------------------------------------------------------
3509    Name: Matthias Blume
3510    Date: 2002/01/18
3511    Tag: blume-20020118-more-ffifiddle
3512    Description:
3513    
3514    Today's FFI fiddling:
3515    
3516      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
3517        C.Set.* that use concrete (MLRep.*) arguments and results instead
3518        of abstract ones.
3519    
3520      - Provided word-style bit operations etc. for "int" representation
3521        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
3522        and Long.
3523    
3524    ----------------------------------------------------------------------
3525    Name: Matthias Blume
3526    Date: 2002/01/18
3527    Tag: blume-20020118-use-x86-fp
3528    Description:
3529    
3530    Now that x86-fast-fp seems to be working, I turned it back on again
3531    by default.  (Seems to work fine now, even with the FFI.)
3532    
3533    Other than that, I added some documentation about the FFI to
3534    src/ml-nlffigen/README and updated the FFI test examples in
3535    src/ml-nlffi-lib/Tests/*.
3536    
3537    ----------------------------------------------------------------------
3538    Name: Allen Leung
3539    Date: 2002/01/17
3540    Tag: leunga-20020117-x86-fast-fp-call
3541    Description:
3542    
3543       1. Fixed a problem with handling return fp values when x86's fast fp
3544          mode is turned on.
3545    
3546       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
3547          of %f32.
3548    
3549       3. Added a constructor INT32lit to the ast of MLRISC tools.
3550    
3551    ----------------------------------------------------------------------
3552    Name: Matthias Blume
3553    Date: 2002/01/16
3554    Tag: blume-20020116-ffifiddle
3555    Description:
3556    
3557    More fiddling with the FFI interface:
3558    
3559     - Make constness 'c instead of rw wherever possible.  This eliminates
3560       the need for certain explicit coercions.  (However, due to ML's
3561       value polymorphism, there will still be many cases where explicit
3562       coercions are necessary.  Phantom types are not the whole answer
3563       to modeling a subtyping relationship in ML.)
3564    
3565     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
3566    
3567     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
3568       clumsy workaround for varargs problem.)
3569    
3570    ----------------------------------------------------------------------
3571    Name: Lal George
3572    Date: 2002/01/15
3573    Tag: <none>
3574    Description:
3575    
3576    1. Since COPY instructions are no longer native to the architecture,
3577       a generic functor can be used to implement the expandCopies function.
3578    
3579    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
3580       TEXT segment.
3581    
3582    ----------------------------------------------------------------------
3583    Name: Matthias Blume
3584    Date: 2002/01/15
3585    Tag: blume-20020115-ffiupdates
3586    Description:
3587    
3588    1. Fix for bug resulting in single-precision float values being returned
3589       incorrectly from FFI calls.
3590    
3591    2. Small modifications to C FFI API:
3592    
3593        - memory-allocation routines return straight objects (no options)
3594          and raise an exception in out-of-memory situations
3595        - unsafe extensions to cast between function pointers and pointers
3596          from/to ints
3597        - added structure C_Debug as an alternative to structure C where
3598          pointer-dereferencing (|*| and |*!) always check for null-pointers
3599        - added open_lib' to DynLinkage;  open_lib' works like open_lib
3600          but also takes a (possibly empty) list of existing library handles
3601          that the current library depends on
3602    
3603    ----------------------------------------------------------------------
3604    Name: Matthias Blume
3605    Date: 2002/01/10
3606    Tag: blume-20020110-newffigen
3607    Description:
3608    
3609    1. Updates to portable graph code.
3610    
3611    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
3612       more scalable now so that even huge interfaces such as the one
3613       for GTK compile in finite time and space. :-)
3614       See src/ml-nlffigen/README for details on what's new.
3615    
3616    ----------------------------------------------------------------------
3617    Name: Lal George
3618    Date: 2001/01/09 14:31:35 EST 2002
3619    Tag: george-20011206-rm-native-copy
3620    Description:
3621    
3622            Removed the native COPY and FCOPY instructions
3623            from all the architectures and replaced it with the
3624            explicit COPY instruction from the previous commit.
3625    
3626            It is now possible to simplify many of the optimizations
3627            modules that manipulate copies. This has not been
3628            done in this change.
