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 796, Tue Mar 6 00:04:33 2001 UTC revision 1067, Fri Feb 15 17:08:17 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
17    Date: 2002/02/15 12:08:00 EST
18    Tag: blume-20020215-iptrs
19    Description:
20    
21    Two improvements to ml-nlffigen:
22    
23      1. Write files only if they do not exist or if their current contents
24         do not coincide with what's being written.  (That is, avoid messing
25         with the time stamps unless absolutely necessary.)
26    
27      2. Implement a "repository" mechanism for generated files related
28         to "incomplete pointer types".   See the README file for details.
29    
30    ----------------------------------------------------------------------
31    Name: Matthias Blume
32    Date: 2002/02/14 11:50:00 EST
33    Tag: blume-20020214-quote
34    Description:
35    
36    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
37    because of the new and improved tag generation scheme.  (Thanks to Allen
38    Leung for pointing it out.)
39    
40    ----------------------------------------------------------------------
41    Name: Lal George
42    Date: 2002/02/14 09:55:27 EST 2002
43    Tag: george-20020214-isabelle-bug
44    Description:
45    
46    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
47    of Isabelle on the x86.
48    
49    From Allen:
50    -----------
51     I've found the problem:
52    
53         in ra-core.sml, I use the counter "blocked" to keep track of the
54         true number of elements in the freeze queue.  When the counter goes
55         to zero, I skip examining the queue.  But I've messed up the
56         bookkeeping in combine():
57    
58             else ();
59             case !ucol of
60               PSEUDO => (if !cntv > 0 then
61                     (if !cntu > 0 then blocked := !blocked - 1 else ();
62                                        ^^^^^^^^^^^^^^^^^^^^^^^
63                      moveu := mergeMoveList(!movev, !moveu)
64                     )
65                  else ();
66    
67         combine() is called to coalesce two nodes u and v.
68         I think I was thinking that if the move counts of u and v are both
69         greater than zero then after they are coalesced then one node is
70         removed from the freeze queue.  Apparently I was thinking that
71         both u and v are of low degree, but that's clearly not necessarily true.
72    
73    
74    02/12/2002:
75        Here's the patch.  HOL now compiles.
76    
77        I don't know how this impact on performance (compile
78        time or runtime).  This bug caused the RA (especially on the x86)
79        to go thru the potential spill phase when there are still nodes on the
80        freeze queue.
81    
82    
83    
84    
85    ----------------------------------------------------------------------
86    Name: Matthias Blume
87    Date: 2002/02/13 22:40:00 EST
88    Tag: blume-20020213-fptr-rtti
89    Description:
90    
91    Fixed a bug in ml-nlffigen that was introduced with one of the previous
92    updates.
93    
94    ----------------------------------------------------------------------
95    Name: Matthias Blume
96    Date: 2002/02/13 16:41:00 EST
97    Tag: blume-20020213-cmlpq
98    Description:
99    
100    Added new priority queue export symbols (which have just been added to
101    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
102    not compile and the installer would choke.)
103    
104    ----------------------------------------------------------------------
105    Name: Matthias Blume
106    Date: 2002/02/13 16:15:00 EST
107    Tag: blume-20020213-various
108    Description:
109    
110    1. More tweaks to ml-nlffigen:
111    
112       - better internal datastructures (resulting in slight speedup)
113       - "-match" option requires exact match
114       - "localized" gensym counters (untagged structs/unions nested within
115         other structs/unions or within typedefs get a fresh counter; their
116         tag will be prefixed by a concatenation of their parents' tags)
117       - bug fixes (related to calculation of transitive closure of types
118         to be included in the output)
119    
120    2. Minor Basis updates:
121    
122       - added implementations for List.collate and Option.app
123    
124    ----------------------------------------------------------------------
125    Name: Matthias Blume
126    Date: 2002/02/11 15:55:00 EST
127    Tag: blume-20020211-gensym
128    Description:
129    
130    Added a "-gensym" option to command line of ml-nlffigen.  This can be
131    used to specify a "stem" -- a string that is inserted in all "gensym'd"
132    names (ML structure names that correspond to unnamed C structs, unions,
133    and enums), so that separate runs of ml-nlffigen do not clash.
134    
135    ----------------------------------------------------------------------
136    Name: Matthias Blume
137    Date: 2002/02/11 12:05:00 EST
138    Tag: blume-20020211-gensml
139    Description:
140    
141    A quick fix for a problem with GenSML (in the pgraph-util library):
142    Make generation of toplevel "local" optional.  (Strictly speaking,
143    signature definitions within "local" are not legal SML.)
144    
145    Other than that: updates to INSTALL and cm/TODO.
146    
147    ----------------------------------------------------------------------
148    Name: Matthias Blume
149    Date: 2002/02/08 15:00:00 EST
150    Tag: blume-20020208-uniquepid
151    Description:
152    
153    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
154    
155    1. The installer (config/install.sh) has gotten smarter:
156    
157         - Configuration options are a bit easier to specify now
158           (in config/targets).
159         - Bug in recognizing .tar.bz2 files fixed.
160         - Installer automatically resolves dependencies between
161           configuration options (e.g., if you ask for eXene, you will
162           also get cml -- regardless whether you asked for it or not).
163         - Installer can run in "quieter mode" by setting the environment
164           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
165           "completely silent", though.
166         - Build HashCons library as part of smlnj-lib.
167    
168    2. A new scheme for assigning persistent identifiers to compilation
169       units (and, by extension, to types etc.) has been put into place.
170       This fixes a long-standing bug where types and even dynamic values
171       can get internally confused, thereby compromising type safety
172       (abstraction) and dynamic correctness.  See
173    
174         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
175    
176       for an example of how things could go wrong until now.
177    
178       The downside of the new scheme is that pids are not quite as
179       persistent as they used to be: CM will generate a fresh pid
180       for every compilation unit that it thinks it sees for the first
181       time.  That means that if you compile starting from a clean, fresh
182       source tree at two different times, you end up with different
183       binaries.
184    
185       Cutoff recompilation, however, has not been compromised because
186       CM keeps pid information in special caches between runs.
187    
188    ----------------------------------------------------------------------
189    Name: Lal George
190    Date: 2002/02/07 15:34:13 EST 2002
191    Tag: <none>
192    Description:
193    
194    Compilers that generate assembly code may produce  global labels
195    whose value is resolved at link time. The various peephole optimization
196    modules did not take this in account.
197    
198    TODO. The Labels.addrOf function should really return an option
199    type so that clients are forced to deal with this issue, rather
200    than an exception being raised.
201    
202    ----------------------------------------------------------------------
203    Name: Lal George
204    Date: 2002/02/06 13:55:02 EST
205    Tag: george-20020206-ra-breakup
206    Description:
207    
208    1. A bug fix from Allen.
209    
210        A typo causes extra fstp %st(0)'s to be generated at compensation
211        edges, which might cause stack underflow traps at runtime.  This
212        occurs in fft where there are extraneous fstps right before the 'into'
213        trap instruction (in this case they are harmless since none of the
214        integers overflow.)
215    
216    2. Pulled out various utility modules that were embedded in the modules
217       of the register allocator. I need these modules for other purposes, but
218       they are not complete enough to put into a library (just yet).
219    ----------------------------------------------------------------------
220    Name: Matthias Blume
221    Date: 2002/01/31 16:05:00 EST
222    Tag: blume-20020131-sparc-ccalls
223    Description:
224    
225    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
226       of extra stack space by mistake.  Fixed.
227    
228    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
229    
230    ----------------------------------------------------------------------
231    Name: Allen Leung
232    Date: 2002/01/30
233    Tag: leunga-20020130-nowhere-bug-fix
234    Description:
235    
236       MLRISC bug fixes:
237       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
238       2. MachineInt.fromString was negating its value.
239    
240    ----------------------------------------------------------------------
241    Name: Matthias Blume
242    Date: 2002/01/29
243    Tag: blume-20020129-INSTALL
244    Description:
245    
246    - Added somewhat detailed installation instructions (file INSTALL).
247    - Fixed curl-detection bug in config/install.sh.
248    - It is now possible to select the URL getter using the URLGETTER
249      environment variable:
250    
251          not set / "unknown"      --> automatic detection (script tries wget,
252                                       curl, and lynx)
253          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
254                                       how to properly invoke them)
255          other                    --> use $URLGETTER directly, it must take
256                                       precisely two command-line arguments
257                                       (source URL and destination file name)
258    
259    ----------------------------------------------------------------------
260    Name: Matthias Blume
261    Date: 2002/01/28
262    Tag: blume-20020128-sparc-ccalls
263    Description:
264    
265    - Fixed problem with calculation of "used" registers in sparc-c-calls.
