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

SCM Repository

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

Diff of /sml/trunk/HISTORY

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

revision 578, Tue Mar 14 05:16:29 2000 UTC revision 993, Fri Nov 23 02:37:10 2001 UTC
# Line 11  Line 11 
11  Date:  Date:
12  Tag: <post-commit CVS tag>  Tag: <post-commit CVS tag>
13  Description:  Description:
14    
15    ----------------------------------------------------------------------
16    Name: Matthias Blume
17    Date: 2001/11/22 21:40:00 EST
18    Tag: Release_110_37
19    Description:
20    
21    Release 110.37.  This time for real.
22    
23    ----------------------------------------------------------------------
24    Name: Matthias Blume
25    Date: 2001/11/21 16:35:00 EST
26    Tag: blume-20011121-foot-in-mouth
27    Description:
28    
29    Removed the "Release_110_37" tag because of a serious bug.
30    This will be re-tagged once the bug is fixed.
31    
32    ----------------------------------------------------------------------
33    Name: Matthias Blume
34    Date: 2001/11/21 16:14:00 EST
35    Tag: blume-20011121-forgottenfile
36    Description:
37    
38    Forgot to add a file.  (Just a .tex-file -- part of
39    the CM manual source.)
40    
41    ----------------------------------------------------------------------
42    Name: Matthias Blume
43    Date: 2001/11/21 16:10:00 EST
44    Tag: blume-20011121-invalid_110_37
45    Description:
46    
47    Note: I removed the original tag "Release_110_37" from this commit
48          because we found a serious bug in all non-x86 backends.
49          - Matthias
50    
51    1. Modifications to the SML/NJ code generator and to the runtime system
52       so that code object name strings are directly inserted into code
53       objects at code generation time.  The only business the runtime system
54       has with this is now to read the name strings on occasions.
55       (The encoding of the name string has also changed somewhat.)
56    
57    2. CM now implements a simple "set calculus" for specifying export lists.
58       In particular, it is now possible to refer to the export lists of
59       other libraries/groups/sources and form unions as well as differences.
60       See the latest CM manual for details.
61    
62    3. An separate notion of "proxy" libraries has again be eliminated from
63       CM's model.  (Proxy libraries are now simply a special case of using
64       the export list calculus.)
65    
66    4. Some of the existing libraries now take advantage of the new set
67       calculus.
68       (Notice that not all libraries have been converted because some
69       of the existing .cm-files are supposed to be backward compatible
70       with 110.0.x.)
71    
72    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
73       "General.exnMessage"!  The former relies on a certain hook to be
74       initialized, and that often does not happen in the stand-alone case.)
75    
76    ----------------------------------------------------------------------
77    Name: Lal George
78    Date: 2001/11/21  13:56:18 EST
79    Tag: george-2001121-pseudo-ops
80    Description:
81    
82      Implemented a complete redesign of MLRISC pseudo-ops. Now there
83      ought to never be any question of incompatabilities with
84      pseudo-op syntax expected by host assemblers.
85    
86      For now, only modules supporting GAS syntax are implemented
87      but more should follow, such as MASM, and vendor assembler
88      syntax, e.g. IBM as, Sun as, etc.
89    
90    ----------------------------------------------------------------------
91    Name: Matthias Blume
92    Date: 2001/11/14 11:52:00 EST
93    Tag: blume-20011114-srcname
94    Description:
95    
96    1. Routed the name of the current source file to mlriscgen where it
97       should be directly emitted into the code object.  (This last part
98       is yet to be done.)
99    
100    2. Some cleanup of the pgraph code to make it match the proposal that
101       I put out the other day.  (The proposal notwithstanding, things are
102       still in flux here.)
103    
104    ----------------------------------------------------------------------
105    Name: Lal George
106    Date: 2001/11/14 09:44:04 EST
107    Tag:
108    Description:
109    
110      Fix for a backpatching bug reported by Allen.
111    
112      Because the boundary between short and long span-dependent
113      instructions is +/- 128, there are an astounding number of
114      span-dependent instructions whose size is over estimated.
115    
116      Allen came up with the idea of letting the size of span
117      dependent instructions be non-monotonic, for a maxIter
118      number of times, after which the size must be monotonically
119      increasing.
120    
121      This table shows the number of span-dependent instructions
122      whose size was over-estimated as a function of maxIter, for the
123      file Parse/parse/ml.grm.sml:
124    
125         maxIter            # of instructions:
126            10                      687
127            20                      438
128            30                      198
129            40                        0
130    
131      In compiling the compiler, there is no significant difference in
132      compilation speed between maxIter=10 and maxIter=40. Actually,
133      my measurements showed that maxIter=40 was a tad faster than
134      maxIter=10! Also 96% of the  files in the compiler reach a fix
135      point within 13 iterations, so fixing maxIter at 40, while high,
136      is okay.
137    
138    ----------------------------------------------------------------------
139    Name: Matthias Blume
140    Date: 2001/10/31 15:25:00 EST
141    Tag: blume-20011031-pgraph
142    Description:
143    
144    CKIT:
145    * Changed the "Function" constructor of type Ast.ctype to carry optional
146      argument identifiers.
147    * Changed the return type of TypeUtil.getFunction accordingly.
148    * Type equality ignores the argument names.
149    * TypeUtil.composite tries to preserve argument names but gives up quickly
150      if there is a mismatch.
151    
152    installation script:
153    * attempts to use "curl" if available (unless "wget" is available as well)
154    
155    CM:
156    * has an experimental implementation of "portable graphs" which I will
157      soon propose as an implementation-independent library format
158    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
159    
160    NLFFI-LIB:
161    * some cleanup (all cosmetic)
162    
163    NLFFIGEN:
164    * temporarily disabled the mechanism that suppresses ML output for
165      C definitions whose identifiers start with an underscore character
166    * generate val bindings for enum constants
167    * user can request that only one style (light or heavy) is being used;
168      default is to use both (command-line arguments: -heavy and -light)
169    * fixed bug in handling of function types involving incomplete pointers
170    * generate ML entry points that take record arguments (i.e., using
171      named arguments) for C functions that have a prototype with named
172      arguments
173      (see changes to CKIT)
174    
175    ----------------------------------------------------------------------
176    Name: Allen Leung
177    Date: 2001/10/27 20:34:00 EDT
178    Tag: leunga-20011027-x86-fast-fp-call
179    Description:
180    
181       Fixed the bug described in blume-20010920-slowfp.
182    
183       The fix involves
184          1. generating FCOPYs in FSTP in ia32-svid
185          2. marking a CALL with the appropriate annotation
186    
187    ----------------------------------------------------------------------
188    Name: Matthias Blume
189    Date: 2001/10/16 11:32:00 EDT
190    Tag: blume-20011016-netbsd
191    Description:
192    
193    Underscore patch from Chris Richards (fixing problem with compiling
194    runtime system under recent NetBSD).
195    
196    ----------------------------------------------------------------------
197    Name: Allen Leung
198    Date: 2001/10/12 17:18:32 EDT 2001
199    Tag: leung-20011012-x86-printflowgraph
200    Description:
201    
202    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
203    
204    ----------------------------------------------------------------------
205    Name: Lal George
206    Date: 2001/10/11 23:51:34 EDT
207    Tag: george-20011011-too-many-instrs
208    Description:
209    
210    The representation of a program point never expected to see more
211    than 65536 instructions in a basic block!
212    
213    ----------------------------------------------------------------------
214    Name: Lal George
215    Date: 2001/10/09 09:41:37 EDT
216    Tag: george-20011008-mlrisc-labels
217    Description:
218    
219    Changed the machine description files to support printing of
220    local and global labels in assembly code, based on host assembler
221    conventions.
222    
223    ----------------------------------------------------------------------
224    Name: Matthias Blume
225    Date: 2001/09/25 15:25:00 EDT
226    Tag: blume-20010925-exninfo
227    Description:
228    
229    I provided a non-hook implementation of exnName (at the toplevel) and
230    made the "dummy" implementation of exnMessage (at the toplevel) more
231    useful: if nothing gets "hooked in", then at least you are going to
232    see the exception name and a message indicating why you don't see more.
233    
234    [For the time being, programs that need exnMessage and want to use
235    ml-build should either use General.exnMessage (strongly recommended) or
236    refer to structure General at some other point so that CM sees a
237    static dependency.]
238    
239    [Similar remarks go for "print" and "use":  If you want to use their
240    functionality in stand-alone programs generated by ml-build, then use
241    TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
242    
243    ----------------------------------------------------------------------
244    Name: Matthias Blume
245    Date: 2001/09/20 17:28:00 EDT
246    Tag: blume-20010920-slowfp
247    Description:
248    
249    Allen says that x86-fast-fp is not safe yet, so I turned it off again...
250    
251    ----------------------------------------------------------------------
252    Name: Matthias Blume
253    Date: 2001/09/20 17:20:00 EDT
254    Tag: blume-20010920-canonicalpaths
255    Description:
256    
257    0. Updated the BOOT file (something that I forgot to do earlier).
258    
259    1. Small internal change to CM so that it avoids "/../" in filenames
260       as much as possible (but only where it is safe).
261    
262    2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
263       that arguments that contain delimiters are passed through correctly.
264       This change also means that all "special" arguments of the form
265       @SMLxxx... must come first.
266    
267    3. Changed install script to put relative anchor names for tool commands
268       into pathconfig.
269    
270    ----------------------------------------------------------------------
271    Name: Matthias Blume
272    >>>>>>> 1.169
273    Date: 2001/09/18 15:35:00 EDT
274    Tag: blume-20010918-readme11036
275    Description:
276    
277    Added README files.
278    
279    ----------------------------------------------------------------------
280    Name: Matthias Blume
281    Date: 2001/09/18 11:45:00 EDT
282    Tag: Release_110_36 (retag)
283    Description:
284    
285    Fixed mistake in config/preloads. Retagged as 110.36.
286    
287    ----------------------------------------------------------------------
288    Name: Matthias Blume
289    Date: 2001/09/18 09:40:00 EDT
290    Tag: Release_110_36_orig (tag changed)
291    Description:
292    
293    New version (110.36).  New bootfiles.
294    
295    ----------------------------------------------------------------------
296    Name: Matthias Blume
297    Date: 2001/09/14 16:15:00 EDT
298    Tag: blume-20010914-x86fastfp
299    Description:
300    
301    John committed some changes that Allen made, in particular a (hopefully)
302    correctly working version of the x86-fp module.
303    
304    I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
305    flag to "true".  Everything seems to compile to a fixpoint ok, and
306    "mandelbrot" speeds up by about 15%.
307    
308    ----------------------------------------------------------------------
309    Name: Matthias Blume
310    Date: 2001/09/13 11:20:00 EDT
311    Tag: blume-20010913-minimal
312    Description:
313    
314    1. Stefan Monnier's patch to fix a miscompilation problem that
315       was brought to light by John Reppy's work on Moby.
316    
317    2. Implemented a minimal "structure Compiler" that contains just
318       "version" and "architecture".  The minimal version will be
319       available when the full version is not.  This is for backward-
320       compatibility with code that wants to test Compiler.version.
321    
322    ----------------------------------------------------------------------
323    Name: Matthias Blume
324    Date: 2001/08/28 14:03:00 EDT
325    Tag: blume-20010828-ml-lex
326    Description:
327    
328    Fix for bug 1581, received from Neophytos Michael.
329    
330    ----------------------------------------------------------------------
331    Name: Matthias Blume
332    Date: 2001/08/27 11:20:00 EDT
333    Tag: blume-20010827-readme11035
334    Description:
335    
336    Fleshed out the README file for 110.35.
337    
338    ----------------------------------------------------------------------
339    Name: Matthias Blume
340    Date: 2001/08/24 17:10:00 EDT
341    Tag: Release_110_35
342    Description:
343    
344    New version number (110.35).  New bootfiles.
345    
346    ----------------------------------------------------------------------
347    Name: Lal George
348    Date: 2001/08/24 13:47:18 EDT 2001
349    Tag: george-20010824-MLRISC-graphs
350    Description:
351    
352     removed clusters from MLRISC completely and replaced with graphs.
353    
354    ----------------------------------------------------------------------
355    Name: Matthias Blume
356    Date: 2001/08/23 17:50:00 EDT
357    Tag: blume-20010823-toplevel
358    Description:
359    
360    - some reorganization of the code that implements various kinds of
361      environments in the compiler (static, dynamic, symbolic, combined)
362    - re-implemented the EnvRef module so that evalStream works properly
363      (if the stream contains references to "use", "CM.make", etc.)
364    - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
365    
366    ----------------------------------------------------------------------
367    Name: Matthias Blume
368    Date: 2001/08/20 15:50 EDT
369    Tag: blume20010820-slipup
370    Description:
371    
372    I forgot to commit a few files.  Here they are...
373    
374    ----------------------------------------------------------------------
375    Name: Matthias Blume
376    Date: 2001/08/20 15:35:00 EDT
377    Tag: blume-20010820-debugprof
378    Description:
379    
380    !!!! NEW BOOTFILES !!!!
381    
382    This is another round of reorganizing the compiler sources.  This
383    time the main goal was to factor out all the "instrumentation"
384    passes (for profiling and backtracing) into their own library.
385    The difficulty was to do it in such a way that it does not depend
386    on elaborate.cm but only on elabdata.cm.
387    
388    Therefore there have been further changes to both elaborate.cm and
389    elabdata.cm -- more "generic" things have been moved from the former
390    to the latter.  As a result, I was forced to split the assignment
391    of numbers indicating "primtyc"s into two portions: SML-generic and
392    SML/NJ-specific.  Since it would have been awkward to maintain,
393    I bit the bullet and actually _changed_ the mapping between these
394    numbers and primtycs.  The bottom line of this is that you need
395    a new set of bin- and bootfiles.
396    
397    I have built new bootfiles for all architectures, so doing a fresh
398    checkout and config/install.sh should be all you need.
399    
400    The newly created library's name is
401    
402        $smlnj/viscomp/debugprof.cm
403    
404    and its sources live under
405    
406        src/compiler/DebugProf
407    
408    ----------------------------------------------------------------------
409    Name: Matthias Blume
410    Date: 2001/08/15 17:15:00 EDT
411    Tag: blume-20010815-compreorg
412    Description:
413    
414    This is a first cut at reorganizing the CM libraries that make up the
415    core of the compiler.  The idea is to separate out pieces that could
416    be used independently by tools, e.g., the parser, the typechecker, etc.
417    
418    The current status is a step in this direction, but it is not quite
419    satisfactory yet.  Expect more changes in the future.
420    
421    Here is the current (new) organization...
422    
423        What used to be $smlnj/viscomp/core.cm is now divided into
424        six CM libraries:
425    
426             $smlnj/viscomp/basics.cm
427                           /parser.cm
428                           /elabdata.cm
429                           /elaborate.cm
430                           /execute.cm
431                           /core.cm
432    
433        The CM files for these libraries live under src/system/smlnj/viscomp.
434        All these libraries are proxy libraries that contain precisely
435        one CM library component.  Here are the locations of the components
436        (all within the src/compiler tree):
437    
438             Basics/basics.cm
439             Parse/parser.cm
440             ElabData/elabdata.cm
441             Elaborator/elaborate.cm
442             Execution/execute.cm
443             core.cm
444    
445         [This organization is the same that has been used already
446         for a while for the architecture-specific parts of the visible
447         compiler and for the old version of core.cm.]
448    
449         As you will notice, many source files have been moved from their
450         respective original locations to a new home in one of the above
451         subtrees.
452    
453         The division of labor between the new libraries is the following:
454    
455             basics.cm:
456                - Simple, basic definitions that pertain to many (or all) of
457                  the other libraries.
458             parser.cm:
459                - The SML parser, producing output of type Ast.dec.
460                - The type family for Ast is also defined and exported here.
461             elabdata.cm:
462                - The datatypes that describe input and output of the elaborator.
463                  This includes types, absyn, and static environments.
464             elaborator.cm:
465                - The SML/NJ type checker and elaborator.
466                  This maps an Ast.dec (with a given static environment) to
467                  an Absyn.dec (with a new static environment).
468                - This libraries implements certain modules that used to be
469                  structures as functors (to remove dependencies on FLINT).
470             execute.cm:
471                - Everything having to do with executing binary code objects.
472                - Dynamic environments.
473             core.cm:
474                - SML/NJ-specific instantiations of the elaborator and MLRISC.
475                - Top-level modules.
476                - FLINT (this should eventually become its own library)
477    
478    Notes:
479    
480    I am not 100% happy with the way I separated the elaborator (and its
481    data structures) from FLINT.  Two instances of the same problem:
482    
483        1. Data structures contain certain fields that carry FLINT-specific
484           information.  I hacked around this using exn and the property list
485           module from smlnj-lib.  But the fact that there are middle-end
486           specific fields around at all is a bit annoying.
487    
488        2. The elaborator calculates certain FLINT-related information.  I tried
489           to make this as abstract as I could using functorization, but, again,
490           the fact that the elaborator has to perform calculations on behalf
491           of the middle-end at all is not nice.
492    
493        3. Having to used exn and property lists is unfortunate because it
494           weakens type checking.  The other alternative (parameterizing
495           nearly *everything*) is not appealing, though.
496    
497    I removed the "rebinding =" warning hack because due to the new organization
498    it was awkward to maintain it.  As a result, the compiler now issues some of
499    these warnings when compiling init.cmi during bootstrap compilation. On
500    the plus side, you also get a warning when you do, for example:
501       val op = = Int32.+
502    which was not the case up to now.
503    
504    I placed "assign" and "deref" into the _Core structure so that the
505    code that deals with the "lazy" keyword can find them there.  This
506    removes the need for having access to the primitive environment
507    during elaboration.