3629    
3630    ----------------------------------------------------------------------
3631    Name: Lal George
3632    Date: 2001/12/06 16:50:13 EST 2001
3633    Tag: george-20011206-mlrisc-instruction
3634    Description:
3635    
3636    Changed the representation of instructions from being fully abstract
3637    to being partially concrete. That is to say:
3638    
3639      from
3640            type instruction
3641    
3642      to
3643            type instr                              (* machine instruction *)
3644    
3645            datatype instruction =
3646                LIVE of {regs: C.cellset, spilled: C.cellset}
3647              | KILL of {regs: C.cellset, spilled: C.cellset}
3648              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
3649              | ANNOTATION of {i: instruction, a: Annotations.annotation}
3650              | INSTR of instr
3651    
3652    This makes the handling of certain special instructions that appear on
3653    all architectures easier and uniform.
3654    
3655    LIVE and KILL say that a list of registers are live or killed at the
3656    program point where they appear. No spill code is generated when an
3657    element of the 'regs' field is spilled, but the register is moved to
3658    the 'spilled' (which is present, more for debugging than anything else).
3659    
3660    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
3661    We used to generate:
3662    
3663            DEFFREG f1
3664            f1 := f2 + f3
3665            trapb
3666    
3667    but now generate:
3668    
3669            f1 := f2 + f3
3670            trapb
3671            LIVE {regs=[f1,f2,f3], spilled=[]}
3672    
3673    Furthermore, the DEFFREG (hack) required that all floating point instruction
3674    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
3675    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
3676    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
3677    this architecture).
3678    
3679    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
3680    all the architectures. This will result in further simplification of the
3681    register allocator that must be aware of them for coalescing purposes, and
3682    will also simplify certain aspects of the machine description that provides
3683    callbacks related to parallel copies.
3684    
3685    ANNOTATION should be obvious, and now INSTR represents the honest to God
3686    machine instruction set!
3687    
3688    The <arch>/instructions/<arch>Instr.sml files define certain utility
3689    functions for making porting easier -- essentially converting upper case
3690    to lower case. All machine instructions (of type instr) are in upper case,
3691    and the lower case form generates an MLRISC instruction. For example on
3692    the alpha we have:
3693    
3694      datatype instr =
3695         LDA of {r:cell, b:cell, d:operand}
3696       | ...
3697    
3698      val lda : {r:cell, b:cell, d:operand} -> instruction
3699        ...
3700    
3701    where lda is just (INSTR o LDA), etc.
3702    
3703    ----------------------------------------------------------------------
3704    Name: Matthias Blume
3705    Date: 2001/11/22 21:40:00 EST
3706    Tag: Release_110_37
3707    Description:
3708    
3709    Release 110.37.  This time for real.
3710    
3711    ----------------------------------------------------------------------
3712    Name: Matthias Blume
3713    Date: 2001/11/21 16:35:00 EST
3714    Tag: blume-20011121-foot-in-mouth
3715    Description:
3716    
3717    Removed the "Release_110_37" tag because of a serious bug.
3718    This will be re-tagged once the bug is fixed.
3719    
3720    ----------------------------------------------------------------------
3721    Name: Matthias Blume
3722    Date: 2001/11/21 16:14:00 EST
3723    Tag: blume-20011121-forgottenfile
3724    Description:
3725    
3726    Forgot to add a file.  (Just a .tex-file -- part of
3727    the CM manual source.)
3728    
3729    ----------------------------------------------------------------------
3730    Name: Matthias Blume
3731    Date: 2001/11/21 16:10:00 EST
3732    Tag: blume-20011121-invalid_110_37
3733    Description:
3734    
3735    Note: I removed the original tag "Release_110_37" from this commit
3736          because we found a serious bug in all non-x86 backends.
3737          - Matthias
3738    
3739    1. Modifications to the SML/NJ code generator and to the runtime system
3740       so that code object name strings are directly inserted into code
3741       objects at code generation time.  The only business the runtime system
3742       has with this is now to read the name strings on occasions.
3743       (The encoding of the name string has also changed somewhat.)
3744    
3745    2. CM now implements a simple "set calculus" for specifying export lists.
3746       In particular, it is now possible to refer to the export lists of
3747       other libraries/groups/sources and form unions as well as differences.
3748       See the latest CM manual for details.
3749    
3750    3. An separate notion of "proxy" libraries has again be eliminated from
3751       CM's model.  (Proxy libraries are now simply a special case of using
3752       the export list calculus.)
