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

SCM Repository

[smlnj] Diff of /sml/branches/primop-branch-3/NOTES/HISTORY
ViewVC logotype

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

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

revision 570, Wed Mar 8 17:30:13 2000 UTC revision 927, Fri Sep 14 20:19:53 2001 UTC
# Line 13  Line 13 
13  Description:  Description:
14    
15  ----------------------------------------------------------------------  ----------------------------------------------------------------------
16  Name:  Name: Matthias Blume
17  Date:  Date: 2001/09/14 16:15:00 EDT
18  Tag:  Tag: blume-20010914-x86fastfp
19  Description:  Description:
20    
21    John committed some changes that Allen made, in particular a (hopefully)
22    correctly working version of the x86-fp module.
23    
24    I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
25    flag to "true".  Everything seems to compile to a fixpoint ok, and
26    "mandelbrot" speeds up by about 15%.
27    
28    ----------------------------------------------------------------------
29    Name: Matthias Blume
30    Date: 2001/09/13 11:20:00 EDT
31    Tag: blume-20010913-minimal
32    Description:
33    
34    1. Stefan Monnier's patch to fix a miscompilation problem that
35       was brought to light by John Reppy's work on Moby.
36    
37    2. Implemented a minimal "structure Compiler" that contains just
38       "version" and "architecture".  The minimal version will be
39       available when the full version is not.  This is for backward-
40       compatibility with code that wants to test Compiler.version.
41    
42    ----------------------------------------------------------------------
43    Name: Matthias Blume
44    Date: 2001/08/28 14:03:00 EDT
45    Tag: blume-20010828-ml-lex
46    Description:
47    
48    Fix for bug 1581, received from Neophytos Michael.
49    
50    ----------------------------------------------------------------------
51    Name: Matthias Blume
52    Date: 2001/08/27 11:20:00 EDT
53    Tag: blume-20010827-readme11035
54    Description:
55    
56    Fleshed out the README file for 110.35.
57    
58    ----------------------------------------------------------------------
59    Name: Matthias Blume
60    Date: 2001/08/24 17:10:00 EDT
61    Tag: Release_110_35
62    Description:
63    
64    New version number (110.35).  New bootfiles.
65    
66    ----------------------------------------------------------------------
67    Name: Lal George
68    Date: 2001/08/24 13:47:18 EDT 2001
69    Tag: george-20010824-MLRISC-graphs
70    Description:
71    
72     removed clusters from MLRISC completely and replaced with graphs.
73    
74    ----------------------------------------------------------------------
75    Name: Matthias Blume
76    Date: 2001/08/23 17:50:00 EDT
77    Tag: blume-20010823-toplevel
78    Description:
79    
80    - some reorganization of the code that implements various kinds of
81      environments in the compiler (static, dynamic, symbolic, combined)
82    - re-implemented the EnvRef module so that evalStream works properly
83      (if the stream contains references to "use", "CM.make", etc.)
84    - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
85    
86    ----------------------------------------------------------------------
87    Name: Matthias Blume
88    Date: 2001/08/20 15:50 EDT
89    Tag: blume20010820-slipup
90    Description:
91    
92    I forgot to commit a few files.  Here they are...
93    
94    ----------------------------------------------------------------------
95    Name: Matthias Blume
96    Date: 2001/08/20 15:35:00 EDT
97    Tag: blume-20010820-debugprof
98    Description:
99    
100    !!!! NEW BOOTFILES !!!!
101    
102    This is another round of reorganizing the compiler sources.  This
103    time the main goal was to factor out all the "instrumentation"
104    passes (for profiling and backtracing) into their own library.
105    The difficulty was to do it in such a way that it does not depend
106    on elaborate.cm but only on elabdata.cm.
107    
108    Therefore there have been further changes to both elaborate.cm and
109    elabdata.cm -- more "generic" things have been moved from the former
110    to the latter.  As a result, I was forced to split the assignment
111    of numbers indicating "primtyc"s into two portions: SML-generic and
112    SML/NJ-specific.  Since it would have been awkward to maintain,
113    I bit the bullet and actually _changed_ the mapping between these
114    numbers and primtycs.  The bottom line of this is that you need
115    a new set of bin- and bootfiles.
116    
117    I have built new bootfiles for all architectures, so doing a fresh
118    checkout and config/install.sh should be all you need.
119    
120    The newly created library's name is
121    
122        $smlnj/viscomp/debugprof.cm
123    
124    and its sources live under
125    
126        src/compiler/DebugProf
127    
128    ----------------------------------------------------------------------
129    Name: Matthias Blume
130    Date: 2001/08/15 17:15:00 EDT
131    Tag: blume-20010815-compreorg
132    Description:
133    
134    This is a first cut at reorganizing the CM libraries that make up the
135    core of the compiler.  The idea is to separate out pieces that could
136    be used independently by tools, e.g., the parser, the typechecker, etc.
137    
138    The current status is a step in this direction, but it is not quite
139    satisfactory yet.  Expect more changes in the future.
140    
141    Here is the current (new) organization...
142    
143        What used to be $smlnj/viscomp/core.cm is now divided into
144        six CM libraries:
145    
146             $smlnj/viscomp/basics.cm
147                           /parser.cm
148                           /elabdata.cm
149                           /elaborate.cm
150                           /execute.cm
151                           /core.cm
152    
153        The CM files for these libraries live under src/system/smlnj/viscomp.
154        All these libraries are proxy libraries that contain precisely
155        one CM library component.  Here are the locations of the components
156        (all within the src/compiler tree):
157    
158             Basics/basics.cm
159             Parse/parser.cm
160             ElabData/elabdata.cm
161             Elaborator/elaborate.cm
162             Execution/execute.cm
163             core.cm
164    
165         [This organization is the same that has been used already
166         for a while for the architecture-specific parts of the visible
167         compiler and for the old version of core.cm.]
168    
169         As you will notice, many source files have been moved from their
170         respective original locations to a new home in one of the above
171         subtrees.
172    
173         The division of labor between the new libraries is the following:
174    
175             basics.cm:
176                - Simple, basic definitions that pertain to many (or all) of
177                  the other libraries.
178             parser.cm:
179                - The SML parser, producing output of type Ast.dec.
180                - The type family for Ast is also defined and exported here.
181             elabdata.cm:
182                - The datatypes that describe input and output of the elaborator.
183                  This includes types, absyn, and static environments.
184             elaborator.cm:
185                - The SML/NJ type checker and elaborator.
186                  This maps an Ast.dec (with a given static environment) to
187                  an Absyn.dec (with a new static environment).
188                - This libraries implements certain modules that used to be
189                  structures as functors (to remove dependencies on FLINT).
190             execute.cm:
191                - Everything having to do with executing binary code objects.
192                - Dynamic environments.
193             core.cm:
194                - SML/NJ-specific instantiations of the elaborator and MLRISC.
195                - Top-level modules.
196                - FLINT (this should eventually become its own library)
197    
198    Notes:
199    
200    I am not 100% happy with the way I separated the elaborator (and its
201    data structures) from FLINT.  Two instances of the same problem:
202    
203        1. Data structures contain certain fields that carry FLINT-specific
204           information.  I hacked around this using exn and the property list
205           module from smlnj-lib.  But the fact that there are middle-end
206           specific fields around at all is a bit annoying.
207    
208        2. The elaborator calculates certain FLINT-related information.  I tried
209           to make this as abstract as I could using functorization, but, again,
210           the fact that the elaborator has to perform calculations on behalf
211           of the middle-end at all is not nice.
212    
213        3. Having to used exn and property lists is unfortunate because it
214           weakens type checking.  The other alternative (parameterizing
215           nearly *everything*) is not appealing, though.
216    
217    I removed the "rebinding =" warning hack because due to the new organization
218    it was awkward to maintain it.  As a result, the compiler now issues some of
219    these warnings when compiling init.cmi during bootstrap compilation. On
220    the plus side, you also get a warning when you do, for example:
221       val op = = Int32.+
222    which was not the case up to now.
223    
224    I placed "assign" and "deref" into the _Core structure so that the
225    code that deals with the "lazy" keyword can find them there.  This
226    removes the need for having access to the primitive environment
227    during elaboration.
228    
229    ----------------------------------------------------------------------
230    Name: Matthias Blume
231    Date: 2001/08/13
232    Tag: blume-20010813-closures
233    Description:
234    
235    This fix was sent to us by Zhong Shao.  It is supposed to improve the
236    performance of certain loops by avoiding needless closure allocation.
237    
238    ----------------------------------------------------------------------
239    Name: Lal George
240    Date: 2001/07/31 10:03:23 EDT 2001
241    Tag: george-20010731-x86-fmalloc
242    Description: Fixed bug in x86 calls
243    
244        There was a bug where call instructions would mysteriously
245        vanish. The call instruction had to be one that returned
246        a floating point value.
247    
248    ----------------------------------------------------------------------
249    Name: Lal George
250    Date: 2001/07/19 16:36:29 EDT 2001
251    Tag: george-20010719-simple-cells
252    Description:
253    
254    I have dramatically simplified the interface for CELLS in MLRISC.
255    
256    In summary, the cells interface is broken up into three parts:
257    
258      1. CellsBasis : CELLS_BASIS
259    
260            CellsBasis is a top level structure and common for all
261            architectures.  it contains the definitions of basic datatypes
262            and utility  functions over these types.
263    
264      2. functor Cells() : CELLS
265    
266            Cells generates an interface for CELLS that incorporates the
267            specific resources on the target architecture, such as the
268            presence of special register classes, their number and size,
269            and various useful substructures.
270    
271      3. <ARCH>CELLS
272    
273            e.g. SparcCells: SPARCCELLS
274    
275            <ARCH>CELLS usually contains additional bindings for special
276            registers  on the architecture, such as:
277    
278                    val r0 : cell           (* register zero *)
279                    val y : cell            (* Y register *)
280                    val psr : cell          (* processor status register *)
281                    ...
282    
283            The structure returned by applying the Cells functor is opened
284            in this interface.
285    
286    The main implication of all this is that the datatypes for cells is
287    split between CellsBasis and CELLS -- a fairly simple change for user
288    code.
289    
290    In the old scheme the CELLS interface had a definitional binding of
291    the form:
292    
293            signature CELLS = sig
294    
295               structure CellsBasis = CellsBasis
296    
297               ...
298    
299            end
300    
301    With all the sharing constraints that goes on in MLRISC, this old
302    design  quickly leads to errors such as:
303    
304            "structure definition spec inside of sharing ... "
305    
306    
307    and appears to require an unacceptable amount of sharing and where
308    constraint hackery.
309    
310    I think this error message (the interaction of definitional specs and
311    sharing) requires more explanation on our web page.
312    
313    ----------------------------------------------------------------------
314    Name: Matthias Blume
315    Date: 2001/07/19 15:00:00 EDT
316    Tag: blume-20010719-libreorg
317    Description:
318    
319    This update puts together a fairly extensive but straightforward change
320    to the way the libraries that implement the interactive system are
321    organized:
322    
323       The biggest change is the elimination of structure Compiler.  As a
324       replacement for this structure, there is now a CM library
325       (known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
326       that exports all the substructures of the original structure Compiler
327       directly.  So instead of saying Compiler.Foo.bar one now simply
328       says Foo.bar.  (The CM libraries actually export a collection of
329       structures that is richer than the collection of substructures of
330       structure Compiler.)
331    
332       To make the transition smooth, there is a separate library called
333       $smlnj/compiler/compiler.cm which puts together and exports the
334       original structure Compiler (or at least something very close to it).
335    
336       There are five members of the original structure Compiler
337       that are not exported directly but which instead became members
338       of a new structure Backend (described by signature BACKEND).  These are:
339       structure Profile (: PROFILE), structure Compile (: COMPILE), structure
340       Interact (: INTERACT), structure Machine (: MACHINE), and val
341       architecture (: string).
342    
343       Structure Compiler.Version has become structure CompilerVersion.
344    
345       Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
346       by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
347       or x86, respectively.
348       Each of these exports the same frontend structures that
349       $smlnj/compiler.cm exports.  But they do not have a structure Backend
350       and instead export some structure <Arch>Backend where <Arch> is Alpha32,
351       Hppa, PPC, Sparc, or X86, respectively.
352    
353       Library $smlnj/compiler/all.cm exports the union of the exports of
354       $smlnj/compiler/<arch>.cm
355    
356       There are no structures <Arch>Compiler anymore, use
357       $smlnj/compiler/<arch>.cm instead.
358    
359       Library host-compiler-0.cm is gone.  Instead, the internal library
360       that instantiates CM is now called cm0.cm.  Selection of the host
361       compiler (backend) is no longer done here but. (Responsibility for it
362       now lies with $smlnj/compiler/current.cm.  This seems to be more
363       logical.)
364    
365       Many individual files have been moved or renamed.  Some files have
366       been split into multiple files, and some "dead" files have been deleted.
367    
368    Aside from these changes to library organization, there are also changes
369    to the way the code itself is organized:
370    
371       Structure Binfile has been re-implemented in such a way that it no
372       longer needs any knowledge of the compiler.  It exclusively deals
373       with the details of binfile layout.  It no longer invokes the
374       compiler (for the purpose of creating new prospective binfile
375       content), and it no longer has any knowledge of how to interpret
376       pickles.
377    
378       Structure Compile (: COMPILE) has been stripped down to the bare
379       essentials of compilation.  It no longer deals with linking/execution.
380       The interface has been cleaned up considerably.
381    
382       Utility routines for dealing with linking and execution have been
383       moved into their own substructures.
384    
385       (The ultimate goal of these changes is to provide a light-weight
386       binfile loader/linker (at least for, e.g., stable libraries) that
387       does not require CM or the compiler to be present.)
388    
389    CM documentation has been updated to reflect the changes to library
390    organization.
391    
392    ----------------------------------------------------------------------
393    Name: Matthias Blume
394    Date: 2001/07/10 17:30:00 EDT
395    Tag: Release_110_34
396    Description:
397    
398    Minor tweak to 110.34 (re-tagged):
399    
400      - README.html file added to CVS repository
401      - runtime compiles properly under FreeBSD 3.X and 4.X
402    
403    ----------------------------------------------------------------------
404    Name: Matthias Blume
405    Date: 2001/07/10 17:30:00 EDT
406    Tag: Release_110_34
407    Description:
408    
409    New version number (110.34). New bootfiles.
410    
411    ----------------------------------------------------------------------
412    Name: Matthias Blume
413    Date: 2001/07/09 16:00:00 EDT
414    Tag: blume-20010709-more-varargs
415    Description:
416    
417    I changed the handling of varargs in ml-nlffigen again:
418    The ellipsis ... will now simply be ignored (with an accompanying warning).
419    
420    The immediate effect is that you can actually call a varargs function
421    from ML -- but you can't actually supply any arguments beyond the ones
422    specified explicitly.  (For example, you can call printf with its format
423    string, but you cannot pass additional arguments.)
424    
425    This behavior is only marginally more useful than the one before, but
426    it has the advantage that a function or, more importantly, a function
427    type never gets dropped on the floor, thus avoiding follow-up problems with
428    other types that refer to the offending one.
429    
430    ----------------------------------------------------------------------
431    Name: Matthias Blume
432    Date: 2001/07/09 11:25:00 EDT
433    Tag: blume-20010709-varargs
434    Description:
435    
436    1. ckit-lib.cm now exports structure Error
437    2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
438       with a warning accompanied by a source location.  Moreover, it
439       merely skips the offending function or type and proceeds with the
440       rest of its work.u  As a result, one can safely feed C code containing
441       "..." to ml-nlffigen.
442    3. There are some internal improvements to CM, providing slightly
443       more general string substitutions in the tools subsystem.
444    
445    ----------------------------------------------------------------------
446    Name: Matthias Blume
447    Date: 2001/06/27 15:10:00 EDT
448    Tag: blume-20010627-concur
449    Description:
450    
451    Fixed a small bug in CM's handling of parallel compilation.
452    (You could observe the bug by Control-C-interrupting an ordinary
453    CMB.make or CM.stabilize and then attaching some compile servers.
454    The result was that all of a sudden the previously interrupted
455    compilation would continue on its own.  This was because of
456    an over-optimization: CM did not bother to clean out certain queues
457    when no servers were attached "anyway", resulting in the contents
458    of these queues to grab control when new servers did get attached.)
459    
460    There is also another minor update to the CM manual.
461    
462    ----------------------------------------------------------------------
463    Name: Matthias Blume
464    Date: 2001/06/26 16:15:00 EDT
465    Tag: blume-20010626-cmdoc
466    Description:
467    
468    Minor typo fixed in CM manual (syntax diagram for libraries).
469    
470    ----------------------------------------------------------------------
471    Name: Matthias Blume
472    Date: 2001/06/25 22:55:00 EDT
473    Tag: blume-20010625-x86pc
474    Description:
475    
476    Fixed a nasty bug in the X86 assembly code that caused signal
477    handlers to fail (crash) randomly.
