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

SCM Repository

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

Diff of /sml/trunk/HISTORY

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

revision 916, Fri Aug 24 21:13:58 2001 UTC revision 1045, Mon Jan 28 22:13:13 2002 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  Name: Matthias Blume
17    Date: 2002/01/28
18    Tag: blume-20020128-sparc-ccalls
19    Description:
20    
21    - Fixed problem with calculation of "used" registers in sparc-c-calls.
22    - Make use of the allocParam argument in sparc-c-calls.
23    
24    ----------------------------------------------------------------------
25    Name: Matthias Blume
26    Date: 2002/01/28
27    Tag: blume-20020128-allocParam
28    Description:
29    
30    John Reppy:  Changes c-calls API to accept client-callback for
31    allocating extra stack space.
32    me: Corresponding changes to mlriscGen (using a dummy argument that
33        does not change the current behavior).
34    
35    ----------------------------------------------------------------------
36    Name: Matthias Blume
37    Date: 2002/01/28 12:00:00
38    Tag: Release_110_38
39    Description:
40    
41    This time for real!!!
42    
43    ----------------------------------------------------------------------
44    Name: Matthias Blume
45    Date: 2002/01/28 10:56:00 EST
46    Tag: blume-20020128-retraction
47    Description:
48    
49    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
50       with blume-Release_110_38-retracted.)
51    
52    1. Fixed a problem with incorrect rounding modes in real64.sml.
53       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
54    
55    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
56       and enums fixed.  The naming of corresponding ML identifiers should
57       now be consistent again.
58    
59    ----------------------------------------------------------------------
60    Name: Allen Leung
61    Date: 2002/01/27
62    Tag: leunga-20020127-nowhere
63    Description:
64    
65       Added a target called nowhere in the configuration scripts.
66       Enabling this will build the MLRISC 'nowhere' tool (for translating
67       programs with where-clauses into legal SML code) during installation.
68    
69    ----------------------------------------------------------------------
70    Name: Matthias Blume
71    Date: 2002/01/25 21:27:00 EST
72    Tag: blume-Release_110_38-retracted
73    Description:
74    
75    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
76    
77    README will be added later.
78    
79    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
80    (Release_110_38) removed.  Reason: Last-minute bug fixes.
81    
82    ----------------------------------------------------------------------
83    Name: Matthias Blume
84    Date: 2002/01/25
85    Tag: blume-20020125-ffi
86    Description:
87    
88    A large number of tweaks and improvements to ml-nlffi-lib and
89    ml-nlffigen:
90    
91       - ML represenation types have been streamlined
92       - getter and setter functions work with concrete values, not abstract
93         ones where possible
94       - ml-nlffigen command line more flexible (see README file there)
95       - some bugs have been fixed (hopefully)
96    
97    ----------------------------------------------------------------------
98    Name: Lal George
99    Date: 2002/01/24
100    Tag: george-20020124-risc-ra-interface
101    Description:
102    
103       There is a dramatic simplification in the interface to the
104       register allocator for RISC architectures as a result of making
105       parallel copy instructions explicit.
106    
107    ----------------------------------------------------------------------
108    Name: Matthias Blume
109    Date: 2002/01/22
110    Tag: blume-20020122-x86-ccalls
111    Description:
112    
113    Bug fix for c-calls on x86 (having to do with how char- and
114    short-arguments are being handled).
115    
116    ----------------------------------------------------------------------
117    Name: Matthias Blume
118    Date: 2002/01/21
119    Tag: blume-20020121-ff
120    Description:
121    
122    Another day of fiddling with the FFI...
123    
124    1. Bug fix/workaround:  CKIT does not complain about negative array
125       dimensions, so ml-nlffigen has to guard itself against this possibility.
126       (Otherwise a negative dimension would send it into an infinite loop.)
127    
128    2. Some of the abstract types (light objects, light pointers, most "base"
129       types) in structure C are now eqtypes.
130    
131    3. Added constructors and test functions for NULL function pointers.
132    
133    ----------------------------------------------------------------------
134    Name: Matthias Blume
135    Date: 2002/01/18
136    Tag: blume-20020118-ready-for-new-release
137    Description:
138    
139    Made config/srcarchiveurl point to a new place.  (Will provide boot
140    files shortly.)
141    
142    Maybe we christen this to be 110.38?
143    
144    ----------------------------------------------------------------------
145    Name: Matthias Blume
146    Date: 2002/01/18
147    Tag: blume-20020118-more-ffifiddle
148    Description:
149    
150    Today's FFI fiddling:
151    
152      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
153        C.Set.* that use concrete (MLRep.*) arguments and results instead
154        of abstract ones.