266    - Make use of the allocParam argument in sparc-c-calls.
267    
268    ----------------------------------------------------------------------
269    Name: Matthias Blume
270    Date: 2002/01/28
271    Tag: blume-20020128-allocParam
272    Description:
273    
274    John Reppy:  Changes c-calls API to accept client-callback for
275    allocating extra stack space.
276    me: Corresponding changes to mlriscGen (using a dummy argument that
277        does not change the current behavior).
278    
279    ----------------------------------------------------------------------
280    Name: Matthias Blume
281    Date: 2002/01/28 12:00:00
282    Tag: Release_110_38
283    Description:
284    
285    This time for real!!!
286    
287    ----------------------------------------------------------------------
288    Name: Matthias Blume
289    Date: 2002/01/28 10:56:00 EST
290    Tag: blume-20020128-retraction
291    Description:
292    
293    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
294       with blume-Release_110_38-retracted.)
295    
296    1. Fixed a problem with incorrect rounding modes in real64.sml.
297       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
298    
299    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
300       and enums fixed.  The naming of corresponding ML identifiers should
301       now be consistent again.
302    
303    ----------------------------------------------------------------------
304    Name: Allen Leung
305    Date: 2002/01/27
306    Tag: leunga-20020127-nowhere
307    Description:
308    
309       Added a target called nowhere in the configuration scripts.
310       Enabling this will build the MLRISC 'nowhere' tool (for translating
311       programs with where-clauses into legal SML code) during installation.
312    
313    ----------------------------------------------------------------------
314    Name: Matthias Blume
315    Date: 2002/01/25 21:27:00 EST
316    Tag: blume-Release_110_38-retracted
317    Description:
318    
319    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
320    
321    README will be added later.
322    
323    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
324    (Release_110_38) removed.  Reason: Last-minute bug fixes.
325    
326    ----------------------------------------------------------------------
327    Name: Matthias Blume
328    Date: 2002/01/25
329    Tag: blume-20020125-ffi
330    Description:
331    
332    A large number of tweaks and improvements to ml-nlffi-lib and
333    ml-nlffigen:
334    
335       - ML represenation types have been streamlined
336       - getter and setter functions work with concrete values, not abstract
337         ones where possible
338       - ml-nlffigen command line more flexible (see README file there)
339       - some bugs have been fixed (hopefully)
340    
341    ----------------------------------------------------------------------
342    Name: Lal George
343    Date: 2002/01/24
344    Tag: george-20020124-risc-ra-interface
345    Description:
346    
347       There is a dramatic simplification in the interface to the
348       register allocator for RISC architectures as a result of making
349       parallel copy instructions explicit.
350    
351    ----------------------------------------------------------------------
352    Name: Matthias Blume
353    Date: 2002/01/22
354    Tag: blume-20020122-x86-ccalls
355    Description:
356    
357    Bug fix for c-calls on x86 (having to do with how char- and
358    short-arguments are being handled).
359    
360    ----------------------------------------------------------------------
361    Name: Matthias Blume
362    Date: 2002/01/21
363    Tag: blume-20020121-ff
364    Description:
365    
366    Another day of fiddling with the FFI...
367    
368    1. Bug fix/workaround:  CKIT does not complain about negative array
369       dimensions, so ml-nlffigen has to guard itself against this possibility.
370       (Otherwise a negative dimension would send it into an infinite loop.)
371    
372    2. Some of the abstract types (light objects, light pointers, most "base"
373       types) in structure C are now eqtypes.
374    
375    3. Added constructors and test functions for NULL function pointers.
376    
377    ----------------------------------------------------------------------
378    Name: Matthias Blume
379    Date: 2002/01/18
380    Tag: blume-20020118-ready-for-new-release
381    Description:
382    
383    Made config/srcarchiveurl point to a new place.  (Will provide boot
384    files shortly.)
385    
386    Maybe we christen this to be 110.38?
387    
388    ----------------------------------------------------------------------
389    Name: Matthias Blume
390    Date: 2002/01/18
391    Tag: blume-20020118-more-ffifiddle
392    Description:
393    
394    Today's FFI fiddling:
395    
396      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
397        C.Set.* that use concrete (MLRep.*) arguments and results instead
398        of abstract ones.
399    
400      - Provided word-style bit operations etc. for "int" representation
401        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
402        and Long.
403    
404    ----------------------------------------------------------------------
405    Name: Matthias Blume
406    Date: 2002/01/18
407    Tag: blume-20020118-use-x86-fp
408    Description:
409    
410    Now that x86-fast-fp seems to be working, I turned it back on again
411    by default.  (Seems to work fine now, even with the FFI.)
412    
413    Other than that, I added some documentation about the FFI to
414    src/ml-nlffigen/README and updated the FFI test examples in
415    src/ml-nlffi-lib/Tests/*.
416    
417    ----------------------------------------------------------------------
418    Name: Allen Leung
419    Date: 2002/01/17
420    Tag: leunga-20020117-x86-fast-fp-call
421    Description:
422    
423       1. Fixed a problem with handling return fp values when x86's fast fp
424          mode is turned on.
425    
426       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
427          of %f32.
428    
429       3. Added a constructor INT32lit to the ast of MLRISC tools.
430    
431    ----------------------------------------------------------------------
432    Name: Matthias Blume
433    Date: 2002/01/16
434    Tag: blume-20020116-ffifiddle
435    Description:
436    
437    More fiddling with the FFI interface:
438    
439     - Make constness 'c instead of rw wherever possible.  This eliminates
440       the need for certain explicit coercions.  (However, due to ML's
441       value polymorphism, there will still be many cases where explicit
442       coercions are necessary.  Phantom types are not the whole answer
443       to modeling a subtyping relationship in ML.)
444    
445     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
446    
447     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
448       clumsy workaround for varargs problem.)
449    
450    ----------------------------------------------------------------------
451    Name: Lal George
452    Date: 2002/01/15
453    Tag: <none>
454    Description:
455    
456    1. Since COPY instructions are no longer native to the architecture,
457       a generic functor can be used to implement the expandCopies function.
458    
459    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
460       TEXT segment.
461    
462    ----------------------------------------------------------------------
463    Name: Matthias Blume
464    Date: 2002/01/15
465    Tag: blume-20020115-ffiupdates
466    Description:
467    
468    1. Fix for bug resulting in single-precision float values being returned
469       incorrectly from FFI calls.
470    
471    2. Small modifications to C FFI API:
472    
473        - memory-allocation routines return straight objects (no options)
474          and raise an exception in out-of-memory situations
475        - unsafe extensions to cast between function pointers and pointers
476          from/to ints
477        - added structure C_Debug as an alternative to structure C where
478          pointer-dereferencing (|*| and |*!) always check for null-pointers
479        - added open_lib' to DynLinkage;  open_lib' works like open_lib
480          but also takes a (possibly empty) list of existing library handles
481          that the current library depends on
482    
483    ----------------------------------------------------------------------
484    Name: Matthias Blume
485    Date: 2002/01/10
486    Tag: blume-20020110-newffigen
487    Description:
488    
489    1. Updates to portable graph code.
490    
491    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
492       more scalable now so that even huge interfaces such as the one
493       for GTK compile in finite time and space. :-)
494       See src/ml-nlffigen/README for details on what's new.
495    
496    ----------------------------------------------------------------------
497    Name: Lal George
498    Date: 2001/01/09 14:31:35 EST 2002
499    Tag: george-20011206-rm-native-copy
500    Description:
501    
502            Removed the native COPY and FCOPY instructions
503            from all the architectures and replaced it with the
504            explicit COPY instruction from the previous commit.
505    
506            It is now possible to simplify many of the optimizations
507            modules that manipulate copies. This has not been
508            done in this change.
509    
510    ----------------------------------------------------------------------
511    Name: Lal George
512    Date: 2001/12/06 16:50:13 EST 2001
513    Tag: george-20011206-mlrisc-instruction
514    Description:
515    
516    Changed the representation of instructions from being fully abstract
517    to being partially concrete. That is to say:
518    
519      from
520            type instruction
521    
522      to
523            type instr                              (* machine instruction *)
524    
525            datatype instruction =
526                LIVE of {regs: C.cellset, spilled: C.cellset}
527              | KILL of {regs: C.cellset, spilled: C.cellset}
528              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
529              | ANNOTATION of {i: instruction, a: Annotations.annotation}
530              | INSTR of instr
531    
532    This makes the handling of certain special instructions that appear on
533    all architectures easier and uniform.