3753    
3754    4. Some of the existing libraries now take advantage of the new set
3755       calculus.
3756       (Notice that not all libraries have been converted because some
3757       of the existing .cm-files are supposed to be backward compatible
3758       with 110.0.x.)
3759    
3760    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
3761       "General.exnMessage"!  The former relies on a certain hook to be
3762       initialized, and that often does not happen in the stand-alone case.)
3763    
3764    ----------------------------------------------------------------------
3765    Name: Lal George
3766    Date: 2001/11/21  13:56:18 EST
3767    Tag: george-2001121-pseudo-ops
3768    Description:
3769    
3770      Implemented a complete redesign of MLRISC pseudo-ops. Now there
3771      ought to never be any question of incompatabilities with
3772      pseudo-op syntax expected by host assemblers.
3773    
3774      For now, only modules supporting GAS syntax are implemented
3775      but more should follow, such as MASM, and vendor assembler
3776      syntax, e.g. IBM as, Sun as, etc.
3777    
3778    ----------------------------------------------------------------------
3779    Name: Matthias Blume
3780    Date: 2001/11/14 11:52:00 EST
3781    Tag: blume-20011114-srcname
3782    Description:
3783    
3784    1. Routed the name of the current source file to mlriscgen where it
3785       should be directly emitted into the code object.  (This last part
3786       is yet to be done.)
3787    
3788    2. Some cleanup of the pgraph code to make it match the proposal that
3789       I put out the other day.  (The proposal notwithstanding, things are
3790       still in flux here.)
3791    
3792    ----------------------------------------------------------------------
3793    Name: Lal George
3794    Date: 2001/11/14 09:44:04 EST
3795    Tag:
3796    Description:
3797    
3798      Fix for a backpatching bug reported by Allen.
3799    
3800      Because the boundary between short and long span-dependent
3801      instructions is +/- 128, there are an astounding number of
3802      span-dependent instructions whose size is over estimated.
3803    
3804      Allen came up with the idea of letting the size of span
3805      dependent instructions be non-monotonic, for a maxIter
3806      number of times, after which the size must be monotonically
3807      increasing.
3808    
3809      This table shows the number of span-dependent instructions
3810      whose size was over-estimated as a function of maxIter, for the
3811      file Parse/parse/ml.grm.sml:
3812    
3813         maxIter            # of instructions:
3814            10                      687
3815            20                      438
3816            30                      198
3817            40                        0
3818    
3819      In compiling the compiler, there is no significant difference in
3820      compilation speed between maxIter=10 and maxIter=40. Actually,
3821      my measurements showed that maxIter=40 was a tad faster than
3822      maxIter=10! Also 96% of the  files in the compiler reach a fix
3823      point within 13 iterations, so fixing maxIter at 40, while high,
3824      is okay.
3825    
3826    ----------------------------------------------------------------------
3827    Name: Matthias Blume
3828    Date: 2001/10/31 15:25:00 EST
3829    Tag: blume-20011031-pgraph
3830    Description:
3831    
3832    CKIT:
3833    * Changed the "Function" constructor of type Ast.ctype to carry optional
3834      argument identifiers.
3835    * Changed the return type of TypeUtil.getFunction accordingly.
3836    * Type equality ignores the argument names.
3837    * TypeUtil.composite tries to preserve argument names but gives up quickly
3838      if there is a mismatch.
3839    
3840    installation script:
3841    * attempts to use "curl" if available (unless "wget" is available as well)
3842    
3843    CM:
3844    * has an experimental implementation of "portable graphs" which I will
3845      soon propose as an implementation-independent library format
3846    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
3847    
3848    NLFFI-LIB:
3849    * some cleanup (all cosmetic)
3850    
3851    NLFFIGEN:
3852    * temporarily disabled the mechanism that suppresses ML output for
3853      C definitions whose identifiers start with an underscore character
3854    * generate val bindings for enum constants
3855    * user can request that only one style (light or heavy) is being used;
3856      default is to use both (command-line arguments: -heavy and -light)
3857    * fixed bug in handling of function types involving incomplete pointers
3858    * generate ML entry points that take record arguments (i.e., using
3859      named arguments) for C functions that have a prototype with named
3860      arguments
3861      (see changes to CKIT)