155    
156      - Provided word-style bit operations etc. for "int" representation
157        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
158        and Long.
159    
160    ----------------------------------------------------------------------
161    Name: Matthias Blume
162    Date: 2002/01/18
163    Tag: blume-20020118-use-x86-fp
164    Description:
165    
166    Now that x86-fast-fp seems to be working, I turned it back on again
167    by default.  (Seems to work fine now, even with the FFI.)
168    
169    Other than that, I added some documentation about the FFI to
170    src/ml-nlffigen/README and updated the FFI test examples in
171    src/ml-nlffi-lib/Tests/*.
172    
173    ----------------------------------------------------------------------
174    Name: Allen Leung
175    Date: 2002/01/17
176    Tag: leunga-20020117-x86-fast-fp-call
177    Description:
178    
179       1. Fixed a problem with handling return fp values when x86's fast fp
180          mode is turned on.
181    
182       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
183          of %f32.
184    
185       3. Added a constructor INT32lit to the ast of MLRISC tools.
186    
187    ----------------------------------------------------------------------
188    Name: Matthias Blume
189    Date: 2002/01/16
190    Tag: blume-20020116-ffifiddle
191    Description:
192    
193    More fiddling with the FFI interface:
194    
195     - Make constness 'c instead of rw wherever possible.  This eliminates
196       the need for certain explicit coercions.  (However, due to ML's
197       value polymorphism, there will still be many cases where explicit
198       coercions are necessary.  Phantom types are not the whole answer
199       to modeling a subtyping relationship in ML.)
200    
201     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
202    
203     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
204       clumsy workaround for varargs problem.)
205    
206    ----------------------------------------------------------------------
207    Name: Lal George
208    Date: 2002/01/15
209    Tag: <none>
210    Description:
211    
212    1. Since COPY instructions are no longer native to the architecture,
213       a generic functor can be used to implement the expandCopies function.
214    
215    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
216       TEXT segment.
217    
218    ----------------------------------------------------------------------
219    Name: Matthias Blume
220    Date: 2002/01/15
221    Tag: blume-20020115-ffiupdates
222    Description:
223    
224    1. Fix for bug resulting in single-precision float values being returned
225       incorrectly from FFI calls.
226    
227    2. Small modifications to C FFI API:
228    
229        - memory-allocation routines return straight objects (no options)
230          and raise an exception in out-of-memory situations
231        - unsafe extensions to cast between function pointers and pointers
232          from/to ints
233        - added structure C_Debug as an alternative to structure C where
234          pointer-dereferencing (|*| and |*!) always check for null-pointers
235        - added open_lib' to DynLinkage;  open_lib' works like open_lib
236          but also takes a (possibly empty) list of existing library handles
237          that the current library depends on
238    
239    ----------------------------------------------------------------------
240    Name: Matthias Blume
241    Date: 2002/01/10
242    Tag: blume-20020110-newffigen
243    Description:
244    
245    1. Updates to portable graph code.
246    
247    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
248       more scalable now so that even huge interfaces such as the one
249       for GTK compile in finite time and space. :-)
250       See src/ml-nlffigen/README for details on what's new.
251    
252    ----------------------------------------------------------------------
253    Name: Lal George
254    Date: 2001/01/09 14:31:35 EST 2002
255    Tag: george-20011206-rm-native-copy
256    Description:
257    
258            Removed the native COPY and FCOPY instructions
259            from all the architectures and replaced it with the
260            explicit COPY instruction from the previous commit.
261    
262            It is now possible to simplify many of the optimizations
263            modules that manipulate copies. This has not been
264            done in this change.
265    
266    ----------------------------------------------------------------------
267    Name: Lal George
268    Date: 2001/12/06 16:50:13 EST 2001
269    Tag: george-20011206-mlrisc-instruction
270    Description:
271    
272    Changed the representation of instructions from being fully abstract
273    to being partially concrete. That is to say:
274    
275      from
276            type instruction
277    
278      to
279            type instr                              (* machine instruction *)
280    
281            datatype instruction =
282                LIVE of {regs: C.cellset, spilled: C.cellset}
283              | KILL of {regs: C.cellset, spilled: C.cellset}
284              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
285              | ANNOTATION of {i: instruction, a: Annotations.annotation}
286              | INSTR of instr
287    
288    This makes the handling of certain special instructions that appear on
289    all architectures easier and uniform.
290    
291    LIVE and KILL say that a list of registers are live or killed at the
292    program point where they appear. No spill code is generated when an
293    element of the 'regs' field is spilled, but the register is moved to
294    the 'spilled' (which is present, more for debugging than anything else).