478    
479    ----------------------------------------------------------------------
480    Name: Matthias Blume
481    Date: 2001/06/25 12:05:00 EDT
482    Tag: blume-20010625-nlffigen
483    Description:
484    
485    This update fixes a number of minor bugs in ml-nlffigen as reported by
486    Nick Carter <nbc@andrew.cmu.edu>.
487    
488      1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
489      2. Default names for generated files are now derived from the name of
490         the C file *without its directory*.  In particular, this causes generated
491         files to be placed locally even if the C file is in some system directory.
492      3. Default names for generated signatures and structures are also derived
493         from the C file name without its directory.  This avoids silly things
494         like "structure GL/GL".
495         (Other silly names are still possible because ml-nlffigen does not do
496          a thorough check of whether generated names are legal ML identifiers.
497          When in doubt, use command line arguments to force particular names.)
498    
499    ----------------------------------------------------------------------
500    Name: Matthias Blume
501    Date: 2001/06/21 12:25:00 EDT
502    Tag: blume-20010621-eXene
503    Description:
504    
505    eXene now compiles and (sort of) works again.
506    
507    The library name (for version > 110.33) is $/eXene.cm.
508    
509    I also added an new example in src/eXene/examples/nbody.  See the
510    README file there for details.
511    
512    ----------------------------------------------------------------------
513    Name: Matthias Blume
514    Date: 2001/06/20 16:40:00 EDT
515    Tag: blume-20010620-cml
516    Description:
517    
518    CML now compiles and works again.
519    
520    Libraries (for version > 110.33):
521    
522      $cml/cml.cm            Main CML library.
523      $cml/basis.cm          CML's version of $/basis.cm.
524      $cml/cml-internal.cm   Internal helper library.
525      $cml/core-cml.cm       Internal helper library.
526      $cml-lib/trace-cml.cm  Tracing facility.
527      $cml-lib/smlnj-lib.cm  CML's version of $/smlnj-lib.cm
528    
529    The installer (config/install.sh) has been taught how to properly
530    install this stuff.
531    
532    ----------------------------------------------------------------------
533    Name: Matthias Blume
534    Date: 2001/06/19 17:55:00 EDT
535    Tag: blume-20010619-instantiate
536    Description:
537    
538    This un-breaks the fix for bug 1432.
539    (The bug was originally fixed in 110.9 but I broke it again some
540    time after that.)
541    
542    ----------------------------------------------------------------------
543    Name: Matthias Blume
544    Date: 2001/06/19 17:25:00 EDT
545    Tag: blume-20010619-signals
546    Description:
547    
548    This should (hopefully) fix the long-standing signal handling bug.
549    (The runtime system was constructing a continuation record with an
550    incorrect descriptor which would cause the GC to drop data on the floor...)
551    
552    ----------------------------------------------------------------------
553    Name: Matthias Blume
554    Date: 2001/06/15 15:05:00 EDT
555    Tag: blume-20010615-moresparc
556    Description:
557    
558    Here is a short late-hour update related to Sparc c-calls:
559    
560     -- made handling of double-word arguments a bit smarter
561    
562     -- instruction selection phase tries to collapse certain clumsily
563        constructed ML-Trees; typical example:
564    
565            ADD(ty,ADD(_,e,LI d1),LI d2)  ->  ADD(ty,e,LI(d1+d2))
566    
567        This currently has no further impact on SML/NJ since mlriscGen does
568        not seem to generate such patterns in the first place, and c-calls
569        (which did generate them in the beginning) has meanwhile been fixed
570        so as to avoid them as well.
571    
572    ----------------------------------------------------------------------
573    Name: Matthias Blume
574    Date: 2001/06/15 15:05:00 EDT
575    Tag: blume-20010615-sparc
576    Description:
577    
578    The purpose of this update is to provide an implementation of NLFFI
579    on Sparc machines.
580    
581    Here are the changes in detail:
582    
583       * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
584       the Sparc implementation of the c-calls API.
585       * The Sparc backend of SML/NJ has been modified to uniformely use %fp
586       for accessing the ML frame.  Thus, we have a real frame pointer and
587       can freely modify %sp without need for an omit-frame-ptr phase.
588       The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
589       this case.
590       * ml-nlffigen has been taught to produce code for different architectures
591       and calling conventions.
592       * In a way similar to what was done in the x86 case, the Sparc
593       backend uses its own specific extension to mltree.  (For example,
594       it needs to be able to generate UNIMP instructions which are part
595       of the calling convention.)
596       * ml-nlffi-lib was reorganized to make it more modular (in particular,
597       to make it easier to plug in new machine- and os-dependent parts).
598    
599    There are some other fairly unrelated bug fixes and cleanups as well:
600    
601       * I further hacked the .cm files for MLRISC tools (like MDLGen) so
602       that they properly share their libraries with existing SML/NJ libraries.
603       * I fixed a minor cosmetic bug in CM, supressing certain spurious
604       follow-up error messages.
605       * Updates to CM/CMB documentation.
606    
607    TODO items:
608    
609       * MLRISC should use a different register as its asmTemp on the Sparc.
610         (The current %o2 is a really bad choice because it is part of the
611         calling conventions, so things might interfere in unexpected ways.)
612    
613    ----------------------------------------------------------------------
614    Name: Matthias Blume
615    Date: 2001/06/07
616    Tag: blume-20010607-calls
617    Description:
618    
619    A number of internal changes related to C calls and calling conventions:
620    
621    1. ML-Tree CALL statements now carry a "pops" field.  It indicates the
622       number of bytes popped implicitly (by the callee).  In most cases
623       this field is 0 but on x86/win32 it is some non-zero value.  This
624       is information provided for the benefit of the "omit-frameptr" pass.
625    2. The CALL instruction on the x86 carries a similar "pops" field.
626       The instruction selection phase copies its value from the ML-Tree
627       CALL statement.
628    3. On all other architectures, the instruction selection phase checks
629       whether "pops=0" and complains if not.
630    4. The c-calls implementation for x86 now accepts two calling conventions:
631       "ccall" and "stdcall".  When "ccall" is selected, the caller cleans
632       up after the call and pops is set to 0.  For "stdcall", the caller
633       does nothing, leaving the cleanup to the callee; pops is set to
634       the number of bytes that were pushed onto the stack.
635    5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
636       distinguish between "ccall" and "stdcall".
637    6. The UNIMP instruction has been added to the supported Sparc instruction
638       set. (This is needed for implementing the official C calling convention
639       on this architecture.)
640    7. I fixed some of the .cm files under src/MLRISC/Tools to make them
641       work with the latest CM.
642    
643    ----------------------------------------------------------------------
644    Name: Matthias Blume
645    Date: 2001/06/05 15:10:00 EDT
646    Tag: blume-20010605-cm-index
647    Description:
648    
649    0. The "lambdasplit" parameter for class "sml" in CM has been documented.
650    
651    1. CM can now generate "index files".  These are human-readable files
652       that list on a per-.cm-file basis each toplevel symbol defined or
653       imported.  The location of the index file for
654       <p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
655       To enable index-file generation, set CM.Control.generate_index to true
656       or export an environment-symbol: export CM_GENERATE_INDEX=true.
657    
658       The CM manual has been updated accordingly.
659    
660    2. I made some slight modifications to the c-calls API in MLRISC.
661    
662         a) There is now a callback to support saving/restoring of
663            dedicated but caller-save registers around the actual call
664            instruction.
665         b) One can optionally specify a comment-annotation for the
666            call instruction.
667    
668    3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
669       (For example, the comment annotation shows the C prototype of
670        the function being called.)
671    
672    ----------------------------------------------------------------------
673    Name: Matthias Blume
674    Date: 2001/06/01 13:30:00 EDT
675    Tag: blume-20010601-nlffi-cleanup
676    Description:
677    
678    This is mostly a cleanup of MLFFI stuff:
679    
680       - some signature files have been put into a more exposed place
681       - the ugly 'f type parameter is gone (simplifies types tremendously!)
682       - ml-nlffigen changed accordingly
683       - tutorial updated
684    
685    Other changes:
686    
687       - author's affiliation in CM manual(s) updated
688       - some more recognized keywords added to Allen's sml.sty
689    
690    ----------------------------------------------------------------------
691    Name: Matthias Blume
692    Date: 2001/05/25 15:30:00 EDT
693    Tag: blume-20010525-iptr
694    Description:
695    
696      - put the official 110.33-README (as it appears on the ftp server) under
697        CVS
698      - fixed a small bug related to incomplete pointer types in
699        ml-nlffigen
700      - small cosmetic change to the ml-nlffi-lib's "arr" type constructor
701        (it does not need the 'f type parameter)
702    
703    ----------------------------------------------------------------------
704    Name: Matthias Blume
705    Date: 2001/05/23 14:30:00 EDT
706    Tag: Release_110_33
707    Description:
708    
709    New version number (110.33).  New bootfiles.
710    
711    ----------------------------------------------------------------------
712    Name: Matthias Blume
713    Date: 2001/05/22 18:06:00 EDT
714    Tag: blume-20010522-targets
715    Description:
716    
717    Made install.sh use file config/targets.customized if it exists, falling
718    back to config/targets if it doesn't.  This way one can have a customized
719    version of the targets file without touching the "real thing", thus
720    eliminating the constant fear of accidentally checking something bogus
721    back into the CVS repository...  (File config/targets.customized must
722    not be added to the repository!)
723    
724    ----------------------------------------------------------------------
725    Name: Matthias Blume
726    Date: 2001/05/22 16:30:00 EDT
727    Tag: blume-20010522-minitut
728    Description:
729    
730    1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
731       struct returns.
732    2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt.  This is some very
733       incomplete, preliminary documentation for NLFFI.
734    
735    ----------------------------------------------------------------------
736    Name: Matthias Blume
737    Date: 2001/05/14 11:30:00 EDT
738    Tag: blume-20010514-script
739    Description:
740    
741    Some bugs in install script fixed.
742    
743    In addition to that I also made a slight change to the NLFFI API:
744    Functors generated by ml-nlffigen now take the dynamic library as a
745    straight functor argument, not as a suspended one.  (The original
746    functor code used to force the suspension right away anyway, so there
747    was nothing gained by this complication of the interface.)
748    
749    ----------------------------------------------------------------------
750    Name: Matthias Blume
751    Date: 2001/05/11 14:35:00 EDT
752    Tag: blume-20010511-ml-nlffi
753    Description:
754    
755    I finally took the plunge and added my new FFI code to the main
756    repository.  For x86-linux it is now ready for prime-time.
757    
758    There are two new subdirectories of "src":
759    
760      - ml-nlffi-lib:
761           The utility library for programs using the FFI interface.
762           Here is the implementation of $/c.cm and its associated low-level
763           partners $/c-int.cm and $/memory.cm.
764      - ml-nlffigen:
765           A stand-alone program for generating ML glue code from C source
766           code.
767    
768    Building ml-nlffigen requires $/ckit-lib.cm.
769    
770    The config/install.sh script has been updates to do the Right Thing
771    (hopefully).
772    
773    Notice that the source tree for the C-Kit will not be put under "src"
774    but directly under the installation root directory.  (This is the
775    structure that currently exists on the CVS server when you check out
776    module "sml".)  Fortunately, config/install.sh knows about this oddity.
777    
778    Bugs: No documentation yet.
779    
780    ----------------------------------------------------------------------
781    Name: Matthias Blume
782    Date: 2001/05/09 16:35:00 EDT
783    Tag: blume-20010509-cpscontract
784    Description:
785    
786    Fixed a bug in the accounting code in cpsopt/contract.sml.  (The
787    wrapper/unwrapper elimination did not decrement usage counts and some
788    dead variables got overlooked by the dead-up logic.)
789    
790    ----------------------------------------------------------------------
791    Name: Lal George
792    Date: 2001/05/08  17:26:09 EDT
793    Tag: george-20010508-omit-frameptr
794    Description:
795    
796    Changes to implement the omit-frame-pointer optimization to support
797    raw C calls. For now, there is only support on the Intel x86, but
798    other architectures will follow as more experience is gained with this.
799    
800    
801    ----------------------------------------------------------------------
802    Name: Matthias Blume
803    Date: 2001/05/07 14:40:00 EDT
804    Tag: blume-20010507-proxies
805    Description:
806    
807    I made into "proxy libraries" all libraries that qualify for such a
808    change.  (A qualifying library is a library that has another library or
809    groups as its sole member and repeats that member's export list
810    verbatim.  A proxy library avoids this repetition by omitting its export
811    list, effectively inheriting the list that its (only) member exports.
812    See the CM manual for more explanation.)
813    The main effect is that explicit export lists for these libraries
814    do not have to be kepts in sync, making maintenance a bit easier.
815    
816    I also added copyright notices to many .cm-files.
817    
818    Last but not least, I made a new set of bootfiles.
819    
820    ----------------------------------------------------------------------
821    Name: Matthias Blume
822    Date: 2001/05/04 17:00:00 EDT
823    Tag: blume-20010504-cm-lsplit
824    Description:
825    
826    0. John merged pending changes to $/smlnj-lib.cm
827    
828    1. Allen's previous change accidentally backed out of one of Lal's
829       earlier changes.  I undid this mistake (re-introducing Lal's change).
830    
831    2. I used the new topOrder' function from graph-scc.sml (from $/smlnj-lib.cm)
832       within the compiler where applicable.  There is some code simplification
833       because of that.
834    
835    3. The "split" phase (in FLINT) is now part of the default list of phases.
836       Compiler.Control.LambdaSplitting.* can be used to globally control the
837       lambda-splitting (cross-module-inlining) engine.  In addition to that,
838       it can now also be controlled on a per-source basis: CM has been taught
839       a new tool parameter applicable to ML source files.
840    
841       - To turn lambda-splitting off completely:
842            local open Compiler.Control.LambdaSplitting in
843                val _ = set Off
844            end
845       - To make "no lambda-splitting" the global default (but allow per-source
846         overriding); this is the initial setting:
847            local open Compiler.Control.LambdaSplitting in
848                val _ = set (Default NONE)
849            end
850       - To make "lambda-splitting with aggressiveness a" the global default
851         (and allow per-source overriding):
852            local open Compiler.Control.LambdaSplitting in
853                val _ = set (Default (SOME a))
854            end
855    
856       - To turn lambda-splitting off for a given ML souce file (say: a.sml)
857         write (in the respective .cm-file):
858            a.sml (lambdasplitting:off)
859       - To turn lambda-splitting for a.sml on with minimal aggressiveness:
860            a.sml (lambdasplitting:on)
861       - To turn lambda-splitting for a.sml on with aggressiveness <a> (where
862         <a> is a decimal non-negative integer):
863            a.sml (lambdasplitting:<a>)
864       - To turn lambda-splitting for a.sml on with maximal aggressiveness:
865            a.sml (lambdasplitting:infinity)
866       - To use the global default for a.sml:
867            a.sml (lambdasplitting:default)
868         or simply
869            a.sml
870    
871    ----------------------------------------------------------------------
872    Name: Allen Leung
873    Date: 2001/05/04 01:57:00 EDT
874    Tag: leunga-20010504-sync
875    Description:
876    
877      MLRISC features.
878    
879      1. Fix to CMPXCHG instructions.
880      2. Changed RA interface to allow annotations in callbacks.
881      3. Added a new method to the stream interface to allow annotations updates.
882    
883    ----------------------------------------------------------------------
884    Name: Matthias Blume
885    Date: 2001/05/01 11:45:00 EDT
886    Tag: blume-20010501-pcedittmp
887    Description:
888    
889    Changed install.sh to use the current working directory instead of
890    /usr/tmp for a temporary file (pcedittmp).  The previous choice
891    of /usr/tmp caused trouble with MacOS X because of file premission
892    problems.
893    
894    ----------------------------------------------------------------------
895    Name: Matthias Blume
896    Date: 2001/04/20 11:10:00 EDT
897    Tag: blume-20010420-inMLflag
898    Description:
899    
900     - added vp_limitPtrMask to vproc-state.h
901       (for use by the raw-C-calls mechanism to implement proper interrupt
902        handling)
903     - made the ML compiler aware of various data-structure offsets so it
904       can generate code for accessing the vp_inML flag and vp_limitPtrMask
905     - tweaked mlriscGen.sml to have it emit interrupt-handling code for
906       raw C-calls
907    
908    ----------------------------------------------------------------------
909    Name: Lal George
910    Date: 2001/04/20 09:15:28 EDT
911    Tag: george-20010420-macosX
912    Description:
913    
914     - Changes to port to Mac OS X; Darwin.
915    
916     - In the process I found that sqrt was broken on the PPC, because the
917       fsqrt instruction is not implemented.