534    
535    LIVE and KILL say that a list of registers are live or killed at the
536    program point where they appear. No spill code is generated when an
537    element of the 'regs' field is spilled, but the register is moved to
538    the 'spilled' (which is present, more for debugging than anything else).
539    
540    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
541    We used to generate:
542    
543            DEFFREG f1
544            f1 := f2 + f3
545            trapb
546    
547    but now generate:
548    
549            f1 := f2 + f3
550            trapb
551            LIVE {regs=[f1,f2,f3], spilled=[]}
552    
553    Furthermore, the DEFFREG (hack) required that all floating point instruction
554    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
555    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
556    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
557    this architecture).
558    
559    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
560    all the architectures. This will result in further simplification of the
561    register allocator that must be aware of them for coalescing purposes, and
562    will also simplify certain aspects of the machine description that provides
563    callbacks related to parallel copies.
564    
565    ANNOTATION should be obvious, and now INSTR represents the honest to God
566    machine instruction set!
567    
568    The <arch>/instructions/<arch>Instr.sml files define certain utility
569    functions for making porting easier -- essentially converting upper case
570    to lower case. All machine instructions (of type instr) are in upper case,
571    and the lower case form generates an MLRISC instruction. For example on
572    the alpha we have:
573    
574      datatype instr =
575         LDA of {r:cell, b:cell, d:operand}
576       | ...
577    
578      val lda : {r:cell, b:cell, d:operand} -> instruction
579        ...
580    
581    where lda is just (INSTR o LDA), etc.
582    
583    ----------------------------------------------------------------------
584    Name: Matthias Blume
585    Date: 2001/11/22 21:40:00 EST
586    Tag: Release_110_37
587    Description:
588    
589    Release 110.37.  This time for real.
590    
591    ----------------------------------------------------------------------
592    Name: Matthias Blume
593    Date: 2001/11/21 16:35:00 EST
594    Tag: blume-20011121-foot-in-mouth
595    Description:
596    
597    Removed the "Release_110_37" tag because of a serious bug.
598    This will be re-tagged once the bug is fixed.
599    
600    ----------------------------------------------------------------------
601    Name: Matthias Blume
602    Date: 2001/11/21 16:14:00 EST
603    Tag: blume-20011121-forgottenfile
604    Description:
605    
606    Forgot to add a file.  (Just a .tex-file -- part of
607    the CM manual source.)
608    
609    ----------------------------------------------------------------------
610    Name: Matthias Blume
611    Date: 2001/11/21 16:10:00 EST
612    Tag: blume-20011121-invalid_110_37
613    Description:
614    
615    Note: I removed the original tag "Release_110_37" from this commit
616          because we found a serious bug in all non-x86 backends.
617          - Matthias
618    
619    1. Modifications to the SML/NJ code generator and to the runtime system
620       so that code object name strings are directly inserted into code
621       objects at code generation time.  The only business the runtime system
622       has with this is now to read the name strings on occasions.
623       (The encoding of the name string has also changed somewhat.)
624    
625    2. CM now implements a simple "set calculus" for specifying export lists.
626       In particular, it is now possible to refer to the export lists of
627       other libraries/groups/sources and form unions as well as differences.
628       See the latest CM manual for details.
629    
630    3. An separate notion of "proxy" libraries has again be eliminated from
631       CM's model.  (Proxy libraries are now simply a special case of using
632       the export list calculus.)
633    
634    4. Some of the existing libraries now take advantage of the new set
635       calculus.
636       (Notice that not all libraries have been converted because some
637       of the existing .cm-files are supposed to be backward compatible
638       with 110.0.x.)
639    
640    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
641       "General.exnMessage"!  The former relies on a certain hook to be
642       initialized, and that often does not happen in the stand-alone case.)
643    
644    ----------------------------------------------------------------------
645    Name: Lal George
646    Date: 2001/11/21  13:56:18 EST
647    Tag: george-2001121-pseudo-ops
648    Description:
649    
650      Implemented a complete redesign of MLRISC pseudo-ops. Now there
651      ought to never be any question of incompatabilities with
652      pseudo-op syntax expected by host assemblers.
653    
654      For now, only modules supporting GAS syntax are implemented
655      but more should follow, such as MASM, and vendor assembler
656      syntax, e.g. IBM as, Sun as, etc.
657    
658    ----------------------------------------------------------------------
659    Name: Matthias Blume
660    Date: 2001/11/14 11:52:00 EST
661    Tag: blume-20011114-srcname
662    Description:
663    
664    1. Routed the name of the current source file to mlriscgen where it
665       should be directly emitted into the code object.  (This last part
666       is yet to be done.)
667    
668    2. Some cleanup of the pgraph code to make it match the proposal that
669       I put out the other day.  (The proposal notwithstanding, things are
670       still in flux here.)
671    
672    ----------------------------------------------------------------------
673    Name: Lal George
674    Date: 2001/11/14 09:44:04 EST
675    Tag:
676    Description:
677    
678      Fix for a backpatching bug reported by Allen.
679    
680      Because the boundary between short and long span-dependent
681      instructions is +/- 128, there are an astounding number of
682      span-dependent instructions whose size is over estimated.
683    
684      Allen came up with the idea of letting the size of span
685      dependent instructions be non-monotonic, for a maxIter
686      number of times, after which the size must be monotonically
687      increasing.
688    
689      This table shows the number of span-dependent instructions
690      whose size was over-estimated as a function of maxIter, for the
691      file Parse/parse/ml.grm.sml:
692    
693         maxIter            # of instructions:
694            10                      687
695            20                      438
696            30                      198
697            40                        0
698    
699      In compiling the compiler, there is no significant difference in
700      compilation speed between maxIter=10 and maxIter=40. Actually,
701      my measurements showed that maxIter=40 was a tad faster than
702      maxIter=10! Also 96% of the  files in the compiler reach a fix
703      point within 13 iterations, so fixing maxIter at 40, while high,
704      is okay.
705    
706    ----------------------------------------------------------------------
707    Name: Matthias Blume
708    Date: 2001/10/31 15:25:00 EST
709    Tag: blume-20011031-pgraph
710    Description:
711    
712    CKIT:
713    * Changed the "Function" constructor of type Ast.ctype to carry optional
714      argument identifiers.
715    * Changed the return type of TypeUtil.getFunction accordingly.
716    * Type equality ignores the argument names.
717    * TypeUtil.composite tries to preserve argument names but gives up quickly
718      if there is a mismatch.
719    
720    installation script:
721    * attempts to use "curl" if available (unless "wget" is available as well)
722    
723    CM:
724    * has an experimental implementation of "portable graphs" which I will
725      soon propose as an implementation-independent library format
726    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
727    
728    NLFFI-LIB:
729    * some cleanup (all cosmetic)
730    
731    NLFFIGEN:
732    * temporarily disabled the mechanism that suppresses ML output for
733      C definitions whose identifiers start with an underscore character
734    * generate val bindings for enum constants
735    * user can request that only one style (light or heavy) is being used;
736      default is to use both (command-line arguments: -heavy and -light)
737    * fixed bug in handling of function types involving incomplete pointers
738    * generate ML entry points that take record arguments (i.e., using
739      named arguments) for C functions that have a prototype with named
740      arguments
741      (see changes to CKIT)
742    
743    ----------------------------------------------------------------------
744    Name: Allen Leung
745    Date: 2001/10/27 20:34:00 EDT
746    Tag: leunga-20011027-x86-fast-fp-call
747    Description:
748    
749       Fixed the bug described in blume-20010920-slowfp.
750    
751       The fix involves
752          1. generating FCOPYs in FSTP in ia32-svid
753          2. marking a CALL with the appropriate annotation
754    
755    ----------------------------------------------------------------------
756    Name: Matthias Blume
757    Date: 2001/10/16 11:32:00 EDT
758    Tag: blume-20011016-netbsd
759    Description:
760    
761    Underscore patch from Chris Richards (fixing problem with compiling
762    runtime system under recent NetBSD).
763    
764    ----------------------------------------------------------------------
765    Name: Allen Leung
766    Date: 2001/10/12 17:18:32 EDT 2001
767    Tag: leung-20011012-x86-printflowgraph
768    Description:
769    
770    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
771    
772    ----------------------------------------------------------------------
773    Name: Lal George
774    Date: 2001/10/11 23:51:34 EDT
775    Tag: george-20011011-too-many-instrs
776    Description:
777    
778    The representation of a program point never expected to see more
779    than 65536 instructions in a basic block!