295    
296    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
297    We used to generate:
298    
299            DEFFREG f1
300            f1 := f2 + f3
301            trapb
302    
303    but now generate:
304    
305            f1 := f2 + f3
306            trapb
307            LIVE {regs=[f1,f2,f3], spilled=[]}
308    
309    Furthermore, the DEFFREG (hack) required that all floating point instruction
310    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
311    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
312    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
313    this architecture).
314    
315    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
316    all the architectures. This will result in further simplification of the
317    register allocator that must be aware of them for coalescing purposes, and
318    will also simplify certain aspects of the machine description that provides
319    callbacks related to parallel copies.
320    
321    ANNOTATION should be obvious, and now INSTR represents the honest to God
322    machine instruction set!
323    
324    The <arch>/instructions/<arch>Instr.sml files define certain utility
325    functions for making porting easier -- essentially converting upper case
326    to lower case. All machine instructions (of type instr) are in upper case,
327    and the lower case form generates an MLRISC instruction. For example on
328    the alpha we have:
329    
330      datatype instr =
331         LDA of {r:cell, b:cell, d:operand}
332       | ...
333    
334      val lda : {r:cell, b:cell, d:operand} -> instruction
335        ...
336    
337    where lda is just (INSTR o LDA), etc.
338    
339    ----------------------------------------------------------------------
340    Name: Matthias Blume
341    Date: 2001/11/22 21:40:00 EST
342    Tag: Release_110_37
343    Description:
344    
345    Release 110.37.  This time for real.
346    
347    ----------------------------------------------------------------------
348    Name: Matthias Blume
349    Date: 2001/11/21 16:35:00 EST
350    Tag: blume-20011121-foot-in-mouth
351    Description:
352    
353    Removed the "Release_110_37" tag because of a serious bug.
354    This will be re-tagged once the bug is fixed.
355    
356    ----------------------------------------------------------------------
357    Name: Matthias Blume
358    Date: 2001/11/21 16:14:00 EST
359    Tag: blume-20011121-forgottenfile
360    Description:
361    
362    Forgot to add a file.  (Just a .tex-file -- part of
363    the CM manual source.)
364    
365    ----------------------------------------------------------------------
366    Name: Matthias Blume
367    Date: 2001/11/21 16:10:00 EST
368    Tag: blume-20011121-invalid_110_37
369    Description:
370    
371    Note: I removed the original tag "Release_110_37" from this commit
372          because we found a serious bug in all non-x86 backends.
373          - Matthias
374    
375    1. Modifications to the SML/NJ code generator and to the runtime system
376       so that code object name strings are directly inserted into code
377       objects at code generation time.  The only business the runtime system
378       has with this is now to read the name strings on occasions.
379       (The encoding of the name string has also changed somewhat.)
380    
381    2. CM now implements a simple "set calculus" for specifying export lists.
382       In particular, it is now possible to refer to the export lists of
383       other libraries/groups/sources and form unions as well as differences.
384       See the latest CM manual for details.
385    
386    3. An separate notion of "proxy" libraries has again be eliminated from
387       CM's model.  (Proxy libraries are now simply a special case of using
388       the export list calculus.)
389    
390    4. Some of the existing libraries now take advantage of the new set
391       calculus.
392       (Notice that not all libraries have been converted because some
393       of the existing .cm-files are supposed to be backward compatible
394       with 110.0.x.)
395    
396    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
397       "General.exnMessage"!  The former relies on a certain hook to be
398       initialized, and that often does not happen in the stand-alone case.)
399    
400    ----------------------------------------------------------------------
401    Name: Lal George
402    Date: 2001/11/21  13:56:18 EST
403    Tag: george-2001121-pseudo-ops
404    Description:
405    
406      Implemented a complete redesign of MLRISC pseudo-ops. Now there
407      ought to never be any question of incompatabilities with
408      pseudo-op syntax expected by host assemblers.
409    
410      For now, only modules supporting GAS syntax are implemented
411      but more should follow, such as MASM, and vendor assembler
412      syntax, e.g. IBM as, Sun as, etc.
413    
414    ----------------------------------------------------------------------
415    Name: Matthias Blume
416    Date: 2001/11/14 11:52:00 EST
417    Tag: blume-20011114-srcname
418    Description:
419    
420    1. Routed the name of the current source file to mlriscgen where it
421       should be directly emitted into the code object.  (This last part
422       is yet to be done.)
423    
424    2. Some cleanup of the pgraph code to make it match the proposal that
425       I put out the other day.  (The proposal notwithstanding, things are
426       still in flux here.)