508    
509    ----------------------------------------------------------------------
510    Name: Matthias Blume
511    Date: 2001/08/13
512    Tag: blume-20010813-closures
513    Description:
514    
515    This fix was sent to us by Zhong Shao.  It is supposed to improve the
516    performance of certain loops by avoiding needless closure allocation.
517    
518    ----------------------------------------------------------------------
519    Name: Lal George
520    Date: 2001/07/31 10:03:23 EDT 2001
521    Tag: george-20010731-x86-fmalloc
522    Description: Fixed bug in x86 calls
523    
524        There was a bug where call instructions would mysteriously
525        vanish. The call instruction had to be one that returned
526        a floating point value.
527    
528    ----------------------------------------------------------------------
529    Name: Lal George
530    Date: 2001/07/19 16:36:29 EDT 2001
531    Tag: george-20010719-simple-cells
532    Description:
533    
534    I have dramatically simplified the interface for CELLS in MLRISC.
535    
536    In summary, the cells interface is broken up into three parts:
537    
538      1. CellsBasis : CELLS_BASIS
539    
540            CellsBasis is a top level structure and common for all
541            architectures.  it contains the definitions of basic datatypes
542            and utility  functions over these types.
543    
544      2. functor Cells() : CELLS
545    
546            Cells generates an interface for CELLS that incorporates the
547            specific resources on the target architecture, such as the
548            presence of special register classes, their number and size,
549            and various useful substructures.
550    
551      3. <ARCH>CELLS
552    
553            e.g. SparcCells: SPARCCELLS
554    
555            <ARCH>CELLS usually contains additional bindings for special
556            registers  on the architecture, such as:
557    
558                    val r0 : cell           (* register zero *)
559                    val y : cell            (* Y register *)
560                    val psr : cell          (* processor status register *)
561                    ...
562    
563            The structure returned by applying the Cells functor is opened
564            in this interface.
565    
566    The main implication of all this is that the datatypes for cells is
567    split between CellsBasis and CELLS -- a fairly simple change for user
568    code.
569    
570    In the old scheme the CELLS interface had a definitional binding of
571    the form:
572    
573            signature CELLS = sig
574    
575               structure CellsBasis = CellsBasis
576    
577               ...
578    
579            end
580    
581    With all the sharing constraints that goes on in MLRISC, this old
582    design  quickly leads to errors such as:
583    
584            "structure definition spec inside of sharing ... "
585    
586    
587    and appears to require an unacceptable amount of sharing and where
588    constraint hackery.
589    
590    I think this error message (the interaction of definitional specs and
591    sharing) requires more explanation on our web page.
592    
593    ----------------------------------------------------------------------
594    Name: Matthias Blume
595    Date: 2001/07/19 15:00:00 EDT
596    Tag: blume-20010719-libreorg
597    Description:
598    
599    This update puts together a fairly extensive but straightforward change
600    to the way the libraries that implement the interactive system are
601    organized:
602    
603       The biggest change is the elimination of structure Compiler.  As a
604       replacement for this structure, there is now a CM library
605       (known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
606       that exports all the substructures of the original structure Compiler
607       directly.  So instead of saying Compiler.Foo.bar one now simply
608       says Foo.bar.  (The CM libraries actually export a collection of
609       structures that is richer than the collection of substructures of
610       structure Compiler.)
611    
612       To make the transition smooth, there is a separate library called
613       $smlnj/compiler/compiler.cm which puts together and exports the
614       original structure Compiler (or at least something very close to it).
615    
616       There are five members of the original structure Compiler
617       that are not exported directly but which instead became members
618       of a new structure Backend (described by signature BACKEND).  These are:
619       structure Profile (: PROFILE), structure Compile (: COMPILE), structure
620       Interact (: INTERACT), structure Machine (: MACHINE), and val
621       architecture (: string).
622    
623       Structure Compiler.Version has become structure CompilerVersion.
624    
625       Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
626       by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
627       or x86, respectively.
628       Each of these exports the same frontend structures that
629       $smlnj/compiler.cm exports.  But they do not have a structure Backend
630       and instead export some structure <Arch>Backend where <Arch> is Alpha32,
631       Hppa, PPC, Sparc, or X86, respectively.
632    
633       Library $smlnj/compiler/all.cm exports the union of the exports of
634       $smlnj/compiler/<arch>.cm
635    
636       There are no structures <Arch>Compiler anymore, use
637       $smlnj/compiler/<arch>.cm instead.
638    
639       Library host-compiler-0.cm is gone.  Instead, the internal library
640       that instantiates CM is now called cm0.cm.  Selection of the host
641       compiler (backend) is no longer done here but. (Responsibility for it
642       now lies with $smlnj/compiler/current.cm.  This seems to be more
643       logical.)
644    
645       Many individual files have been moved or renamed.  Some files have
646       been split into multiple files, and some "dead" files have been deleted.
647    
648    Aside from these changes to library organization, there are also changes
649    to the way the code itself is organized:
650    
651       Structure Binfile has been re-implemented in such a way that it no
652       longer needs any knowledge of the compiler.  It exclusively deals
653       with the details of binfile layout.  It no longer invokes the
654       compiler (for the purpose of creating new prospective binfile
655       content), and it no longer has any knowledge of how to interpret
656       pickles.
657    
658       Structure Compile (: COMPILE) has been stripped down to the bare
659       essentials of compilation.  It no longer deals with linking/execution.
660       The interface has been cleaned up considerably.
661    
662       Utility routines for dealing with linking and execution have been
663       moved into their own substructures.
664    
665       (The ultimate goal of these changes is to provide a light-weight
666       binfile loader/linker (at least for, e.g., stable libraries) that
667       does not require CM or the compiler to be present.)
668    
669    CM documentation has been updated to reflect the changes to library
670    organization.
671    
672    ----------------------------------------------------------------------
673    Name: Matthias Blume
674    Date: 2001/07/10 17:30:00 EDT
675    Tag: Release_110_34
676    Description:
677    
678    Minor tweak to 110.34 (re-tagged):
679    
680      - README.html file added to CVS repository
681      - runtime compiles properly under FreeBSD 3.X and 4.X
682    
683    ----------------------------------------------------------------------
684    Name: Matthias Blume
685    Date: 2001/07/10 17:30:00 EDT
686    Tag: Release_110_34
687    Description:
688    
689    New version number (110.34). New bootfiles.
690    
691    ----------------------------------------------------------------------
692    Name: Matthias Blume
693    Date: 2001/07/09 16:00:00 EDT
694    Tag: blume-20010709-more-varargs
695    Description:
696    
697    I changed the handling of varargs in ml-nlffigen again:
698    The ellipsis ... will now simply be ignored (with an accompanying warning).
699    
700    The immediate effect is that you can actually call a varargs function
701    from ML -- but you can't actually supply any arguments beyond the ones
702    specified explicitly.  (For example, you can call printf with its format
703    string, but you cannot pass additional arguments.)
704    
705    This behavior is only marginally more useful than the one before, but
706    it has the advantage that a function or, more importantly, a function
707    type never gets dropped on the floor, thus avoiding follow-up problems with
708    other types that refer to the offending one.
709    
710    ----------------------------------------------------------------------
711    Name: Matthias Blume
712    Date: 2001/07/09 11:25:00 EDT
713    Tag: blume-20010709-varargs
714    Description:
715    
716    1. ckit-lib.cm now exports structure Error
717    2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
718       with a warning accompanied by a source location.  Moreover, it
719       merely skips the offending function or type and proceeds with the
720       rest of its work.u  As a result, one can safely feed C code containing
721       "..." to ml-nlffigen.
722    3. There are some internal improvements to CM, providing slightly
723       more general string substitutions in the tools subsystem.
724    
725    ----------------------------------------------------------------------
726    Name: Matthias Blume
727    Date: 2001/06/27 15:10:00 EDT
728    Tag: blume-20010627-concur
729    Description:
730    
731    Fixed a small bug in CM's handling of parallel compilation.
732    (You could observe the bug by Control-C-interrupting an ordinary
733    CMB.make or CM.stabilize and then attaching some compile servers.
734    The result was that all of a sudden the previously interrupted
735    compilation would continue on its own.  This was because of
736    an over-optimization: CM did not bother to clean out certain queues
737    when no servers were attached "anyway", resulting in the contents
738    of these queues to grab control when new servers did get attached.)
739    
740    There is also another minor update to the CM manual.
741    
742    ----------------------------------------------------------------------
743    Name: Matthias Blume
744    Date: 2001/06/26 16:15:00 EDT
745    Tag: blume-20010626-cmdoc
746    Description:
747    
748    Minor typo fixed in CM manual (syntax diagram for libraries).
749    
750    ----------------------------------------------------------------------
751    Name: Matthias Blume
752    Date: 2001/06/25 22:55:00 EDT
753    Tag: blume-20010625-x86pc
754    Description:
755    
756    Fixed a nasty bug in the X86 assembly code that caused signal
757    handlers to fail (crash) randomly.
758    
759    ----------------------------------------------------------------------
760    Name: Matthias Blume
761    Date: 2001/06/25 12:05:00 EDT
762    Tag: blume-20010625-nlffigen
763    Description:
764    
765    This update fixes a number of minor bugs in ml-nlffigen as reported by
766    Nick Carter <nbc@andrew.cmu.edu>.
767    
768      1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
769      2. Default names for generated files are now derived from the name of
770         the C file *without its directory*.  In particular, this causes generated
771         files to be placed locally even if the C file is in some system directory.
772      3. Default names for generated signatures and structures are also derived
773         from the C file name without its directory.  This avoids silly things
774         like "structure GL/GL".
775         (Other silly names are still possible because ml-nlffigen does not do
776          a thorough check of whether generated names are legal ML identifiers.
777          When in doubt, use command line arguments to force particular names.)
778    
779    ----------------------------------------------------------------------
780    Name: Matthias Blume
781    Date: 2001/06/21 12:25:00 EDT
782    Tag: blume-20010621-eXene
783    Description:
784    
785    eXene now compiles and (sort of) works again.
786    
787    The library name (for version > 110.33) is $/eXene.cm.
788    
789    I also added an new example in src/eXene/examples/nbody.  See the
790    README file there for details.
791    
792    ----------------------------------------------------------------------
793    Name: Matthias Blume
794    Date: 2001/06/20 16:40:00 EDT
795    Tag: blume-20010620-cml
796    Description:
797    
798    CML now compiles and works again.
799    
800    Libraries (for version > 110.33):
801    
802      $cml/cml.cm            Main CML library.
803      $cml/basis.cm          CML's version of $/basis.cm.
804      $cml/cml-internal.cm   Internal helper library.
805      $cml/core-cml.cm       Internal helper library.
806      $cml-lib/trace-cml.cm  Tracing facility.
807      $cml-lib/smlnj-lib.cm  CML's version of $/smlnj-lib.cm
808    
809    The installer (config/install.sh) has been taught how to properly
810    install this stuff.
811    
812    ----------------------------------------------------------------------
813    Name: Matthias Blume
814    Date: 2001/06/19 17:55:00 EDT
815    Tag: blume-20010619-instantiate
816    Description:
817    
818    This un-breaks the fix for bug 1432.
819    (The bug was originally fixed in 110.9 but I broke it again some
820    time after that.)
821    
822    ----------------------------------------------------------------------
823    Name: Matthias Blume
824    Date: 2001/06/19 17:25:00 EDT
825    Tag: blume-20010619-signals
826    Description:
827    
828    This should (hopefully) fix the long-standing signal handling bug.
829    (The runtime system was constructing a continuation record with an
830    incorrect descriptor which would cause the GC to drop data on the floor...)
831    
832    ----------------------------------------------------------------------
833    Name: Matthias Blume
834    Date: 2001/06/15 15:05:00 EDT
835    Tag: blume-20010615-moresparc
836    Description:
837    
838    Here is a short late-hour update related to Sparc c-calls:
839    
840     -- made handling of double-word arguments a bit smarter
841    
842     -- instruction selection phase tries to collapse certain clumsily
843        constructed ML-Trees; typical example:
844    
845            ADD(ty,ADD(_,e,LI d1),LI d2)  ->  ADD(ty,e,LI(d1+d2))
846    
847        This currently has no further impact on SML/NJ since mlriscGen does
848        not seem to generate such patterns in the first place, and c-calls
849        (which did generate them in the beginning) has meanwhile been fixed
850        so as to avoid them as well.
851    
852    ----------------------------------------------------------------------
853    Name: Matthias Blume
854    Date: 2001/06/15 15:05:00 EDT
855    Tag: blume-20010615-sparc
856    Description:
857    
858    The purpose of this update is to provide an implementation of NLFFI
859    on Sparc machines.
860    
861    Here are the changes in detail:
862    
863       * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
864       the Sparc implementation of the c-calls API.
865       * The Sparc backend of SML/NJ has been modified to uniformely use %fp
866       for accessing the ML frame.  Thus, we have a real frame pointer and
867       can freely modify %sp without need for an omit-frame-ptr phase.
868       The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
869       this case.
870       * ml-nlffigen has been taught to produce code for different architectures
871       and calling conventions.
872       * In a way similar to what was done in the x86 case, the Sparc
873       backend uses its own specific extension to mltree.  (For example,
874       it needs to be able to generate UNIMP instructions which are part
875       of the calling convention.)
876       * ml-nlffi-lib was reorganized to make it more modular (in particular,
877       to make it easier to plug in new machine- and os-dependent parts).
878    
879    There are some other fairly unrelated bug fixes and cleanups as well:
880    
881       * I further hacked the .cm files for MLRISC tools (like MDLGen) so
882       that they properly share their libraries with existing SML/NJ libraries.
883       * I fixed a minor cosmetic bug in CM, supressing certain spurious
884       follow-up error messages.
885       * Updates to CM/CMB documentation.
886    
887    TODO items:
888    
889       * MLRISC should use a different register as its asmTemp on the Sparc.
890         (The current %o2 is a really bad choice because it is part of the
891         calling conventions, so things might interfere in unexpected ways.)
892    
893    ----------------------------------------------------------------------
894    Name: Matthias Blume
895    Date: 2001/06/07
896    Tag: blume-20010607-calls
897    Description:
898    
899    A number of internal changes related to C calls and calling conventions:
900    
901    1. ML-Tree CALL statements now carry a "pops" field.  It indicates the
902       number of bytes popped implicitly (by the callee).  In most cases
903       this field is 0 but on x86/win32 it is some non-zero value.  This
904       is information provided for the benefit of the "omit-frameptr" pass.
905    2. The CALL instruction on the x86 carries a similar "pops" field.
906       The instruction selection phase copies its value from the ML-Tree
907       CALL statement.
908    3. On all other architectures, the instruction selection phase checks
909       whether "pops=0" and complains if not.
910    4. The c-calls implementation for x86 now accepts two calling conventions:
911       "ccall" and "stdcall".  When "ccall" is selected, the caller cleans
912       up after the call and pops is set to 0.  For "stdcall", the caller
913       does nothing, leaving the cleanup to the callee; pops is set to
914       the number of bytes that were pushed onto the stack.
915    5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
916       distinguish between "ccall" and "stdcall".
917    6. The UNIMP instruction has been added to the supported Sparc instruction
918       set. (This is needed for implementing the official C calling convention
919       on this architecture.)
920    7. I fixed some of the .cm files under src/MLRISC/Tools to make them
921       work with the latest CM.
922    
923    ----------------------------------------------------------------------
924    Name: Matthias Blume
925    Date: 2001/06/05 15:10:00 EDT
926    Tag: blume-20010605-cm-index
927    Description:
928    
929    0. The "lambdasplit" parameter for class "sml" in CM has been documented.
930    
931    1. CM can now generate "index files".  These are human-readable files
932       that list on a per-.cm-file basis each toplevel symbol defined or
933       imported.  The location of the index file for
934       <p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
935       To enable index-file generation, set CM.Control.generate_index to true
936       or export an environment-symbol: export CM_GENERATE_INDEX=true.
937    
938       The CM manual has been updated accordingly.
939    
940    2. I made some slight modifications to the c-calls API in MLRISC.
941    
942         a) There is now a callback to support saving/restoring of
943            dedicated but caller-save registers around the actual call
944            instruction.
945         b) One can optionally specify a comment-annotation for the
946            call instruction.
947    
948    3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
949       (For example, the comment annotation shows the C prototype of
950        the function being called.)
951    
952    ----------------------------------------------------------------------
953    Name: Matthias Blume
954    Date: 2001/06/01 13:30:00 EDT
955    Tag: blume-20010601-nlffi-cleanup
956    Description:
957    
958    This is mostly a cleanup of MLFFI stuff:
959    
960       - some signature files have been put into a more exposed place
961       - the ugly 'f type parameter is gone (simplifies types tremendously!)
962       - ml-nlffigen changed accordingly
963       - tutorial updated
964    
965    Other changes:
966    
967       - author's affiliation in CM manual(s) updated
968       - some more recognized keywords added to Allen's sml.sty
969    
970    ----------------------------------------------------------------------
971    Name: Matthias Blume
972    Date: 2001/05/25 15:30:00 EDT
973    Tag: blume-20010525-iptr
974    Description:
975    
976      - put the official 110.33-README (as it appears on the ftp server) under
977        CVS
978      - fixed a small bug related to incomplete pointer types in
979        ml-nlffigen
980      - small cosmetic change to the ml-nlffi-lib's "arr" type constructor
981        (it does not need the 'f type parameter)
982    
983    ----------------------------------------------------------------------
984    Name: Matthias Blume
985    Date: 2001/05/23 14:30:00 EDT
986    Tag: Release_110_33
987    Description:
988    
989    New version number (110.33).  New bootfiles.
990    
991    ----------------------------------------------------------------------
992    Name: Matthias Blume
993    Date: 2001/05/22 18:06:00 EDT
994    Tag: blume-20010522-targets
995    Description:
996    
997    Made install.sh use file config/targets.customized if it exists, falling
998    back to config/targets if it doesn't.  This way one can have a customized
999    version of the targets file without touching the "real thing", thus
1000    eliminating the constant fear of accidentally checking something bogus
1001    back into the CVS repository...  (File config/targets.customized must
1002    not be added to the repository!)