918    
919    ----------------------------------------------------------------------
920    Name: Matthias Blume
921    Date: 2001/04/18 12:45:00 EDT
922    Tag: blume-20010418-ccalls
923    Description:
924    
925     - fixed two off-by-4 errors in the x86-specific c-calls implementation
926       (this bug prevented structure arguments containing pointers from being
927        passed correctly)
928     - changed the raw-C-call code in mlriscGen.sml in such a way that
929       structure arguments are represented as a pointer to the beginning
930       of the structure (instead of having a series of synthesized arguments,
931       one for each structure member)
932    
933     - made makeml script's verbosity level configurable via environment
934       variable (MAKEML_VERBOSITY)
935    
936     - eliminated placeholder implementations for f32l, w16s, i16s, and f32s
937       in rawmem-x86.sml; we are now using the real thing
938    
939    ----------------------------------------------------------------------
940    Name: Matthias Blume
941    Date: 2001/03/22 16:25:00 EST
942    Tag: blume-20010322-bootfiles
943    Description:
944    
945    Created a new set of bootfiles (for your automatic installation convenience).
946    
947    ----------------------------------------------------------------------
948    Name: Matthias Blume
949    Date: 2001/03/22 15:10:00 EST
950    Tag: blume-20010322-rawmem-parcm
951    Description:
952    
953    1. All "raw memory access" primitives for the new FFI are implemented now
954       (at least on the x86).
955    2. Some further cleanup of CM's parallel make mechanism.
956    
957    ----------------------------------------------------------------------
958    Name: Matthias Blume
959    Date: 2001/03/19 17:53:00 EST
960    Tag: blume-20010319-parallel
961    Description:
962    
963    Parallel make (using compile servers) now works again.
964    
965    To this end, CM.stabilize and CMB.make have been modified to work in
966    two passes when compile servers are attached:
967       1. Compile everything, do not perform stabilization; this pass
968          uses compile servers
969       2. Stabilize everything; this pass does not use compile servers
970    If there are no compile servers, the two passes are combined into one
971    (as before).  Splitting the passes increases the inherent parallelism
972    in the dependency graph because the entire graph including all
973    libraries is available at the same time.  This, in turn, improves
974    server utilization.  The downside is that the master process will
975    have to do some extra work after compilation is done (because for
976    technical reasons it must re-read all the binfiles during stabilization).
977    
978    ----------------------------------------------------------------------
979    Name: Matthias Blume
980    Date: 2001/03/16 12:22:00 EST
981    Tag: blume-20010316-bootfiles
982    Description:
983    
984    Created a new set of bootfiles (for your automatic installation convenience).
985    
986    ----------------------------------------------------------------------
987    Name: Matthias Blume
988    Date: 2001/03/16 11:00:00 EST
989    Tag: blume-20010316-MLTREE-fixup
990    Description:
991    
992    This is a minor fixup for an (untagged) earlier commit by Allen.
993    (A file was missing).
994    
995    ----------------------------------------------------------------------
996    Name: Allen Leung
997    Date: Mon Mar  5 18:54:57 EST 2001
998    Tag: leunga-20010305-cut-support
999    
1000    1. New support for alternative control-flow in MLTREE.
1001       Currently we support
1002    
1003          FLOW_TO(CALL ...., [k1,...,kn])
1004    
1005       This is needed for 'cuts to' in C-- and try/handle-like constructs
1006       in Moby
1007    
1008       New assembler flag "asm-show-cutsto" to turn on control-flow debugging.
1009    
1010    2. Register Allocator
1011    
1012       Changes in interface [from Fermin, John]
1013    
1014    3. Alpha 8-bit SLL support [Fermin]
1015    
1016    4. All architectures
1017    
1018       A new module (ClusterExpandCopies) for expanding parallel copies.
1019    
1020    ----------------------------------------------------------------------
1021    Name: Allen Leung
1022    Date: 2001/02/27 23:07:00 EST
1023    Tag: leunga-20010227-minor-stuff
1024    
1025    1. Alpha bug fix for CMOVNE
1026    2. Handle mltree COND(..,FCMP ...,...)
1027    3. Bug fix in simplifier
1028    
1029    ----------------------------------------------------------------------
1030    Name: Matthias Blume
1031    Date: 2001/01/30 17:50:00 EST
1032    Tag: blume-20010130-sync
1033    Description:
1034    
1035    This is just a minor update to sync my devel branch with the main brach.
1036    The only visible change is the addition of some README files.
1037    
1038    ----------------------------------------------------------------------
1039    Name: Matthias Blume
1040    Date: 2001/01/12 23:30:00 JST
1041    Tag: blume-20010112-bootfiles
1042    Description:
1043    
1044    Made a new set of bootfiles that goes with the current state of the
1045    repository.
1046    
1047    ----------------------------------------------------------------------
1048    Name: Matthias Blume
1049    Date: 2001/01/12 21:20:00 JST
1050    Tag: blume-20010112-sync
1051    Description:
1052    
1053    I am just flushing out some minor changes that had accumulated in
1054    my private branch in order to sync with the main tree.  (This is
1055    mainly because I had CVS trouble when trying to merge _into_ my
1056    private branch.)
1057    
1058    Most people should be completely unaffected by this.
1059    
1060    ----------------------------------------------------------------------
1061    Name: Allen Leung
1062    Date: Thu Jan 11 21:03:00 EST 2001
1063    Tag: leunga-20010111-labexp=mltree
1064    Description:
1065    
1066    1.  Removed the type LabelExp and replace it by MLTree.
1067    2.  Rewritten mltree-simplify with the pattern matcher tool.
1068    3.  There were some bugs in alpha code generator which would break
1069        64-bit code generation.
1070    4.  Redo the tools to generate code with the
1071    5.  The CM files in MLRISC (and in src/system/smlnj/MLRISC)
1072        are now generated by perl scripts.
1073    
1074    ----------------------------------------------------------------------
1075    Name: Matthias Blume
1076    Date: 2001/01/10 21:55:00 JST
1077    Tag: blume-20010110-rcc
1078    Description:
1079    
1080    The RCC stuff now seems to work (but only on the x86).
1081    This required hacking of the c-calls interface (and -implementation) in
1082    MLRISC.
1083    
1084    Normal compiler users should be unaffected.
1085    
1086    ----------------------------------------------------------------------
1087    Name: Matthias Blume
1088    Date: 2001/01/09 01:20:00 JST
1089    Tag: blume-20010109-rcc
1090    Description:
1091    
1092    This is a fairly big patch, flushing out a large number of pending
1093    changes that I made to my development copy over the last couple of days.
1094    
1095    Of practical relevance at this moment is a workaround for a pickling
1096    bug that Allen ran into the other day.  The cause of the bug itself is
1097    still unknown and it might be hard to fix it properly, but the
1098    workaround has some merits of its own (namely somewhat reducing pickling
1099    overhead for certain libraries).  Therefore, I think this solution should
1100    be satisfactory at this time.
1101    
1102    The rest of the changes (i.e., the vast majority) has to do with my
1103    ongoing efforts of providing direct support for C function calls from
1104    ML.  At the moment there is a new primop "RAW_CCALL", typing magic
1105    in types/cproto.sml (invoked from FLINT/trans/translate.sml), a new
1106    case in the FLINT CPS datatype (RCC), changes to cps/convert.sml to
1107    translate uses of RAW_CCALL into RCC, and changes to mlriscGen.sml to
1108    handle RCC.
1109    
1110    The last part (the changes to mlriscGen.sml) are still known to be
1111    wrong on the x86 and not implemented on all other architectures.  But
1112    the infrastructure is in place. I had to change a few functor
1113    signatures in the backend to be able to route the CCalls interface
1114    from MLRISC there, and I had to specialize the mltree type (on the
1115    x86) to include the necessary extensions. (The extensions themselves
1116    were already there and redy to go in MLRISC/x86).
1117    
1118    Everything should be very happy as soon as someone helps me with
1119    mlriscGen.sml...
1120    
1121    In any case, nothing of this should matter to anyone as long as the
1122    new primop is not being used (which is going to be the case unless you
1123    find it where I hid it :). The rest of the compiler is completely
1124    unaffected.
1125    
1126    ----------------------------------------------------------------------
1127    Name: Matthias Blume
1128    Date: 2001/01/05 00:30:00 JST
1129    Tag: blume-20010105-primops
1130    Description:
1131    
1132    Added some experimental support for work that I am doing right now.
1133    These changes mostly concern added primops, but there is also a new
1134    experimental C library in the runtime system (but currently not enabled
1135    anywhere except on Linux/X86).
1136    
1137    In the course of adding primops (and playing with them), I discovered that
1138    Zhong's INL_PRIM hack (no type info for certain primops) was, in fact, badly
1139    broken.  (Zhong was very right he labeled this stuff as "major gross hack".)
1140    To recover, I made type information in INL_PRIM mandatory and changed
1141    prim.sml as well as built-in.sml accordingly.  The InLine structure now
1142    has complete, correct type information (i.e., no bottom types).
1143    
1144    Since all these changes mean that we need new binfiles, I also bumped the
1145    version number to 110.32.1.
1146    
1147    ----------------------------------------------------------------------
1148    Name: Matthias Blume
1149    Date: 2000/12/30 22:10:00 JST
1150    Tag: blume-20001230-various
1151    Description:
1152    
1153    Added proxy libraries for MLRISC and let MLRISC libraries refer
1154    to each other using path anchors.  (See CM manual for explanation.)
1155    
1156    Updated CM documentation.
1157    
1158    Fixed some bugs in CM.
1159    
1160    Implemented "proxy" libraries (= syntactic sugar for CM).
1161    
1162    Added "-quiet" option to makeml and changed runtime system accordingly.
1163    
1164    Added cleanup handler for exportML to reset timers and compiler stats.
1165    
1166    ----------------------------------------------------------------------
1167    Name: Lal George
1168    Date: 2000/12/22 22:22:58 EST 2000
1169    Tag: Release_110_32
1170    Description:
1171    
1172            Infinite precision used throughout MLRISC.
1173            see MLRISC/mltree/machine-int.sig
1174    
1175    ----------------------------------------------------------------------
1176    Name: Matthias Blume
1177    Date: 2000/12/22 23:16:00 JST
1178    Tag: blume-20001222-warn
1179    Description:
1180    
1181    Corrected wording and formatting of some CM warning message which I
1182    broke in my previous patch.
1183    
1184    ----------------------------------------------------------------------
1185    Name: Matthias Blume
1186    Date: 2000/12/22 21:20:00 JST
1187    Tag: blume-20001222-anchorenv
1188    Description:
1189    
1190    Fixed CM's handling of anchor environments in connection with CMB.make.
1191    
1192    ----------------------------------------------------------------------
1193    Name: Matthias Blume
1194    Date: 2000/12/22 13:15:00 JST
1195    Tag: blume-20001222-cleanup
1196    Description:
1197    
1198    Removed src/cm/ffi which does not (and did not) belong here.
1199    
1200    ----------------------------------------------------------------------
1201    Name: Matthias Blume
1202    Date: 2000/12/21 23:55:00 JST
1203    Tag: blume-20001221-exn
1204    Description:
1205    
1206    Probably most important: CM no longer silently swallows all exceptions
1207    in the compiler.
1208    Plus: some other minor CM changes.  For example, CM now reports some
1209    sizes for generated binfiles (code, data, envpickle, lambdapickle).
1210    
1211    ----------------------------------------------------------------------
1212    Name: Matthias Blume
1213    Date: 2000/12/15 00:01:05 JST
1214    Tag: blume-20001215-dirtool
1215    Description:
1216    
1217    - "dir" tool added.
1218    - improvements and cleanup to Tools structure
1219    - documentation updates
1220    
1221    ----------------------------------------------------------------------
1222    Name: Allen Leung
1223    Date: Thu Dec 14 03:45:24 EST 2000
1224    Description:
1225    Tag:  leunga-20001214-int-inf
1226    Description:
1227    
1228       In IntInf, added these standard functions, which are missing from our
1229    implementation:
1230    
1231        andb : int * int -> int
1232        xorb : int * int -> int
1233        orb  : int * int -> int
1234        notb : int -> int
1235         <<   : int * word -> int
1236        ~>>  : int * word -> int
1237    
1238       Not tested, I hope they are correct.
1239    
1240    ----------------------------------------------------------------------
1241    Name: Allen Leung
1242    Date: Fri Dec  8 19:23:26 EST 2000
1243    Description:
1244    Tag:  leunga-20001208-nowhere
1245    Description:
1246    
1247      Slight improvements to the 'nowhere' tool to handle OR-patterns,
1248    to generate better error messages etc.  Plus a brief manual.
1249    
1250    ----------------------------------------------------------------------
1251    Name: Lal George
1252    Date: 2000/12/08 09:54:02 EST 2000
1253    Tag: Release_110_31
1254    Description:
1255    
1256    - Version 110.31
1257    ----------------------------------------------------------------------
1258    Name: Allen Leung
1259    Date: Thu Dec  7 22:01:04 EST 2000
1260    Tag:  leunga-20001207-cell-monster-hack
1261    Description:
1262    
1263    Major MLRISC internal changes.  Affect all clients.
1264    Summary:
1265    
1266    1.  Type CELLS.cell = int is now replaced by a datatype.
1267        As a result, the old regmap is now gone.  Almost all interfaces
1268        in MLRISC change as a consequence.
1269    
1270    2.  A new brand version of machine description tool (v3.0) that generates
1271        modules expecting the new interface.  The old version is removed.
1272    
1273    3.  The RA interface has been further abstracted into two new functors.
1274        RISC_RA and X86RA.  These functors have much simpler interfaces.
1275        [See also directory MLRISC/demo.]
1276    
1277    4.  Some other new source->source code generation tools are available:
1278    
1279        a. MLRISC/Tools/RewriteGen -- generate rewriters from rules.
1280        b. MLRISC/Tools/WhereGen -- expands conditional pattern matching rules.
1281           I use this tool to generate the peephole optimizers---with the new
1282           cell type changes, peephole rules are becoming difficult to write
1283           without conditional pattern matching.
1284    
1285    5.  More Intmap -> IntHashTable change.  Previous changes by Matthias didn't
1286        cover the entire MLRISC source tree so many things broke.
1287    
1288    6.  CM files have been moved to the subdirectory MLRISC/cm.
1289        They are moved because there are a lot of them and they clutter up the
1290        root dir.
1291    
1292    7.  More detailed documentation to come...
1293    
1294        NOTE: To rebuild from 110.30 (ftp distribution), you'll have to do
1295        a makeml -rebuild first.  This is because of other other
1296        changes that Matthias has made (see below).
1297    
1298    
1299    ----------------------------------------------------------------------
1300    Name: Matthias Blume
1301    Date: 2000/11/30 23:12:00 JST
1302    Tag: blume-20001130-filereorg
1303    Description:
1304    
1305    Some manual updates and some file reorganizations in CM.
1306    
1307    ----------------------------------------------------------------------
1308    Name: Matthias Blume
1309    Date: 2000/11/24 17:45:00 JST
1310    Tag: blume-20001124-link
1311    Description:
1312    
1313    Drastically improved link traversal code for the case that the dynamic
1314    value was already loaded at bootstrap time.  As a result, CM and CMB
1315    now both load blazingly fast -- even on a very slow machine.  Also,
1316    memory consumption has been further reduced by this.
1317    
1318    Warning: The format of the PIDMAP file has changed.  THerefore, to
1319    bootstrap you have to do this:
1320    
1321    1. Run CMB.make
1322    2. Make a symbolic link for the boot directory:
1323         ln -s sml.boot.ARCH-OS xxx
1324    3. "Rebuild" the boot directory:
1325         ./makeml -boot xxx -rebuild sml ; rm xxx
1326    4. Boot normally:
1327          ./makeml
1328    
1329    ----------------------------------------------------------------------
1330    Name: Matthias Blume
1331    Date: 2000/11/21 21:20:00 JST
1332    Tag: blume-20001121-tools
1333    Description:
1334    
1335    Continued hacking on autoloading problem -- with success this time.
1336    Also changed tool-plugin mechanism.  See new CM manual.
1337    
1338    ----------------------------------------------------------------------
1339    Name: Matthias Blume
1340    Date: 2000/11/19 14:30:00 JST
1341    Tag:  blume-20001119-autoload
1342    Description:
1343    
1344    Some hacking to make autoloading faster.  Success for CMB, no success
1345    so far for CM.  There is a reduced structure CM' that autoloads faster.
1346    (This is a temporary, non-documented hack to be eliminated again when
1347    the general problem is solved.)
1348    
1349    ----------------------------------------------------------------------
1350    Name: Matthias Blume
1351    Date: 2000/11/17 14:10:00 JST
1352    Tag: blume-20001117-pickle-lib
1353    Description:
1354    
1355    1. Eliminated comp-lib.cm
1356    2. Made pickle-lib.cm
1357    3. Eliminated all uses of intset.sml (from comp-lib.cm)
1358    4. Replaced all uses of intmap.{sig,sml} (from comp-lib.cm) with
1359       equivalent constructs from smlnj-lib.cm (INtHashTable).
1360    5. Point 4. also goes for those uses of intmap.* in MLRISC.
1361       Duplicated intmap modules thrown out.
1362    6. Hunted down all duplicated SCC code and replaced it with
1363       equivalent stuff (GraphSCCFn from smlnj-lib.cm).