780    
781    ----------------------------------------------------------------------
782    Name: Lal George
783    Date: 2001/10/09 09:41:37 EDT
784    Tag: george-20011008-mlrisc-labels
785    Description:
786    
787    Changed the machine description files to support printing of
788    local and global labels in assembly code, based on host assembler
789    conventions.
790    
791    ----------------------------------------------------------------------
792    Name: Matthias Blume
793    Date: 2001/09/25 15:25:00 EDT
794    Tag: blume-20010925-exninfo
795    Description:
796    
797    I provided a non-hook implementation of exnName (at the toplevel) and
798    made the "dummy" implementation of exnMessage (at the toplevel) more
799    useful: if nothing gets "hooked in", then at least you are going to
800    see the exception name and a message indicating why you don't see more.
801    
802    [For the time being, programs that need exnMessage and want to use
803    ml-build should either use General.exnMessage (strongly recommended) or
804    refer to structure General at some other point so that CM sees a
805    static dependency.]
806    
807    [Similar remarks go for "print" and "use":  If you want to use their
808    functionality in stand-alone programs generated by ml-build, then use
809    TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
810    
811    ----------------------------------------------------------------------
812    Name: Matthias Blume
813    Date: 2001/09/20 17:28:00 EDT
814    Tag: blume-20010920-slowfp
815    Description:
816    
817    Allen says that x86-fast-fp is not safe yet, so I turned it off again...
818    
819    ----------------------------------------------------------------------
820    Name: Matthias Blume
821    Date: 2001/09/20 17:20:00 EDT
822    Tag: blume-20010920-canonicalpaths
823    Description:
824    
825    0. Updated the BOOT file (something that I forgot to do earlier).
826    
827    1. Small internal change to CM so that it avoids "/../" in filenames
828       as much as possible (but only where it is safe).
829    
830    2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
831       that arguments that contain delimiters are passed through correctly.
832       This change also means that all "special" arguments of the form
833       @SMLxxx... must come first.
834    
835    3. Changed install script to put relative anchor names for tool commands
836       into pathconfig.
837    
838    ----------------------------------------------------------------------
839    Name: Matthias Blume
840    Date: 2001/09/18 15:35:00 EDT
841    Tag: blume-20010918-readme11036
842    Description:
843    
844    Added README files.
845    
846    ----------------------------------------------------------------------
847    Name: Matthias Blume
848    Date: 2001/09/18 11:45:00 EDT
849    Tag: Release_110_36 (retag)
850    Description:
851    
852    Fixed mistake in config/preloads. Retagged as 110.36.
853    
854    ----------------------------------------------------------------------
855    Name: Matthias Blume
856    Date: 2001/09/18 09:40:00 EDT
857    Tag: Release_110_36_orig (tag changed)
858    Description:
859    
860    New version (110.36).  New bootfiles.
861    
862    ----------------------------------------------------------------------
863    Name: Matthias Blume
864    Date: 2001/09/14 16:15:00 EDT
865    Tag: blume-20010914-x86fastfp
866    Description:
867    
868    John committed some changes that Allen made, in particular a (hopefully)
869    correctly working version of the x86-fp module.
870    
871    I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
872    flag to "true".  Everything seems to compile to a fixpoint ok, and
873    "mandelbrot" speeds up by about 15%.
874    
875    ----------------------------------------------------------------------
876    Name: Matthias Blume
877    Date: 2001/09/13 11:20:00 EDT
878    Tag: blume-20010913-minimal
879    Description:
880    
881    1. Stefan Monnier's patch to fix a miscompilation problem that
882       was brought to light by John Reppy's work on Moby.
883    
884    2. Implemented a minimal "structure Compiler" that contains just
885       "version" and "architecture".  The minimal version will be
886       available when the full version is not.  This is for backward-
887       compatibility with code that wants to test Compiler.version.
888    
889    ----------------------------------------------------------------------
890    Name: Matthias Blume
891    Date: 2001/08/28 14:03:00 EDT
892    Tag: blume-20010828-ml-lex
893    Description:
894    
895    Fix for bug 1581, received from Neophytos Michael.
896    
897    ----------------------------------------------------------------------
898    Name: Matthias Blume
899    Date: 2001/08/27 11:20:00 EDT
900    Tag: blume-20010827-readme11035
901    Description:
902    
903    Fleshed out the README file for 110.35.
904    
905    ----------------------------------------------------------------------
906    Name: Matthias Blume
907    Date: 2001/08/24 17:10:00 EDT
908    Tag: Release_110_35
909    Description:
910    
911    New version number (110.35).  New bootfiles.
912    
913    ----------------------------------------------------------------------
914    Name: Lal George
915    Date: 2001/08/24 13:47:18 EDT 2001
916    Tag: george-20010824-MLRISC-graphs
917    Description:
918    
919     removed clusters from MLRISC completely and replaced with graphs.
920    
921    ----------------------------------------------------------------------
922    Name: Matthias Blume
923    Date: 2001/08/23 17:50:00 EDT
924    Tag: blume-20010823-toplevel
925    Description:
926    
927    - some reorganization of the code that implements various kinds of
928      environments in the compiler (static, dynamic, symbolic, combined)
929    - re-implemented the EnvRef module so that evalStream works properly
930      (if the stream contains references to "use", "CM.make", etc.)
931    - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
932    
933    ----------------------------------------------------------------------
934    Name: Matthias Blume
935    Date: 2001/08/20 15:50 EDT
936    Tag: blume20010820-slipup
937    Description:
938    
939    I forgot to commit a few files.  Here they are...
940    
941    ----------------------------------------------------------------------
942    Name: Matthias Blume
943    Date: 2001/08/20 15:35:00 EDT
944    Tag: blume-20010820-debugprof
945    Description:
946    
947    !!!! NEW BOOTFILES !!!!
948    
949    This is another round of reorganizing the compiler sources.  This
950    time the main goal was to factor out all the "instrumentation"
951    passes (for profiling and backtracing) into their own library.
952    The difficulty was to do it in such a way that it does not depend
953    on elaborate.cm but only on elabdata.cm.
954    
955    Therefore there have been further changes to both elaborate.cm and
956    elabdata.cm -- more "generic" things have been moved from the former
957    to the latter.  As a result, I was forced to split the assignment
958    of numbers indicating "primtyc"s into two portions: SML-generic and
959    SML/NJ-specific.  Since it would have been awkward to maintain,
960    I bit the bullet and actually _changed_ the mapping between these
961    numbers and primtycs.  The bottom line of this is that you need
962    a new set of bin- and bootfiles.
963    
964    I have built new bootfiles for all architectures, so doing a fresh
965    checkout and config/install.sh should be all you need.
966    
967    The newly created library's name is
968    
969        $smlnj/viscomp/debugprof.cm
970    
971    and its sources live under
972    
973        src/compiler/DebugProf
974    
975    ----------------------------------------------------------------------
976    Name: Matthias Blume
977    Date: 2001/08/15 17:15:00 EDT
978    Tag: blume-20010815-compreorg
979    Description:
980    
981    This is a first cut at reorganizing the CM libraries that make up the
982    core of the compiler.  The idea is to separate out pieces that could
983    be used independently by tools, e.g., the parser, the typechecker, etc.
984    
985    The current status is a step in this direction, but it is not quite
986    satisfactory yet.  Expect more changes in the future.
987    
988    Here is the current (new) organization...
989    
990        What used to be $smlnj/viscomp/core.cm is now divided into
991        six CM libraries:
992    
993             $smlnj/viscomp/basics.cm
994                           /parser.cm
995                           /elabdata.cm
996                           /elaborate.cm
997                           /execute.cm
998                           /core.cm
999    
1000        The CM files for these libraries live under src/system/smlnj/viscomp.
1001        All these libraries are proxy libraries that contain precisely
1002        one CM library component.  Here are the locations of the components
1003        (all within the src/compiler tree):
1004    
1005             Basics/basics.cm
1006             Parse/parser.cm
1007             ElabData/elabdata.cm
1008             Elaborator/elaborate.cm
1009             Execution/execute.cm
1010             core.cm
1011    
1012         [This organization is the same that has been used already
1013         for a while for the architecture-specific parts of the visible
1014         compiler and for the old version of core.cm.]
1015    
1016         As you will notice, many source files have been moved from their
1017         respective original locations to a new home in one of the above
1018         subtrees.