1003    
1004    ----------------------------------------------------------------------
1005    Name: Matthias Blume
1006    Date: 2001/05/22 16:30:00 EDT
1007    Tag: blume-20010522-minitut
1008    Description:
1009    
1010    1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
1011       struct returns.
1012    2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt.  This is some very
1013       incomplete, preliminary documentation for NLFFI.
1014    
1015    ----------------------------------------------------------------------
1016    Name: Matthias Blume
1017    Date: 2001/05/14 11:30:00 EDT
1018    Tag: blume-20010514-script
1019    Description:
1020    
1021    Some bugs in install script fixed.
1022    
1023    In addition to that I also made a slight change to the NLFFI API:
1024    Functors generated by ml-nlffigen now take the dynamic library as a
1025    straight functor argument, not as a suspended one.  (The original
1026    functor code used to force the suspension right away anyway, so there
1027    was nothing gained by this complication of the interface.)
1028    
1029    ----------------------------------------------------------------------
1030    Name: Matthias Blume
1031    Date: 2001/05/11 14:35:00 EDT
1032    Tag: blume-20010511-ml-nlffi
1033    Description:
1034    
1035    I finally took the plunge and added my new FFI code to the main
1036    repository.  For x86-linux it is now ready for prime-time.
1037    
1038    There are two new subdirectories of "src":
1039    
1040      - ml-nlffi-lib:
1041           The utility library for programs using the FFI interface.
1042           Here is the implementation of $/c.cm and its associated low-level
1043           partners $/c-int.cm and $/memory.cm.
1044      - ml-nlffigen:
1045           A stand-alone program for generating ML glue code from C source
1046           code.
1047    
1048    Building ml-nlffigen requires $/ckit-lib.cm.
1049    
1050    The config/install.sh script has been updates to do the Right Thing
1051    (hopefully).
1052    
1053    Notice that the source tree for the C-Kit will not be put under "src"
1054    but directly under the installation root directory.  (This is the
1055    structure that currently exists on the CVS server when you check out
1056    module "sml".)  Fortunately, config/install.sh knows about this oddity.
1057    
1058    Bugs: No documentation yet.
1059    
1060    ----------------------------------------------------------------------
1061    Name: Matthias Blume
1062    Date: 2001/05/09 16:35:00 EDT
1063    Tag: blume-20010509-cpscontract
1064    Description:
1065    
1066    Fixed a bug in the accounting code in cpsopt/contract.sml.  (The
1067    wrapper/unwrapper elimination did not decrement usage counts and some
1068    dead variables got overlooked by the dead-up logic.)
1069    
1070    ----------------------------------------------------------------------
1071    Name: Lal George
1072    Date: 2001/05/08  17:26:09 EDT
1073    Tag: george-20010508-omit-frameptr
1074    Description:
1075    
1076    Changes to implement the omit-frame-pointer optimization to support
1077    raw C calls. For now, there is only support on the Intel x86, but
1078    other architectures will follow as more experience is gained with this.
1079    
1080    
1081    ----------------------------------------------------------------------
1082    Name: Matthias Blume
1083    Date: 2001/05/07 14:40:00 EDT
1084    Tag: blume-20010507-proxies
1085    Description:
1086    
1087    I made into "proxy libraries" all libraries that qualify for such a
1088    change.  (A qualifying library is a library that has another library or
1089    groups as its sole member and repeats that member's export list
1090    verbatim.  A proxy library avoids this repetition by omitting its export
1091    list, effectively inheriting the list that its (only) member exports.
1092    See the CM manual for more explanation.)
1093    The main effect is that explicit export lists for these libraries
1094    do not have to be kepts in sync, making maintenance a bit easier.
1095    
1096    I also added copyright notices to many .cm-files.
1097    
1098    Last but not least, I made a new set of bootfiles.
1099    
1100    ----------------------------------------------------------------------
1101    Name: Matthias Blume
1102    Date: 2001/05/04 17:00:00 EDT
1103    Tag: blume-20010504-cm-lsplit
1104    Description:
1105    
1106    0. John merged pending changes to $/smlnj-lib.cm
1107    
1108    1. Allen's previous change accidentally backed out of one of Lal's
1109       earlier changes.  I undid this mistake (re-introducing Lal's change).
1110    
1111    2. I used the new topOrder' function from graph-scc.sml (from $/smlnj-lib.cm)
1112       within the compiler where applicable.  There is some code simplification
1113       because of that.
1114    
1115    3. The "split" phase (in FLINT) is now part of the default list of phases.
1116       Compiler.Control.LambdaSplitting.* can be used to globally control the
1117       lambda-splitting (cross-module-inlining) engine.  In addition to that,
1118       it can now also be controlled on a per-source basis: CM has been taught
1119       a new tool parameter applicable to ML source files.
1120    
1121       - To turn lambda-splitting off completely:
1122            local open Compiler.Control.LambdaSplitting in
1123                val _ = set Off
1124            end
1125       - To make "no lambda-splitting" the global default (but allow per-source
1126         overriding); this is the initial setting:
1127            local open Compiler.Control.LambdaSplitting in
1128                val _ = set (Default NONE)
1129            end
1130       - To make "lambda-splitting with aggressiveness a" the global default
1131         (and allow per-source overriding):
1132            local open Compiler.Control.LambdaSplitting in
1133                val _ = set (Default (SOME a))
1134            end
1135    
1136       - To turn lambda-splitting off for a given ML souce file (say: a.sml)
1137         write (in the respective .cm-file):
1138            a.sml (lambdasplitting:off)
1139       - To turn lambda-splitting for a.sml on with minimal aggressiveness:
1140            a.sml (lambdasplitting:on)
1141       - To turn lambda-splitting for a.sml on with aggressiveness <a> (where
1142         <a> is a decimal non-negative integer):
1143            a.sml (lambdasplitting:<a>)
1144       - To turn lambda-splitting for a.sml on with maximal aggressiveness:
1145            a.sml (lambdasplitting:infinity)
1146       - To use the global default for a.sml:
1147            a.sml (lambdasplitting:default)
1148         or simply
1149            a.sml
1150    
1151    ----------------------------------------------------------------------
1152    Name: Allen Leung
1153    Date: 2001/05/04 01:57:00 EDT
1154    Tag: leunga-20010504-sync
1155    Description:
1156    
1157      MLRISC features.
1158    
1159      1. Fix to CMPXCHG instructions.
1160      2. Changed RA interface to allow annotations in callbacks.
1161      3. Added a new method to the stream interface to allow annotations updates.
1162    
1163    ----------------------------------------------------------------------
1164    Name: Matthias Blume
1165    Date: 2001/05/01 11:45:00 EDT
1166    Tag: blume-20010501-pcedittmp
1167    Description:
1168    
1169    Changed install.sh to use the current working directory instead of
1170    /usr/tmp for a temporary file (pcedittmp).  The previous choice
1171    of /usr/tmp caused trouble with MacOS X because of file premission
1172    problems.
1173    
1174    ----------------------------------------------------------------------
1175    Name: Matthias Blume
1176    Date: 2001/04/20 11:10:00 EDT
1177    Tag: blume-20010420-inMLflag
1178    Description:
1179    
1180     - added vp_limitPtrMask to vproc-state.h
1181       (for use by the raw-C-calls mechanism to implement proper interrupt
1182        handling)
1183     - made the ML compiler aware of various data-structure offsets so it
1184       can generate code for accessing the vp_inML flag and vp_limitPtrMask
1185     - tweaked mlriscGen.sml to have it emit interrupt-handling code for
1186       raw C-calls
1187    
1188    ----------------------------------------------------------------------
1189    Name: Lal George
1190    Date: 2001/04/20 09:15:28 EDT
1191    Tag: george-20010420-macosX
1192    Description:
1193    
1194     - Changes to port to Mac OS X; Darwin.
1195    
1196     - In the process I found that sqrt was broken on the PPC, because the
1197       fsqrt instruction is not implemented.
1198    
1199    ----------------------------------------------------------------------
1200    Name: Matthias Blume
1201    Date: 2001/04/18 12:45:00 EDT
1202    Tag: blume-20010418-ccalls
1203    Description:
1204    
1205     - fixed two off-by-4 errors in the x86-specific c-calls implementation
1206       (this bug prevented structure arguments containing pointers from being
1207        passed correctly)
1208     - changed the raw-C-call code in mlriscGen.sml in such a way that
1209       structure arguments are represented as a pointer to the beginning
1210       of the structure (instead of having a series of synthesized arguments,
1211       one for each structure member)
1212    
1213     - made makeml script's verbosity level configurable via environment
1214       variable (MAKEML_VERBOSITY)
1215    
1216     - eliminated placeholder implementations for f32l, w16s, i16s, and f32s
1217       in rawmem-x86.sml; we are now using the real thing
1218    
1219    ----------------------------------------------------------------------
1220    Name: Matthias Blume
1221    Date: 2001/03/22 16:25:00 EST
1222    Tag: blume-20010322-bootfiles
1223    Description:
1224    
1225    Created a new set of bootfiles (for your automatic installation convenience).
1226    
1227    ----------------------------------------------------------------------
1228    Name: Matthias Blume
1229    Date: 2001/03/22 15:10:00 EST
1230    Tag: blume-20010322-rawmem-parcm
1231    Description:
1232    
1233    1. All "raw memory access" primitives for the new FFI are implemented now
1234       (at least on the x86).
1235    2. Some further cleanup of CM's parallel make mechanism.
1236    
1237    ----------------------------------------------------------------------
1238    Name: Matthias Blume
1239    Date: 2001/03/19 17:53:00 EST
1240    Tag: blume-20010319-parallel
1241    Description:
1242    
1243    Parallel make (using compile servers) now works again.
1244    
1245    To this end, CM.stabilize and CMB.make have been modified to work in
1246    two passes when compile servers are attached:
1247       1. Compile everything, do not perform stabilization; this pass
1248          uses compile servers
1249       2. Stabilize everything; this pass does not use compile servers
1250    If there are no compile servers, the two passes are combined into one
1251    (as before).  Splitting the passes increases the inherent parallelism
1252    in the dependency graph because the entire graph including all
1253    libraries is available at the same time.  This, in turn, improves
1254    server utilization.  The downside is that the master process will
1255    have to do some extra work after compilation is done (because for
1256    technical reasons it must re-read all the binfiles during stabilization).
1257    
1258    ----------------------------------------------------------------------
1259    Name: Matthias Blume
1260    Date: 2001/03/16 12:22:00 EST
1261    Tag: blume-20010316-bootfiles
1262    Description:
1263    
1264    Created a new set of bootfiles (for your automatic installation convenience).
1265    
1266    ----------------------------------------------------------------------
1267    Name: Matthias Blume
1268    Date: 2001/03/16 11:00:00 EST
1269    Tag: blume-20010316-MLTREE-fixup
1270    Description:
1271    
1272    This is a minor fixup for an (untagged) earlier commit by Allen.
1273    (A file was missing).
1274    
1275    ----------------------------------------------------------------------
1276    Name: Allen Leung
1277    Date: Mon Mar  5 18:54:57 EST 2001
1278    Tag: leunga-20010305-cut-support
1279    
1280    1. New support for alternative control-flow in MLTREE.
1281       Currently we support
1282    
1283          FLOW_TO(CALL ...., [k1,...,kn])
1284    
1285       This is needed for 'cuts to' in C-- and try/handle-like constructs
1286       in Moby
1287    
1288       New assembler flag "asm-show-cutsto" to turn on control-flow debugging.
1289    
1290    2. Register Allocator
1291    
1292       Changes in interface [from Fermin, John]
1293    
1294    3. Alpha 8-bit SLL support [Fermin]
1295    
1296    4. All architectures
1297    
1298       A new module (ClusterExpandCopies) for expanding parallel copies.
1299    
1300    ----------------------------------------------------------------------
1301    Name: Allen Leung
1302    Date: 2001/02/27 23:07:00 EST
1303    Tag: leunga-20010227-minor-stuff
1304    
1305    1. Alpha bug fix for CMOVNE
1306    2. Handle mltree COND(..,FCMP ...,...)
1307    3. Bug fix in simplifier
1308    
1309    ----------------------------------------------------------------------
1310    Name: Matthias Blume
1311    Date: 2001/01/30 17:50:00 EST
1312    Tag: blume-20010130-sync
1313    Description:
1314    
1315    This is just a minor update to sync my devel branch with the main brach.
1316    The only visible change is the addition of some README files.
1317    
1318    ----------------------------------------------------------------------
1319    Name: Matthias Blume
1320    Date: 2001/01/12 23:30:00 JST
1321    Tag: blume-20010112-bootfiles
1322    Description:
1323    
1324    Made a new set of bootfiles that goes with the current state of the
1325    repository.
1326    
1327    ----------------------------------------------------------------------
1328    Name: Matthias Blume
1329    Date: 2001/01/12 21:20:00 JST
1330    Tag: blume-20010112-sync
1331    Description:
1332    
1333    I am just flushing out some minor changes that had accumulated in
1334    my private branch in order to sync with the main tree.  (This is
1335    mainly because I had CVS trouble when trying to merge _into_ my
1336    private branch.)
1337    
1338    Most people should be completely unaffected by this.
1339    
1340    ----------------------------------------------------------------------
1341    Name: Allen Leung
1342    Date: Thu Jan 11 21:03:00 EST 2001
1343    Tag: leunga-20010111-labexp=mltree
1344    Description:
1345    
1346    1.  Removed the type LabelExp and replace it by MLTree.
1347    2.  Rewritten mltree-simplify with the pattern matcher tool.
1348    3.  There were some bugs in alpha code generator which would break
1349        64-bit code generation.
1350    4.  Redo the tools to generate code with the
1351    5.  The CM files in MLRISC (and in src/system/smlnj/MLRISC)
1352        are now generated by perl scripts.
1353    
1354    ----------------------------------------------------------------------
1355    Name: Matthias Blume
1356    Date: 2001/01/10 21:55:00 JST
1357    Tag: blume-20010110-rcc
1358    Description:
1359    
1360    The RCC stuff now seems to work (but only on the x86).
1361    This required hacking of the c-calls interface (and -implementation) in
1362    MLRISC.
1363    
1364    Normal compiler users should be unaffected.
1365    
1366    ----------------------------------------------------------------------
1367    Name: Matthias Blume
1368    Date: 2001/01/09 01:20:00 JST
1369    Tag: blume-20010109-rcc
1370    Description:
1371    
1372    This is a fairly big patch, flushing out a large number of pending
1373    changes that I made to my development copy over the last couple of days.
1374    
1375    Of practical relevance at this moment is a workaround for a pickling
1376    bug that Allen ran into the other day.  The cause of the bug itself is
1377    still unknown and it might be hard to fix it properly, but the
1378    workaround has some merits of its own (namely somewhat reducing pickling
1379    overhead for certain libraries).  Therefore, I think this solution should
1380    be satisfactory at this time.
1381    
1382    The rest of the changes (i.e., the vast majority) has to do with my
1383    ongoing efforts of providing direct support for C function calls from
1384    ML.  At the moment there is a new primop "RAW_CCALL", typing magic
1385    in types/cproto.sml (invoked from FLINT/trans/translate.sml), a new
1386    case in the FLINT CPS datatype (RCC), changes to cps/convert.sml to
1387    translate uses of RAW_CCALL into RCC, and changes to mlriscGen.sml to
1388    handle RCC.
1389    
1390    The last part (the changes to mlriscGen.sml) are still known to be
1391    wrong on the x86 and not implemented on all other architectures.  But
1392    the infrastructure is in place. I had to change a few functor
1393    signatures in the backend to be able to route the CCalls interface
1394    from MLRISC there, and I had to specialize the mltree type (on the
1395    x86) to include the necessary extensions. (The extensions themselves
1396    were already there and redy to go in MLRISC/x86).
1397    
1398    Everything should be very happy as soon as someone helps me with
1399    mlriscGen.sml...
1400    
1401    In any case, nothing of this should matter to anyone as long as the
1402    new primop is not being used (which is going to be the case unless you
1403    find it where I hid it :). The rest of the compiler is completely
1404    unaffected.
1405    
1406    ----------------------------------------------------------------------
1407    Name: Matthias Blume
1408    Date: 2001/01/05 00:30:00 JST
1409    Tag: blume-20010105-primops
1410    Description:
1411    
1412    Added some experimental support for work that I am doing right now.
1413    These changes mostly concern added primops, but there is also a new
1414    experimental C library in the runtime system (but currently not enabled
1415    anywhere except on Linux/X86).
1416    
1417    In the course of adding primops (and playing with them), I discovered that
1418    Zhong's INL_PRIM hack (no type info for certain primops) was, in fact, badly
1419    broken.  (Zhong was very right he labeled this stuff as "major gross hack".)
1420    To recover, I made type information in INL_PRIM mandatory and changed
1421    prim.sml as well as built-in.sml accordingly.  The InLine structure now
1422    has complete, correct type information (i.e., no bottom types).
1423    
1424    Since all these changes mean that we need new binfiles, I also bumped the
1425    version number to 110.32.1.