1364    7. Rewrote Feedback module.
1365    8. Moved sortedlist.sml into viscomp-lib.cm.  Eventually it
1366       should be thrown out and equivalent modules from smlnj-lib.cm
1367       should be used (IntRedBlackSet, IntListSet, ...).
1368    
1369    Confirmed that compiler compiles to fixpoint.
1370    
1371    ----------------------------------------------------------------------
1372    Name: Allen Leung
1373    Date: 2000/11/10 18:00:00
1374    Tag: leunga-20001110-new-x86-fp
1375    
1376    A new x86 floating point code generator has been added.
1377    By default this is turned off.  To turn this on, do:
1378    
1379        CM.autoload "$smlnj/compiler.cm";
1380        Compiler.Control.MLRISC.getFlag "x86-fast-fp" := true;
1381    
1382    Changes:
1383    
1384    1.  Changed FTAN to FPTAN so that the assembly output is correct.
1385    2.  Changed the extension callback for FTANGENT to generate:
1386    
1387              fptan
1388              fstp  %st(0)
1389        instead of
1390              fptan
1391              fstpl ftempmem
1392    
1393    3.  Numerous assembly fixes for x86.
1394    
1395    5.  Cleaned up the machine code output module x86/x86MC.sml and added
1396        support for a whole bunch of instructions and addressing modes:
1397    
1398          fadd/fsub/fsubr/fmul/fdiv/fdivr  %st, %st(n)
1399          faddp/fsubp/fsubrp/fmulp/fdivp/fdivrp  %st, %st(n)
1400          fadd/fsub/fsubr/fmul/fdiv/fdivr  %st(n), %st
1401          fiadd/fisub/fisubr/fimul/fidiv/fidivr mem
1402          fxch %st(n)
1403          fld %st(n)
1404          fst %st(n)
1405          fst mem
1406          fstp %st(n)
1407          fucom %st(n)
1408          fucomp %st(n)
1409    
1410        All these are now generated when the fast fp mode is turned on.
1411    
1412    6.  Removed the dedicated registers %st(0), ..., %st(7) from X86CpsRegs
1413    
1414    ----------------------------------------------------------------------
1415    Name: Matthias Blume
1416    Date: 2000/11/09 11:20:00 JST
1417    Tag: blume-20001109-scc
1418    Description:
1419    
1420    Eliminated some code duplication:
1421    
1422    1. Added "where" clause to GraphSCCFn in SML/NJ Library.
1423       (Otherwise the functor is useless.)
1424    2. Used GraphSCCFn where SCCUtilFun was used previously.
1425    3. Got rid of SCCUtilFun (in comp-lib.cm).
1426    
1427    ----------------------------------------------------------------------
1428    Name: Lal George
1429    Date: 2000/11/06 09:02:21 EST 2000
1430    Tag: Release_110_30
1431    Description:
1432    
1433    - Version 110.30
1434    ----------------------------------------------------------------------
1435    Name: Matthias Blume
1436    Date: 2000/11/04 14:45:00
1437    Tag: blume-20001104-mlbuild
1438    Description:
1439    
1440    - Made ml-build faster on startup.
1441    - Documentation fixes.
1442    
1443    ----------------------------------------------------------------------
1444    Name: Matthias Blume
1445    Date: 2000/11/02 17:00:00 JST
1446    Tag: blume-20001102-condcomp
1447    Description:
1448    
1449    - Small tweaks to pickler -- new BOOTFILES!
1450    - Version bumped to 110.29.2.
1451    - Added conditional compilation facility to init.cmi (see comment there).
1452    ----------------------------------------------------------------------
1453    Name: Allen Leung
1454    Date: 2000/10/23 19:31:00
1455    Tag: leunga-20001023-demo-ra
1456    
1457    1. Minor RA changes that improves spilling on x86 (affects Moby and C-- only)
1458    2. Test programs for the graph library updated
1459    3. Some new MLRISC demo programs added
1460    
1461    ----------------------------------------------------------------------
1462    Name: Matthias Blume
1463    Date: 2000/08/31 22:15:00 JST
1464    Tag: blume-20001017-errmsg
1465    Description:
1466    
1467    More error message grief: Where there used to be no messages, there
1468    now were some that had bogus error regions.  Fixed.
1469    
1470    ----------------------------------------------------------------------
1471    Name: Matthias Blume
1472    Date: 2000/08/31 17:30:00 JST
1473    Tag: blume-20001017-v110p29p1
1474    Description:
1475    
1476    I made a version 110.29.1 with new bootfiles.
1477    
1478    Changes:  Modified pickler/unpickler for faster and leaner unpickling.
1479              CM documentation changes and a small bugfix in CM's error reporting.
1480    
1481    ----------------------------------------------------------------------
1482    Name: Lal George
1483    Date: 2000/09/27 14:42:35 EDT
1484    Tag: george-20000927-nodestatus
1485    Description:
1486    
1487    Changed the type of the nodestatus, so that:
1488    
1489            SPILLED(~1)             is now SPILLED
1490            SPILLED(m) where m>=0   is now MEMREG(m)
1491            SPILLED(s) where s<~1   is now SPILL_LOC(~s)
1492    
1493    ----------------------------------------------------------------------
1494    Name: Matthias Blume
1495    Date: 2000/09/07 14:45:00 JST
1496    Tag: blume-20000907-cmerrmsg
1497    Description:
1498    
1499    Small tweak to CM to avoid getting ML syntax error messages twice.
1500    
1501    ----------------------------------------------------------------------
1502    Name: Matthias Blume
1503    Date: 2000/08/31 18:00:00 JST
1504    Tag: blume-20000831-cvsbootfiles
1505    Description:
1506    
1507    New URL for boot files (because the 110.29 files on the BL server do
1508    now work correctly with my updated install scripts for yacc and lex).
1509    
1510    ----------------------------------------------------------------------
1511    Name: Matthias Blume
1512    Date: 2000/08/08 12:33:00 JST
1513    Tag: blume-20000808-manual
1514    Description:
1515    
1516    Tiny update to CM manual.
1517    
1518    ----------------------------------------------------------------------
1519    Name: Allen Leung
1520    Date: 2000/08/7 19:31:00
1521    Tag: leunga-20000807-a-whole-bunch-of-stuff
1522    
1523      Moby, C--, SSA, x86, machine descriptions etc.  Should only affect C--
1524    and Mobdy.
1525    
1526    1.  x86
1527    
1528       a.  Fixes to peephole module by John and Dan.
1529       b.  Assembly fix to SETcc by Allen.
1530       c.  Fix to c-call by John.
1531       d.  Fix to spilling by John.  (This one deals with the missing FSTPT case)
1532       e.  Instruction selection optimization to SETcc as suggested by John.
1533    
1534           For example,
1535    
1536            MV(32, x, COND(32, CMP(32, LT, a, b), LI 1, LI 0))
1537    
1538           should generate:
1539    
1540            MOVL a, x
1541            SUBL b, x
1542            SHRL 31, x
1543    
1544    2.  IR stuff
1545    
1546         A bunch of new DJ-graph related algorithms added.  These
1547         speed up SSA construction.
1548    
1549    3.  SSA + Scheduling
1550    
1551         Added code for SSA and scheduling to the repository
1552    
1553    ----------------------------------------------------------------------
1554    Name: Lal George
1555    Date: 2000/07/27 11:53:14 EDT
1556    
1557    Tag: lal-20000727-linux-ppc
1558    Description:
1559    
1560     Made changes to support Linux PPC.
1561     p.s. I have confirmation that the 110.29 boot files work fine.
1562    
1563    ----------------------------------------------------------------------
1564    Name: Matthias Blume
1565    Date: 2000/07/27 17:40:00 JST
1566    Tag: blume-20000727-scripts
1567    Description:
1568    
1569    !!!! WARNING !!!!
1570    You must recompile the runtime system!
1571    !!!! WARNING !!!!
1572    
1573    This is basically another round of script-enhancements:
1574    
1575    1. sml, ml-build, and ml-makedepend accept options -D and -U to define
1576       and undefine CM preprocessor symbols.
1577    
1578    2. ml-build avoids generating a new heap image if it finds that the
1579       existing one is still ok.  (The condition is that no ML file had to
1580       be recompiled and all ML files are found to be older that the heap
1581       file.)
1582    
1583       To make this work smoothly, I also hacked the runtime system as
1584       well as SMLofNJ.SysInfo to get access to the heap image suffix
1585       (.sparc-solaris, ...) that is currently being used.
1586    
1587       Moreover, the signature of CM.mk_standalone has changed.  See the
1588       CM manual.
1589    
1590    3. ml-makedepend accepts additional options -n, -a, and -o.  (See the
1591       CM manual for details.)
1592    
1593    4. More CM manual updates:
1594        - all of the above has been documented.
1595        - there is now a section describing the (CM-related) command line
1596          arguments that are accepted by the "sml" command
1597    
1598    ----------------------------------------------------------------------
1599    Name: Matthias Blume
1600    Date: 2000/07/25 16:20:00 JST
1601    Tag: blume-20000725-makedepend
1602    Description:
1603    
1604    Added a script called ml-makedepend.  This can be used in makefiles
1605    for Unix' make in a way very similar to the "makedepend" command for
1606    C.
1607    
1608    The script internally uses function CM.sources.
1609    
1610    Synopsis:
1611    
1612        ml-makedepend [-f makefile] cmfile targetname
1613    
1614    The default for the makefile is "makefile" (or "Makefile" should
1615    "makefile" not exist).
1616    
1617    ml-makedepend adds a cmfile/targetname-specific section to this
1618    makefile (after removing the previous version of this section).  The
1619    section contains a single dependency specification with targetname on
1620    the LHS (targetname is an arbitrary name), and a list of files derived
1621    from the cmfile on the RHS.  Some of the files on the RHS are
1622    ARCH/OPSYS-specific.  Therefore, ml-makedepend inserts references to
1623    "make" variables $(ARCH) and $(OPSYS) in place of the corresponding
1624    path names.  The makefile writer is responsible for making sure that
1625    these variables have correct at the time "make" is invoked.
1626    
1627    ----------------------------------------------------------------------
1628    Name: Matthias Blume
1629    Date: 2000/07/22 23:30:00 JST
1630    Tag: blume-20000722-urlupdate
1631    Description:
1632    
1633    Changed BOOT and config/srcarchiveurl to point to BL server:
1634    
1635        ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.29/
1636    
1637    ----------------------------------------------------------------------
1638    Name: Matthias Blume
1639    Date: 2000/07/18 18:00:00 JST
1640    Tag: blume-20000718-Version_110_29
1641    Description:
1642    
1643    1. Updated src/compiler/TopLevel/main/version.sml to version 110.29
1644    
1645    2. Updated config/version to 110.29
1646    
1647    3. Updated config/srcarchiveurl
1648    
1649    3. New boot files!
1650       ftp://ftp.cs.princeton.edu/pub/people/blume/sml/110.29-autofetch
1651    
1652    ----------------------------------------------------------------------
1653    Name: Matthias Blume
1654    Date: 2000/07/11 13:58:00 JST
1655    Tag: blume-20000711-doctypo
1656    Description:
1657    
1658    Fixed a few typos in CM manual.
1659    
1660    ----------------------------------------------------------------------
1661    Name: Allen Leung
1662    Date: 2000/06/15 00:38:00
1663    Tag: leunga-20000704-sparc-x86
1664    
1665    1. x86 peephole improvement sp += k; sp -= k => nop  [from John]
1666    2. fix to x86 RET bug [found by Dan Grossman]
1667    3. sparc assembly bug fix for ticc instructions [found by Fermin]
1668    
1669       Affects c-- and moby only
1670    
1671    ----------------------------------------------------------------------
1672    Name: Matthias Blume
1673    Date: 2000/07/04 15:26:00
1674    Tag: blume-20000704-trigger
1675    Description:
1676    
1677    1. Improvements to CM manual.
1678    2. SMLofNJ.Internals.BTrace.trigger reinstated as an alternative way
1679       of getting a back-trace.  The function, when called, raises an
1680       internal exception which explicitly carries the full back-trace history,
1681       so it is unaffected by any intervening handle-raise pairs ("trivial"
1682       or not).  The interactive loop will print that history once it arrives
1683       at top level.
1684       Short of having all exceptions implicitly carry the full history, the
1685       recommended way of using this facility is:
1686         - compile your program with instrumentation "on"
1687         - run it, when it raises an exception, look at the history
1688         - if the history is "cut off" because of some handler, go and modify
1689           your program so that it explicitly calls BTrace.trigger
1690         - recompile (still instrumented), and rerun; look at the full history
1691    
1692    ----------------------------------------------------------------------
1693    Name: Matthias Blume
1694    Date: 2000/07/03 15:36:00 JST
1695    Tag: blume-20000702-manual
1696    Description:
1697    
1698    Small corrections and updates to CM manual.
1699    
1700    ----------------------------------------------------------------------
1701    Name: Matthias Blume
1702    Date: 2000/06/29 16:04:00 JST
1703    Tag: blume-20000629-yacctool
1704    Description:
1705    
1706    Changes:
1707    
1708    1. Class "mlyacc" now takes separate arguments to pass options to
1709       generated .sml- and .sig-files independently.
1710    2. Corresponding CM manual updates.
1711    3. BTrace module now also reports call sites.  (However, for loop clusters
1712       it only shows from where the cluster was entered.)  There are associated
1713       modifications to core.sml, internals.{sig,sml}, btrace.sml, and btimp.sml.
1714    
1715    ----------------------------------------------------------------------
1716    Name: Matthias Blume
1717    Date: 2000/06/27 16:51:00 JST
1718    Tag: blume-20000627-noweb
1719    Description:
1720    
1721    Changes:
1722    
1723     1. Implemented "subdir" and "witness" options for noweb tool.
1724        This caused some slight internal changes in CM's tool implementation.
1725     2. Fixed bug in "tool plugin" mechanism.  This is essentially cleaning
1726        some remaining issues from earlier path anchor changes.
1727     3. Updated CM manual accordingly.
1728    
1729     4. Changed implementation of back-tracing so that I now consider it
1730        ready for prime-time.
1731    
1732        In particular, you don't have to explicitly trigger the back-trace
1733        anymore.  Instead, if you are running BTrace-instrumented code and
1734        there is an uncaught exception (regardless of whether or not it was
1735        raised in instrumented code), the top-level evalloop will print
1736        the back-trace.
1737    
1738        Features:
1739    
1740          - Instrumented and uninstrumented code work together seemlessly.
1741            (Of course, uninstrumented code is never mentioned in actual
1742             back-traces.)
1743    
1744          - Asymptotic time- and space-complexity of instrumented code is
1745            equal to that of uninstrumented code.  (This means that
1746            tail-recursion is preserved by the instrumentation phase.)
1747    
1748          - Modules whose code has been instrumented in different sessions
1749            work together without problem.
1750    
1751          - There is no penalty whatsoever on uninstrumented code.
1752    
1753          - There is no penalty on "raise" expressions, even in
1754            instrumented code.
1755    
1756        A potential bug (or perhaps it is a feature, too):
1757    
1758          A back-trace reaches no further than the outermost instrumented
1759          non-trivial "raise".  Here, a "trivial" raise is one that is the
1760          sole RHS of a "handle" rule.  Thus, back-traces reach trough
1761    
1762               <exp> handle e => raise e
1763    
1764          and even
1765    
1766               <exp> handle Foo => raise Bar
1767    
1768          and, of course, through
1769    
1770               <exp> handle Foo => ...
1771    
1772         if the exception was not Foo.
1773    
1774         Back-traces always reach right through any un-instrumented code
1775         including any of its "handle" expressions, trivial or not.
1776    
1777       To try this out, do the following:
1778    
1779         - Erase all existing binfiles for your program.
1780           (You may keep binfiles for those modules where you think you
1781            definitely don't need back-tracing.)
1782         - Turn on back-trace instrumentation:
1783              SMLofNJ.Internals.BTrace.mode (SOME true);
1784         - Recompile your program.  (I.e., run "CM.make" or "use".)
1785         - You may now turn instrumentation off again (if you want):
1786              SMLofNJ.Internals.BTrace.mode (SOME false);
1787         - Run your program as usual.  If it raises an exception that
1788           reaches the interactive toplevel, then a back-trace will
1789           automatically be printed.  After that, the toplevel loop
1790           will print the exception history as usual.
1791    
1792    ----------------------------------------------------------------------
1793    Name: Matthias Blume
1794    Date: 2000/06/26 09:56:46 JST
1795    Tag: blume-20000626-setup
1796    Description:
1797    
1798    CM: - setup-parameter to "sml" added; this can be used to run arbitrary
1799          ML code before and after compiling a file (e.g., to set compiler
1800          flags)
1801    
1802    Compiler: - improved btrace API (in core.sml, internals.{sig,sml})
1803              - associated changes to btrace.sml (BTrace instrumentation pass)
1804              - cleaner implementation of btimp.sml (BTrace tracing and report
1805                module)
1806    
1807    CM manual: * new path encoding documented
1808               * description of setup-parameter to "sml" added
1809    
1810    The biggest user-visible change to back-tracing is that it is no
1811    longer necessary to compile all traced modules within the same
1812    session.  (This was a real limitation.)