3862    
3863    ----------------------------------------------------------------------
3864    Name: Allen Leung
3865    Date: 2001/10/27 20:34:00 EDT
3866    Tag: leunga-20011027-x86-fast-fp-call
3867    Description:
3868    
3869       Fixed the bug described in blume-20010920-slowfp.
3870    
3871       The fix involves
3872          1. generating FCOPYs in FSTP in ia32-svid
3873          2. marking a CALL with the appropriate annotation
3874    
3875    ----------------------------------------------------------------------
3876    Name: Matthias Blume
3877    Date: 2001/10/16 11:32:00 EDT
3878    Tag: blume-20011016-netbsd
3879    Description:
3880    
3881    Underscore patch from Chris Richards (fixing problem with compiling
3882    runtime system under recent NetBSD).
3883    
3884    ----------------------------------------------------------------------
3885    Name: Allen Leung
3886    Date: 2001/10/12 17:18:32 EDT 2001
3887    Tag: leung-20011012-x86-printflowgraph
3888    Description:
3889    
3890    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
3891    
3892    ----------------------------------------------------------------------
3893    Name: Lal George
3894    Date: 2001/10/11 23:51:34 EDT
3895    Tag: george-20011011-too-many-instrs
3896    Description:
3897    
3898    The representation of a program point never expected to see more
3899    than 65536 instructions in a basic block!
3900    
3901    ----------------------------------------------------------------------
3902    Name: Lal George
3903    Date: 2001/10/09 09:41:37 EDT
3904    Tag: george-20011008-mlrisc-labels
3905    Description:
3906    
3907    Changed the machine description files to support printing of
3908    local and global labels in assembly code, based on host assembler
3909    conventions.
3910    
3911    ----------------------------------------------------------------------
3912    Name: Matthias Blume
3913    Date: 2001/09/25 15:25:00 EDT
3914    Tag: blume-20010925-exninfo
3915    Description:
3916    
3917    I provided a non-hook implementation of exnName (at the toplevel) and
3918    made the "dummy" implementation of exnMessage (at the toplevel) more
3919    useful: if nothing gets "hooked in", then at least you are going to
3920    see the exception name and a message indicating why you don't see more.
3921    
3922    [For the time being, programs that need exnMessage and want to use
3923    ml-build should either use General.exnMessage (strongly recommended) or
3924    refer to structure General at some other point so that CM sees a
3925    static dependency.]
3926    
3927    [Similar remarks go for "print" and "use":  If you want to use their
3928    functionality in stand-alone programs generated by ml-build, then use
3929    TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
3930    
3931    ----------------------------------------------------------------------
3932    Name: Matthias Blume
3933    Date: 2001/09/20 17:28:00 EDT
3934    Tag: blume-20010920-slowfp
3935    Description:
3936    
3937    Allen says that x86-fast-fp is not safe yet, so I turned it off again...
3938    
3939    ----------------------------------------------------------------------
3940    Name: Matthias Blume
3941    Date: 2001/09/20 17:20:00 EDT
3942    Tag: blume-20010920-canonicalpaths
3943    Description:
3944    
3945    0. Updated the BOOT file (something that I forgot to do earlier).
3946    
3947    1. Small internal change to CM so that it avoids "/../" in filenames
3948       as much as possible (but only where it is safe).
3949    
3950    2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
3951       that arguments that contain delimiters are passed through correctly.
3952       This change also means that all "special" arguments of the form
3953       @SMLxxx... must come first.
3954    
3955    3. Changed install script to put relative anchor names for tool commands
3956       into pathconfig.
3957    
3958    ----------------------------------------------------------------------
3959    Name: Matthias Blume
3960    Date: 2001/09/18 15:35:00 EDT
3961    Tag: blume-20010918-readme11036
3962    Description:
3963    
3964    Added README files.
3965    
3966    ----------------------------------------------------------------------
3967    Name: Matthias Blume
3968    Date: 2001/09/18 11:45:00 EDT
3969    Tag: Release_110_36 (retag)
3970    Description:
3971    
3972    Fixed mistake in config/preloads. Retagged as 110.36.
3973    
3974    ----------------------------------------------------------------------
3975    Name: Matthias Blume
3976    Date: 2001/09/18 09:40:00 EDT
3977    Tag: Release_110_36_orig (tag changed)
3978    Description:
3979    
3980    New version (110.36).  New bootfiles.
3981    
3982    ----------------------------------------------------------------------
3983    Name: Matthias Blume
3984    Date: 2001/09/14 16:15:00 EDT
3985    Tag: blume-20010914-x86fastfp
3986    Description:
3987    
3988    John committed some changes that Allen made, in particular a (hopefully)
3989    correctly working version of the x86-fp module.