427    
428    ----------------------------------------------------------------------
429    Name: Lal George
430    Date: 2001/11/14 09:44:04 EST
431    Tag:
432    Description:
433    
434      Fix for a backpatching bug reported by Allen.
435    
436      Because the boundary between short and long span-dependent
437      instructions is +/- 128, there are an astounding number of
438      span-dependent instructions whose size is over estimated.
439    
440      Allen came up with the idea of letting the size of span
441      dependent instructions be non-monotonic, for a maxIter
442      number of times, after which the size must be monotonically
443      increasing.
444    
445      This table shows the number of span-dependent instructions
446      whose size was over-estimated as a function of maxIter, for the
447      file Parse/parse/ml.grm.sml:
448    
449         maxIter            # of instructions:
450            10                      687
451            20                      438
452            30                      198
453            40                        0
454    
455      In compiling the compiler, there is no significant difference in
456      compilation speed between maxIter=10 and maxIter=40. Actually,
457      my measurements showed that maxIter=40 was a tad faster than
458      maxIter=10! Also 96% of the  files in the compiler reach a fix
459      point within 13 iterations, so fixing maxIter at 40, while high,
460      is okay.
461    
462    ----------------------------------------------------------------------
463    Name: Matthias Blume
464    Date: 2001/10/31 15:25:00 EST
465    Tag: blume-20011031-pgraph
466    Description:
467    
468    CKIT:
469    * Changed the "Function" constructor of type Ast.ctype to carry optional
470      argument identifiers.
471    * Changed the return type of TypeUtil.getFunction accordingly.
472    * Type equality ignores the argument names.
473    * TypeUtil.composite tries to preserve argument names but gives up quickly
474      if there is a mismatch.
475    
476    installation script:
477    * attempts to use "curl" if available (unless "wget" is available as well)
478    
479    CM:
480    * has an experimental implementation of "portable graphs" which I will
481      soon propose as an implementation-independent library format
482    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
483    
484    NLFFI-LIB:
485    * some cleanup (all cosmetic)
486    
487    NLFFIGEN:
488    * temporarily disabled the mechanism that suppresses ML output for
489      C definitions whose identifiers start with an underscore character
490    * generate val bindings for enum constants
491    * user can request that only one style (light or heavy) is being used;
492      default is to use both (command-line arguments: -heavy and -light)
493    * fixed bug in handling of function types involving incomplete pointers
494    * generate ML entry points that take record arguments (i.e., using
495      named arguments) for C functions that have a prototype with named
496      arguments
497      (see changes to CKIT)
498    
499    ----------------------------------------------------------------------
500    Name: Allen Leung
501    Date: 2001/10/27 20:34:00 EDT
502    Tag: leunga-20011027-x86-fast-fp-call
503    Description:
504    
505       Fixed the bug described in blume-20010920-slowfp.
506    
507       The fix involves
508          1. generating FCOPYs in FSTP in ia32-svid
509          2. marking a CALL with the appropriate annotation
510    
511    ----------------------------------------------------------------------
512    Name: Matthias Blume
513    Date: 2001/10/16 11:32:00 EDT
514    Tag: blume-20011016-netbsd
515    Description:
516    
517    Underscore patch from Chris Richards (fixing problem with compiling
518    runtime system under recent NetBSD).
519    
520    ----------------------------------------------------------------------
521    Name: Allen Leung
522    Date: 2001/10/12 17:18:32 EDT 2001
523    Tag: leung-20011012-x86-printflowgraph
524    Description:
525    
526    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
527    
528    ----------------------------------------------------------------------
529    Name: Lal George
530    Date: 2001/10/11 23:51:34 EDT
531    Tag: george-20011011-too-many-instrs
532    Description:
533    
534    The representation of a program point never expected to see more
535    than 65536 instructions in a basic block!
536    
537    ----------------------------------------------------------------------
538    Name: Lal George
539    Date: 2001/10/09 09:41:37 EDT
540    Tag: george-20011008-mlrisc-labels
541    Description:
542    
543    Changed the machine description files to support printing of
544    local and global labels in assembly code, based on host assembler
545    conventions.
546    
547    ----------------------------------------------------------------------
548    Name: Matthias Blume
549    Date: 2001/09/25 15:25:00 EDT
550    Tag: blume-20010925-exninfo
551    Description:
552    
553    I provided a non-hook implementation of exnName (at the toplevel) and
554    made the "dummy" implementation of exnMessage (at the toplevel) more
555    useful: if nothing gets "hooked in", then at least you are going to
556    see the exception name and a message indicating why you don't see more.