1019    
1020         The division of labor between the new libraries is the following:
1021    
1022             basics.cm:
1023                - Simple, basic definitions that pertain to many (or all) of
1024                  the other libraries.
1025             parser.cm:
1026                - The SML parser, producing output of type Ast.dec.
1027                - The type family for Ast is also defined and exported here.
1028             elabdata.cm:
1029                - The datatypes that describe input and output of the elaborator.
1030                  This includes types, absyn, and static environments.
1031             elaborator.cm:
1032                - The SML/NJ type checker and elaborator.
1033                  This maps an Ast.dec (with a given static environment) to
1034                  an Absyn.dec (with a new static environment).
1035                - This libraries implements certain modules that used to be
1036                  structures as functors (to remove dependencies on FLINT).
1037             execute.cm:
1038                - Everything having to do with executing binary code objects.
1039                - Dynamic environments.
1040             core.cm:
1041                - SML/NJ-specific instantiations of the elaborator and MLRISC.
1042                - Top-level modules.
1043                - FLINT (this should eventually become its own library)
1044    
1045    Notes:
1046    
1047    I am not 100% happy with the way I separated the elaborator (and its
1048    data structures) from FLINT.  Two instances of the same problem:
1049    
1050        1. Data structures contain certain fields that carry FLINT-specific
1051           information.  I hacked around this using exn and the property list
1052           module from smlnj-lib.  But the fact that there are middle-end
1053           specific fields around at all is a bit annoying.
1054    
1055        2. The elaborator calculates certain FLINT-related information.  I tried
1056           to make this as abstract as I could using functorization, but, again,
1057           the fact that the elaborator has to perform calculations on behalf
1058           of the middle-end at all is not nice.
1059    
1060        3. Having to used exn and property lists is unfortunate because it
1061           weakens type checking.  The other alternative (parameterizing
1062           nearly *everything*) is not appealing, though.
1063    
1064    I removed the "rebinding =" warning hack because due to the new organization
1065    it was awkward to maintain it.  As a result, the compiler now issues some of
1066    these warnings when compiling init.cmi during bootstrap compilation. On
1067    the plus side, you also get a warning when you do, for example:
1068       val op = = Int32.+
1069    which was not the case up to now.
1070    
1071    I placed "assign" and "deref" into the _Core structure so that the
1072    code that deals with the "lazy" keyword can find them there.  This
1073    removes the need for having access to the primitive environment
1074    during elaboration.
1075    
1076    ----------------------------------------------------------------------
1077    Name: Matthias Blume
1078    Date: 2001/08/13
1079    Tag: blume-20010813-closures
1080    Description:
1081    
1082    This fix was sent to us by Zhong Shao.  It is supposed to improve the
1083    performance of certain loops by avoiding needless closure allocation.
1084    
1085    ----------------------------------------------------------------------
1086    Name: Lal George
1087    Date: 2001/07/31 10:03:23 EDT 2001
1088    Tag: george-20010731-x86-fmalloc
1089    Description: Fixed bug in x86 calls
1090    
1091        There was a bug where call instructions would mysteriously
1092        vanish. The call instruction had to be one that returned
1093        a floating point value.
1094    
1095    ----------------------------------------------------------------------
1096    Name: Lal George
1097    Date: 2001/07/19 16:36:29 EDT 2001
1098    Tag: george-20010719-simple-cells
1099    Description:
1100    
1101    I have dramatically simplified the interface for CELLS in MLRISC.
1102    
1103    In summary, the cells interface is broken up into three parts:
1104    
1105      1. CellsBasis : CELLS_BASIS
1106    
1107            CellsBasis is a top level structure and common for all
1108            architectures.  it contains the definitions of basic datatypes
1109            and utility  functions over these types.
1110    
1111      2. functor Cells() : CELLS
1112    
1113            Cells generates an interface for CELLS that incorporates the
1114            specific resources on the target architecture, such as the
1115            presence of special register classes, their number and size,
1116            and various useful substructures.
1117    
1118      3. <ARCH>CELLS
1119    
1120            e.g. SparcCells: SPARCCELLS
1121    
1122            <ARCH>CELLS usually contains additional bindings for special
1123            registers  on the architecture, such as:
1124    
1125                    val r0 : cell           (* register zero *)
1126                    val y : cell            (* Y register *)
1127                    val psr : cell          (* processor status register *)
1128                    ...
1129    
1130            The structure returned by applying the Cells functor is opened
1131            in this interface.
1132    
1133    The main implication of all this is that the datatypes for cells is
1134    split between CellsBasis and CELLS -- a fairly simple change for user
1135    code.
1136    
1137    In the old scheme the CELLS interface had a definitional binding of
1138    the form:
1139    
1140            signature CELLS = sig
1141    
1142               structure CellsBasis = CellsBasis
1143    
1144               ...
1145    
1146            end
1147    
1148    With all the sharing constraints that goes on in MLRISC, this old
1149    design  quickly leads to errors such as:
1150    
1151            "structure definition spec inside of sharing ... "
1152    
1153    
1154    and appears to require an unacceptable amount of sharing and where
1155    constraint hackery.
1156    
1157    I think this error message (the interaction of definitional specs and
1158    sharing) requires more explanation on our web page.
1159    
1160    ----------------------------------------------------------------------
1161    Name: Matthias Blume
1162    Date: 2001/07/19 15:00:00 EDT
1163    Tag: blume-20010719-libreorg
1164    Description:
1165    
1166    This update puts together a fairly extensive but straightforward change
1167    to the way the libraries that implement the interactive system are
1168    organized:
1169    
1170       The biggest change is the elimination of structure Compiler.  As a
1171       replacement for this structure, there is now a CM library
1172       (known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
1173       that exports all the substructures of the original structure Compiler
1174       directly.  So instead of saying Compiler.Foo.bar one now simply
1175       says Foo.bar.  (The CM libraries actually export a collection of
1176       structures that is richer than the collection of substructures of
1177       structure Compiler.)
1178    
1179       To make the transition smooth, there is a separate library called
1180       $smlnj/compiler/compiler.cm which puts together and exports the
1181       original structure Compiler (or at least something very close to it).
1182    
1183       There are five members of the original structure Compiler
1184       that are not exported directly but which instead became members
1185       of a new structure Backend (described by signature BACKEND).  These are:
1186       structure Profile (: PROFILE), structure Compile (: COMPILE), structure
1187       Interact (: INTERACT), structure Machine (: MACHINE), and val
1188       architecture (: string).
1189    
1190       Structure Compiler.Version has become structure CompilerVersion.
1191    
1192       Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
1193       by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
1194       or x86, respectively.
1195       Each of these exports the same frontend structures that
1196       $smlnj/compiler.cm exports.  But they do not have a structure Backend
1197       and instead export some structure <Arch>Backend where <Arch> is Alpha32,
1198       Hppa, PPC, Sparc, or X86, respectively.
1199    
1200       Library $smlnj/compiler/all.cm exports the union of the exports of
1201       $smlnj/compiler/<arch>.cm
1202    
1203       There are no structures <Arch>Compiler anymore, use
1204       $smlnj/compiler/<arch>.cm instead.
1205    
1206       Library host-compiler-0.cm is gone.  Instead, the internal library
1207       that instantiates CM is now called cm0.cm.  Selection of the host
1208       compiler (backend) is no longer done here but. (Responsibility for it
1209       now lies with $smlnj/compiler/current.cm.  This seems to be more
1210       logical.)
1211    
1212       Many individual files have been moved or renamed.  Some files have
1213       been split into multiple files, and some "dead" files have been deleted.
1214    
1215    Aside from these changes to library organization, there are also changes
1216    to the way the code itself is organized:
1217    
1218       Structure Binfile has been re-implemented in such a way that it no
1219       longer needs any knowledge of the compiler.  It exclusively deals
1220       with the details of binfile layout.  It no longer invokes the
1221       compiler (for the purpose of creating new prospective binfile
1222       content), and it no longer has any knowledge of how to interpret
1223       pickles.
1224    
1225       Structure Compile (: COMPILE) has been stripped down to the bare
1226       essentials of compilation.  It no longer deals with linking/execution.
1227       The interface has been cleaned up considerably.
1228    
1229       Utility routines for dealing with linking and execution have been
1230       moved into their own substructures.
1231    
1232       (The ultimate goal of these changes is to provide a light-weight
1233       binfile loader/linker (at least for, e.g., stable libraries) that
1234       does not require CM or the compiler to be present.)
1235    
1236    CM documentation has been updated to reflect the changes to library
1237    organization.