1426    
1427    ----------------------------------------------------------------------
1428    Name: Matthias Blume
1429    Date: 2000/12/30 22:10:00 JST
1430    Tag: blume-20001230-various
1431    Description:
1432    
1433    Added proxy libraries for MLRISC and let MLRISC libraries refer
1434    to each other using path anchors.  (See CM manual for explanation.)
1435    
1436    Updated CM documentation.
1437    
1438    Fixed some bugs in CM.
1439    
1440    Implemented "proxy" libraries (= syntactic sugar for CM).
1441    
1442    Added "-quiet" option to makeml and changed runtime system accordingly.
1443    
1444    Added cleanup handler for exportML to reset timers and compiler stats.
1445    
1446    ----------------------------------------------------------------------
1447    Name: Lal George
1448    Date: 2000/12/22 22:22:58 EST 2000
1449    Tag: Release_110_32
1450    Description:
1451    
1452            Infinite precision used throughout MLRISC.
1453            see MLRISC/mltree/machine-int.sig
1454    
1455    ----------------------------------------------------------------------
1456    Name: Matthias Blume
1457    Date: 2000/12/22 23:16:00 JST
1458    Tag: blume-20001222-warn
1459    Description:
1460    
1461    Corrected wording and formatting of some CM warning message which I
1462    broke in my previous patch.
1463    
1464    ----------------------------------------------------------------------
1465    Name: Matthias Blume
1466    Date: 2000/12/22 21:20:00 JST
1467    Tag: blume-20001222-anchorenv
1468    Description:
1469    
1470    Fixed CM's handling of anchor environments in connection with CMB.make.
1471    
1472    ----------------------------------------------------------------------
1473    Name: Matthias Blume
1474    Date: 2000/12/22 13:15:00 JST
1475    Tag: blume-20001222-cleanup
1476    Description:
1477    
1478    Removed src/cm/ffi which does not (and did not) belong here.
1479    
1480    ----------------------------------------------------------------------
1481    Name: Matthias Blume
1482    Date: 2000/12/21 23:55:00 JST
1483    Tag: blume-20001221-exn
1484    Description:
1485    
1486    Probably most important: CM no longer silently swallows all exceptions
1487    in the compiler.
1488    Plus: some other minor CM changes.  For example, CM now reports some
1489    sizes for generated binfiles (code, data, envpickle, lambdapickle).
1490    
1491    ----------------------------------------------------------------------
1492    Name: Matthias Blume
1493    Date: 2000/12/15 00:01:05 JST
1494    Tag: blume-20001215-dirtool
1495    Description:
1496    
1497    - "dir" tool added.
1498    - improvements and cleanup to Tools structure
1499    - documentation updates
1500    
1501    ----------------------------------------------------------------------
1502    Name: Allen Leung
1503    Date: Thu Dec 14 03:45:24 EST 2000
1504    Description:
1505    Tag:  leunga-20001214-int-inf
1506    Description:
1507    
1508       In IntInf, added these standard functions, which are missing from our
1509    implementation:
1510    
1511        andb : int * int -> int
1512        xorb : int * int -> int
1513        orb  : int * int -> int
1514        notb : int -> int
1515         <<   : int * word -> int
1516        ~>>  : int * word -> int
1517    
1518       Not tested, I hope they are correct.
1519    
1520    ----------------------------------------------------------------------
1521    Name: Allen Leung
1522    Date: Fri Dec  8 19:23:26 EST 2000
1523    Description:
1524    Tag:  leunga-20001208-nowhere
1525    Description:
1526    
1527      Slight improvements to the 'nowhere' tool to handle OR-patterns,
1528    to generate better error messages etc.  Plus a brief manual.
1529    
1530    ----------------------------------------------------------------------
1531    Name: Lal George
1532    Date: 2000/12/08 09:54:02 EST 2000
1533    Tag: Release_110_31
1534    Description:
1535    
1536    - Version 110.31
1537    ----------------------------------------------------------------------
1538    Name: Allen Leung
1539    Date: Thu Dec  7 22:01:04 EST 2000
1540    Tag:  leunga-20001207-cell-monster-hack
1541    Description:
1542    
1543    Major MLRISC internal changes.  Affect all clients.
1544    Summary:
1545    
1546    1.  Type CELLS.cell = int is now replaced by a datatype.
1547        As a result, the old regmap is now gone.  Almost all interfaces
1548        in MLRISC change as a consequence.
1549    
1550    2.  A new brand version of machine description tool (v3.0) that generates
1551        modules expecting the new interface.  The old version is removed.
1552    
1553    3.  The RA interface has been further abstracted into two new functors.
1554        RISC_RA and X86RA.  These functors have much simpler interfaces.
1555        [See also directory MLRISC/demo.]
1556    
1557    4.  Some other new source->source code generation tools are available:
1558    
1559        a. MLRISC/Tools/RewriteGen -- generate rewriters from rules.
1560        b. MLRISC/Tools/WhereGen -- expands conditional pattern matching rules.
1561           I use this tool to generate the peephole optimizers---with the new
1562           cell type changes, peephole rules are becoming difficult to write
1563           without conditional pattern matching.
1564    
1565    5.  More Intmap -> IntHashTable change.  Previous changes by Matthias didn't
1566        cover the entire MLRISC source tree so many things broke.
1567    
1568    6.  CM files have been moved to the subdirectory MLRISC/cm.
1569        They are moved because there are a lot of them and they clutter up the
1570        root dir.
1571    
1572    7.  More detailed documentation to come...
1573    
1574        NOTE: To rebuild from 110.30 (ftp distribution), you'll have to do
1575        a makeml -rebuild first.  This is because of other other
1576        changes that Matthias has made (see below).
1577    
1578    
1579    ----------------------------------------------------------------------
1580    Name: Matthias Blume
1581    Date: 2000/11/30 23:12:00 JST
1582    Tag: blume-20001130-filereorg
1583    Description:
1584    
1585    Some manual updates and some file reorganizations in CM.
1586    
1587    ----------------------------------------------------------------------
1588    Name: Matthias Blume
1589    Date: 2000/11/24 17:45:00 JST
1590    Tag: blume-20001124-link
1591    Description:
1592    
1593    Drastically improved link traversal code for the case that the dynamic
1594    value was already loaded at bootstrap time.  As a result, CM and CMB
1595    now both load blazingly fast -- even on a very slow machine.  Also,
1596    memory consumption has been further reduced by this.
1597    
1598    Warning: The format of the PIDMAP file has changed.  THerefore, to
1599    bootstrap you have to do this:
1600    
1601    1. Run CMB.make
1602    2. Make a symbolic link for the boot directory:
1603         ln -s sml.boot.ARCH-OS xxx
1604    3. "Rebuild" the boot directory:
1605         ./makeml -boot xxx -rebuild sml ; rm xxx
1606    4. Boot normally:
1607          ./makeml
1608    
1609    ----------------------------------------------------------------------
1610    Name: Matthias Blume
1611    Date: 2000/11/21 21:20:00 JST
1612    Tag: blume-20001121-tools
1613    Description:
1614    
1615    Continued hacking on autoloading problem -- with success this time.
1616    Also changed tool-plugin mechanism.  See new CM manual.
1617    
1618    ----------------------------------------------------------------------
1619    Name: Matthias Blume
1620    Date: 2000/11/19 14:30:00 JST
1621    Tag:  blume-20001119-autoload
1622    Description:
1623    
1624    Some hacking to make autoloading faster.  Success for CMB, no success
1625    so far for CM.  There is a reduced structure CM' that autoloads faster.
1626    (This is a temporary, non-documented hack to be eliminated again when
1627    the general problem is solved.)
1628    
1629    ----------------------------------------------------------------------
1630    Name: Matthias Blume
1631    Date: 2000/11/17 14:10:00 JST
1632    Tag: blume-20001117-pickle-lib
1633    Description:
1634    
1635    1. Eliminated comp-lib.cm
1636    2. Made pickle-lib.cm
1637    3. Eliminated all uses of intset.sml (from comp-lib.cm)
1638    4. Replaced all uses of intmap.{sig,sml} (from comp-lib.cm) with
1639       equivalent constructs from smlnj-lib.cm (INtHashTable).
1640    5. Point 4. also goes for those uses of intmap.* in MLRISC.
1641       Duplicated intmap modules thrown out.
1642    6. Hunted down all duplicated SCC code and replaced it with
1643       equivalent stuff (GraphSCCFn from smlnj-lib.cm).
1644    7. Rewrote Feedback module.
1645    8. Moved sortedlist.sml into viscomp-lib.cm.  Eventually it
1646       should be thrown out and equivalent modules from smlnj-lib.cm
1647       should be used (IntRedBlackSet, IntListSet, ...).
1648    
1649    Confirmed that compiler compiles to fixpoint.
1650    
1651    ----------------------------------------------------------------------
1652    Name: Allen Leung
1653    Date: 2000/11/10 18:00:00
1654    Tag: leunga-20001110-new-x86-fp
1655    
1656    A new x86 floating point code generator has been added.
1657    By default this is turned off.  To turn this on, do:
1658    
1659        CM.autoload "$smlnj/compiler.cm";
1660        Compiler.Control.MLRISC.getFlag "x86-fast-fp" := true;
1661    
1662    Changes:
1663    
1664    1.  Changed FTAN to FPTAN so that the assembly output is correct.
1665    2.  Changed the extension callback for FTANGENT to generate:
1666    
1667              fptan
1668              fstp  %st(0)
1669        instead of
1670              fptan
1671              fstpl ftempmem
1672    
1673    3.  Numerous assembly fixes for x86.
1674    
1675    5.  Cleaned up the machine code output module x86/x86MC.sml and added
1676        support for a whole bunch of instructions and addressing modes:
1677    
1678          fadd/fsub/fsubr/fmul/fdiv/fdivr  %st, %st(n)
1679          faddp/fsubp/fsubrp/fmulp/fdivp/fdivrp  %st, %st(n)
1680          fadd/fsub/fsubr/fmul/fdiv/fdivr  %st(n), %st
1681          fiadd/fisub/fisubr/fimul/fidiv/fidivr mem
1682          fxch %st(n)
1683          fld %st(n)
1684          fst %st(n)
1685          fst mem
1686          fstp %st(n)
1687          fucom %st(n)
1688          fucomp %st(n)
1689    
1690        All these are now generated when the fast fp mode is turned on.
1691    
1692    6.  Removed the dedicated registers %st(0), ..., %st(7) from X86CpsRegs
1693    
1694    ----------------------------------------------------------------------
1695    Name: Matthias Blume
1696    Date: 2000/11/09 11:20:00 JST
1697    Tag: blume-20001109-scc
1698    Description:
1699    
1700    Eliminated some code duplication:
1701    
1702    1. Added "where" clause to GraphSCCFn in SML/NJ Library.
1703       (Otherwise the functor is useless.)
1704    2. Used GraphSCCFn where SCCUtilFun was used previously.
1705    3. Got rid of SCCUtilFun (in comp-lib.cm).
1706    
1707    ----------------------------------------------------------------------
1708    Name: Lal George
1709    Date: 2000/11/06 09:02:21 EST 2000
1710    Tag: Release_110_30
1711    Description:
1712    
1713    - Version 110.30
1714    ----------------------------------------------------------------------
1715    Name: Matthias Blume
1716    Date: 2000/11/04 14:45:00
1717    Tag: blume-20001104-mlbuild
1718    Description:
1719    
1720    - Made ml-build faster on startup.
1721    - Documentation fixes.
1722    
1723    ----------------------------------------------------------------------
1724    Name: Matthias Blume
1725    Date: 2000/11/02 17:00:00 JST
1726    Tag: blume-20001102-condcomp
1727    Description:
1728    
1729    - Small tweaks to pickler -- new BOOTFILES!
1730    - Version bumped to 110.29.2.
1731    - Added conditional compilation facility to init.cmi (see comment there).
1732    ----------------------------------------------------------------------
1733    Name: Allen Leung
1734    Date: 2000/10/23 19:31:00
1735    Tag: leunga-20001023-demo-ra
1736    
1737    1. Minor RA changes that improves spilling on x86 (affects Moby and C-- only)
1738    2. Test programs for the graph library updated
1739    3. Some new MLRISC demo programs added
1740    
1741    ----------------------------------------------------------------------
1742    Name: Matthias Blume
1743    Date: 2000/08/31 22:15:00 JST
1744    Tag: blume-20001017-errmsg
1745    Description:
1746    
1747    More error message grief: Where there used to be no messages, there
1748    now were some that had bogus error regions.  Fixed.
1749    
1750    ----------------------------------------------------------------------
1751    Name: Matthias Blume
1752    Date: 2000/08/31 17:30:00 JST
1753    Tag: blume-20001017-v110p29p1
1754    Description:
1755    
1756    I made a version 110.29.1 with new bootfiles.
1757    
1758    Changes:  Modified pickler/unpickler for faster and leaner unpickling.
1759              CM documentation changes and a small bugfix in CM's error reporting.
1760    
1761    ----------------------------------------------------------------------
1762    Name: Lal George
1763    Date: 2000/09/27 14:42:35 EDT
1764    Tag: george-20000927-nodestatus
1765    Description:
1766    
1767    Changed the type of the nodestatus, so that:
1768    
1769            SPILLED(~1)             is now SPILLED
1770            SPILLED(m) where m>=0   is now MEMREG(m)
1771            SPILLED(s) where s<~1   is now SPILL_LOC(~s)
1772    
1773    ----------------------------------------------------------------------
1774    Name: Matthias Blume
1775    Date: 2000/09/07 14:45:00 JST
1776    Tag: blume-20000907-cmerrmsg
1777    Description:
1778    
1779    Small tweak to CM to avoid getting ML syntax error messages twice.
1780    
1781    ----------------------------------------------------------------------
1782    Name: Matthias Blume
1783    Date: 2000/08/31 18:00:00 JST
1784    Tag: blume-20000831-cvsbootfiles
1785    Description:
1786    
1787    New URL for boot files (because the 110.29 files on the BL server do
1788    now work correctly with my updated install scripts for yacc and lex).
1789    
1790    ----------------------------------------------------------------------
1791    Name: Matthias Blume
1792    Date: 2000/08/08 12:33:00 JST
1793    Tag: blume-20000808-manual
1794    Description:
1795    
1796    Tiny update to CM manual.
1797    
1798    ----------------------------------------------------------------------
1799    Name: Allen Leung
1800    Date: 2000/08/7 19:31:00
1801    Tag: leunga-20000807-a-whole-bunch-of-stuff
1802    
1803      Moby, C--, SSA, x86, machine descriptions etc.  Should only affect C--
1804    and Mobdy.
1805    
1806    1.  x86
1807    
1808       a.  Fixes to peephole module by John and Dan.
1809       b.  Assembly fix to SETcc by Allen.
1810       c.  Fix to c-call by John.
1811       d.  Fix to spilling by John.  (This one deals with the missing FSTPT case)
1812       e.  Instruction selection optimization to SETcc as suggested by John.
1813    
1814           For example,
1815    
1816            MV(32, x, COND(32, CMP(32, LT, a, b), LI 1, LI 0))
1817    
1818           should generate:
1819    
1820            MOVL a, x
1821            SUBL b, x
1822            SHRL 31, x
1823    
1824    2.  IR stuff
1825    
1826         A bunch of new DJ-graph related algorithms added.  These
1827         speed up SSA construction.
1828    
1829    3.  SSA + Scheduling
1830    
1831         Added code for SSA and scheduling to the repository
1832    
1833    ----------------------------------------------------------------------
1834    Name: Lal George
1835    Date: 2000/07/27 11:53:14 EDT
1836    
1837    Tag: lal-20000727-linux-ppc
1838    Description:
1839    
1840     Made changes to support Linux PPC.
1841     p.s. I have confirmation that the 110.29 boot files work fine.
1842    
1843    ----------------------------------------------------------------------
1844    Name: Matthias Blume
1845    Date: 2000/07/27 17:40:00 JST
1846    Tag: blume-20000727-scripts
1847    Description:
1848    
1849    !!!! WARNING !!!!
1850    You must recompile the runtime system!
1851    !!!! WARNING !!!!
1852    
1853    This is basically another round of script-enhancements:
1854    
1855    1. sml, ml-build, and ml-makedepend accept options -D and -U to define
1856       and undefine CM preprocessor symbols.
1857    
1858    2. ml-build avoids generating a new heap image if it finds that the
1859       existing one is still ok.  (The condition is that no ML file had to
1860       be recompiled and all ML files are found to be older that the heap
1861       file.)
1862    
1863       To make this work smoothly, I also hacked the runtime system as
1864       well as SMLofNJ.SysInfo to get access to the heap image suffix
1865       (.sparc-solaris, ...) that is currently being used.
1866    
1867       Moreover, the signature of CM.mk_standalone has changed.  See the
1868       CM manual.
1869    
1870    3. ml-makedepend accepts additional options -n, -a, and -o.  (See the
1871       CM manual for details.)
1872    
1873    4. More CM manual updates:
1874        - all of the above has been documented.
1875        - there is now a section describing the (CM-related) command line
1876          arguments that are accepted by the "sml" command
1877    
1878    ----------------------------------------------------------------------
1879    Name: Matthias Blume
1880    Date: 2000/07/25 16:20:00 JST
1881    Tag: blume-20000725-makedepend
1882    Description:
1883    
1884    Added a script called ml-makedepend.  This can be used in makefiles
1885    for Unix' make in a way very similar to the "makedepend" command for
1886    C.
1887    
1888    The script internally uses function CM.sources.
1889    
1890    Synopsis:
1891    
1892        ml-makedepend [-f makefile] cmfile targetname
1893    
1894    The default for the makefile is "makefile" (or "Makefile" should
1895    "makefile" not exist).