1813    
1814    ----------------------------------------------------------------------
1815    Name: Matthias Blume
1816    Date: 2000/06/24 12:40:00 JST
1817    Tag: blume-20000624-startup
1818    Description:
1819    
1820    Fixes startup slowdown problem.  (I was calling SrcPath.sync a _tad_
1821    bit too often -- to put it mildly. :)
1822    
1823    ----------------------------------------------------------------------
1824    Name: Matthias Blume
1825    Date: 2000/06/23 18:20:00 JST
1826    Tag: blume-20000623-btrace
1827    Description:
1828    
1829    This updates adds a backtrace facility to aid programmers in debugging
1830    their programs.  This involves the following changes:
1831    
1832    1. Module system/smlnj/init/core.sml (structure _Core) now has hooks for
1833       keeping track of the current call stack.  When programs are compiled
1834       in a special mode, the compiler will insert calls to these hooks
1835       into the user program.
1836       "Hook" means that it is possible for different implementations of
1837       back-tracing to register themselves (at different times).
1838    
1839    2. compiler/MiscUtil/profile/btrace.sml implements the annotation phase
1840       as an Absyn.dec->Absyn.dec rewrite.  Normally this phase is turned off.
1841       It can be turned on using this call:
1842         SMLofNJ.Internals.BTrace.mode (SOME true);
1843       Turning it off again:
1844         SMLofNJ.Internals.BTrace.mode (SOME false);
1845       Querying the current status:
1846         SMLofNJ.Internals.BTrace.mode NONE;
1847       Annotated programs are about twice as big as normal ones, and they
1848       run a factor of 2 to 4 slower with a dummy back-trace plugin (one
1849       where all hooks do nothing).  The slowdown with a plugin that is
1850       actually useful (such as the one supplied by default) is even greater,
1851       but in the case of the default plugin it is still only an constant
1852       factor (amortized).
1853    
1854    3. system/Basis/Implementation/NJ/internals.{sig,sml} have been augmented
1855       with a sub-structure BTrace for controlling back-tracing.  In particular,
1856       the above-mentioned function "mode" controls whether the annotation
1857       phase is invoked by the compiler.  Another important function is
1858       "trigger": when called it aborts the current execution and causes
1859       the top-level loop to print a full back-trace.
1860    
1861    4. compiler/MiscUtil/profile/btimp.sml is the current default plugin
1862       for back-tracing.  It keeps track of the dynamic call stack and in
1863       addition to that it keeps a partial history at each "level" of that
1864       stack.  For example, if a tail-calls b, b tail-calls c, and c tail-calls
1865       d and b (at separate times, dynamically), then the report will show:
1866    
1867       GOTO   d
1868             /c
1869       GOTO  \b
1870       CALL   a
1871    
1872       This shows that there was an initial non-tail call of a, then a
1873       tail-call to b or c, looping behavior in a cluster of functions that
1874       consist of b and c, and then a goto from that cluster (i.e., either from
1875       b or from c) to d.
1876    
1877       Note that (depending on the user program) the amount of information
1878       that the back-trace module has to keep track of at each level is bounded
1879       by a constant.  Thus, the whole implementation has the same asymptotical
1880       complexity as the original program (both in space and in time).
1881    
1882    5. compiler/TopLevel/interact/evalloop.sml has been modified to
1883       handle the special exception SMLofNJ.Internals.BTrace.BTrace
1884       which is raised by the "trigger" function mentioned above.
1885    
1886    Notes on usage:
1887    
1888    - Annotated code works well together with unannotated code:
1889    Unannotated calls simply do not show up at all in the backtrace.
1890    
1891    - It is not a good idea to let modules that were annotated during
1892    different sessions run at the same time.  This is because the compiler
1893    chooses small integers to identify individual functions, and there
1894    will be clashes if different modules were compiled in separate sessions.
1895    (Nothing will crash, and you will even be told about the clashes, but
1896    back-trace information will in general not be useful.)
1897    
1898    - Back-tracing can be confused by callcc and capture.
1899    
1900    - The only way of getting a back-trace right now is to explicitly
1901    invoke the "trigger" function from your user program.  Eventually, we
1902    should make every exception carry back-trace information (if
1903    available).  But since this creates more overhead at "raise"-time
1904    (similar to the current exnHistory overhead), I have not yet
1905    implemented this.  (The implementation will be rather easy.)  With
1906    exceptions carrying back-trace information, this facility will be even
1907    more useful because users don't need to modify their programs...
1908    
1909    - While it is possible to compile the compiler with back-trace
1910    annotations turned on (I did it to get some confidence in
1911    correctness), you must make absolutely sure that core.sml and
1912    btimp.sml are compiled WITHOUT annotation!  (core.sml cannot actually
1913    be compiled with annotation because there is no core access yet, but
1914    if you compile btimp.sml with annotation, then the system will go into
1915    an infinite recursion and crash.)
1916    Since CM currently does not know about BTrace, the only way to turn
1917    annotations on and off for different modules of the compiler is to
1918    interrupt CMB.make, change the settings, and re-invoke it.  Of course,
1919    this is awkward and clumsy.
1920    
1921    Sample sessions:
1922    
1923    Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
1924    - SMLofNJ.Internals.BTrace.mode (SOME true);
1925    [autoloading]
1926    [autoloading done]
1927    val it = false : bool
1928    - structure X = struct
1929    -     fun main n = let
1930    -         fun a (x, 0) = d x
1931    -           | a (x, n) = b (x, n - 1)
1932    -         and b (x, n) = c (x, n)
1933    -         and c (x, n) = a (x, n)
1934    -         and d x = e (x, 3)
1935    -         and e (x, 0) = f x
1936    -           | e (x, n) = e (x, n - 1)
1937    -         and f 0 = SMLofNJ.Internals.BTrace.trigger ()
1938    -           | f n = n * g (n - 1)
1939    -         and g n = a (n, 3)
1940    -     in
1941    -         f n
1942    -     end
1943    - end;
1944    structure X : sig val main : int -> int end
1945    - X.main 3;
1946    *** BACK-TRACE ***
1947    GOTO   stdIn:4.2-13.20: X.main[2].f
1948    GOTO-( stdIn:4.2-13.20: X.main[2].e
1949    GOTO   stdIn:4.2-13.20: X.main[2].d
1950         / stdIn:4.2-13.20: X.main[2].a
1951         | stdIn:4.2-13.20: X.main[2].b
1952    GOTO-\ stdIn:4.2-13.20: X.main[2].c
1953    CALL   stdIn:4.2-13.20: X.main[2].g
1954    GOTO   stdIn:4.2-13.20: X.main[2].f
1955    GOTO-( stdIn:4.2-13.20: X.main[2].e
1956    GOTO   stdIn:4.2-13.20: X.main[2].d
1957         / stdIn:4.2-13.20: X.main[2].a
1958         | stdIn:4.2-13.20: X.main[2].b
1959    GOTO-\ stdIn:4.2-13.20: X.main[2].c
1960    CALL   stdIn:4.2-13.20: X.main[2].g
1961    GOTO   stdIn:4.2-13.20: X.main[2].f
1962    GOTO-( stdIn:4.2-13.20: X.main[2].e
1963    GOTO   stdIn:4.2-13.20: X.main[2].d
1964         / stdIn:4.2-13.20: X.main[2].a
1965         | stdIn:4.2-13.20: X.main[2].b
1966    GOTO-\ stdIn:4.2-13.20: X.main[2].c
1967    CALL   stdIn:4.2-13.20: X.main[2].g
1968    GOTO   stdIn:4.2-13.20: X.main[2].f
1969    CALL   stdIn:2.15-17.4: X.main[2]
1970    -
1971    
1972    (Note that because of a FLINt bug the above code currently does not
1973    compile without BTrace turned on.)
1974    
1975    Here is another example, using my modified Tiger compiler:
1976    
1977    Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
1978    - SMLofNJ.Internals.BTrace.mode (SOME true);
1979    [autoloading]
1980    [autoloading done]
1981    val it = false : bool
1982    - CM.make "sources.cm";
1983    [autoloading]
1984    ...
1985    [autoloading done]
1986    [scanning sources.cm]
1987    [parsing (sources.cm):parse.sml]
1988    [creating directory CM/SKEL ...]
1989    [parsing (sources.cm):tiger.lex.sml]
1990    ...
1991    [wrote CM/sparc-unix/semant.sml]
1992    [compiling (sources.cm):main.sml]
1993    [wrote CM/sparc-unix/main.sml]
1994    [New bindings added.]
1995    val it = true : bool
1996    - Main.compile ("../testcases/merge.tig", "foo.out");
1997    *** BACK-TRACE ***
1998    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trvar
1999    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
2000    CALL   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
2001    GOTO   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
2002    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
2003    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
2004    CALL   lib/semant.sml:488.3-505.6: SemantFun[2].transDec.trdec[2].transBody[2]
2005         / lib/semant.sml:411.65-543.8: SemantFun[2].transDec
2006    CALL-\ lib/semant.sml:413.2-540.9: SemantFun[2].transDec.trdec[2]
2007    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
2008    CALL   lib/semant.sml:8.52-558.4: SemantFun[2].transProg[2]
2009    CALL   main.sml:1.18-118.4: Main.compile[2]
2010    -
2011    
2012    ----------------------------------------------------------------------
2013    Name: Matthias Blumen
2014    Date: 2000/06/21 18:00:00 JST
2015    Tag: blume-20000621-manual
2016    Description:
2017    
2018    CM manual update: Path environments documented.
2019    
2020    ----------------------------------------------------------------------
2021    Name: Matthias Blume
2022    Date: 2000/06/19 13:40:00
2023    Tag: blume-20000619-manual
2024    Description:
2025    
2026    CM manual and system/README update.  This only covers the fact that
2027    there are no more implicit anchors.  (Path environments and the "bind"
2028    option to "cm" have yet to be documented.)
2029    
2030    ----------------------------------------------------------------------
2031    Name: Matthias Blume
2032    Date: 2000/06/19 11:05:00 JST
2033    Tag: blume-20000619-chdir-bugfix
2034    Description:
2035    
2036    Fixed a bug in new SrcPath module that sometimes led to a bad chDir call.
2037    
2038    ----------------------------------------------------------------------
2039    Name: Matthias Blume
2040    Date: 2000/06/18 22:00:10 JST
2041    Tag: blume-20000618-implicit-anchors-really-gone
2042    Description:
2043    
2044    I updates the previous HISTORY entry where I forgot to mention that
2045    implicit anchors are no longer with us.
2046    
2047    The current update also gets rid of the (now useless) controller
2048    CM.Control.implicit_anchors.
2049    
2050    ----------------------------------------------------------------------
2051    Name: Matthias Blume
2052    Date: 2000/06/16 17:30:00 JST
2053    Tag: blume-20000616-anchorenv
2054    Description:
2055    
2056    This patch implements the long anticipated (just kidding :) "anchor
2057    environment" mechanism.  In the course of doing this, I also
2058    re-implemented CM's internal "SrcPath" module from scratch.  The new
2059    one should be more robust in certain boundary cases.  In any case, it
2060    is a lot cleaner than its predecessor (IMHO).
2061    
2062    This time, although there is yet another boot file format change, I
2063    kept the unpickler backward-compatible.  As a result, no new bootfiles
2064    are necessary and bootstrapping is straightforward.  (You cannot read
2065    new bootfiles into an old system, but the other way around is no
2066    problem.)
2067    
2068    Visible changes:
2069    
2070    ** 0. Implicit path anchors (without the leading $-symbol) are no
2071    longer recognized at all. This means that such path names are not
2072    illegal either.  For example, the name basis.cm simply refers to a
2073    local file called "basis.cm" (i.e, the name is an ordinary path
2074    relative to .cm-files directory).  Or, to put it differently, only
2075    names that start with $ are anchored paths.
2076    
2077    ** 1. The $<singlearc> abbreviation for $/<singlearc> has finally
2078    vanished.
2079    
2080    John (Reppy) had critizised this as soon as I originally proposed and
2081    implemented it, but at that time I did not really deeply believe
2082    him. :) Now I came full-circle because I need the $<singlearc> syntax
2083    in another place where it cannot be seen as an abbreviation for
2084    $/<singlearc>.  To avoid the confusion, $<singlearc> now means what it
2085    seems to mean (i.e., it "expands" into the corresponding anchor
2086    value).
2087    
2088    However, when paths are used as members in CM description files, it
2089    continues to be true that there must be at least another arc after the
2090    anchor.  This is now enforced separately during semantic analysis
2091    (i.e., from a lexical/syntactical point of view, the notation is ok.)
2092    
2093    ** 2. The "cm" class now accepts an option "bind".  The option's value
2094    is a sub-option list of precisely two items -- one labeled "anchor"
2095    and the other one labeled "value".  As you might expect, "anchor" is
2096    used to specify an anchor name to be bound, and "value" specifies what
2097    the anchor is being bound to.
2098    
2099    The value must be a directory name and can be given in either standard
2100    syntax (including the possibility that it is itself an anchored path)
2101    or native syntax.
2102    
2103    Examples:
2104    
2105       foo.cm (bind:(anchor:bar value:$mystuff/bar))
2106       lib.cm (bind:(anchor:a value:"H:\\x\\y\\z"))  (* only works under windows *)
2107    
2108    and so on.
2109    
2110    The meaning of this is that the .cm-file will be processed with an
2111    augmented anchor environment where the given anchor(s) is/are bound to
2112    the given values(s).
2113    
2114    The rationale for having this feature is this: Suppose you are trying
2115    to use two different (already stable) libraries a.cm and b.cm (that
2116    you perhaps didn't write yourself).  Further, suppose each of these
2117    two libraries internally uses its own auxiliary library $aux/lib.cm.
2118    Normally you would now have a problem because the anchor "lib" can not
2119    be bound to more than one value globally.  Therefore, the project that
2120    uses both a.cm and b.cm must locally redirect the anchor to some other
2121    place:
2122    
2123       a.cm (bind:(anchor:lib value:/usr/lib/smlnj/a-stuff))
2124       b.cm (bind:(anchor:lib value:/usr/lib/smlnj/b-stuff))
2125    
2126    This hard-wires $lib/aux.cm to /usr/lib/smlnj/a-stuff/aux.cm or
2127    /usr/lib/smlnj/b-stuff/aux.cm, respectively.
2128    
2129    Hard-wiring path names is a bit inflexible (and CM will verbosely warn
2130    you when you do so at the time of CM.stabilize).  Therefore, you can
2131    also use an anchored path as the value:
2132    
2133      a.cm (bind:(anchor:lib value:$a-lib))
2134      b.cm (bind:(anchor:lib value:$b-lib))
2135    
2136    Now you can globally configure (using the usual CM.Anchor.anchor or
2137    pathconfig machinery) bindings for "a-lib" and "b-lib".  Since "lib"
2138    itself is always locally bound, setting it globally is no longer
2139    meaningful or necessary (but it does not hurt either).  In fact, "lib"
2140    can still be used as a global anchor for separate purposes.  As a
2141    matter of fact, one can locally define "lib" in terms of a global
2142    "lib":
2143    
2144      a.cm (bind:(anchor:lib value:$lib/a))
2145      b.cm (bind:(anchor:lib value:$lib/b))
2146    
2147    ** 3: The encoding of path names has changed.  This affects the way
2148    path names are shown in CM's progress report and also the internal
2149    protocol encoding used for parallel make.
2150    
2151    The encoding now uses one or more ':'-separated segments.  Each
2152    segments corresponds to a file that has been specified relative to the
2153    file given by its preceding segment.  The first segment is either
2154    relative to the CWD, absolute, or anchored.  Each segment itself is
2155    basically a Unix pathname; all segments but the first are relative.
2156    
2157    Example:
2158    
2159       $foo/bar/baz.cm:a/b/c.sml
2160    
2161    This path denotes the file bar/a/b/c.sml relative to the directory
2162    denoted by anchor "foo".  Notice that the encoding also includes
2163    baz.cm which is the .cm-file that listed a/b/c.sml.  As usual, such
2164    paths are resolved relative to the .cm-files directory, so baz.cm must
2165    be ignored to get the "real" pathname.
2166    
2167    To make this fact more obvious, CM puts the names of such "virtual
2168    arcs" into parentheses when they appear in progress reports. (No
2169    parentheses will appear in the internal protocol encoding.)  Thus,
2170    what you really see is:
2171    
2172      $foo/bar/(baz.cm):a/b/c.sml
2173    
2174    I find this notation to be much more informative than before.
2175    
2176    Another new feature of the encoding is that special characters
2177    including parentheses, colons, (back)slashes, and white space are
2178    written as \ddd (where ddd is the decimal encoding of the character).