1238    
1239    ----------------------------------------------------------------------
1240    Name: Matthias Blume
1241    Date: 2001/07/10 17:30:00 EDT
1242    Tag: Release_110_34
1243    Description:
1244    
1245    Minor tweak to 110.34 (re-tagged):
1246    
1247      - README.html file added to CVS repository
1248      - runtime compiles properly under FreeBSD 3.X and 4.X
1249    
1250    ----------------------------------------------------------------------
1251    Name: Matthias Blume
1252    Date: 2001/07/10 17:30:00 EDT
1253    Tag: Release_110_34
1254    Description:
1255    
1256    New version number (110.34). New bootfiles.
1257    
1258    ----------------------------------------------------------------------
1259    Name: Matthias Blume
1260    Date: 2001/07/09 16:00:00 EDT
1261    Tag: blume-20010709-more-varargs
1262    Description:
1263    
1264    I changed the handling of varargs in ml-nlffigen again:
1265    The ellipsis ... will now simply be ignored (with an accompanying warning).
1266    
1267    The immediate effect is that you can actually call a varargs function
1268    from ML -- but you can't actually supply any arguments beyond the ones
1269    specified explicitly.  (For example, you can call printf with its format
1270    string, but you cannot pass additional arguments.)
1271    
1272    This behavior is only marginally more useful than the one before, but
1273    it has the advantage that a function or, more importantly, a function
1274    type never gets dropped on the floor, thus avoiding follow-up problems with
1275    other types that refer to the offending one.
1276    
1277    ----------------------------------------------------------------------
1278    Name: Matthias Blume
1279    Date: 2001/07/09 11:25:00 EDT
1280    Tag: blume-20010709-varargs
1281    Description:
1282    
1283    1. ckit-lib.cm now exports structure Error
1284    2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
1285       with a warning accompanied by a source location.  Moreover, it
1286       merely skips the offending function or type and proceeds with the
1287       rest of its work.u  As a result, one can safely feed C code containing
1288       "..." to ml-nlffigen.
1289    3. There are some internal improvements to CM, providing slightly
1290       more general string substitutions in the tools subsystem.
1291    
1292    ----------------------------------------------------------------------
1293    Name: Matthias Blume
1294    Date: 2001/06/27 15:10:00 EDT
1295    Tag: blume-20010627-concur
1296    Description:
1297    
1298    Fixed a small bug in CM's handling of parallel compilation.
1299    (You could observe the bug by Control-C-interrupting an ordinary
1300    CMB.make or CM.stabilize and then attaching some compile servers.
1301    The result was that all of a sudden the previously interrupted
1302    compilation would continue on its own.  This was because of
1303    an over-optimization: CM did not bother to clean out certain queues
1304    when no servers were attached "anyway", resulting in the contents
1305    of these queues to grab control when new servers did get attached.)
1306    
1307    There is also another minor update to the CM manual.
1308    
1309    ----------------------------------------------------------------------
1310    Name: Matthias Blume
1311    Date: 2001/06/26 16:15:00 EDT
1312    Tag: blume-20010626-cmdoc
1313    Description:
1314    
1315    Minor typo fixed in CM manual (syntax diagram for libraries).
1316    
1317    ----------------------------------------------------------------------
1318    Name: Matthias Blume
1319    Date: 2001/06/25 22:55:00 EDT
1320    Tag: blume-20010625-x86pc
1321    Description:
1322    
1323    Fixed a nasty bug in the X86 assembly code that caused signal
1324    handlers to fail (crash) randomly.
1325    
1326    ----------------------------------------------------------------------
1327    Name: Matthias Blume
1328    Date: 2001/06/25 12:05:00 EDT
1329    Tag: blume-20010625-nlffigen
1330    Description:
1331    
1332    This update fixes a number of minor bugs in ml-nlffigen as reported by
1333    Nick Carter <nbc@andrew.cmu.edu>.
1334    
1335      1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
1336      2. Default names for generated files are now derived from the name of
1337         the C file *without its directory*.  In particular, this causes generated
1338         files to be placed locally even if the C file is in some system directory.
1339      3. Default names for generated signatures and structures are also derived
1340         from the C file name without its directory.  This avoids silly things
1341         like "structure GL/GL".
1342         (Other silly names are still possible because ml-nlffigen does not do
1343          a thorough check of whether generated names are legal ML identifiers.
1344          When in doubt, use command line arguments to force particular names.)
1345    
1346    ----------------------------------------------------------------------
1347    Name: Matthias Blume
1348    Date: 2001/06/21 12:25:00 EDT
1349    Tag: blume-20010621-eXene
1350    Description:
1351    
1352    eXene now compiles and (sort of) works again.
1353    
1354    The library name (for version > 110.33) is $/eXene.cm.
1355    
1356    I also added an new example in src/eXene/examples/nbody.  See the
1357    README file there for details.
1358    
1359    ----------------------------------------------------------------------
1360    Name: Matthias Blume
1361    Date: 2001/06/20 16:40:00 EDT
1362    Tag: blume-20010620-cml
1363    Description:
1364    
1365    CML now compiles and works again.
1366    
1367    Libraries (for version > 110.33):
1368    
1369      $cml/cml.cm            Main CML library.
1370      $cml/basis.cm          CML's version of $/basis.cm.
1371      $cml/cml-internal.cm   Internal helper library.
1372      $cml/core-cml.cm       Internal helper library.
1373      $cml-lib/trace-cml.cm  Tracing facility.
1374      $cml-lib/smlnj-lib.cm  CML's version of $/smlnj-lib.cm
1375    
1376    The installer (config/install.sh) has been taught how to properly
1377    install this stuff.
1378    
1379    ----------------------------------------------------------------------
1380    Name: Matthias Blume
1381    Date: 2001/06/19 17:55:00 EDT
1382    Tag: blume-20010619-instantiate
1383    Description:
1384    
1385    This un-breaks the fix for bug 1432.
1386    (The bug was originally fixed in 110.9 but I broke it again some
1387    time after that.)
1388    
1389    ----------------------------------------------------------------------
1390    Name: Matthias Blume
1391    Date: 2001/06/19 17:25:00 EDT
1392    Tag: blume-20010619-signals
1393    Description:
1394    
1395    This should (hopefully) fix the long-standing signal handling bug.
1396    (The runtime system was constructing a continuation record with an
1397    incorrect descriptor which would cause the GC to drop data on the floor...)
1398    
1399    ----------------------------------------------------------------------
1400    Name: Matthias Blume
1401    Date: 2001/06/15 15:05:00 EDT
1402    Tag: blume-20010615-moresparc
1403    Description:
1404    
1405    Here is a short late-hour update related to Sparc c-calls:
1406    
1407     -- made handling of double-word arguments a bit smarter
1408    
1409     -- instruction selection phase tries to collapse certain clumsily
1410        constructed ML-Trees; typical example:
1411    
1412            ADD(ty,ADD(_,e,LI d1),LI d2)  ->  ADD(ty,e,LI(d1+d2))
1413    
1414        This currently has no further impact on SML/NJ since mlriscGen does
1415        not seem to generate such patterns in the first place, and c-calls
1416        (which did generate them in the beginning) has meanwhile been fixed
1417        so as to avoid them as well.
1418    
1419    ----------------------------------------------------------------------
1420    Name: Matthias Blume
1421    Date: 2001/06/15 15:05:00 EDT
1422    Tag: blume-20010615-sparc
1423    Description:
1424    
1425    The purpose of this update is to provide an implementation of NLFFI
1426    on Sparc machines.
1427    
1428    Here are the changes in detail:
1429    
1430       * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
1431       the Sparc implementation of the c-calls API.
1432       * The Sparc backend of SML/NJ has been modified to uniformely use %fp
1433       for accessing the ML frame.  Thus, we have a real frame pointer and
1434       can freely modify %sp without need for an omit-frame-ptr phase.
1435       The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
1436       this case.
1437       * ml-nlffigen has been taught to produce code for different architectures
1438       and calling conventions.
1439       * In a way similar to what was done in the x86 case, the Sparc
1440       backend uses its own specific extension to mltree.  (For example,
1441       it needs to be able to generate UNIMP instructions which are part
1442       of the calling convention.)
1443       * ml-nlffi-lib was reorganized to make it more modular (in particular,
1444       to make it easier to plug in new machine- and os-dependent parts).
1445    
1446    There are some other fairly unrelated bug fixes and cleanups as well:
1447    
1448       * I further hacked the .cm files for MLRISC tools (like MDLGen) so
1449       that they properly share their libraries with existing SML/NJ libraries.