1896    
1897    ml-makedepend adds a cmfile/targetname-specific section to this
1898    makefile (after removing the previous version of this section).  The
1899    section contains a single dependency specification with targetname on
1900    the LHS (targetname is an arbitrary name), and a list of files derived
1901    from the cmfile on the RHS.  Some of the files on the RHS are
1902    ARCH/OPSYS-specific.  Therefore, ml-makedepend inserts references to
1903    "make" variables $(ARCH) and $(OPSYS) in place of the corresponding
1904    path names.  The makefile writer is responsible for making sure that
1905    these variables have correct at the time "make" is invoked.
1906    
1907    ----------------------------------------------------------------------
1908    Name: Matthias Blume
1909    Date: 2000/07/22 23:30:00 JST
1910    Tag: blume-20000722-urlupdate
1911    Description:
1912    
1913    Changed BOOT and config/srcarchiveurl to point to BL server:
1914    
1915        ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.29/
1916    
1917    ----------------------------------------------------------------------
1918    Name: Matthias Blume
1919    Date: 2000/07/18 18:00:00 JST
1920    Tag: blume-20000718-Version_110_29
1921    Description:
1922    
1923    1. Updated src/compiler/TopLevel/main/version.sml to version 110.29
1924    
1925    2. Updated config/version to 110.29
1926    
1927    3. Updated config/srcarchiveurl
1928    
1929    3. New boot files!
1930       ftp://ftp.cs.princeton.edu/pub/people/blume/sml/110.29-autofetch
1931    
1932    ----------------------------------------------------------------------
1933    Name: Matthias Blume
1934    Date: 2000/07/11 13:58:00 JST
1935    Tag: blume-20000711-doctypo
1936    Description:
1937    
1938    Fixed a few typos in CM manual.
1939    
1940    ----------------------------------------------------------------------
1941    Name: Allen Leung
1942    Date: 2000/06/15 00:38:00
1943    Tag: leunga-20000704-sparc-x86
1944    
1945    1. x86 peephole improvement sp += k; sp -= k => nop  [from John]
1946    2. fix to x86 RET bug [found by Dan Grossman]
1947    3. sparc assembly bug fix for ticc instructions [found by Fermin]
1948    
1949       Affects c-- and moby only
1950    
1951    ----------------------------------------------------------------------
1952    Name: Matthias Blume
1953    Date: 2000/07/04 15:26:00
1954    Tag: blume-20000704-trigger
1955    Description:
1956    
1957    1. Improvements to CM manual.
1958    2. SMLofNJ.Internals.BTrace.trigger reinstated as an alternative way
1959       of getting a back-trace.  The function, when called, raises an
1960       internal exception which explicitly carries the full back-trace history,
1961       so it is unaffected by any intervening handle-raise pairs ("trivial"
1962       or not).  The interactive loop will print that history once it arrives
1963       at top level.
1964       Short of having all exceptions implicitly carry the full history, the
1965       recommended way of using this facility is:
1966         - compile your program with instrumentation "on"
1967         - run it, when it raises an exception, look at the history
1968         - if the history is "cut off" because of some handler, go and modify
1969           your program so that it explicitly calls BTrace.trigger
1970         - recompile (still instrumented), and rerun; look at the full history
1971    
1972    ----------------------------------------------------------------------
1973    Name: Matthias Blume
1974    Date: 2000/07/03 15:36:00 JST
1975    Tag: blume-20000702-manual
1976    Description:
1977    
1978    Small corrections and updates to CM manual.
1979    
1980    ----------------------------------------------------------------------
1981    Name: Matthias Blume
1982    Date: 2000/06/29 16:04:00 JST
1983    Tag: blume-20000629-yacctool
1984    Description:
1985    
1986    Changes:
1987    
1988    1. Class "mlyacc" now takes separate arguments to pass options to
1989       generated .sml- and .sig-files independently.
1990    2. Corresponding CM manual updates.
1991    3. BTrace module now also reports call sites.  (However, for loop clusters
1992       it only shows from where the cluster was entered.)  There are associated
1993       modifications to core.sml, internals.{sig,sml}, btrace.sml, and btimp.sml.
1994    
1995    ----------------------------------------------------------------------
1996    Name: Matthias Blume
1997    Date: 2000/06/27 16:51:00 JST
1998    Tag: blume-20000627-noweb
1999    Description:
2000    
2001    Changes:
2002    
2003     1. Implemented "subdir" and "witness" options for noweb tool.
2004        This caused some slight internal changes in CM's tool implementation.
2005     2. Fixed bug in "tool plugin" mechanism.  This is essentially cleaning
2006        some remaining issues from earlier path anchor changes.
2007     3. Updated CM manual accordingly.
2008    
2009     4. Changed implementation of back-tracing so that I now consider it
2010        ready for prime-time.
2011    
2012        In particular, you don't have to explicitly trigger the back-trace
2013        anymore.  Instead, if you are running BTrace-instrumented code and
2014        there is an uncaught exception (regardless of whether or not it was
2015        raised in instrumented code), the top-level evalloop will print
2016        the back-trace.
2017    
2018        Features:
2019    
2020          - Instrumented and uninstrumented code work together seemlessly.
2021            (Of course, uninstrumented code is never mentioned in actual
2022             back-traces.)
2023    
2024          - Asymptotic time- and space-complexity of instrumented code is
2025            equal to that of uninstrumented code.  (This means that
2026            tail-recursion is preserved by the instrumentation phase.)
2027    
2028          - Modules whose code has been instrumented in different sessions
2029            work together without problem.
2030    
2031          - There is no penalty whatsoever on uninstrumented code.
2032    
2033          - There is no penalty on "raise" expressions, even in
2034            instrumented code.
2035    
2036        A potential bug (or perhaps it is a feature, too):
2037    
2038          A back-trace reaches no further than the outermost instrumented
2039          non-trivial "raise".  Here, a "trivial" raise is one that is the
2040          sole RHS of a "handle" rule.  Thus, back-traces reach trough
2041    
2042               <exp> handle e => raise e
2043    
2044          and even
2045    
2046               <exp> handle Foo => raise Bar
2047    
2048          and, of course, through
2049    
2050               <exp> handle Foo => ...
2051    
2052         if the exception was not Foo.
2053    
2054         Back-traces always reach right through any un-instrumented code
2055         including any of its "handle" expressions, trivial or not.
2056    
2057       To try this out, do the following:
2058    
2059         - Erase all existing binfiles for your program.
2060           (You may keep binfiles for those modules where you think you
2061            definitely don't need back-tracing.)
2062         - Turn on back-trace instrumentation:
2063              SMLofNJ.Internals.BTrace.mode (SOME true);
2064         - Recompile your program.  (I.e., run "CM.make" or "use".)
2065         - You may now turn instrumentation off again (if you want):
2066              SMLofNJ.Internals.BTrace.mode (SOME false);
2067         - Run your program as usual.  If it raises an exception that
2068           reaches the interactive toplevel, then a back-trace will
2069           automatically be printed.  After that, the toplevel loop
2070           will print the exception history as usual.
2071    
2072    ----------------------------------------------------------------------
2073    Name: Matthias Blume
2074    Date: 2000/06/26 09:56:46 JST
2075    Tag: blume-20000626-setup
2076    Description:
2077    
2078    CM: - setup-parameter to "sml" added; this can be used to run arbitrary
2079          ML code before and after compiling a file (e.g., to set compiler
2080          flags)
2081    
2082    Compiler: - improved btrace API (in core.sml, internals.{sig,sml})
2083              - associated changes to btrace.sml (BTrace instrumentation pass)
2084              - cleaner implementation of btimp.sml (BTrace tracing and report
2085                module)
2086    
2087    CM manual: * new path encoding documented
2088               * description of setup-parameter to "sml" added
2089    
2090    The biggest user-visible change to back-tracing is that it is no
2091    longer necessary to compile all traced modules within the same
2092    session.  (This was a real limitation.)
2093    
2094    ----------------------------------------------------------------------
2095    Name: Matthias Blume
2096    Date: 2000/06/24 12:40:00 JST
2097    Tag: blume-20000624-startup
2098    Description:
2099    
2100    Fixes startup slowdown problem.  (I was calling SrcPath.sync a _tad_
2101    bit too often -- to put it mildly. :)
2102    
2103    ----------------------------------------------------------------------
2104    Name: Matthias Blume
2105    Date: 2000/06/23 18:20:00 JST
2106    Tag: blume-20000623-btrace
2107    Description:
2108    
2109    This updates adds a backtrace facility to aid programmers in debugging
2110    their programs.  This involves the following changes:
2111    
2112    1. Module system/smlnj/init/core.sml (structure _Core) now has hooks for
2113       keeping track of the current call stack.  When programs are compiled
2114       in a special mode, the compiler will insert calls to these hooks
2115       into the user program.
2116       "Hook" means that it is possible for different implementations of
2117       back-tracing to register themselves (at different times).
2118    
2119    2. compiler/MiscUtil/profile/btrace.sml implements the annotation phase
2120       as an Absyn.dec->Absyn.dec rewrite.  Normally this phase is turned off.
2121       It can be turned on using this call:
2122         SMLofNJ.Internals.BTrace.mode (SOME true);
2123       Turning it off again:
2124         SMLofNJ.Internals.BTrace.mode (SOME false);
2125       Querying the current status:
2126         SMLofNJ.Internals.BTrace.mode NONE;
2127       Annotated programs are about twice as big as normal ones, and they
2128       run a factor of 2 to 4 slower with a dummy back-trace plugin (one
2129       where all hooks do nothing).  The slowdown with a plugin that is
2130       actually useful (such as the one supplied by default) is even greater,
2131       but in the case of the default plugin it is still only an constant
2132       factor (amortized).
2133    
2134    3. system/Basis/Implementation/NJ/internals.{sig,sml} have been augmented
2135       with a sub-structure BTrace for controlling back-tracing.  In particular,
2136       the above-mentioned function "mode" controls whether the annotation
2137       phase is invoked by the compiler.  Another important function is
2138       "trigger": when called it aborts the current execution and causes
2139       the top-level loop to print a full back-trace.
2140    
2141    4. compiler/MiscUtil/profile/btimp.sml is the current default plugin
2142       for back-tracing.  It keeps track of the dynamic call stack and in
2143       addition to that it keeps a partial history at each "level" of that
2144       stack.  For example, if a tail-calls b, b tail-calls c, and c tail-calls
2145       d and b (at separate times, dynamically), then the report will show:
2146    
2147       GOTO   d
2148             /c
2149       GOTO  \b
2150       CALL   a
2151    
2152       This shows that there was an initial non-tail call of a, then a
2153       tail-call to b or c, looping behavior in a cluster of functions that
2154       consist of b and c, and then a goto from that cluster (i.e., either from
2155       b or from c) to d.
2156    
2157       Note that (depending on the user program) the amount of information
2158       that the back-trace module has to keep track of at each level is bounded
2159       by a constant.  Thus, the whole implementation has the same asymptotical
2160       complexity as the original program (both in space and in time).
2161    
2162    5. compiler/TopLevel/interact/evalloop.sml has been modified to
2163       handle the special exception SMLofNJ.Internals.BTrace.BTrace
2164       which is raised by the "trigger" function mentioned above.
2165    
2166    Notes on usage:
2167    
2168    - Annotated code works well together with unannotated code:
2169    Unannotated calls simply do not show up at all in the backtrace.
2170    
2171    - It is not a good idea to let modules that were annotated during
2172    different sessions run at the same time.  This is because the compiler
2173    chooses small integers to identify individual functions, and there
2174    will be clashes if different modules were compiled in separate sessions.
2175    (Nothing will crash, and you will even be told about the clashes, but
2176    back-trace information will in general not be useful.)
2177    
2178    - Back-tracing can be confused by callcc and capture.
2179    
2180    - The only way of getting a back-trace right now is to explicitly
2181    invoke the "trigger" function from your user program.  Eventually, we
2182    should make every exception carry back-trace information (if
2183    available).  But since this creates more overhead at "raise"-time
2184    (similar to the current exnHistory overhead), I have not yet
2185    implemented this.  (The implementation will be rather easy.)  With
2186    exceptions carrying back-trace information, this facility will be even
2187    more useful because users don't need to modify their programs...
2188    
2189    - While it is possible to compile the compiler with back-trace
2190    annotations turned on (I did it to get some confidence in
2191    correctness), you must make absolutely sure that core.sml and
2192    btimp.sml are compiled WITHOUT annotation!  (core.sml cannot actually
2193    be compiled with annotation because there is no core access yet, but
2194    if you compile btimp.sml with annotation, then the system will go into
2195    an infinite recursion and crash.)
2196    Since CM currently does not know about BTrace, the only way to turn
2197    annotations on and off for different modules of the compiler is to
2198    interrupt CMB.make, change the settings, and re-invoke it.  Of course,
2199    this is awkward and clumsy.
2200    
2201    Sample sessions:
2202    
2203    Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
2204    - SMLofNJ.Internals.BTrace.mode (SOME true);
2205    [autoloading]
2206    [autoloading done]
2207    val it = false : bool
2208    - structure X = struct
2209    -     fun main n = let
2210    -         fun a (x, 0) = d x
2211    -           | a (x, n) = b (x, n - 1)
2212    -         and b (x, n) = c (x, n)
2213    -         and c (x, n) = a (x, n)
2214    -         and d x = e (x, 3)
2215    -         and e (x, 0) = f x
2216    -           | e (x, n) = e (x, n - 1)
2217    -         and f 0 = SMLofNJ.Internals.BTrace.trigger ()
2218    -           | f n = n * g (n - 1)
2219    -         and g n = a (n, 3)
2220    -     in
2221    -         f n
2222    -     end
2223    - end;
2224    structure X : sig val main : int -> int end
2225    - X.main 3;
2226    *** BACK-TRACE ***
2227    GOTO   stdIn:4.2-13.20: X.main[2].f
2228    GOTO-( stdIn:4.2-13.20: X.main[2].e
2229    GOTO   stdIn:4.2-13.20: X.main[2].d
2230         / stdIn:4.2-13.20: X.main[2].a
2231         | stdIn:4.2-13.20: X.main[2].b
2232    GOTO-\ stdIn:4.2-13.20: X.main[2].c
2233    CALL   stdIn:4.2-13.20: X.main[2].g
2234    GOTO   stdIn:4.2-13.20: X.main[2].f
2235    GOTO-( stdIn:4.2-13.20: X.main[2].e
2236    GOTO   stdIn:4.2-13.20: X.main[2].d
2237         / stdIn:4.2-13.20: X.main[2].a
2238         | stdIn:4.2-13.20: X.main[2].b
2239    GOTO-\ stdIn:4.2-13.20: X.main[2].c
2240    CALL   stdIn:4.2-13.20: X.main[2].g
2241    GOTO   stdIn:4.2-13.20: X.main[2].f
2242    GOTO-( stdIn:4.2-13.20: X.main[2].e
2243    GOTO   stdIn:4.2-13.20: X.main[2].d
2244         / stdIn:4.2-13.20: X.main[2].a
2245         | stdIn:4.2-13.20: X.main[2].b
2246    GOTO-\ stdIn:4.2-13.20: X.main[2].c
2247    CALL   stdIn:4.2-13.20: X.main[2].g
2248    GOTO   stdIn:4.2-13.20: X.main[2].f
2249    CALL   stdIn:2.15-17.4: X.main[2]
2250    -
2251    
2252    (Note that because of a FLINt bug the above code currently does not
2253    compile without BTrace turned on.)
2254    
2255    Here is another example, using my modified Tiger compiler:
2256    
2257    Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
2258    - SMLofNJ.Internals.BTrace.mode (SOME true);
2259    [autoloading]
2260    [autoloading done]
2261    val it = false : bool
2262    - CM.make "sources.cm";
2263    [autoloading]
2264    ...
2265    [autoloading done]
2266    [scanning sources.cm]
2267    [parsing (sources.cm):parse.sml]
2268    [creating directory CM/SKEL ...]
2269    [parsing (sources.cm):tiger.lex.sml]
2270    ...
2271    [wrote CM/sparc-unix/semant.sml]
2272    [compiling (sources.cm):main.sml]
2273    [wrote CM/sparc-unix/main.sml]
2274    [New bindings added.]
2275    val it = true : bool
2276    - Main.compile ("../testcases/merge.tig", "foo.out");
2277    *** BACK-TRACE ***
2278    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trvar
2279    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
2280    CALL   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
2281    GOTO   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
2282    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
2283    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
2284    CALL   lib/semant.sml:488.3-505.6: SemantFun[2].transDec.trdec[2].transBody[2]
2285         / lib/semant.sml:411.65-543.8: SemantFun[2].transDec
2286    CALL-\ lib/semant.sml:413.2-540.9: SemantFun[2].transDec.trdec[2]
2287    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
2288    CALL   lib/semant.sml:8.52-558.4: SemantFun[2].transProg[2]
2289    CALL   main.sml:1.18-118.4: Main.compile[2]
2290    -
2291    
2292    ----------------------------------------------------------------------
2293    Name: Matthias Blumen
2294    Date: 2000/06/21 18:00:00 JST
2295    Tag: blume-20000621-manual
2296    Description:
2297    
2298    CM manual update: Path environments documented.
2299    
2300    ----------------------------------------------------------------------
2301    Name: Matthias Blume
2302    Date: 2000/06/19 13:40:00
2303    Tag: blume-20000619-manual
2304    Description:
2305    
2306    CM manual and system/README update.  This only covers the fact that
2307    there are no more implicit anchors.  (Path environments and the "bind"
2308    option to "cm" have yet to be documented.)