2179    
2180    *** The CM manual still needs to be updated.
2181    
2182    ----------------------------------------------------------------------
2183    Name: Allen Leung
2184    Date: 2000/06/15 00:38:00
2185    Tag: leunga-20000615-x86-peephole
2186    
2187    x86 Peephole fix by Fermin.  Affects c-- and moby only.
2188    
2189    ----------------------------------------------------------------------
2190    Name: Matthias Blume
2191    Date: 2000/06/12 11:40:00
2192    Tag: blume-20000612-parmakefix
2193    Description:
2194    
2195    More cleanup after changing the file naming scheme: This time I
2196    repaired the parallel make mechanism for CMB.make which I broke earlier.
2197    
2198    ----------------------------------------------------------------------
2199    Name: Allen Leung
2200    Date: 2000/06/09 01:25:00
2201    Tag: leunga-20000609-various
2202    
2203    None of these things should affect normal SML/NJ operations
2204    
2205    1. Peephole improvements provided by Fermin (c--)
2206    2. New annotation DEFUSE for adding extra dependence (moby)
2207    3. New X86 LOCK instructions (moby)
2208    4. New machine description language for reservation tables (scheduling)
2209    5. Fixes to various optimization/analysis modules (branch chaining, dominator
2210       trees etc.)
2211    6. I've changed the CM files so that they can work with versions
2212       110.0.6, 110.25 and 110.28
2213    
2214    ----------------------------------------------------------------------
2215    Name: Matthias Blume
2216    Date: 2000/06/09 12:40:00
2217    Tag: blume-20000609-log
2218    Description:
2219    
2220    - Removed all(?) remaining RCS Log entries from sources.
2221    
2222    - Fixed bug in ml-yacc and ml-lex sources (use explicit anchors for
2223      anchored paths).
2224    
2225    ----------------------------------------------------------------------
2226    Name: Matthias Blume
2227    Date: 2000/06/07 17:00:00 JST
2228    Tag: blume-20000607-no-implicit-anchors
2229    Description:
2230    
2231    1. This update changes the default setting for
2232    CM.Control.implicit_anchors from true to false.  This means that
2233    implicit anchors are no longer permitted by default.  I also tried to
2234    make sure that nothing else still relies on implicit anchors.
2235    (This is the next step on the schedule towards a CM that does not even
2236    have the notion of implicit anchors anymore.)
2237    
2238    2. More CM manual updates.
2239    
2240    3. I managed to track down and fix the pickling bug I mentioned last
2241    time.  Because of the previously existing workaround, this entails no
2242    immediate practical changes.
2243    
2244    ----------------------------------------------------------------------
2245    Name: Matthias Blume
2246    Date: 2000/06/06 11:15:00 JST
2247    Tag: blume-20000606-lazierpickle
2248    Description:
2249    
2250    !!!! NEW BOOT FILES !!!!
2251    
2252    * The main purpose of this update is to make library pickles lazier in
2253    order to reduce the initial space penalty for autoloading a library.
2254    As a result, it is now possible to have $smlnj/compiler.cm
2255    pre-registered.  This should take care of the many complaints or
2256    inquiries about missing structure Compiler.  This required changes to
2257    CM's internal data structures and small tweaks to some algorithms.
2258    
2259    As a neat additional effect, it is no longer necessary (for the sake
2260    of lean heap image files) to distinguish between a "minimal" CM and a
2261    "full" CM.  Now, there is only one CM (i.e., the "full" version:
2262    $smlnj/cm.cm aka $smlnj/cm/full.cm), and it is always available at the
2263    interactive top level. ($smlnj/cm/minimal.cm is gone.)
2264    
2265    To make the life of compiler-hackers easier, "makeml" now also
2266    pre-registers $smlnj/cmb.cm (aka $smlnj/cmb/current.cm).  In other
2267    words, after you bootstrap a new sml for the first time, you will not
2268    have to autoload $smlnj/cmb.cm again afterwards.  (The first time
2269    around you will still have to do it, though.)
2270    
2271    * A second change consists of major updates to the CM manual.  There
2272    are now several appendices with summary information and also a full
2273    specification of the CM description file syntax.
2274    
2275    * In directory src/system I added the script "allcross".  This script
2276    invokes sml and cross-compiles the compiler for all supported
2277    architectures.  (Useful when providing a new set of boot files.)
2278    
2279    * There seems to be a latent bug in my "lazy pickles" mechanism.  I
2280    added a small tweak to pickle-util.sml to work around this problem,
2281    but it is not a proper fix yet.  I will investigate further.  (The
2282    effect of the bug was an inflation of library pickle size.)
2283    
2284    * Version number increased to 110.28.1 (to avoid compatibility problems).
2285    
2286    ----------------------------------------------------------------------
2287    Name: Allen Leung
2288    Date: 2000/05/25 17:28 EDT
2289    Tag: leunga-20000525-ra
2290    Description:
2291    
2292      Fixed a bug in freezing phase of the register allocator.
2293    
2294    ----------------------------------------------------------------------
2295    Name: Allen Leung
2296    Date: 2000/05/15 22:53 EDT
2297    Tag: leunga-20000515-alpha-x86-ra
2298    Description:
2299    
2300      1. Alpha
2301    
2302          Slight cleanup.  Removed the instruction SGNXL
2303    
2304      2. X86
2305    
2306          Added the following instructions to the instruction set:
2307    
2308            ROLx, RORx,
2309            BTx, BTSx, BTLx, BTRx,
2310            XCHGx, and variants with the LOCK prefix
2311    
2312      3. Register Allocation
2313    
2314          The module ra-rewrite-with-renaming has been improved.
2315    
2316      These have no effect on SML/NJ.
2317    
2318    ----------------------------------------------------------------------
2319    Name: Matthias Blume
2320    Date: 2000/05/15 16:20:00 JST
2321    Tag: blume-20000515-lightrebuild
2322    Description:
2323    
2324    1. I added an alternative to "-rebuild" to "makeml".  The difference is
2325       that prior to calling CMB.make' the CM-variable "LIGHT" will be
2326       defined.  In effect, the command will not build any cross-compiler
2327       backends and therefore finish more quickly.
2328    
2329       The "fixpt" script also takes a "-light" switch to be able to use
2330       this new facility while compiling for a fixpoint.
2331    
2332    2. I replaced all mentions of anchored paths in group owner specifications
2333       with simple relative paths (usually starting with "..").
2334       The rationale is that a library's internal workings should not be
2335       compromised by the lack of some anchor.  (An anchor is necessary
2336       for someone who wants to refer to the library by an anchored path,
2337       but it should not be necessary to build the same library in the first
2338       place.)
2339    
2340    3. I changed the way CM's tool mechanism determines the shell command
2341       string used for things like ml-yacc etc. so that it does not break
2342       when CM.Control.implicit_anchors is turned off.
2343    
2344    ----------------------------------------------------------------------
2345    Name: Matthias Blume
2346    Date: 2000/05/12 18:20:00 JST
2347    Tag: blume-20000512-ml-build
2348    Description:
2349    
2350    Fixed a bug in config/_ml-build that prevented ml-yacc and ml-lex from
2351    getting installed properly (by config/install.sh).
2352    
2353    ----------------------------------------------------------------------
2354    Name: Matthias Blume
2355    Date: 2000/05/12 17:30:00 JST
2356    Tag: blume-20000512-anchors
2357    Description:
2358    
2359    !!! NEW BOOT FILES !!!
2360    
2361    This change is in preparation of fading out support for "implicitly
2362    anchored path names".  I went through all sources and used the
2363    explicit (and relatively new) $-notation.  See system/README and the
2364    CM manual for more info on this.
2365    
2366    I also modified the anchoring scheme for some things such as "smlnj",
2367    "MLRISC", "cm", etc. to take advantage of the fact that explicit
2368    anchors are more expressive: anchor name and first arc do not have to
2369    coincide.  This entails the following user-visible change:
2370    
2371    You have to write $smlnj/foo/bar instead of smlnj/foo/bar.  In
2372    particular, when you fire up sml with a command-line argument, say,
2373    e.g.:
2374    
2375       sml '$smlnj/cmb.cm'
2376    
2377    At the ML toplevel prompt:
2378    
2379       CM.autoload "$smlnj/cmb.cm";
2380    
2381    There is also a new controller in CM.Control that can be used to turn
2382    off all remaining support for implicit anchors by saying:
2383    
2384        CM.autoload "$smlnj/
2385        #set CM.Control.implicit_anchors false;
2386    
2387    This causes CM to reject implicitly anchored paths.  This is (for the
2388    time being) less permissive than the "final" version where there will
2389    be no more such implicit anchors and relative paths will be just that:
2390    relative.
2391    
2392    The next step (version after next version?) will be to make the
2393    default for CM.Control.implicit_anchors false.  After the dust has
2394    settled, I can then produce the "final" version of this...
2395    
2396    Note: Since bootstrapping is a bit tricky, I provided new boot files.
2397    
2398    ----------------------------------------------------------------------
2399    Name: Matthias Blume
2400    Date: 2000/05/11 16:30:00 JST
2401    Tag: blume-20000511-sources
2402    Description:
2403    
2404    The main change is that I added function CM.sources as a generalized
2405    version of the earlier CM.makedepend.  This entails the following
2406    additional changes:
2407    
2408      - CM.makedepend has been dropped.
2409    
2410      - CM manual has been updated.
2411    
2412      - TOOLS signature and API have been changed.
2413    
2414    ----------------------------------------------------------------------
2415    Name: Allen Leung
2416    Date: 2000/05/10 21:17 EDT
2417    Tag: leunga-20000510-moby-c--ssa
2418    Description:
2419    
2420      Various bug fixes and new features for C--, Moby and MLRISC optimizations.
2421    None of these affect SML/NJ.
2422    
2423    1. Register Allocation
2424    
2425        a. A new ra spilling module (ra/ra-spill-with-renaming) is implemented.
2426           This module tries to remove local (i.e. basic block level) redundancies
2427           during spilling.
2428    
2429        b. A new framework for performing region based register allocation.
2430           Not yet entirely functional.
2431    
2432    2. X86
2433    
2434       a. DefUse for POP was missing the stack pointer [found by Lal]
2435       b. Reload for CALL was incorrect in X86Spill [found by John]
2436       c. Various fixes in X86Spill so that it can be used correctly for
2437          the new spilling module.
2438    
2439    3. SSA/IR
2440    
2441       a. New module ir/dj-dataflow.sml implements elimination based
2442          data flow analysis.
2443    
2444    4. MLRiscGen
2445    
2446       a. Fix for gc type annotation
2447    
2448    5. MDGen
2449    
2450       Various fixes for machine description -> ml code translation.  For ssa
2451       only.
2452    
2453    ----------------------------------------------------------------------
2454    Name: Allen Leung
2455    Date: 2000/05/08 22:17 EDT
2456    Tag: leunga-20000508-labexp
2457    Description:
2458    
2459      Fermin has found a few assembly problems with constant expressions
2460      generated in LabelExp.  Mostly, the problems involve extra parentheses,
2461      which choke on dumb assemblers.  This is his fix.
2462    
2463    ----------------------------------------------------------------------
2464    Name: Dave MacQueen
2465    Date: 2000/04/09 14:00 EDT
2466    Tag: dbm-20000502-Version_110_28
2467    Description:
2468    
2469    1. Updated src/compiler/TopLevel/main/version.sml to version 110.28
2470    
2471    2. Updated config/version to 110.28
2472    
2473    3. Updated config/srcarchiveurl
2474    
2475    3. New boot files!
2476       ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.28/
2477    
2478    ----------------------------------------------------------------------
2479    Name: Matthias Blume
2480    Date: 2000/05/01 19:05:00 JST
2481    Tag: blume-20000501-noweb
2482    Description:
2483    
2484    A new noweb tool has been added.  The existing system is entirely
2485    unaffected by this, but some CM users have asked for renewed noweb
2486    support.  Everything is documented in the CM manual.
2487    
2488    New (plugin) libraries:
2489    
2490       noweb-tool.cm
2491       nw-ext.cm
2492    
2493    ----------------------------------------------------------------------
2494    Name: Dave MacQueen
2495    Date: 2000/04/30 12:40PM EDT
2496    Tag: dbm-20000430-bug_fixes
2497    Description:
2498    
2499    1. Fix for bug 1498
2500       smlnj/src/system/Basis/Implementation/Unsafe/object.sig
2501       smlnj/src/system/Basis/Implementation/Unsafe/object.sml
2502         added toRealArray function
2503       smlnj/src/compiler/MiscUtil/print/ppobj.sml
2504         added check for tag Obj.RealArray to array printing case in ppObj
2505    
2506    2. Fix for bug 1510
2507       smlnj/src/compiler/Semant/types/typesutil.sml
2508         fixed definition of dummyargs (used by equalTycon) so that
2509         dummy args are distinct types
2510    
2511    ----------------------------------------------------------------------
2512    Name: Matthias Blume
2513    Date: 2000/04/30 01:00:00 JST
2514    Tag: blume-20000430-versions
2515    Description:
2516    
2517    1. CM version numbering added.  This is an implementation of Lal's
2518       proposal for adding version numbers and version checking to .cm
2519       files.  Lal said that his proposal was just that -- a proposal.
2520       For the time being I went ahead and implemented it so that people
2521       can comment on it.  Everything is completely backward-compatible
2522       (except for the stable library format, i.e., new bootfiles!).
2523    
2524       As usual, see the CM manual for details.
2525    
2526    2. An alternative syntax for anchored paths has been implemented.
2527       Dave has recently voiced the same concerns that I had when I did
2528       this, so there should be some support.  My take is that eventually
2529       I will let support for the current syntax (where anchors are
2530       "implicit") fade out in favor of the new, explicit syntax.
2531       In order to be backward-compatible, both old and new syntax are
2532       currently supported.
2533    
2534       Again, see the CM manual for details.
2535    
2536    3. Parallel make is trying to be slightly smarter:  When the master
2537       process finds a "bottleneck", i.e., when there is only one
2538       compilation unit that can be compiled and everybody else is
2539       waiting on it, then it will simply compile it directly instead
2540       of clumsily telling one of the slaves to do it.
2541    
2542    4. Support for "unsharing" added.  This is necessary in order to be
2543       able to have two different versions of the same library running
2544       at the same time (e.g., for trying out a new MLRISC while still
2545       having the old MLRISC linked into the current compiler, etc.)
2546       See the CM manual.
2547    
2548    5. Simple "makedepend" functionality added for generating Makefile
2549       dependency information.  (This is rather crude at the moment.
2550       Expect some changes here in the future.)
2551    
2552    6. ".fun" added as a recognized suffix for ML files. Also documented
2553       explicitly in the manual that the fallback behavior (unknown suffix
2554       -> ML file) is not an official feature!
2555    
2556    7. Small changes to the pickler for stable libraries.
2557    
2558    8. Several internal changes to CM (for cleanup/improvement).
2559    
2560    
2561    !!!! NEW BINFILES !!!!
2562    
2563    ----------------------------------------------------------------------
2564    Name: Matthias Blume
2565    Date: 2000/04/28 17:30:00 JST
2566    Tag: blume-20000428-pathconfig
2567    Description:
2568    
2569    1. I changed config/install.sh to remove duplicate entries from the
2570       lib/pathconfig file at the end.  Moreover, the final version of
2571       lib/pathconfig is sorted alphabetically.  The same (sorting) is done
2572       in src/system/installml.
2573    
2574    2. The config/install.sh script now consistently uses relative
2575       pathnames in lib/pathconfig whenever the anchor is in the lib
2576       directory.  (So far this was true for the libraries that come
2577       pre-compiled and bundled as part of the bootfiles but not for
2578       libraries that are compiled by the script itself.)
2579    
2580    ----------------------------------------------------------------------
2581    Name: Matthias Blume
2582    Date: 2000/04/26 13:10:00 JST
2583    Tag: blume-20000426-fun_suffix
2584    Description:
2585    
2586    Added ".fun" as a recognized file name suffix (for ML code).
2587    
2588    ----------------------------------------------------------------------
2589    Name: Allen Leung
2590    Date: 2000/04/25 17:00:00 EST
2591    Tag: leunga-20000425-alpha-ra
2592    Description:
2593    
2594    1. Alpha
2595    
2596        PSEUDOARITH was missing in AlphaRewrite.  This causes an endless loop
2597    in C--.
2598    
2599    2. RA
2600    
2601       Added a flag "ra-dump-size" to print out the size of the flowgraph
2602       and the interference graph.
2603    
2604    ----------------------------------------------------------------------
2605    Name: Dave MacQueen
2606    Date: 2000/04/25/
2607    Tag: dbm-20000425-mlyacc_doc_examples
2608    Description:
2609      Updated mlyacc.tex sections 5 and 7 for SML '97 and CM.
2610      Updated all three examples in src/ml-yacc/examples to run
2611      under 110.* using CM.make.