3990    
3991    I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
3992    flag to "true".  Everything seems to compile to a fixpoint ok, and
3993    "mandelbrot" speeds up by about 15%.
3994    
3995    ----------------------------------------------------------------------
3996    Name: Matthias Blume
3997    Date: 2001/09/13 11:20:00 EDT
3998    Tag: blume-20010913-minimal
3999    Description:
4000    
4001    1. Stefan Monnier's patch to fix a miscompilation problem that
4002       was brought to light by John Reppy's work on Moby.
4003    
4004    2. Implemented a minimal "structure Compiler" that contains just
4005       "version" and "architecture".  The minimal version will be
4006       available when the full version is not.  This is for backward-
4007       compatibility with code that wants to test Compiler.version.
4008    
4009    ----------------------------------------------------------------------
4010    Name: Matthias Blume
4011    Date: 2001/08/28 14:03:00 EDT
4012    Tag: blume-20010828-ml-lex
4013    Description:
4014    
4015    Fix for bug 1581, received from Neophytos Michael.
4016    
4017    ----------------------------------------------------------------------
4018    Name: Matthias Blume
4019    Date: 2001/08/27 11:20:00 EDT
4020    Tag: blume-20010827-readme11035
4021    Description:
4022    
4023    Fleshed out the README file for 110.35.
4024    
4025    ----------------------------------------------------------------------
4026    Name: Matthias Blume
4027    Date: 2001/08/24 17:10:00 EDT
4028    Tag: Release_110_35
4029    Description:
4030    
4031    New version number (110.35).  New bootfiles.
4032    
4033    ----------------------------------------------------------------------
4034    Name: Lal George
4035    Date: 2001/08/24 13:47:18 EDT 2001
4036    Tag: george-20010824-MLRISC-graphs
4037    Description:
4038    
4039     removed clusters from MLRISC completely and replaced with graphs.
4040    
4041    ----------------------------------------------------------------------
4042    Name: Matthias Blume
4043    Date: 2001/08/23 17:50:00 EDT
4044    Tag: blume-20010823-toplevel
4045    Description:
4046    
4047    - some reorganization of the code that implements various kinds of
4048      environments in the compiler (static, dynamic, symbolic, combined)
4049    - re-implemented the EnvRef module so that evalStream works properly
4050      (if the stream contains references to "use", "CM.make", etc.)
4051    - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
4052    
4053    ----------------------------------------------------------------------
4054    Name: Matthias Blume
4055    Date: 2001/08/20 15:50 EDT
4056    Tag: blume20010820-slipup
4057    Description:
4058    
4059    I forgot to commit a few files.  Here they are...
4060    
4061    ----------------------------------------------------------------------
4062    Name: Matthias Blume
4063    Date: 2001/08/20 15:35:00 EDT
4064    Tag: blume-20010820-debugprof
4065    Description:
4066    
4067    !!!! NEW BOOTFILES !!!!
4068    
4069    This is another round of reorganizing the compiler sources.  This
4070    time the main goal was to factor out all the "instrumentation"
4071    passes (for profiling and backtracing) into their own library.
4072    The difficulty was to do it in such a way that it does not depend
4073    on elaborate.cm but only on elabdata.cm.
4074    
4075    Therefore there have been further changes to both elaborate.cm and
4076    elabdata.cm -- more "generic" things have been moved from the former
4077    to the latter.  As a result, I was forced to split the assignment
4078    of numbers indicating "primtyc"s into two portions: SML-generic and
4079    SML/NJ-specific.  Since it would have been awkward to maintain,
4080    I bit the bullet and actually _changed_ the mapping between these
4081    numbers and primtycs.  The bottom line of this is that you need
4082    a new set of bin- and bootfiles.
4083    
4084    I have built new bootfiles for all architectures, so doing a fresh
4085    checkout and config/install.sh should be all you need.
4086    
4087    The newly created library's name is
4088    
4089        $smlnj/viscomp/debugprof.cm
4090    
4091    and its sources live under
4092    
4093        src/compiler/DebugProf
4094