2309    
2310    ----------------------------------------------------------------------
2311    Name: Matthias Blume
2312    Date: 2000/06/19 11:05:00 JST
2313    Tag: blume-20000619-chdir-bugfix
2314    Description:
2315    
2316    Fixed a bug in new SrcPath module that sometimes led to a bad chDir call.
2317    
2318    ----------------------------------------------------------------------
2319    Name: Matthias Blume
2320    Date: 2000/06/18 22:00:10 JST
2321    Tag: blume-20000618-implicit-anchors-really-gone
2322    Description:
2323    
2324    I updates the previous HISTORY entry where I forgot to mention that
2325    implicit anchors are no longer with us.
2326    
2327    The current update also gets rid of the (now useless) controller
2328    CM.Control.implicit_anchors.
2329    
2330    ----------------------------------------------------------------------
2331    Name: Matthias Blume
2332    Date: 2000/06/16 17:30:00 JST
2333    Tag: blume-20000616-anchorenv
2334    Description:
2335    
2336    This patch implements the long anticipated (just kidding :) "anchor
2337    environment" mechanism.  In the course of doing this, I also
2338    re-implemented CM's internal "SrcPath" module from scratch.  The new
2339    one should be more robust in certain boundary cases.  In any case, it
2340    is a lot cleaner than its predecessor (IMHO).
2341    
2342    This time, although there is yet another boot file format change, I
2343    kept the unpickler backward-compatible.  As a result, no new bootfiles
2344    are necessary and bootstrapping is straightforward.  (You cannot read
2345    new bootfiles into an old system, but the other way around is no
2346    problem.)
2347    
2348    Visible changes:
2349    
2350    ** 0. Implicit path anchors (without the leading $-symbol) are no
2351    longer recognized at all. This means that such path names are not
2352    illegal either.  For example, the name basis.cm simply refers to a
2353    local file called "basis.cm" (i.e, the name is an ordinary path
2354    relative to .cm-files directory).  Or, to put it differently, only
2355    names that start with $ are anchored paths.
2356    
2357    ** 1. The $<singlearc> abbreviation for $/<singlearc> has finally
2358    vanished.
2359    
2360    John (Reppy) had critizised this as soon as I originally proposed and
2361    implemented it, but at that time I did not really deeply believe
2362    him. :) Now I came full-circle because I need the $<singlearc> syntax
2363    in another place where it cannot be seen as an abbreviation for
2364    $/<singlearc>.  To avoid the confusion, $<singlearc> now means what it
2365    seems to mean (i.e., it "expands" into the corresponding anchor
2366    value).
2367    
2368    However, when paths are used as members in CM description files, it
2369    continues to be true that there must be at least another arc after the
2370    anchor.  This is now enforced separately during semantic analysis
2371    (i.e., from a lexical/syntactical point of view, the notation is ok.)
2372    
2373    ** 2. The "cm" class now accepts an option "bind".  The option's value
2374    is a sub-option list of precisely two items -- one labeled "anchor"
2375    and the other one labeled "value".  As you might expect, "anchor" is
2376    used to specify an anchor name to be bound, and "value" specifies what
2377    the anchor is being bound to.
2378    
2379    The value must be a directory name and can be given in either standard
2380    syntax (including the possibility that it is itself an anchored path)
2381    or native syntax.
2382    
2383    Examples:
2384    
2385       foo.cm (bind:(anchor:bar value:$mystuff/bar))
2386       lib.cm (bind:(anchor:a value:"H:\\x\\y\\z"))  (* only works under windows *)
2387    
2388    and so on.
2389    
2390    The meaning of this is that the .cm-file will be processed with an
2391    augmented anchor environment where the given anchor(s) is/are bound to
2392    the given values(s).
2393    
2394    The rationale for having this feature is this: Suppose you are trying
2395    to use two different (already stable) libraries a.cm and b.cm (that
2396    you perhaps didn't write yourself).  Further, suppose each of these
2397    two libraries internally uses its own auxiliary library $aux/lib.cm.
2398    Normally you would now have a problem because the anchor "lib" can not
2399    be bound to more than one value globally.  Therefore, the project that
2400    uses both a.cm and b.cm must locally redirect the anchor to some other
2401    place:
2402    
2403       a.cm (bind:(anchor:lib value:/usr/lib/smlnj/a-stuff))
2404       b.cm (bind:(anchor:lib value:/usr/lib/smlnj/b-stuff))
2405    
2406    This hard-wires $lib/aux.cm to /usr/lib/smlnj/a-stuff/aux.cm or
2407    /usr/lib/smlnj/b-stuff/aux.cm, respectively.
2408    
2409    Hard-wiring path names is a bit inflexible (and CM will verbosely warn
2410    you when you do so at the time of CM.stabilize).  Therefore, you can
2411    also use an anchored path as the value:
2412    
2413      a.cm (bind:(anchor:lib value:$a-lib))
2414      b.cm (bind:(anchor:lib value:$b-lib))
2415    
2416    Now you can globally configure (using the usual CM.Anchor.anchor or
2417    pathconfig machinery) bindings for "a-lib" and "b-lib".  Since "lib"
2418    itself is always locally bound, setting it globally is no longer
2419    meaningful or necessary (but it does not hurt either).  In fact, "lib"
2420    can still be used as a global anchor for separate purposes.  As a
2421    matter of fact, one can locally define "lib" in terms of a global
2422    "lib":
2423    
2424      a.cm (bind:(anchor:lib value:$lib/a))
2425      b.cm (bind:(anchor:lib value:$lib/b))
2426    
2427    ** 3: The encoding of path names has changed.  This affects the way
2428    path names are shown in CM's progress report and also the internal
2429    protocol encoding used for parallel make.
2430    
2431    The encoding now uses one or more ':'-separated segments.  Each
2432    segments corresponds to a file that has been specified relative to the
2433    file given by its preceding segment.  The first segment is either
2434    relative to the CWD, absolute, or anchored.  Each segment itself is
2435    basically a Unix pathname; all segments but the first are relative.
2436    
2437    Example:
2438    
2439       $foo/bar/baz.cm:a/b/c.sml
2440    
2441    This path denotes the file bar/a/b/c.sml relative to the directory
2442    denoted by anchor "foo".  Notice that the encoding also includes
2443    baz.cm which is the .cm-file that listed a/b/c.sml.  As usual, such
2444    paths are resolved relative to the .cm-files directory, so baz.cm must
2445    be ignored to get the "real" pathname.
2446    
2447    To make this fact more obvious, CM puts the names of such "virtual
2448    arcs" into parentheses when they appear in progress reports. (No
2449    parentheses will appear in the internal protocol encoding.)  Thus,
2450    what you really see is:
2451    
2452      $foo/bar/(baz.cm):a/b/c.sml
2453    
2454    I find this notation to be much more informative than before.
2455    
2456    Another new feature of the encoding is that special characters
2457    including parentheses, colons, (back)slashes, and white space are
2458    written as \ddd (where ddd is the decimal encoding of the character).
2459    
2460    *** The CM manual still needs to be updated.
2461    
2462    ----------------------------------------------------------------------
2463    Name: Allen Leung
2464    Date: 2000/06/15 00:38:00
2465    Tag: leunga-20000615-x86-peephole
2466    
2467    x86 Peephole fix by Fermin.  Affects c-- and moby only.
2468    
2469    ----------------------------------------------------------------------
2470    Name: Matthias Blume
2471    Date: 2000/06/12 11:40:00
2472    Tag: blume-20000612-parmakefix
2473    Description:
2474    
2475    More cleanup after changing the file naming scheme: This time I
2476    repaired the parallel make mechanism for CMB.make which I broke earlier.
2477    
2478    ----------------------------------------------------------------------
2479    Name: Allen Leung
2480    Date: 2000/06/09 01:25:00
2481    Tag: leunga-20000609-various
2482    
2483    None of these things should affect normal SML/NJ operations
2484    
2485    1. Peephole improvements provided by Fermin (c--)
2486    2. New annotation DEFUSE for adding extra dependence (moby)
2487    3. New X86 LOCK instructions (moby)
2488    4. New machine description language for reservation tables (scheduling)
2489    5. Fixes to various optimization/analysis modules (branch chaining, dominator
2490       trees etc.)
2491    6. I've changed the CM files so that they can work with versions
2492       110.0.6, 110.25 and 110.28
2493    
2494    ----------------------------------------------------------------------
2495    Name: Matthias Blume
2496    Date: 2000/06/09 12:40:00
2497    Tag: blume-20000609-log
2498    Description:
2499    
2500    - Removed all(?) remaining RCS Log entries from sources.
2501    
2502    - Fixed bug in ml-yacc and ml-lex sources (use explicit anchors for
2503      anchored paths).
2504    
2505    ----------------------------------------------------------------------
2506    Name: Matthias Blume
2507    Date: 2000/06/07 17:00:00 JST
2508    Tag: blume-20000607-no-implicit-anchors
2509    Description:
2510    
2511    1. This update changes the default setting for
2512    CM.Control.implicit_anchors from true to false.  This means that
2513    implicit anchors are no longer permitted by default.  I also tried to
2514    make sure that nothing else still relies on implicit anchors.
2515    (This is the next step on the schedule towards a CM that does not even
2516    have the notion of implicit anchors anymore.)
2517    
2518    2. More CM manual updates.
2519    
2520    3. I managed to track down and fix the pickling bug I mentioned last
2521    time.  Because of the previously existing workaround, this entails no
2522    immediate practical changes.
2523    
2524    ----------------------------------------------------------------------
2525    Name: Matthias Blume
2526    Date: 2000/06/06 11:15:00 JST
2527    Tag: blume-20000606-lazierpickle
2528    Description:
2529    
2530    !!!! NEW BOOT FILES !!!!
2531    
2532    * The main purpose of this update is to make library pickles lazier in
2533    order to reduce the initial space penalty for autoloading a library.
2534    As a result, it is now possible to have $smlnj/compiler.cm
2535    pre-registered.  This should take care of the many complaints or
2536    inquiries about missing structure Compiler.  This required changes to
2537    CM's internal data structures and small tweaks to some algorithms.
2538    
2539    As a neat additional effect, it is no longer necessary (for the sake
2540    of lean heap image files) to distinguish between a "minimal" CM and a
2541    "full" CM.  Now, there is only one CM (i.e., the "full" version:
2542    $smlnj/cm.cm aka $smlnj/cm/full.cm), and it is always available at the
2543    interactive top level. ($smlnj/cm/minimal.cm is gone.)
2544    
2545    To make the life of compiler-hackers easier, "makeml" now also
2546    pre-registers $smlnj/cmb.cm (aka $smlnj/cmb/current.cm).  In other
2547    words, after you bootstrap a new sml for the first time, you will not
2548    have to autoload $smlnj/cmb.cm again afterwards.  (The first time
2549    around you will still have to do it, though.)
2550    
2551    * A second change consists of major updates to the CM manual.  There
2552    are now several appendices with summary information and also a full
2553    specification of the CM description file syntax.
2554    
2555    * In directory src/system I added the script "allcross".  This script
2556    invokes sml and cross-compiles the compiler for all supported
2557    architectures.  (Useful when providing a new set of boot files.)
2558    
2559    * There seems to be a latent bug in my "lazy pickles" mechanism.  I
2560    added a small tweak to pickle-util.sml to work around this problem,
2561    but it is not a proper fix yet.  I will investigate further.  (The
2562    effect of the bug was an inflation of library pickle size.)
2563    
2564    * Version number increased to 110.28.1 (to avoid compatibility problems).
2565    
2566    ----------------------------------------------------------------------
2567    Name: Allen Leung
2568    Date: 2000/05/25 17:28 EDT
2569    Tag: leunga-20000525-ra
2570    Description:
2571    
2572      Fixed a bug in freezing phase of the register allocator.
2573    
2574    ----------------------------------------------------------------------
2575    Name: Allen Leung
2576    Date: 2000/05/15 22:53 EDT
2577    Tag: leunga-20000515-alpha-x86-ra
2578    Description:
2579    
2580      1. Alpha
2581    
2582          Slight cleanup.  Removed the instruction SGNXL
2583    
2584      2. X86
2585    
2586          Added the following instructions to the instruction set:
2587    
2588            ROLx, RORx,
2589            BTx, BTSx, BTLx, BTRx,
2590            XCHGx, and variants with the LOCK prefix
2591    
2592      3. Register Allocation
2593    
2594          The module ra-rewrite-with-renaming has been improved.
2595    
2596      These have no effect on SML/NJ.
2597    
2598    ----------------------------------------------------------------------
2599    Name: Matthias Blume
2600    Date: 2000/05/15 16:20:00 JST
2601    Tag: blume-20000515-lightrebuild
2602    Description:
2603    
2604    1. I added an alternative to "-rebuild" to "makeml".  The difference is
2605       that prior to calling CMB.make' the CM-variable "LIGHT" will be
2606       defined.  In effect, the command will not build any cross-compiler
2607       backends and therefore finish more quickly.
2608    
2609       The "fixpt" script also takes a "-light" switch to be able to use
2610       this new facility while compiling for a fixpoint.
2611    
2612    2. I replaced all mentions of anchored paths in group owner specifications
2613       with simple relative paths (usually starting with "..").
2614       The rationale is that a library's internal workings should not be
2615       compromised by the lack of some anchor.  (An anchor is necessary
2616       for someone who wants to refer to the library by an anchored path,
2617       but it should not be necessary to build the same library in the first
2618       place.)
2619    
2620    3. I changed the way CM's tool mechanism determines the shell command
2621       string used for things like ml-yacc etc. so that it does not break
2622       when CM.Control.implicit_anchors is turned off.
2623    
2624    ----------------------------------------------------------------------
2625    Name: Matthias Blume
2626    Date: 2000/05/12 18:20:00 JST
2627    Tag: blume-20000512-ml-build
2628    Description:
2629    
2630    Fixed a bug in config/_ml-build that prevented ml-yacc and ml-lex from
2631    getting installed properly (by config/install.sh).
2632    
2633    ----------------------------------------------------------------------
2634    Name: Matthias Blume
2635    Date: 2000/05/12 17:30:00 JST
2636    Tag: blume-20000512-anchors
2637    Description:
2638    
2639    !!! NEW BOOT FILES !!!
2640    
2641    This change is in preparation of fading out support for "implicitly
2642    anchored path names".  I went through all sources and used the
2643    explicit (and relatively new) $-notation.  See system/README and the
2644    CM manual for more info on this.
2645    
2646    I also modified the anchoring scheme for some things such as "smlnj",
2647    "MLRISC", "cm", etc. to take advantage of the fact that explicit
2648    anchors are more expressive: anchor name and first arc do not have to
2649    coincide.  This entails the following user-visible change:
2650    
2651    You have to write $smlnj/foo/bar instead of smlnj/foo/bar.  In
2652    particular, when you fire up sml with a command-line argument, say,
2653    e.g.:
2654    
2655       sml '$smlnj/cmb.cm'
2656    
2657    At the ML toplevel prompt:
2658    
2659       CM.autoload "$smlnj/cmb.cm";
2660    
2661    There is also a new controller in CM.Control that can be used to turn
2662    off all remaining support for implicit anchors by saying:
2663    
2664        CM.autoload "$smlnj/
2665        #set CM.Control.implicit_anchors false;
2666    
2667    This causes CM to reject implicitly anchored paths.  This is (for the
2668    time being) less permissive than the "final" version where there will
2669    be no more such implicit anchors and relative paths will be just that:
2670    relative.
2671    
2672    The next step (version after next version?) will be to make the
2673    default for CM.Control.implicit_anchors false.  After the dust has
2674    settled, I can then produce the "final" version of this...
2675    
2676    Note: Since bootstrapping is a bit tricky, I provided new boot files.
2677    
2678    ----------------------------------------------------------------------
2679    Name: Matthias Blume
2680    Date: 2000/05/11 16:30:00 JST
2681    Tag: blume-20000511-sources
2682    Description:
2683    
2684    The main change is that I added function CM.sources as a generalized
2685    version of the earlier CM.makedepend.  This entails the following
2686    additional changes:
2687    
2688      - CM.makedepend has been dropped.
2689    
2690      - CM manual has been updated.
2691    
2692      - TOOLS signature and API have been changed.
2693    
2694    ----------------------------------------------------------------------
2695    Name: Allen Leung
2696    Date: 2000/05/10 21:17 EDT
2697    Tag: leunga-20000510-moby-c--ssa
2698    Description:
2699    
2700      Various bug fixes and new features for C--, Moby and MLRISC optimizations.
2701    None of these affect SML/NJ.
2702    
2703    1. Register Allocation
2704    
2705        a. A new ra spilling module (ra/ra-spill-with-renaming) is implemented.
2706           This module tries to remove local (i.e. basic block level) redundancies
2707           during spilling.
2708    
2709        b. A new framework for performing region based register allocation.
2710           Not yet entirely functional.
2711    
2712    2. X86
2713    
2714       a. DefUse for POP was missing the stack pointer [found by Lal]
2715       b. Reload for CALL was incorrect in X86Spill [found by John]
2716       c. Various fixes in X86Spill so that it can be used correctly for
2717          the new spilling module.
2718    
2719    3. SSA/IR
2720    
2721       a. New module ir/dj-dataflow.sml implements elimination based
2722          data flow analysis.
2723    
2724    4. MLRiscGen
2725    
2726       a. Fix for gc type annotation
2727    
2728    5. MDGen
2729    
2730       Various fixes for machine description -> ml code translation.  For ssa
2731       only.