557    
558    [For the time being, programs that need exnMessage and want to use
559    ml-build should either use General.exnMessage (strongly recommended) or
560    refer to structure General at some other point so that CM sees a
561    static dependency.]
562    
563    [Similar remarks go for "print" and "use":  If you want to use their
564    functionality in stand-alone programs generated by ml-build, then use
565    TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
566    
567    ----------------------------------------------------------------------
568    Name: Matthias Blume
569    Date: 2001/09/20 17:28:00 EDT
570    Tag: blume-20010920-slowfp
571    Description:
572    
573    Allen says that x86-fast-fp is not safe yet, so I turned it off again...
574    
575    ----------------------------------------------------------------------
576    Name: Matthias Blume
577    Date: 2001/09/20 17:20:00 EDT
578    Tag: blume-20010920-canonicalpaths
579    Description:
580    
581    0. Updated the BOOT file (something that I forgot to do earlier).
582    
583    1. Small internal change to CM so that it avoids "/../" in filenames
584       as much as possible (but only where it is safe).
585    
586    2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
587       that arguments that contain delimiters are passed through correctly.
588       This change also means that all "special" arguments of the form
589       @SMLxxx... must come first.
590    
591    3. Changed install script to put relative anchor names for tool commands
592       into pathconfig.
593    
594    ----------------------------------------------------------------------
595    Name: Matthias Blume
596    >>>>>>> 1.169
597    Date: 2001/09/18 15:35:00 EDT
598    Tag: blume-20010918-readme11036
599    Description:
600    
601    Added README files.
602    
603    ----------------------------------------------------------------------
604    Name: Matthias Blume
605    Date: 2001/09/18 11:45:00 EDT
606    Tag: Release_110_36 (retag)
607    Description:
608    
609    Fixed mistake in config/preloads. Retagged as 110.36.
610    
611    ----------------------------------------------------------------------
612    Name: Matthias Blume
613    Date: 2001/09/18 09:40:00 EDT
614    Tag: Release_110_36_orig (tag changed)
615    Description:
616    
617    New version (110.36).  New bootfiles.
618    
619    ----------------------------------------------------------------------
620    Name: Matthias Blume
621    Date: 2001/09/14 16:15:00 EDT
622    Tag: blume-20010914-x86fastfp
623    Description:
624    
625    John committed some changes that Allen made, in particular a (hopefully)
626    correctly working version of the x86-fp module.
627    
628    I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
629    flag to "true".  Everything seems to compile to a fixpoint ok, and
630    "mandelbrot" speeds up by about 15%.
631    
632    ----------------------------------------------------------------------
633    Name: Matthias Blume
634    Date: 2001/09/13 11:20:00 EDT
635    Tag: blume-20010913-minimal
636    Description:
637    
638    1. Stefan Monnier's patch to fix a miscompilation problem that
639       was brought to light by John Reppy's work on Moby.
640    
641    2. Implemented a minimal "structure Compiler" that contains just
642       "version" and "architecture".  The minimal version will be
643       available when the full version is not.  This is for backward-
644       compatibility with code that wants to test Compiler.version.
645    
646    ----------------------------------------------------------------------
647    Name: Matthias Blume
648    Date: 2001/08/28 14:03:00 EDT
649    Tag: blume-20010828-ml-lex
650    Description:
651    
652    Fix for bug 1581, received from Neophytos Michael.
653    
654    ----------------------------------------------------------------------
655    Name: Matthias Blume
656    Date: 2001/08/27 11:20:00 EDT
657    Tag: blume-20010827-readme11035
658    Description:
659    
660    Fleshed out the README file for 110.35.
661    
662    ----------------------------------------------------------------------
663    Name: Matthias Blume
664  Date: 2001/08/24 17:10:00 EDT  Date: 2001/08/24 17:10:00 EDT
665  Tag: Release_110_35  Tag: Release_110_35
666  Description:  Description:
# Line 2923  Line 3570 
3570       elaborator). There were a lot of changes during my "linkpath" trials       elaborator). There were a lot of changes during my "linkpath" trials
3571       that could have been reverted to their original state but weren't.       that could have been reverted to their original state but weren't.
3572       Please, don't be too harsh on me for messing with this code a bit more       Please, don't be too harsh on me for messing with this code a bit more
3573       than what was strictly necessary...  (I _did_ resist the tempation       than what was strictly necessary...  (I _did_ resist the temptation
3574       of doing any "global reformatting" to avoid an untimely death at       of doing any "global reformatting" to avoid an untimely death at
3575       Dave's hands. :)       Dave's hands. :)
3576    

Legend:
Removed from v.916  
changed lines
  Added in v.1045

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