2612    
2613    ----------------------------------------------------------------------
2614    Name: Allen Leung
2615    Date: 2000/04/20 23:04:00 EST
2616    Tag: leunga-20000420-ssa-c---stuff
2617    Description:
2618    
2619      This update synchronizes my repository with Yale's.  Most of these
2620    changes, however, do not affect SML/NJ at all (the RA is an exception).
2621    
2622    1. Register Allocator
2623    
2624       a. An improvement in the interference graph construction:
2625          Given a copy
2626    
2627                s <- t
2628    
2629          no interference edge between s and t is added for this definition of s.
2630    
2631       b. I've added two new spill heuristic modules that Fermin and I developed
2632          (in the new library RA.cm). These are unused in SML/NJ but maybe
2633          useful for others (Moby?)
2634    
2635    2. X86
2636    
2637       a. Various fixes in the backend provided by Fermin [C--] and Lal.
2638    
2639    3. Alpha
2640    
2641       a. Added the BSR instruction and code generation that goes with it [C--]
2642       b. Other fixes too numerous to recount provided by Fermin [C--]
2643    
2644    4. Regmaps
2645    
2646       a. The regmaps are not initialized with the identity physical bindings
2647          at creation time.  This is unneeded.
2648    
2649    5. MLRISC Optimizations
2650    
2651       a. The DJ-Graph module can now compute the iterated dominance frontiers
2652          intersects with liveness incrementally in linear time! Woohoo!
2653          This is now used in my new SSA construction algorithm.
2654    
2655       b. THe branch reorganization module is now smarter about linear chains of
2656          basic blocks.
2657    
2658    
2659    ----------------------------------------------------------------------
2660    Name: Matthias Blume
2661    Date: 2000/04/12 13:52:00 JST
2662    Tag: blume_main_v110p27_1
2663    Description:
2664    
2665    Changed install.sh script to handle archive files without version number
2666    and to use "boot.<arch>-<os>" instead of "sml.boot.<arch>-<os>" for the
2667    name of the boot file archive.
2668    
2669    ----------------------------------------------------------------------
2670    Name: Dave MacQueen
2671    Date: 2000/04/09 14:00 EDT
2672    Tag: dbm-20000410-Version_110_27
2673    Description:
2674    
2675    1. Updated src/compiler/TopLevel/main/version.sml to version 110.27
2676    
2677    2. Updated src/config/version to 110.27
2678    
2679    3. New boot files!
2680    
2681    ----------------------------------------------------------------------
2682    Name: Allen Leung
2683    Date: 2000/04/09 19:09:00 EST
2684    Tag: leunga-20000409-misc
2685    Description:
2686    
2687    1.  Yet another fix for x86 assembly for idivl, imull, mull and friends.
2688    
2689    2.  Miscellaneous improvements to MLRISC (unused in sml/nj)
2690    
2691    ----------------------------------------------------------------------
2692    Name: Stefan
2693    Date: 2000/04/07 10:00:00 EDT
2694    Tag: monnier-20000406-branch-handling
2695    Description:
2696    
2697    Improved handling of branches (mostly those generated from
2698    polymorphic equality), removed switchoff and changed the
2699    default optimization settings (more cpsopt and less flintopt).
2700    
2701    ----------------------------------------------------------------------
2702    Name: Allen Leung
2703    Date: 2000/04/06 01:30:00 EST
2704    Tag: leunga-20000406-peephole-x86-SSA-2
2705    Description:
2706    
2707       Forgot a few files.
2708    
2709    ----------------------------------------------------------------------
2710    Name: Allen Leung
2711    Date: 2000/04/06 00:36:00 EST
2712    Tag: leunga-20000406-peephole-x86-SSA
2713    Description:
2714    
2715    1.  New Peephole code
2716    
2717    2.  Minor improvement to X86 instruction selection
2718    
2719    3.  Various fixes to SSA and machine description -> code translator
2720    
2721    ----------------------------------------------------------------------
2722    Name: Matthias Blume
2723    Date: 2000/04/05 12:30:00 JST
2724    Tag: blume_main_v110p26p2_3
2725    Description:
2726    
2727    This update just merges three minor cosmetic updates to CM's sources
2728    to get ready for the 110.27 code freeze on Friday.  No functionality
2729    has changed.
2730    
2731    ----------------------------------------------------------------------
2732    Name: Allen Leung
2733    Date: 2000/04/04 19:39:00 EST
2734    Tag: leunga-20000404-x86-asm
2735    Description:
2736    
2737    1.  Fixed a problem in X86 assembly.
2738    
2739        Things like
2740    
2741           jmp %eax
2742           jmp (%eax)
2743    
2744        should be output as
2745    
2746           jmp *%eax
2747           jmp *(%eax)
2748    
2749    2.  Assembly output
2750    
2751          Added a new flag
2752    
2753              "asm-indent-copies" (default to false)
2754    
2755          When this flag is on, parallel copies will be indented an extra level.
2756    
2757    ----------------------------------------------------------------------
2758    Name: Allen Leung
2759    Date: 2000/04/04 03:18:00 EST
2760    Tag: leunga-20000404-C--Moby
2761    Description:
2762    
2763        All of these fixes are related to C--, Moby, and my own optimization
2764        stuff; so they shouldn't affect SML/NJ.
2765    
2766    1.  X86
2767    
2768        Various fixes related floating point, and extensions.
2769    
2770    2.  Alpha
2771    
2772        Some extra patterns related to loads with signed/zero extension
2773        provided by Fermin.
2774    
2775    3.  Assembly
2776    
2777        When generating assembly, resolve the value of client defined constants,
2778        instead of generating symbolic values.  This is controlled by the
2779        new flag "asm-resolve-constants", which is default to true.
2780    
2781    4.  Machine Descriptions
2782    
2783        a. The precedence parser was slightly broken when parsing infixr symbols.
2784        b. The type generalizing code had the bound variables reversed, resulting
2785           in a problem during arity raising.
2786        c. Various fixes in machine descriptions.
2787    
2788    ----------------------------------------------------------------------
2789    Name: Matthias Blume
2790    Date: 2000/04/03 16:05:00 JST
2791    Tag: blume_main_v110p26p2_2
2792    Description:
2793    
2794    I eliminated coreEnv from compInfo.  Access to the "Core" structure is
2795    now done via the ordinary static environment that is context to each
2796    compilation unit.
2797    
2798    To this end, I arranged that instead of "structure Core" as "structure
2799    _Core" is bound in the pervasive environment.  Core access is done via
2800    _Core (which can never be accidentally rebound because _Core is not a
2801    legal surface-syntax symbol).
2802    
2803    The current solution is much cleaner because the core environment is
2804    now simply part of the pervasive environment which is part of every
2805    compilation unit's context anyway.  In particular, this eliminates all
2806    special-case handling that was necessary until now in order to deal
2807    with dynamic and symbolic parts of the core environment.
2808    
2809    Remaining hackery (to bind the "magic" symbol _Core) is localized in the
2810    compilation manager's bootstrap compiler (actually: in the "init group"
2811    handling).  See the comments in src/system/smlnj/init/init.cmi for
2812    more details.
2813    
2814    I also tried to track down all mentions of "Core" (as string argument
2815    to Symbol.strSymbol) in the compiler and replaced them with a
2816    reference to the new CoreSym.coreSym.  Seems cleaner since the actual
2817    name appears in one place only.
2818    
2819    Binfile and bootfile format have not changed, but the switchover from
2820    the old "init.cmi" to the new one is a bit tricky, so I supplied new
2821    bootfiles anyway.
2822    
2823    ----------------------------------------------------------------------
2824    Name: Allen Leung
2825    Date: 2000/04/02 21:17:00 EST
2826    Tag: leunga-20000402-mltree
2827    Description:
2828    
2829       1. Renamed the constructor CALL in MLTREE by popular demand.
2830       2. Added a bunch of files from my repository.  These are currently
2831          used by other non-SMLNJ backends.
2832    
2833    ----------------------------------------------------------------------
2834    Name: Allen Leung
2835    Date: 2000/03/31 21:15:00 EST
2836    Tag: leunga-20000331-aliasing
2837    Description:
2838    
2839    This update contains a rewritten (and hopefully more correct) module
2840    for extracting aliasing information from CPS.
2841    
2842       To turn on this feature:
2843    
2844            Compiler.Control.CG.memDisambiguate := true
2845    
2846       To pretty print the region information with assembly
2847    
2848           Compiler.Control.MLRISC.getFlag "asm-show-region" := true;
2849    
2850       To control how many levels of aliasing information are printed, use:
2851    
2852           Compiler.Control.MLRISC.getInt "points-to-show-level" := n
2853    
2854       The default of n is 3.
2855    
2856    ----------------------------------------------------------------------
2857    Name: David MacQueen
2858    Date: 2000/03/31 11:15:00 EST
2859    Tag: dbm-20000331-runtime_fix
2860    Description:
2861    
2862    This update contains:
2863    
2864    1. runtime/c-lib/c-libraries.c
2865       includes added in revision 1.2 caused compilation errors on hppa-hpux
2866    
2867    2. fix for bug 1556
2868       system/Basis/Implementation/NJ/internal-signals.sml
2869    
2870    ----------------------------------------------------------------------
2871    Name: Matthias Blume
2872    Date: 2000/03/31 18:00:00 JST
2873    Tag: blume_main_v110p26p2_1
2874    Description:
2875    
2876    This update contains:
2877    
2878    1. A small change to CM's handling of stable libraries:
2879       CM now maintains one "global" modmap that is used for all stable
2880       libraries.  The use of such a global modmap maximizes sharing and
2881       minimizes the need for re-traversing parts of environments during
2882       modmap construction.  (However, this has minor impact since modmap
2883       construction seems to account for just one percent or less of total
2884       compile time.)
2885    
2886    2. I added a "genmap" phase to the statistics.  This is where I got the
2887       "one percent" number (see above).
2888    
2889    3. CM's new tool parameter mechanism just became _even_ better. :)
2890       - The parser understands named parameters and recursive options.
2891       - The "make" and "shell" tools use these new features.
2892         (This makes it a lot easier to cascade these tools.)
2893       - There is a small syntax change: named parameters use a
2894    
2895           <name> : ( <option> ... )            or
2896           <name> : <string>
2897    
2898         syntax.  Previously, named parameters were implemented in an
2899         ad-hoc fashion by each tool individually (by parsing strings)
2900         and had the form
2901    
2902           <name>=<string>
2903    
2904       See the CM manual for a full description of these issues.
2905    
2906    ----------------------------------------------------------------------
2907    Name: Matthias Blume
2908    Date: 2000/03/30 18:00:00 JST
2909    Tag: blume_main_v110p26p2_0
2910    Description:
2911    
2912    !!!!! WARNING !!!!!!
2913    !!  New binfiles  !!
2914    !!!!!!!!!!!!!!!!!!!!
2915    
2916    This update contains:
2917    
2918    1. Moderate changes to CM:
2919    
2920       - Changes to CM's tools mechanism.  In particular, it is now possible
2921       to have tools that accept additional "command line" parameters
2922       (specified in the .cm file at each instance where the tool's class is
2923       used).
2924    
2925       This was done to accommodate the new "make" and "shell" tools which
2926       facilitate fairly seamless hookup to portions of code managed using
2927       Makefiles or Shell scripts.
2928    
2929       There are no classes "shared" or "private" anymore.  Instead, the
2930       sharing annotation is now a parameter to the "sml" class.
2931    
2932       There is a bit of generic machinery for implementing one's own
2933       tools that accept command-line parameters.  However, I am not yet fully
2934       satisfied with that part, so expect changes here in the future.
2935    
2936       All existing tools are described in the CM manual.
2937    
2938       - Slightly better error handling.  (CM now suppresses many followup
2939       error messages that tended to be more annoying than helpful.)
2940    
2941    2. Major changes to the compiler's static environment data structures.
2942    
2943       - no CMStaticEnv anymore.
2944            - no CMEnv, no "BareEnvironment" (actually, _only_ BareEnvironment,
2945              but it is called Environment), no conversions between different
2946              kinds of static environments
2947    
2948       - There is still a notion of a "modmap", but such modmaps are generated
2949         on demand at the time when they are needed.  This sounds slow, but I
2950         sped up the code that generates modmaps enough for this not to lead to
2951         a slowdown of the compiler (at least I didn't detect any).
2952    
2953       - To facilitate rapid modmap generation, static environments now
2954         contain an (optional) "modtree" structure.  Modtree annotations are
2955         constructed by the unpickler during unpickling.  (This means that
2956         the elaborator does not have to worry about modtrees at all.)
2957         Modtrees have the advantage that they are compositional in the same
2958         way as the environment data structure itself is compositional.
2959         As a result, modtrees never hang on to parts of an environment that
2960         has already been rendered "stale" by filtering or rebinding.
2961    
2962       - I went through many, many trials and errors before arriving at the
2963         current solution.  (The initial idea of "linkpaths" did not work.)
2964         But the result of all this is that I have touched a lot of files that
2965         depend on the "modules" and "types" data structures (most of the
2966         elaborator). There were a lot of changes during my "linkpath" trials
2967         that could have been reverted to their original state but weren't.
2968         Please, don't be too harsh on me for messing with this code a bit more
2969         than what was strictly necessary...  (I _did_ resist the temptation
2970         of doing any "global reformatting" to avoid an untimely death at
2971         Dave's hands. :)
2972    
2973       - One positive aspect of the previous point:  At least I made sure that
2974         all files that I touched now compile without warnings (other than
2975         "polyEqual").
2976    
2977       - compiler now tends to run "leaner" (i.e., ties up less memory in
2978         redundant modmaps)
2979    
2980    ----------------------------------------------------------------------
2981    Name: Allen Leung
2982    Date: 2000/03/29 18:00:00
2983    Tag: leunga-20000327-mlriscGen_hppa_alpha_x86
2984    Boot files (optional): ftp://react-ilp.cs.nyu.edu/leunga/110.26.1-sml.boot.x86-unix-20000330.tar.gz
2985    Description:
2986    
2987       This update contains *MAJOR* changes to the way code is generated from CPS
2988    in the module mlriscGen, and in various backend modules.
2989    
2990    CHANGES
2991    =======
2992    
2993    1. MLRiscGen: forward propagation fix.
2994    
2995       There was a bug in forward propagation introduced at about the same time
2996       as the MLRISC x86 backend, which prohibits coalescing to be
2997       performed effectively in loops.
2998    
2999       Effect: speed up of loops in RISC architectures.
3000               By itself, this actually slowed down certain benchmarks on the x86.
3001    
3002    2. MLRiscGen:  forward propagating addresses from consing.
3003    
3004       I've changed the way consing code is generated.  Basically I separated
3005       out the initialization part:
3006    
3007            store tag,   offset(allocptr)
3008            store elem1, offset+4(allocptr)
3009            store elem2, offset+8(allocptr)
3010            ...
3011            store elemn, offset+4n(allocptr)
3012    
3013       and the address computation part:
3014    
3015            celladdr <- offset+4+alloctpr
3016    
3017       and move the address computation part
3018    
3019       Effect:  register pressure is generally lower as a result.  This
3020                makes compilation of certain expressions much faster, such as
3021                long lists with non-trivial elements.
3022    
3023                 [(0,0), (0,0), .... (0,0)]
3024    
3025    3. MLRiscGen: base pointer elimination.
3026    
3027        As part of the linkage mechanism, we generate the sequence:
3028    
3029         L:  ...  <- start of the code fragment
3030    
3031         L1:
3032             base pointer <- linkreg - L1 + L
3033    
3034         The base pointer was then used for computing relocatable addresses
3035       in the code fragment.  Frequently (such as in lots of continuations)
3036       this is not needed.  We now eliminate this sequence whenever possible.
3037    
3038         For compile time efficiency, I'm using a very stupid local heuristic.
3039       But in general, this should be done as a control flow analysis.
3040    
3041       Effect:  Smaller code size.  Speed up of most programs.
3042    
3043    4. Hppa back end
3044    
3045         Long jumps in span dependence resolution used to depend on the existence
3046      of the base pointer.
3047    
3048         A jump to a long label L was expanded into the following sequence:
3049    
3050          LDIL %hi(L-8192), %r29
3051          LDO  %lo(L-8192)(%r29), %r29
3052          ADD  %r29, baseptr, %r29
3053          BV,n %r0(%r29)
3054    
3055         In the presence of change (3) above, this will not work.  I've changed
3056       it so that the following sequence of instructions are generated, which
3057       doesn't mention the base pointer at all:
3058    
3059             BL,n  L', %r29           /* branch and link, L' + 4 -> %r29 */
3060        L':  ADDIL L-(L'+4), %r29     /* Compute address of L */
3061             BV,n  %r0(%r29)          /* Jump */
3062    
3063    5. Alpha back end
3064    
3065          New alpha instructions LDB/LDW have been added, as per Fermin's
3066       suggestions.   This is unrelated to all other changes.
3067    
3068    6. X86 back end
3069    
3070         I've changed andl to testl in the floating point test sequence
3071         whenever appropriate.  The Intel optimization guide states that
3072         testl is preferable to andl.