2732    
2733    ----------------------------------------------------------------------
2734    Name: Allen Leung
2735    Date: 2000/05/08 22:17 EDT
2736    Tag: leunga-20000508-labexp
2737    Description:
2738    
2739      Fermin has found a few assembly problems with constant expressions
2740      generated in LabelExp.  Mostly, the problems involve extra parentheses,
2741      which choke on dumb assemblers.  This is his fix.
2742    
2743    ----------------------------------------------------------------------
2744    Name: Dave MacQueen
2745    Date: 2000/04/09 14:00 EDT
2746    Tag: dbm-20000502-Version_110_28
2747    Description:
2748    
2749    1. Updated src/compiler/TopLevel/main/version.sml to version 110.28
2750    
2751    2. Updated config/version to 110.28
2752    
2753    3. Updated config/srcarchiveurl
2754    
2755    3. New boot files!
2756       ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.28/
2757    
2758    ----------------------------------------------------------------------
2759    Name: Matthias Blume
2760    Date: 2000/05/01 19:05:00 JST
2761    Tag: blume-20000501-noweb
2762    Description:
2763    
2764    A new noweb tool has been added.  The existing system is entirely
2765    unaffected by this, but some CM users have asked for renewed noweb
2766    support.  Everything is documented in the CM manual.
2767    
2768    New (plugin) libraries:
2769    
2770       noweb-tool.cm
2771       nw-ext.cm
2772    
2773    ----------------------------------------------------------------------
2774    Name: Dave MacQueen
2775    Date: 2000/04/30 12:40PM EDT
2776    Tag: dbm-20000430-bug_fixes
2777    Description:
2778    
2779    1. Fix for bug 1498
2780       smlnj/src/system/Basis/Implementation/Unsafe/object.sig
2781       smlnj/src/system/Basis/Implementation/Unsafe/object.sml
2782         added toRealArray function
2783       smlnj/src/compiler/MiscUtil/print/ppobj.sml
2784         added check for tag Obj.RealArray to array printing case in ppObj
2785    
2786    2. Fix for bug 1510
2787       smlnj/src/compiler/Semant/types/typesutil.sml
2788         fixed definition of dummyargs (used by equalTycon) so that
2789         dummy args are distinct types
2790    
2791    ----------------------------------------------------------------------
2792    Name: Matthias Blume
2793    Date: 2000/04/30 01:00:00 JST
2794    Tag: blume-20000430-versions
2795    Description:
2796    
2797    1. CM version numbering added.  This is an implementation of Lal's
2798       proposal for adding version numbers and version checking to .cm
2799       files.  Lal said that his proposal was just that -- a proposal.
2800       For the time being I went ahead and implemented it so that people
2801       can comment on it.  Everything is completely backward-compatible
2802       (except for the stable library format, i.e., new bootfiles!).
2803    
2804       As usual, see the CM manual for details.
2805    
2806    2. An alternative syntax for anchored paths has been implemented.
2807       Dave has recently voiced the same concerns that I had when I did
2808       this, so there should be some support.  My take is that eventually
2809       I will let support for the current syntax (where anchors are
2810       "implicit") fade out in favor of the new, explicit syntax.
2811       In order to be backward-compatible, both old and new syntax are
2812       currently supported.
2813    
2814       Again, see the CM manual for details.
2815    
2816    3. Parallel make is trying to be slightly smarter:  When the master
2817       process finds a "bottleneck", i.e., when there is only one
2818       compilation unit that can be compiled and everybody else is
2819       waiting on it, then it will simply compile it directly instead
2820       of clumsily telling one of the slaves to do it.
2821    
2822    4. Support for "unsharing" added.  This is necessary in order to be
2823       able to have two different versions of the same library running
2824       at the same time (e.g., for trying out a new MLRISC while still
2825       having the old MLRISC linked into the current compiler, etc.)
2826       See the CM manual.
2827    
2828    5. Simple "makedepend" functionality added for generating Makefile
2829       dependency information.  (This is rather crude at the moment.
2830       Expect some changes here in the future.)
2831    
2832    6. ".fun" added as a recognized suffix for ML files. Also documented
2833       explicitly in the manual that the fallback behavior (unknown suffix
2834       -> ML file) is not an official feature!
2835    
2836    7. Small changes to the pickler for stable libraries.
2837    
2838    8. Several internal changes to CM (for cleanup/improvement).
2839    
2840    
2841    !!!! NEW BINFILES !!!!
2842    
2843    ----------------------------------------------------------------------
2844    Name: Matthias Blume
2845    Date: 2000/04/28 17:30:00 JST
2846    Tag: blume-20000428-pathconfig
2847    Description:
2848    
2849    1. I changed config/install.sh to remove duplicate entries from the
2850       lib/pathconfig file at the end.  Moreover, the final version of
2851       lib/pathconfig is sorted alphabetically.  The same (sorting) is done
2852       in src/system/installml.
2853    
2854    2. The config/install.sh script now consistently uses relative
2855       pathnames in lib/pathconfig whenever the anchor is in the lib
2856       directory.  (So far this was true for the libraries that come
2857       pre-compiled and bundled as part of the bootfiles but not for
2858       libraries that are compiled by the script itself.)
2859    
2860    ----------------------------------------------------------------------
2861    Name: Matthias Blume
2862    Date: 2000/04/26 13:10:00 JST
2863    Tag: blume-20000426-fun_suffix
2864    Description:
2865    
2866    Added ".fun" as a recognized file name suffix (for ML code).
2867    
2868    ----------------------------------------------------------------------
2869    Name: Allen Leung
2870    Date: 2000/04/25 17:00:00 EST
2871    Tag: leunga-20000425-alpha-ra
2872    Description:
2873    
2874    1. Alpha
2875    
2876        PSEUDOARITH was missing in AlphaRewrite.  This causes an endless loop
2877    in C--.
2878    
2879    2. RA
2880    
2881       Added a flag "ra-dump-size" to print out the size of the flowgraph
2882       and the interference graph.
2883    
2884    ----------------------------------------------------------------------
2885    Name: Dave MacQueen
2886    Date: 2000/04/25/
2887    Tag: dbm-20000425-mlyacc_doc_examples
2888    Description:
2889      Updated mlyacc.tex sections 5 and 7 for SML '97 and CM.
2890      Updated all three examples in src/ml-yacc/examples to run
2891      under 110.* using CM.make.
2892    
2893    ----------------------------------------------------------------------
2894    Name: Allen Leung
2895    Date: 2000/04/20 23:04:00 EST
2896    Tag: leunga-20000420-ssa-c---stuff
2897    Description:
2898    
2899      This update synchronizes my repository with Yale's.  Most of these
2900    changes, however, do not affect SML/NJ at all (the RA is an exception).
2901    
2902    1. Register Allocator
2903    
2904       a. An improvement in the interference graph construction:
2905          Given a copy
2906    
2907                s <- t
2908    
2909          no interference edge between s and t is added for this definition of s.
2910    
2911       b. I've added two new spill heuristic modules that Fermin and I developed
2912          (in the new library RA.cm). These are unused in SML/NJ but maybe
2913          useful for others (Moby?)
2914    
2915    2. X86
2916    
2917       a. Various fixes in the backend provided by Fermin [C--] and Lal.
2918    
2919    3. Alpha
2920    
2921       a. Added the BSR instruction and code generation that goes with it [C--]
2922       b. Other fixes too numerous to recount provided by Fermin [C--]
2923    
2924    4. Regmaps
2925    
2926       a. The regmaps are not initialized with the identity physical bindings
2927          at creation time.  This is unneeded.
2928    
2929    5. MLRISC Optimizations
2930    
2931       a. The DJ-Graph module can now compute the iterated dominance frontiers
2932          intersects with liveness incrementally in linear time! Woohoo!
2933          This is now used in my new SSA construction algorithm.
2934    
2935       b. THe branch reorganization module is now smarter about linear chains of
2936          basic blocks.
2937    
2938    
2939    ----------------------------------------------------------------------
2940    Name: Matthias Blume
2941    Date: 2000/04/12 13:52:00 JST
2942    Tag: blume_main_v110p27_1
2943    Description:
2944    
2945    Changed install.sh script to handle archive files without version number
2946    and to use "boot.<arch>-<os>" instead of "sml.boot.<arch>-<os>" for the
2947    name of the boot file archive.
2948    
2949    ----------------------------------------------------------------------
2950    Name: Dave MacQueen
2951    Date: 2000/04/09 14:00 EDT
2952    Tag: dbm-20000410-Version_110_27
2953    Description:
2954    
2955    1. Updated src/compiler/TopLevel/main/version.sml to version 110.27
2956    
2957    2. Updated src/config/version to 110.27
2958    
2959    3. New boot files!
2960    
2961    ----------------------------------------------------------------------
2962    Name: Allen Leung
2963    Date: 2000/04/09 19:09:00 EST
2964    Tag: leunga-20000409-misc
2965    Description:
2966    
2967    1.  Yet another fix for x86 assembly for idivl, imull, mull and friends.
2968    
2969    2.  Miscellaneous improvements to MLRISC (unused in sml/nj)
2970    
2971    ----------------------------------------------------------------------
2972    Name: Stefan
2973    Date: 2000/04/07 10:00:00 EDT
2974    Tag: monnier-20000406-branch-handling
2975    Description:
2976    
2977    Improved handling of branches (mostly those generated from
2978    polymorphic equality), removed switchoff and changed the
2979    default optimization settings (more cpsopt and less flintopt).
2980    
2981    ----------------------------------------------------------------------
2982    Name: Allen Leung
2983    Date: 2000/04/06 01:30:00 EST
2984    Tag: leunga-20000406-peephole-x86-SSA-2
2985    Description:
2986    
2987       Forgot a few files.
2988    
2989    ----------------------------------------------------------------------
2990    Name: Allen Leung
2991    Date: 2000/04/06 00:36:00 EST
2992    Tag: leunga-20000406-peephole-x86-SSA
2993    Description:
2994    
2995    1.  New Peephole code
2996    
2997    2.  Minor improvement to X86 instruction selection
2998    
2999    3.  Various fixes to SSA and machine description -> code translator
3000    
3001    ----------------------------------------------------------------------
3002    Name: Matthias Blume
3003    Date: 2000/04/05 12:30:00 JST
3004    Tag: blume_main_v110p26p2_3
3005    Description:
3006    
3007    This update just merges three minor cosmetic updates to CM's sources
3008    to get ready for the 110.27 code freeze on Friday.  No functionality
3009    has changed.
3010    
3011    ----------------------------------------------------------------------
3012    Name: Allen Leung
3013    Date: 2000/04/04 19:39:00 EST
3014    Tag: leunga-20000404-x86-asm
3015    Description:
3016    
3017    1.  Fixed a problem in X86 assembly.
3018    
3019        Things like
3020    
3021           jmp %eax
3022           jmp (%eax)
3023    
3024        should be output as
3025    
3026           jmp *%eax
3027           jmp *(%eax)
3028    
3029    2.  Assembly output
3030    
3031          Added a new flag
3032    
3033              "asm-indent-copies" (default to false)
3034    
3035          When this flag is on, parallel copies will be indented an extra level.
3036    
3037    ----------------------------------------------------------------------
3038    Name: Allen Leung
3039    Date: 2000/04/04 03:18:00 EST
3040    Tag: leunga-20000404-C--Moby
3041    Description:
3042    
3043        All of these fixes are related to C--, Moby, and my own optimization
3044        stuff; so they shouldn't affect SML/NJ.
3045    
3046    1.  X86
3047    
3048        Various fixes related floating point, and extensions.
3049    
3050    2.  Alpha
3051    
3052        Some extra patterns related to loads with signed/zero extension
3053        provided by Fermin.
3054    
3055    3.  Assembly
3056    
3057        When generating assembly, resolve the value of client defined constants,
3058        instead of generating symbolic values.  This is controlled by the
3059        new flag "asm-resolve-constants", which is default to true.
3060    
3061    4.  Machine Descriptions
3062    
3063        a. The precedence parser was slightly broken when parsing infixr symbols.
3064        b. The type generalizing code had the bound variables reversed, resulting
3065           in a problem during arity raising.
3066        c. Various fixes in machine descriptions.
3067    
3068    ----------------------------------------------------------------------
3069    Name: Matthias Blume
3070    Date: 2000/04/03 16:05:00 JST
3071    Tag: blume_main_v110p26p2_2
3072    Description:
3073    
3074    I eliminated coreEnv from compInfo.  Access to the "Core" structure is
3075    now done via the ordinary static environment that is context to each
3076    compilation unit.
3077    
3078    To this end, I arranged that instead of "structure Core" as "structure
3079    _Core" is bound in the pervasive environment.  Core access is done via
3080    _Core (which can never be accidentally rebound because _Core is not a
3081    legal surface-syntax symbol).
3082    
3083    The current solution is much cleaner because the core environment is
3084    now simply part of the pervasive environment which is part of every
3085    compilation unit's context anyway.  In particular, this eliminates all
3086    special-case handling that was necessary until now in order to deal
3087    with dynamic and symbolic parts of the core environment.
3088    
3089    Remaining hackery (to bind the "magic" symbol _Core) is localized in the
3090    compilation manager's bootstrap compiler (actually: in the "init group"
3091    handling).  See the comments in src/system/smlnj/init/init.cmi for
3092    more details.
3093    
3094    I also tried to track down all mentions of "Core" (as string argument
3095    to Symbol.strSymbol) in the compiler and replaced them with a
3096    reference to the new CoreSym.coreSym.  Seems cleaner since the actual
3097    name appears in one place only.
3098    
3099    Binfile and bootfile format have not changed, but the switchover from
3100    the old "init.cmi" to the new one is a bit tricky, so I supplied new
3101    bootfiles anyway.
3102    
3103    ----------------------------------------------------------------------
3104    Name: Allen Leung
3105    Date: 2000/04/02 21:17:00 EST
3106    Tag: leunga-20000402-mltree
3107    Description:
3108    
3109       1. Renamed the constructor CALL in MLTREE by popular demand.
3110       2. Added a bunch of files from my repository.  These are currently
3111          used by other non-SMLNJ backends.
3112    
3113    ----------------------------------------------------------------------
3114    Name: Allen Leung
3115    Date: 2000/03/31 21:15:00 EST
3116    Tag: leunga-20000331-aliasing
3117    Description:
3118    
3119    This update contains a rewritten (and hopefully more correct) module
3120    for extracting aliasing information from CPS.
3121    
3122       To turn on this feature:
3123    
3124            Compiler.Control.CG.memDisambiguate := true
3125    
3126       To pretty print the region information with assembly
3127    
3128           Compiler.Control.MLRISC.getFlag "asm-show-region" := true;
3129    
3130       To control how many levels of aliasing information are printed, use:
3131    
3132           Compiler.Control.MLRISC.getInt "points-to-show-level" := n
3133    
3134       The default of n is 3.
3135    
3136    ----------------------------------------------------------------------
3137    Name: David MacQueen
3138    Date: 2000/03/31 11:15:00 EST
3139    Tag: dbm-20000331-runtime_fix
3140    Description:
3141    
3142    This update contains:
3143    
3144    1. runtime/c-lib/c-libraries.c
3145       includes added in revision 1.2 caused compilation errors on hppa-hpux
3146    
3147    2. fix for bug 1556
3148       system/Basis/Implementation/NJ/internal-signals.sml
3149    
3150    ----------------------------------------------------------------------
3151    Name: Matthias Blume
3152    Date: 2000/03/31 18:00:00 JST
3153    Tag: blume_main_v110p26p2_1
3154    Description:
3155    
3156    This update contains:
3157    
3158    1. A small change to CM's handling of stable libraries:
3159       CM now maintains one "global" modmap that is used for all stable
3160       libraries.  The use of such a global modmap maximizes sharing and
3161       minimizes the need for re-traversing parts of environments during
3162       modmap construction.  (However, this has minor impact since modmap
3163       construction seems to account for just one percent or less of total
3164       compile time.)
3165    
3166    2. I added a "genmap" phase to the statistics.  This is where I got the
3167       "one percent" number (see above).
3168    
3169    3. CM's new tool parameter mechanism just became _even_ better. :)
3170       - The parser understands named parameters and recursive options.
3171       - The "make" and "shell" tools use these new features.
3172         (This makes it a lot easier to cascade these tools.)
3173       - There is a small syntax change: named parameters use a
3174    
3175           <name> : ( <option> ... )            or
3176           <name> : <string>
3177    
3178         syntax.  Previously, named parameters were implemented in an
3179         ad-hoc fashion by each tool individually (by parsing strings)
3180         and had the form
3181    
3182           <name>=<string>
3183    
3184       See the CM manual for a full description of these issues.
3185    
3186    ----------------------------------------------------------------------
3187    Name: Matthias Blume
3188    Date: 2000/03/30 18:00:00 JST
3189    Tag: blume_main_v110p26p2_0
3190    Description:
3191    
3192    !!!!! WARNING !!!!!!
3193    !!  New binfiles  !!
3194    !!!!!!!!!!!!!!!!!!!!
3195    
3196    This update contains:
3197    
3198    1. Moderate changes to CM:
3199    
3200       - Changes to CM's tools mechanism.  In particular, it is now possible
3201       to have tools that accept additional "command line" parameters
3202       (specified in the .cm file at each instance where the tool's class is
3203       used).
3204    
3205       This was done to accommodate the new "make" and "shell" tools which
3206       facilitate fairly seamless hookup to portions of code managed using
3207       Makefiles or Shell scripts.
3208    
3209       There are no classes "shared" or "private" anymore.  Instead, the
3210       sharing annotation is now a parameter to the "sml" class.
3211    
3212       There is a bit of generic machinery for implementing one's own
3213       tools that accept command-line parameters.  However, I am not yet fully
3214       satisfied with that part, so expect changes here in the future.