1450       * I fixed a minor cosmetic bug in CM, supressing certain spurious
1451       follow-up error messages.
1452       * Updates to CM/CMB documentation.
1453    
1454    TODO items:
1455    
1456       * MLRISC should use a different register as its asmTemp on the Sparc.
1457         (The current %o2 is a really bad choice because it is part of the
1458         calling conventions, so things might interfere in unexpected ways.)
1459    
1460    ----------------------------------------------------------------------
1461    Name: Matthias Blume
1462    Date: 2001/06/07
1463    Tag: blume-20010607-calls
1464    Description:
1465    
1466    A number of internal changes related to C calls and calling conventions:
1467    
1468    1. ML-Tree CALL statements now carry a "pops" field.  It indicates the
1469       number of bytes popped implicitly (by the callee).  In most cases
1470       this field is 0 but on x86/win32 it is some non-zero value.  This
1471       is information provided for the benefit of the "omit-frameptr" pass.
1472    2. The CALL instruction on the x86 carries a similar "pops" field.
1473       The instruction selection phase copies its value from the ML-Tree
1474       CALL statement.
1475    3. On all other architectures, the instruction selection phase checks
1476       whether "pops=0" and complains if not.
1477    4. The c-calls implementation for x86 now accepts two calling conventions:
1478       "ccall" and "stdcall".  When "ccall" is selected, the caller cleans
1479       up after the call and pops is set to 0.  For "stdcall", the caller
1480       does nothing, leaving the cleanup to the callee; pops is set to
1481       the number of bytes that were pushed onto the stack.
1482    5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
1483       distinguish between "ccall" and "stdcall".
1484    6. The UNIMP instruction has been added to the supported Sparc instruction
1485       set. (This is needed for implementing the official C calling convention
1486       on this architecture.)
1487    7. I fixed some of the .cm files under src/MLRISC/Tools to make them
1488       work with the latest CM.
1489    
1490    ----------------------------------------------------------------------
1491    Name: Matthias Blume
1492    Date: 2001/06/05 15:10:00 EDT
1493    Tag: blume-20010605-cm-index
1494    Description:
1495    
1496    0. The "lambdasplit" parameter for class "sml" in CM has been documented.
1497    
1498    1. CM can now generate "index files".  These are human-readable files
1499       that list on a per-.cm-file basis each toplevel symbol defined or
1500       imported.  The location of the index file for
1501       <p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
1502       To enable index-file generation, set CM.Control.generate_index to true
1503       or export an environment-symbol: export CM_GENERATE_INDEX=true.
1504    
1505       The CM manual has been updated accordingly.
1506    
1507    2. I made some slight modifications to the c-calls API in MLRISC.
1508    
1509         a) There is now a callback to support saving/restoring of
1510            dedicated but caller-save registers around the actual call
1511            instruction.
1512         b) One can optionally specify a comment-annotation for the
1513            call instruction.
1514    
1515    3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
1516       (For example, the comment annotation shows the C prototype of
1517        the function being called.)
1518    
1519    ----------------------------------------------------------------------
1520    Name: Matthias Blume
1521    Date: 2001/06/01 13:30:00 EDT
1522    Tag: blume-20010601-nlffi-cleanup
1523    Description:
1524    
1525    This is mostly a cleanup of MLFFI stuff:
1526    
1527       - some signature files have been put into a more exposed place
1528       - the ugly 'f type parameter is gone (simplifies types tremendously!)
1529       - ml-nlffigen changed accordingly
1530       - tutorial updated
1531    
1532    Other changes:
1533    
1534       - author's affiliation in CM manual(s) updated
1535       - some more recognized keywords added to Allen's sml.sty
1536    
1537    ----------------------------------------------------------------------
1538    Name: Matthias Blume
1539    Date: 2001/05/25 15:30:00 EDT
1540    Tag: blume-20010525-iptr
1541    Description:
1542    
1543      - put the official 110.33-README (as it appears on the ftp server) under
1544        CVS
1545      - fixed a small bug related to incomplete pointer types in
1546        ml-nlffigen
1547      - small cosmetic change to the ml-nlffi-lib's "arr" type constructor
1548        (it does not need the 'f type parameter)
1549    
1550    ----------------------------------------------------------------------
1551    Name: Matthias Blume
1552    Date: 2001/05/23 14:30:00 EDT
1553    Tag: Release_110_33
1554    Description:
1555    
1556    New version number (110.33).  New bootfiles.
1557    
1558    ----------------------------------------------------------------------
1559    Name: Matthias Blume
1560    Date: 2001/05/22 18:06:00 EDT
1561    Tag: blume-20010522-targets
1562    Description:
1563    
1564    Made install.sh use file config/targets.customized if it exists, falling
1565    back to config/targets if it doesn't.  This way one can have a customized
1566    version of the targets file without touching the "real thing", thus
1567    eliminating the constant fear of accidentally checking something bogus
1568    back into the CVS repository...  (File config/targets.customized must
1569    not be added to the repository!)
1570    
1571    ----------------------------------------------------------------------
1572    Name: Matthias Blume
1573    Date: 2001/05/22 16:30:00 EDT
1574    Tag: blume-20010522-minitut
1575    Description:
1576    
1577    1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
1578       struct returns.
1579    2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt.  This is some very
1580       incomplete, preliminary documentation for NLFFI.
1581    
1582    ----------------------------------------------------------------------
1583    Name: Matthias Blume
1584    Date: 2001/05/14 11:30:00 EDT
1585    Tag: blume-20010514-script
1586    Description:
1587    
1588    Some bugs in install script fixed.
1589    
1590    In addition to that I also made a slight change to the NLFFI API:
1591    Functors generated by ml-nlffigen now take the dynamic library as a
1592    straight functor argument, not as a suspended one.  (The original
1593    functor code used to force the suspension right away anyway, so there
1594    was nothing gained by this complication of the interface.)
1595    
1596    ----------------------------------------------------------------------
1597    Name: Matthias Blume
1598    Date: 2001/05/11 14:35:00 EDT
1599    Tag: blume-20010511-ml-nlffi
1600    Description:
1601    
1602    I finally took the plunge and added my new FFI code to the main
1603    repository.  For x86-linux it is now ready for prime-time.
1604    
1605    There are two new subdirectories of "src":
1606    
1607      - ml-nlffi-lib:
1608           The utility library for programs using the FFI interface.
1609           Here is the implementation of $/c.cm and its associated low-level
1610           partners $/c-int.cm and $/memory.cm.
1611      - ml-nlffigen:
1612           A stand-alone program for generating ML glue code from C source
1613           code.
1614    
1615    Building ml-nlffigen requires $/ckit-lib.cm.
1616    
1617    The config/install.sh script has been updates to do the Right Thing
1618    (hopefully).
1619    
1620    Notice that the source tree for the C-Kit will not be put under "src"
1621    but directly under the installation root directory.  (This is the
1622    structure that currently exists on the CVS server when you check out
1623    module "sml".)  Fortunately, config/install.sh knows about this oddity.
1624    
1625    Bugs: No documentation yet.
1626    
1627    ----------------------------------------------------------------------
1628    Name: Matthias Blume
1629    Date: 2001/05/09 16:35:00 EDT
1630    Tag: blume-20010509-cpscontract
1631    Description:
1632    
1633    Fixed a bug in the accounting code in cpsopt/contract.sml.  (The
1634    wrapper/unwrapper elimination did not decrement usage counts and some
1635    dead variables got overlooked by the dead-up logic.)
1636    
1637    ----------------------------------------------------------------------
1638    Name: Lal George
1639    Date: 2001/05/08  17:26:09 EDT
1640    Tag: george-20010508-omit-frameptr
1641    Description:
1642    
1643    Changes to implement the omit-frame-pointer optimization to support
1644    raw C calls. For now, there is only support on the Intel x86, but
1645    other architectures will follow as more experience is gained with this.
1646    
1647    
1648    ----------------------------------------------------------------------
1649    Name: Matthias Blume
1650    Date: 2001/05/07 14:40:00 EDT
1651    Tag: blume-20010507-proxies
1652    Description:
1653    
1654    I made into "proxy libraries" all libraries that qualify for such a
1655    change.  (A qualifying library is a library that has another library or
1656    groups as its sole member and repeats that member's export list
1657    verbatim.  A proxy library avoids this repetition by omitting its export
1658    list, effectively inheriting the list that its (only) member exports.