3073    
3074    7. RA (x86 only)
3075    
3076         I've improved the spill propagation algorithm, using an approximation
3077       of maximal weighted independent sets.   This seems to be necessary to
3078       alleviate the negative effect in light of the slow down in (1).
3079    
3080         I'll write down the algorithm one of these days.
3081    
3082    8. MLRiscGen: frequencies
3083    
3084         I've added an annotation that states that all call gc blocks have zero
3085       execution frequencies.  This improves register allocation on the x86.
3086    
3087    BENCHMARKS
3088    ==========
3089    
3090       I've only perform the comparison on 110.25.
3091    
3092       The platforms are:
3093    
3094        HPPA  A four processor HP machine (E9000) with 5G of memory.
3095        X86   A 300Hhz Pentium II with 128M of memory, and
3096        SPARC An Ultra sparc 2 with 512M of memory.
3097    
3098       I used the following parameters for the SML benchmarks:
3099    
3100                 @SMLalloc
3101         HPPA    256k
3102         SPARC   512k
3103         X86     256k
3104    
3105    COMPILATION TIME
3106    ----------------
3107       Here are the numbers comparing the compilation times of the compilers.
3108       I've only compared 110.25 compiling the new sources versus
3109       a fixpoint version of the new compiler compiling the same.
3110    
3111                     110.25                                  New
3112               Total  Time in RA  Spill+Reload   Total  Time In RA Spill+Reload
3113         HPPA   627s    116s        2684+3584     599s    95s       1003+1879
3114         SPARC  892s    173s        2891+3870     708s    116s      1004+1880
3115         X86    999s    315s       94006+130691   987s    296s    108877+141957
3116    
3117                   110.25         New
3118                Code Size      Code Size
3119         HPPA   8596736         8561421
3120         SPARC  8974299         8785143
3121         X86    9029180         8716783
3122    
3123       So in summary, things are at least as good as before.   Dramatic
3124       reduction in compilation is obtained on the Sparc; I can't explain it,
3125       but it is reproducible.  Perhaps someone should try to reproduce this
3126       on their own machines.
3127    
3128    SML BENCHMARKS
3129    --------------
3130    
3131        On the average, all benchmarks perform at least as well as before.
3132    
3133          HPPA         Compilation Time     Spill+Reload      Run Time
3134                     110.25  New            110.25    New   110.25  New
3135    
3136          barnesHut  3.158  3.015  4.75%    1+1       0+0   2.980  2.922   2.00%
3137              boyer  6.152  5.708  7.77%    0+0       0+0   0.218  0.213   2.34%
3138       count-graphs  1.168  1.120  4.32%    0+0       0+0  22.705 23.073  -1.60%
3139                fft  0.877  0.792 10.74%    1+3       1+3   0.602  0.587   2.56%
3140        knuthBendix  3.180  2.857 11.32%    0+0       0+0   0.675  0.662   2.02%
3141             lexgen  6.190  5.290 17.01%    0+0       0+0   0.913  0.788  15.86%
3142               life  0.803  0.703 14.22%   25+25      0+0   0.153  0.140   9.52%
3143              logic  2.048  2.007  2.08%    6+6       1+1   4.133  4.008   3.12%
3144         mandelbrot  0.077  0.080 -4.17%    0+0       0+0   0.765  0.712   7.49%
3145             mlyacc 22.932 20.937  9.53%  154+181    32+57  0.468  0.430   8.91%
3146            nucleic  5.183  5.060  2.44%    2+2       0+0   0.125  0.120   4.17%
3147      ratio-regions  3.357  3.142  6.84%    0+0       0+0  116.225 113.173 2.70%
3148                ray  1.283  1.290 -0.52%    0+0       0+0   2.887  2.855   1.11%
3149             simple  6.307  6.032  4.56%   28+30      5+7   3.705  3.658   1.28%
3150                tsp  0.888  0.862  3.09%    0+0       0+0   7.040  6.893   2.13%
3151               vliw 24.378 23.455  3.94%  106+127    25+45  2.758  2.707   1.91%
3152      --------------------------------------------------------------------------
3153       Average                     6.12%                                   4.09%
3154    
3155          SPARC        Compilation Time     Spill+Reload      Run Time
3156                     110.25  New            110.25    New   110.25  New
3157    
3158          barnesHut  3.778  3.592  5.20%    2+2       0+0   3.648  3.453    5.65%
3159              boyer  6.632  6.110  8.54%    0+0       0+0   0.258  0.242    6.90%
3160       count-graphs  1.435  1.325  8.30%    0+0       0+0  33.672 34.737   -3.07%
3161                fft  0.980  0.940  4.26%    3+9       2+6   0.838  0.827    1.41%
3162        knuthBendix  3.590  3.138 14.39%    0+0       0+0   0.962  0.967   -0.52%
3163             lexgen  6.593  6.072  8.59%    1+1       0+0   1.077  1.078   -0.15%
3164               life  0.972  0.868 11.90%   26+26      0+0   0.143  0.140    2.38%
3165              logic  2.525  2.387  5.80%    7+7       1+1   5.625  5.158    9.05%
3166         mandelbrot  0.090  0.093 -3.57%    0+0       0+0   0.855  0.728   17.39%
3167             mlyacc 26.732 23.827 12.19%  162+189    32+57  0.550  0.560   -1.79%
3168            nucleic  6.233  6.197  0.59%    3+3       0+0   0.163  0.173   -5.77%
3169      ratio-regions  3.780  3.507  7.79%    0+0       0+0 133.993 131.035   2.26%
3170                ray  1.595  1.550  2.90%    1+1       0+0   3.440  3.418    0.63%
3171             simple  6.972  6.487  7.48%   29+32      5+7   3.523  3.525   -0.05%
3172                tsp  1.115  1.063  4.86%    0+0       0+0   7.393  7.265    1.77%
3173               vliw 27.765 24.818 11.87%  110+135    25+45  2.265  2.135    6.09%
3174      ----------------------------------------------------------------------------
3175       Average                     6.94%                                    2.64%
3176    
3177          X86          Compilation Time     Spill+Reload      Run Time
3178                     110.25  New            110.25    New   110.25  New
3179    
3180          barnesHut  5.530  5.420  2.03%  593+893   597+915   3.532  3.440   2.66%
3181              boyer  8.768  7.747 13.19%  493+199   301+289   0.327  0.297  10.11%
3182       count-graphs  2.040  2.010  1.49%  298+394   315+457  26.578 28.660  -7.26%
3183                fft  1.327  1.302  1.92%  112+209   115+210   1.055  0.962   9.71%
3184        knuthBendix  5.218  5.475 -4.69%  451+598   510+650   0.928  0.932  -0.36%
3185             lexgen  9.970  9.623  3.60% 1014+841  1157+885   0.947  0.928   1.97%
3186               life  1.183  1.183  0.00%  162+182   145+148   0.127  0.103  22.58%
3187              logic  3.285  3.512 -6.45%  514+684   591+836   5.682  5.577   1.88%
3188         mandelbrot  0.147  0.143  2.33%   38+41     33+54    0.703  0.690   1.93%
3189             mlyacc 35.457 32.763  8.22% 3496+4564 3611+4860  0.552  0.550   0.30%
3190            nucleic  7.100  6.888  3.07%  239+168   201+158   0.175  0.173   0.96%
3191      ratio-regions  6.388  6.843 -6.65% 1182+257   981+300  120.142 120.345 -0.17%
3192                ray  2.332  2.338 -0.29%  346+398   402+494   3.593  3.540   1.51%
3193             simple  9.912  9.903  0.08% 1475+941  1579+1168  3.057  3.178  -3.83%
3194                tsp  1.623  1.532  5.98%  266+200   250+211   8.045  7.878   2.12%
3195               vliw 33.947 35.470 -4.29% 2629+2774 2877+3171  2.072  1.890   9.61%
3196      ----------------------------------------------------------------------------
3197       Average                     1.22%                                     3.36%
3198    
3199    ----------------------------------------------------------------------
3200    Name: Allen Leung
3201    Date: 2000/03/23 16:25:00
3202    Tag: leunga-20000323-fix_x86_alpha
3203    Description:
3204    
3205    1. X86 fixes/changes
3206    
3207       a.  The old code generated for SETcc was completely wrong.
3208           The Intel optimization guide is VERY misleading.
3209    
3210    2. ALPHA fixes/changes
3211    
3212       a.  Added the instructions LDBU, LDWU, STB, STW as per Fermin's suggestion.
3213       b.  Added a new mode byteWordLoadStores to the functor parameter to Alpha()
3214       c.  Added reassociation code for address computation.
3215    
3216    ----------------------------------------------------------------------
3217    Name: Allen Leung
3218    Date: 2000/03/22 01:23:00
3219    Tag: leunga-20000322-fix_x86_hppa_ra
3220    Description:
3221    
3222    1. X86 fixes/changes
3223    
3224       a.  x86Rewrite bug with MUL3 (found by Lal)
3225       b.  Added the instructions FSTS, FSTL
3226    
3227    2. PA-RISC fixes/changes
3228    
3229       a.  B label should not be a delay slot candidate!  Why did this work?
3230       b.  ADDT(32, REG(32, r), LI n) now generates one instruction instead of two,
3231           as it should be.
3232       c.  The assembly syntax for fstds and fstdd was wrong.
3233       d.  Added the composite instruction COMICLR/LDO, which is the immediate
3234           operand variant of COMCLR/LDO.
3235    
3236    3. Generic MLRISC
3237    
3238       a.  shuffle.sml rewritten to be slightly more efficient
3239       b.  DIV bug in mltree-simplify fixed (found by Fermin)
3240    
3241    4. Register Allocator
3242    
3243       a.  I now release the interference graph earlier during spilling.
3244           May improve memory usage.
3245    
3246    ----------------------------------------------------------------------
3247    Name: Matthias Blume
3248    Date: 2000/03/14 14:15:32
3249    Tag: blume_main_v110p26p1_2
3250    Description:
3251    
3252    1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an
3253    additional argument called "template" which is an optional string that
3254    specifies the layout of the tool command line.  See the CM manual for
3255    explanation.
3256    
3257    2. A special-purpose tool can be "registered" by simply dropping the
3258    corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same
3259    directory where the .cm file lives that uses this tool.  (The
3260    behavior/misfeature until now was to look for the tool description
3261    files in the current working directory.)  As before, tool description
3262    files could also be anchored -- in which case they can live anywhere
3263    they like.  Following the recent e-mail discussion, this change should
3264    make it easier to have special-purpose tools that are shipped together
3265    with the sources of the program that uses them.
3266    
3267    ----------------------------------------------------------------------
3268    Name: Matthias Blume
3269    Date: 2000/03/10 07:48:34
3270    Tag: blume_main_v110p26p1_1
3271    Description:
3272    
3273    I added a re-written version of Dave's fixpt script to src/system.
3274    Changes relative to the original version:
3275      - sh-ified (not everybody has ksh)
3276      - automatically figures out which architecture it runs on
3277      - uses ./makeml a bit more cleverly
3278      - never invokes ./installml (and, thus, does not clobber your
3279        good and working installation of sml in case something goes wrong)
3280      - accepts max iteration count using option "-iter <n>"
3281      - accepts a "base" name using option "-base <base>"
3282    
3283    It does not build any extraneous heap images but directly rebuilds
3284    bin- and boot-hierarchies using makeml's "-rebuild" switch. Finally,
3285    it can incorporate existing bin- and boot- hierarchies.  For example,
3286    suppose the base is set to "sml" (which is the default).  Then it
3287    successively builds
3288    
3289            sml.bin.<arch>-unix and sml.boot.<arch>-unix
3290    then    sml1.bin.<arch>-unix and sml1.boot.<arch>-unix
3291    then    sml2.bin.<arch>-unix and sml2.boot.<arch>-unix
3292    ...
3293    then    sml<n>.bin.<arch>-unix and sml<n>.boot.<arch>-unix
3294    
3295    and so on.  If any of these already exist, it will just use what's
3296    there.  In particular, many people will have the initial set of bin
3297    and boot files around, so this saves time for at least one full
3298    rebuild.  Having sets of the form <base><k>.{bin,boot}.<arch>-unix for
3299    <k>=1,2,... is normally not a good idea when invoking fixpt.  However,
3300    they might be the result of an earlier partial run of fixpt (which
3301    perhaps got accidentally killed).  In this case, fixpt will quickly
3302    move through what exists before continuing where it left off earlier,
3303    and, thus, saves a lot of time.
3304    
3305    ----------------------------------------------------------------------
3306    Name: Allen Leung
3307    Date: 00/03/10 02:20:00
3308    Tag: leunga-20000310-fix_x86_asm_ra
3309    Description:
3310    
3311    More assembly output problems involving the indexed addressing mode
3312    on the x86 have been found and corrected. Thanks to Fermin Reig for the
3313    fix.
3314    
3315    The interface and implementation of the register allocator have been changed
3316    slightly to accommodate the possibility to skip the register allocation
3317    phases completely and go directly to memory allocation.  This is needed
3318    for C-- use.
3319    
3320    ----------------------------------------------------------------------
3321    Name: Matthias Blume
3322    Date: 00/03/09 10:23:53
3323    Tag: blume_main_v110p26p1_0
3324    Description:
3325    
3326    * Complete re-organization of library names.  Many libraries have been
3327    consolidated so that they share the same path anchor.  For example,
3328    all MLRISC-related libraries are anchored at MLRISC, most libraries that
3329    are SML/NJ-specific are under "smlnj".  Notice that names like
3330    host-cmb.cm or host-compiler.cm no longer exist.  See system/README
3331    for a complete description of the new naming scheme.  Quick reference:
3332    
3333       host-cmb.cm        -> smlnj/cmb.cm
3334       host-compiler.cm   -> smlnj/compiler.cm
3335       full-cm.cm         -> smlnj/cm.cm
3336       <arch>-<os>.cm     -> smlnj/cmb/<arch>-<os>.cm
3337       <arch>-compiler.cm -> smlnj/compiler/<arch>.cm
3338    
3339    * Bug fixes in CM.
3340        - exceptions in user code are being passed through (i.e., reach top level)
3341        - more bugs in paranoia mode fixed
3342        - bug related to checking group owners fixed
3343    
3344    * New install.sh script that automagically fetches archive files:
3345      The new file config/srcarchiveurl must contain the URL of the
3346      (remote) directory that contains bin files (or other source archives).
3347      If install.sh does not find the archive locally, it tries to get
3348      it from that remote directory.
3349      This should simplify installation further:  For machines that have
3350      access to the internet, just fetch <version>-config.tgz, unpack it,
3351      edit config/targets, and go (run config/install.sh).  The script will
3352      fetch everything else that it might need all by itself.
3353    
3354      For CVS users, this mechanism is not relevant for source archives, but
3355      it is convenient for getting new sets of binfiles.
3356    
3357      Archives should be tar files compressed with either gzip, compress, or
3358      bzip2.  The script recognizes .tgz, .tar, tar.gz, tz, .tar.Z, and .tar.bz2.
3359    
3360    ----------------------------------------------------------------------
3361    Name: Matthias Blume
3362    Date: 2000/03/07 04:01:04
3363    Tag: blume_main_v110_26_2
3364    Description:
3365    - size info in BOOTLIST
3366         * no fixed upper limits for number of bootfiles or length of
3367           bootfile names in runtime
3368         * falling back to old behavior if no BOOTLIST size info found
3369    - allocation size heuristics in .run-sml
3370         * tries to read cache size from /proc/cpuinfo (this is important for
3371            small-cache Celeron systems!)
3372    - install.sh robustified
3373    - CM manual updates
3374    - paranoid mode
3375         * no more CMB.deliver() (i.e., all done by CMB.make())
3376         * can re-use existing sml.boot.* files
3377         * init.cmi now treated as library
3378         * library stamps for consistency checks
3379    - sml.boot.<arch>-<os>/PIDMAP file
3380         * This file is read by the CM startup code.  This is used to minimize
3381           the amount of dynamic state that needs to be stowed away for the
3382           purpose of sharing between interactive system and user code.
3383    - CM.Anchor.anchor instead of CM.Anchor.{set,cancel}
3384         * Upon request by Elsa.  Anchors now controlled by get-set-pair
3385           like most other CM state variables.
3386    - Compiler.CMSA eliminated
3387         * No longer supported by CM anyway.
3388    - fixed bugs in pickler that kept biting Stefan
3389         * past refs to past refs (was caused by the possibility that
3390           ad-hoc sharing is more discriminating than hash-cons sharing)
3391         * integer overflow on LargeInt.minInt
3392    - ml-{lex,yacc} build scripts now use new mechanism
3393      for building standalone programs
3394    - fixed several gcc -Wall warnings that were caused by missing header
3395      files, missing initializations, etc., in runtime (not all warnings
3396      eliminated, though)

Legend:
Removed from v.570  
changed lines
  Added in v.927

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