3215    
3216       All existing tools are described in the CM manual.
3217    
3218       - Slightly better error handling.  (CM now suppresses many followup
3219       error messages that tended to be more annoying than helpful.)
3220    
3221    2. Major changes to the compiler's static environment data structures.
3222    
3223       - no CMStaticEnv anymore.
3224            - no CMEnv, no "BareEnvironment" (actually, _only_ BareEnvironment,
3225              but it is called Environment), no conversions between different
3226              kinds of static environments
3227    
3228       - There is still a notion of a "modmap", but such modmaps are generated
3229         on demand at the time when they are needed.  This sounds slow, but I
3230         sped up the code that generates modmaps enough for this not to lead to
3231         a slowdown of the compiler (at least I didn't detect any).
3232    
3233       - To facilitate rapid modmap generation, static environments now
3234         contain an (optional) "modtree" structure.  Modtree annotations are
3235         constructed by the unpickler during unpickling.  (This means that
3236         the elaborator does not have to worry about modtrees at all.)
3237         Modtrees have the advantage that they are compositional in the same
3238         way as the environment data structure itself is compositional.
3239         As a result, modtrees never hang on to parts of an environment that
3240         has already been rendered "stale" by filtering or rebinding.
3241    
3242       - I went through many, many trials and errors before arriving at the
3243         current solution.  (The initial idea of "linkpaths" did not work.)
3244         But the result of all this is that I have touched a lot of files that
3245         depend on the "modules" and "types" data structures (most of the
3246         elaborator). There were a lot of changes during my "linkpath" trials
3247         that could have been reverted to their original state but weren't.
3248         Please, don't be too harsh on me for messing with this code a bit more
3249         than what was strictly necessary...  (I _did_ resist the temptation
3250         of doing any "global reformatting" to avoid an untimely death at
3251         Dave's hands. :)
3252    
3253       - One positive aspect of the previous point:  At least I made sure that
3254         all files that I touched now compile without warnings (other than
3255         "polyEqual").
3256    
3257       - compiler now tends to run "leaner" (i.e., ties up less memory in
3258         redundant modmaps)
3259    
3260    ----------------------------------------------------------------------
3261    Name: Allen Leung
3262    Date: 2000/03/29 18:00:00
3263    Tag: leunga-20000327-mlriscGen_hppa_alpha_x86
3264    Boot files (optional): ftp://react-ilp.cs.nyu.edu/leunga/110.26.1-sml.boot.x86-unix-20000330.tar.gz
3265    Description:
3266    
3267       This update contains *MAJOR* changes to the way code is generated from CPS
3268    in the module mlriscGen, and in various backend modules.
3269    
3270    CHANGES
3271    =======
3272    
3273    1. MLRiscGen: forward propagation fix.
3274    
3275       There was a bug in forward propagation introduced at about the same time
3276       as the MLRISC x86 backend, which prohibits coalescing to be
3277       performed effectively in loops.
3278    
3279       Effect: speed up of loops in RISC architectures.
3280               By itself, this actually slowed down certain benchmarks on the x86.
3281    
3282    2. MLRiscGen:  forward propagating addresses from consing.
3283    
3284       I've changed the way consing code is generated.  Basically I separated
3285       out the initialization part:
3286    
3287            store tag,   offset(allocptr)
3288            store elem1, offset+4(allocptr)
3289            store elem2, offset+8(allocptr)
3290            ...
3291            store elemn, offset+4n(allocptr)
3292    
3293       and the address computation part:
3294    
3295            celladdr <- offset+4+alloctpr
3296    
3297       and move the address computation part
3298    
3299       Effect:  register pressure is generally lower as a result.  This
3300                makes compilation of certain expressions much faster, such as
3301                long lists with non-trivial elements.
3302    
3303                 [(0,0), (0,0), .... (0,0)]
3304    
3305    3. MLRiscGen: base pointer elimination.
3306    
3307        As part of the linkage mechanism, we generate the sequence:
3308    
3309         L:  ...  <- start of the code fragment
3310    
3311         L1:
3312             base pointer <- linkreg - L1 + L
3313    
3314         The base pointer was then used for computing relocatable addresses
3315       in the code fragment.  Frequently (such as in lots of continuations)
3316       this is not needed.  We now eliminate this sequence whenever possible.
3317    
3318         For compile time efficiency, I'm using a very stupid local heuristic.
3319       But in general, this should be done as a control flow analysis.
3320    
3321       Effect:  Smaller code size.  Speed up of most programs.
3322    
3323    4. Hppa back end
3324    
3325         Long jumps in span dependence resolution used to depend on the existence
3326      of the base pointer.
3327    
3328         A jump to a long label L was expanded into the following sequence:
3329    
3330          LDIL %hi(L-8192), %r29
3331          LDO  %lo(L-8192)(%r29), %r29
3332          ADD  %r29, baseptr, %r29
3333          BV,n %r0(%r29)
3334    
3335         In the presence of change (3) above, this will not work.  I've changed
3336       it so that the following sequence of instructions are generated, which
3337       doesn't mention the base pointer at all:
3338    
3339             BL,n  L', %r29           /* branch and link, L' + 4 -> %r29 */
3340        L':  ADDIL L-(L'+4), %r29     /* Compute address of L */
3341             BV,n  %r0(%r29)          /* Jump */
3342    
3343    5. Alpha back end
3344    
3345          New alpha instructions LDB/LDW have been added, as per Fermin's
3346       suggestions.   This is unrelated to all other changes.
3347    
3348    6. X86 back end
3349    
3350         I've changed andl to testl in the floating point test sequence
3351         whenever appropriate.  The Intel optimization guide states that
3352         testl is preferable to andl.
3353    
3354    7. RA (x86 only)
3355    
3356         I've improved the spill propagation algorithm, using an approximation
3357       of maximal weighted independent sets.   This seems to be necessary to
3358       alleviate the negative effect in light of the slow down in (1).
3359    
3360         I'll write down the algorithm one of these days.
3361    
3362    8. MLRiscGen: frequencies
3363    
3364         I've added an annotation that states that all call gc blocks have zero
3365       execution frequencies.  This improves register allocation on the x86.
3366    
3367    BENCHMARKS
3368    ==========
3369    
3370       I've only perform the comparison on 110.25.
3371    
3372       The platforms are:
3373    
3374        HPPA  A four processor HP machine (E9000) with 5G of memory.
3375        X86   A 300Hhz Pentium II with 128M of memory, and
3376        SPARC An Ultra sparc 2 with 512M of memory.
3377    
3378       I used the following parameters for the SML benchmarks:
3379    
3380                 @SMLalloc
3381         HPPA    256k
3382         SPARC   512k
3383         X86     256k
3384    
3385    COMPILATION TIME
3386    ----------------
3387       Here are the numbers comparing the compilation times of the compilers.
3388       I've only compared 110.25 compiling the new sources versus
3389       a fixpoint version of the new compiler compiling the same.
3390    
3391                     110.25                                  New
3392               Total  Time in RA  Spill+Reload   Total  Time In RA Spill+Reload
3393         HPPA   627s    116s        2684+3584     599s    95s       1003+1879
3394         SPARC  892s    173s        2891+3870     708s    116s      1004+1880
3395         X86    999s    315s       94006+130691   987s    296s    108877+141957
3396    
3397                   110.25         New
3398                Code Size      Code Size
3399         HPPA   8596736         8561421
3400         SPARC  8974299         8785143
3401         X86    9029180         8716783
3402    
3403       So in summary, things are at least as good as before.   Dramatic
3404       reduction in compilation is obtained on the Sparc; I can't explain it,
3405       but it is reproducible.  Perhaps someone should try to reproduce this
3406       on their own machines.
3407    
3408    SML BENCHMARKS
3409    --------------
3410    
3411        On the average, all benchmarks perform at least as well as before.
3412    
3413          HPPA         Compilation Time     Spill+Reload      Run Time
3414                     110.25  New            110.25    New   110.25  New
3415    
3416          barnesHut  3.158  3.015  4.75%    1+1       0+0   2.980  2.922   2.00%
3417              boyer  6.152  5.708  7.77%    0+0       0+0   0.218  0.213   2.34%
3418       count-graphs  1.168  1.120  4.32%    0+0       0+0  22.705 23.073  -1.60%
3419                fft  0.877  0.792 10.74%    1+3       1+3   0.602  0.587   2.56%
3420        knuthBendix  3.180  2.857 11.32%    0+0       0+0   0.675  0.662   2.02%
3421             lexgen  6.190  5.290 17.01%    0+0       0+0   0.913  0.788  15.86%
3422               life  0.803  0.703 14.22%   25+25      0+0   0.153  0.140   9.52%
3423              logic  2.048  2.007  2.08%    6+6       1+1   4.133  4.008   3.12%
3424         mandelbrot  0.077  0.080 -4.17%    0+0       0+0   0.765  0.712   7.49%
3425             mlyacc 22.932 20.937  9.53%  154+181    32+57  0.468  0.430   8.91%
3426            nucleic  5.183  5.060  2.44%    2+2       0+0   0.125  0.120   4.17%
3427      ratio-regions  3.357  3.142  6.84%    0+0       0+0  116.225 113.173 2.70%
3428                ray  1.283  1.290 -0.52%    0+0       0+0   2.887  2.855   1.11%
3429             simple  6.307  6.032  4.56%   28+30      5+7   3.705  3.658   1.28%
3430                tsp  0.888  0.862  3.09%    0+0       0+0   7.040  6.893   2.13%
3431               vliw 24.378 23.455  3.94%  106+127    25+45  2.758  2.707   1.91%
3432      --------------------------------------------------------------------------
3433       Average                     6.12%                                   4.09%
3434    
3435          SPARC        Compilation Time     Spill+Reload      Run Time
3436                     110.25  New            110.25    New   110.25  New
3437    
3438          barnesHut  3.778  3.592  5.20%    2+2       0+0   3.648  3.453    5.65%
3439              boyer  6.632  6.110  8.54%    0+0       0+0   0.258  0.242    6.90%
3440       count-graphs  1.435  1.325  8.30%    0+0       0+0  33.672 34.737   -3.07%
3441                fft  0.980  0.940  4.26%    3+9       2+6   0.838  0.827    1.41%
3442        knuthBendix  3.590  3.138 14.39%    0+0       0+0   0.962  0.967   -0.52%
3443             lexgen  6.593  6.072  8.59%    1+1       0+0   1.077  1.078   -0.15%
3444               life  0.972  0.868 11.90%   26+26      0+0   0.143  0.140    2.38%
3445              logic  2.525  2.387  5.80%    7+7       1+1   5.625  5.158    9.05%
3446         mandelbrot  0.090  0.093 -3.57%    0+0       0+0   0.855  0.728   17.39%
3447             mlyacc 26.732 23.827 12.19%  162+189    32+57  0.550  0.560   -1.79%
3448            nucleic  6.233  6.197  0.59%    3+3       0+0   0.163  0.173   -5.77%
3449      ratio-regions  3.780  3.507  7.79%    0+0       0+0 133.993 131.035   2.26%
3450                ray  1.595  1.550  2.90%    1+1       0+0   3.440  3.418    0.63%
3451             simple  6.972  6.487  7.48%   29+32      5+7   3.523  3.525   -0.05%
3452                tsp  1.115  1.063  4.86%    0+0       0+0   7.393  7.265    1.77%
3453               vliw 27.765 24.818 11.87%  110+135    25+45  2.265  2.135    6.09%
3454      ----------------------------------------------------------------------------
3455       Average                     6.94%                                    2.64%
3456    
3457          X86          Compilation Time     Spill+Reload      Run Time
3458                     110.25  New            110.25    New   110.25  New
3459    
3460          barnesHut  5.530  5.420  2.03%  593+893   597+915   3.532  3.440   2.66%
3461              boyer  8.768  7.747 13.19%  493+199   301+289   0.327  0.297  10.11%
3462       count-graphs  2.040  2.010  1.49%  298+394   315+457  26.578 28.660  -7.26%
3463                fft  1.327  1.302  1.92%  112+209   115+210   1.055  0.962   9.71%
3464        knuthBendix  5.218  5.475 -4.69%  451+598   510+650   0.928  0.932  -0.36%
3465             lexgen  9.970  9.623  3.60% 1014+841  1157+885   0.947  0.928   1.97%
3466               life  1.183  1.183  0.00%  162+182   145+148   0.127  0.103  22.58%
3467              logic  3.285  3.512 -6.45%  514+684   591+836   5.682  5.577   1.88%
3468         mandelbrot  0.147  0.143  2.33%   38+41     33+54    0.703  0.690   1.93%
3469             mlyacc 35.457 32.763  8.22% 3496+4564 3611+4860  0.552  0.550   0.30%
3470            nucleic  7.100  6.888  3.07%  239+168   201+158   0.175  0.173   0.96%
3471      ratio-regions  6.388  6.843 -6.65% 1182+257   981+300  120.142 120.345 -0.17%
3472                ray  2.332  2.338 -0.29%  346+398   402+494   3.593  3.540   1.51%
3473             simple  9.912  9.903  0.08% 1475+941  1579+1168  3.057  3.178  -3.83%
3474                tsp  1.623  1.532  5.98%  266+200   250+211   8.045  7.878   2.12%
3475               vliw 33.947 35.470 -4.29% 2629+2774 2877+3171  2.072  1.890   9.61%
3476      ----------------------------------------------------------------------------
3477       Average                     1.22%                                     3.36%
3478    
3479    ----------------------------------------------------------------------
3480    Name: Allen Leung
3481    Date: 2000/03/23 16:25:00
3482    Tag: leunga-20000323-fix_x86_alpha
3483    Description:
3484    
3485    1. X86 fixes/changes
3486    
3487       a.  The old code generated for SETcc was completely wrong.
3488           The Intel optimization guide is VERY misleading.
3489    
3490    2. ALPHA fixes/changes
3491    
3492       a.  Added the instructions LDBU, LDWU, STB, STW as per Fermin's suggestion.
3493       b.  Added a new mode byteWordLoadStores to the functor parameter to Alpha()
3494       c.  Added reassociation code for address computation.
3495    
3496    ----------------------------------------------------------------------
3497    Name: Allen Leung
3498    Date: 2000/03/22 01:23:00
3499    Tag: leunga-20000322-fix_x86_hppa_ra
3500    Description:
3501    
3502    1. X86 fixes/changes
3503    
3504       a.  x86Rewrite bug with MUL3 (found by Lal)
3505       b.  Added the instructions FSTS, FSTL
3506    
3507    2. PA-RISC fixes/changes
3508    
3509       a.  B label should not be a delay slot candidate!  Why did this work?
3510       b.  ADDT(32, REG(32, r), LI n) now generates one instruction instead of two,
3511           as it should be.
3512       c.  The assembly syntax for fstds and fstdd was wrong.
3513       d.  Added the composite instruction COMICLR/LDO, which is the immediate
3514           operand variant of COMCLR/LDO.
3515    
3516    3. Generic MLRISC
3517    
3518       a.  shuffle.sml rewritten to be slightly more efficient
3519       b.  DIV bug in mltree-simplify fixed (found by Fermin)
3520    
3521    4. Register Allocator
3522    
3523       a.  I now release the interference graph earlier during spilling.
3524           May improve memory usage.
3525    
3526  ----------------------------------------------------------------------  ----------------------------------------------------------------------
3527  Name: Matthias Blume  Name: Matthias Blume
3528  Date: 2000/03/14 14:15:32  Date: 2000/03/14 14:15:32
# Line 19  Line 3531 
3531    
3532  1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an  1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an
3533  additional argument called "template" which is an optional string that  additional argument called "template" which is an optional string that
3534  specifiel the layout of the tool command line.  See the CM manual for  specifies the layout of the tool command line.  See the CM manual for
3535  explanation.  explanation.
3536    
3537  2. A special-purpose tool can be "regisitered" by simply dropping the  2. A special-purpose tool can be "registered" by simply dropping the
3538  corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same  corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same
3539  directory where the .cm file lives that uses this tool.  (The  directory where the .cm file lives that uses this tool.  (The
3540  behavior/misfeature until now was to look for the tool description  behavior/misfeature until now was to look for the tool description
# Line 66  Line 3578 
3578  rebuild.  Having sets of the form <base><k>.{bin,boot}.<arch>-unix for  rebuild.  Having sets of the form <base><k>.{bin,boot}.<arch>-unix for
3579  <k>=1,2,... is normally not a good idea when invoking fixpt.  However,  <k>=1,2,... is normally not a good idea when invoking fixpt.  However,
3580  they might be the result of an earlier partial run of fixpt (which  they might be the result of an earlier partial run of fixpt (which
3581  perhaps got accidentially killed).  In this case, fixpt will quickly  perhaps got accidentally killed).  In this case, fixpt will quickly
3582  move through what exists before continuing where it left off earlier,  move through what exists before continuing where it left off earlier,
3583  and, thus, saves a lot of time.  and, thus, saves a lot of time.
3584    
# Line 116  Line 3628 
3628    it from that remote directory.    it from that remote directory.
3629    This should simplify installation further:  For machines that have    This should simplify installation further:  For machines that have
3630    access to the internet, just fetch <version>-config.tgz, unpack it,    access to the internet, just fetch <version>-config.tgz, unpack it,
3631    edit config/targets, and go (run config/install.sh).  The scipt will    edit config/targets, and go (run config/install.sh).  The script will
3632    fetch everything else that it might need all by itself.    fetch everything else that it might need all by itself.
3633    
3634    For CVS users, this mechanism is not relevant for source archives, but    For CVS users, this mechanism is not relevant for source archives, but

Legend:
Removed from v.578  
changed lines
  Added in v.993

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