1659    See the CM manual for more explanation.)
1660    The main effect is that explicit export lists for these libraries
1661    do not have to be kepts in sync, making maintenance a bit easier.
1662    
1663    I also added copyright notices to many .cm-files.
1664    
1665    Last but not least, I made a new set of bootfiles.
1666    
1667    ----------------------------------------------------------------------
1668    Name: Matthias Blume
1669    Date: 2001/05/04 17:00:00 EDT
1670    Tag: blume-20010504-cm-lsplit
1671    Description:
1672    
1673    0. John merged pending changes to $/smlnj-lib.cm
1674    
1675    1. Allen's previous change accidentally backed out of one of Lal's
1676       earlier changes.  I undid this mistake (re-introducing Lal's change).
1677    
1678    2. I used the new topOrder' function from graph-scc.sml (from $/smlnj-lib.cm)
1679       within the compiler where applicable.  There is some code simplification
1680       because of that.
1681    
1682    3. The "split" phase (in FLINT) is now part of the default list of phases.
1683       Compiler.Control.LambdaSplitting.* can be used to globally control the
1684       lambda-splitting (cross-module-inlining) engine.  In addition to that,
1685       it can now also be controlled on a per-source basis: CM has been taught
1686       a new tool parameter applicable to ML source files.
1687    
1688       - To turn lambda-splitting off completely:
1689            local open Compiler.Control.LambdaSplitting in
1690                val _ = set Off
1691            end
1692       - To make "no lambda-splitting" the global default (but allow per-source
1693         overriding); this is the initial setting:
1694            local open Compiler.Control.LambdaSplitting in
1695                val _ = set (Default NONE)
1696            end
1697       - To make "lambda-splitting with aggressiveness a" the global default
1698         (and allow per-source overriding):
1699            local open Compiler.Control.LambdaSplitting in
1700                val _ = set (Default (SOME a))
1701            end
1702    
1703       - To turn lambda-splitting off for a given ML souce file (say: a.sml)
1704         write (in the respective .cm-file):
1705            a.sml (lambdasplitting:off)
1706       - To turn lambda-splitting for a.sml on with minimal aggressiveness:
1707            a.sml (lambdasplitting:on)
1708       - To turn lambda-splitting for a.sml on with aggressiveness <a> (where
1709         <a> is a decimal non-negative integer):
1710            a.sml (lambdasplitting:<a>)
1711       - To turn lambda-splitting for a.sml on with maximal aggressiveness:
1712            a.sml (lambdasplitting:infinity)
1713       - To use the global default for a.sml:
1714            a.sml (lambdasplitting:default)
1715         or simply
1716            a.sml
1717    
1718    ----------------------------------------------------------------------
1719    Name: Allen Leung
1720    Date: 2001/05/04 01:57:00 EDT
1721    Tag: leunga-20010504-sync
1722    Description:
1723    
1724      MLRISC features.
1725    
1726      1. Fix to CMPXCHG instructions.
1727      2. Changed RA interface to allow annotations in callbacks.
1728      3. Added a new method to the stream interface to allow annotations updates.
1729    
1730    ----------------------------------------------------------------------
1731    Name: Matthias Blume
1732    Date: 2001/05/01 11:45:00 EDT
1733    Tag: blume-20010501-pcedittmp
1734    Description:
1735    
1736    Changed install.sh to use the current working directory instead of
1737    /usr/tmp for a temporary file (pcedittmp).  The previous choice
1738    of /usr/tmp caused trouble with MacOS X because of file premission
1739    problems.
1740    
1741    ----------------------------------------------------------------------
1742    Name: Matthias Blume
1743    Date: 2001/04/20 11:10:00 EDT
1744    Tag: blume-20010420-inMLflag
1745    Description:
1746    
1747     - added vp_limitPtrMask to vproc-state.h
1748       (for use by the raw-C-calls mechanism to implement proper interrupt
1749        handling)
1750     - made the ML compiler aware of various data-structure offsets so it
1751       can generate code for accessing the vp_inML flag and vp_limitPtrMask
1752     - tweaked mlriscGen.sml to have it emit interrupt-handling code for
1753       raw C-calls
1754    
1755    ----------------------------------------------------------------------
1756    Name: Lal George
1757    Date: 2001/04/20 09:15:28 EDT
1758    Tag: george-20010420-macosX
1759    Description:
1760    
1761     - Changes to port to Mac OS X; Darwin.
1762    
1763     - In the process I found that sqrt was broken on the PPC, because the
1764       fsqrt instruction is not implemented.
1765    
1766    ----------------------------------------------------------------------
1767    Name: Matthias Blume
1768    Date: 2001/04/18 12:45:00 EDT
1769    Tag: blume-20010418-ccalls
1770    Description:
1771    
1772     - fixed two off-by-4 errors in the x86-specific c-calls implementation
1773       (this bug prevented structure arguments containing pointers from being
1774        passed correctly)
1775     - changed the raw-C-call code in mlriscGen.sml in such a way that
1776       structure arguments are represented as a pointer to the beginning
1777       of the structure (instead of having a series of synthesized arguments,
1778       one for each structure member)
1779    
1780     - made makeml script's verbosity level configurable via environment
1781       variable (MAKEML_VERBOSITY)
1782    
1783     - eliminated placeholder implementations for f32l, w16s, i16s, and f32s
1784       in rawmem-x86.sml; we are now using the real thing
1785    
1786    ----------------------------------------------------------------------
1787    Name: Matthias Blume
1788    Date: 2001/03/22 16:25:00 EST
1789    Tag: blume-20010322-bootfiles
1790    Description:
1791    
1792    Created a new set of bootfiles (for your automatic installation convenience).
1793    
1794    ----------------------------------------------------------------------
1795    Name: Matthias Blume
1796    Date: 2001/03/22 15:10:00 EST
1797    Tag: blume-20010322-rawmem-parcm
1798    Description:
1799    
1800    1. All "raw memory access" primitives for the new FFI are implemented now
1801       (at least on the x86).
1802    2. Some further cleanup of CM's parallel make mechanism.
1803    
1804    ----------------------------------------------------------------------
1805    Name: Matthias Blume
1806    Date: 2001/03/19 17:53:00 EST
1807    Tag: blume-20010319-parallel
1808    Description:
1809    
1810    Parallel make (using compile servers) now works again.
1811    
1812    To this end, CM.stabilize and CMB.make have been modified to work in
1813    two passes when compile servers are attached:
1814       1. Compile everything, do not perform stabilization; this pass
1815          uses compile servers
1816       2. Stabilize everything; this pass does not use compile servers
1817    If there are no compile servers, the two passes are combined into one
1818    (as before).  Splitting the passes increases the inherent parallelism
1819    in the dependency graph because the entire graph including all
1820    libraries is available at the same time.  This, in turn, improves
1821    server utilization.  The downside is that the master process will
1822    have to do some extra work after compilation is done (because for
1823    technical reasons it must re-read all the binfiles during stabilization).
1824    
1825    ----------------------------------------------------------------------
1826    Name: Matthias Blume
1827    Date: 2001/03/16 12:22:00 EST
1828    Tag: blume-20010316-bootfiles
1829    Description:
1830    
1831    Created a new set of bootfiles (for your automatic installation convenience).
1832    
1833    ----------------------------------------------------------------------
1834    Name: Matthias Blume
1835    Date: 2001/03/16 11:00:00 EST
1836    Tag: blume-20010316-MLTREE-fixup
1837    Description:
1838    
1839    This is a minor fixup for an (untagged) earlier commit by Allen.
1840    (A file was missing).
1841    
1842    ----------------------------------------------------------------------
1843  Name: Allen Leung  Name: Allen Leung
1844  Date: Mon Mar  5 18:54:57 EST 2001  Date: Mon Mar  5 18:54:57 EST 2001
1845  Tag: leunga-20010305-cut-support  Tag: leunga-20010305-cut-support
# Line 1986  Line 3813 
3813       elaborator). There were a lot of changes during my "linkpath" trials       elaborator). There were a lot of changes during my "linkpath" trials
3814       that could have been reverted to their original state but weren't.       that could have been reverted to their original state but weren't.
3815       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
3816       than what was strictly necessary...  (I _did_ resist the tempation       than what was strictly necessary...  (I _did_ resist the temptation
3817       of doing any "global reformatting" to avoid an untimely death at       of doing any "global reformatting" to avoid an untimely death at
3818       Dave's hands. :)       Dave's hands. :)
3819    

Legend:
Removed from v.796  
changed lines
  Added in v.1067

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