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 710, Tue Oct 17 08:21:09 2000 UTC revision 1273, Thu Oct 10 17:13:35 2002 UTC
# Line 8  Line 8 
8  The form of an entry should be:  The form of an entry should be:
9    
10  Name:  Name:
11  Date:  Date: yyyy/mm/dd
12  Tag: <post-commit CVS tag>  Tag: <post-commit CVS tag>
13  Description:  Description:
14    
15  ----------------------------------------------------------------------  ----------------------------------------------------------------------
16  Name: Matthias Blume  Name: Matthias Blume
17    Date: 2002/10/10 13:10:00 EDT
18    Tag: blume-20021010-ppc-divs
19    Description:
20    
21    The mltree operator DIVS must be implemented with an overflow check on
22    the PPC because the hardware indicates divide-by-zero using "overflow" as
23    well.
24    
25    ----------------------------------------------------------------------
26    Name: Matthias Blume
27    Date: 2002/07/23 11:45:00
28    Tag: blume-20020723-smlnj-home
29    Description:
30    
31    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
32    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
33    CM_PATHCONFIG is also set), the path configuration file is assumed
34    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
35    the entire tree to some other place and everything will "just work".
36    
37    (Companion commands such as ml-build and ml-makedepend also sense this
38    variable.)
39    
40    ----------------------------------------------------------------------
41    Name: Matthias Blume
42    Date: 2002/07/12 21:19:00 EDT
43    Tag: blume-20020712-liveness
44    Description:
45    
46    Exported two useful "step" functions from liveness module (MLRISC).
47    
48    ----------------------------------------------------------------------
49    Name: Matthias Blume
50    Date: 2002/07/05 16:00 EDT
51    Tag: Release_110_41
52    Description:
53    
54    New working release.  New bootfiles.
55    
56    ----------------------------------------------------------------------
57    Name: Matthias Blume
58    Date: 2002/07/05 10:25:00 EDT
59    Tag: blume-20020705-btimp
60    Description:
61    
62    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
63    other clients can set up backtracing support.
64    
65    ----------------------------------------------------------------------
66    Name: Matthias Blume
67    Date: 2002/06/25 17:23:00 EDT
68    Tag: blume-20020625-fpmax
69    Description:
70    
71    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
72    case.  (The sense of the isNaN test was reversed -- which made min and
73    max always return their first argument.)
74    
75    ----------------------------------------------------------------------
76    Name: Matthias Blume
77    Date: 2002/06/11
78    Tag: blume-20020611-unixpath
79    Description:
80    
81    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
82    
83    ----------------------------------------------------------------------
84    Name: Matthias Blume
85    Date: 2002/06/10 16:35:00 EDT
86    Tag: blume-20020610-ieeereal
87    Description:
88    
89    I back-ported my implementation of IEEEReal.fromString from the
90    idlbasis-devel branch so that we can test it.
91    
92    Another small change is that ppDec tries to give more information
93    than just "<sig>" in the case of functors.  However, this code is
94    broken in some mysterious way if the functor's body's signature
95    has not been declared by ascription but gets inferred from the
96    implementation.  This needs fixing...
97    
98    ----------------------------------------------------------------------
99    Name: Matthias Blume
100    Date: 2002/05/31
101    Tag: blume-20020531-btrace-mode
102    Description:
103    
104    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
105    the wayside when I switched over to using Controls everywhere.)
106    
107    ----------------------------------------------------------------------
108    Name: Lal George
109    Date: 2002/05/23 12:21:40 EDT
110    Tag: george-20020523-visual-labels
111    Description:
112    
113    Labels are now displayed in the graphical output to make
114    the fall-through and target blocks obvious.
115    
116    ----------------------------------------------------------------------
117    Name: Matthias Blume
118    Date: 2002/05/22 11:03:00 EDT
119    Tag: blume-20020522-shrink
120    Description:
121    
122    John tweaked yesterday's fix for 1131 to handle an out-of-memory
123    situation that comes up when allocating huge arrays.
124    
125    ----------------------------------------------------------------------
126    Name: Matthias Blume
127    Date: 2002/05/21 16:00:00 EDT
128    Tag: Release_110_40
129    Description:
130    
131    New working release (110.40).  New bootfiles.
132    
133    [Also: John Reppy fixed GC bug 1131.]
134    
135    ----------------------------------------------------------------------
136    Name: Matthias Blume
137    Date: 2002/05/21 12:35:00 EDT
138    Tag: blume-20020521-cmdoc
139    Description:
140    
141    CM documentation update.
142    
143    ----------------------------------------------------------------------
144    Name: Matthias Blume
145    Date: 2002/05/21 10:55:00 EDT
146    Tag: blume-20020521-misc
147    Description:
148    
149    - John tweaked runtime to be silent on heap export (except when
150      GC messages are on).
151    - I added a few more things (cross-compiling versions of CMB) to
152      config/preloads (as suggestions).
153    
154    ----------------------------------------------------------------------
155    Name: Matthias Blume
156    Date: 2002/05/20 22:25:00 EDT
157    Tag: blume-20020520-controls
158    Description:
159    
160    - Added ControlUtil structure to control-lib.cm.
161    - Use it throughout.
162    - Used Controls facility to define MLRISC controls (as opposed to
163      registering MLRISC control ref cells with Controls after the fact)
164    - Fixed messed-up controls priorities.
165    
166    * Removed again all the stuff from config/preloads that one wouldn't
167      be able to preload at the time the initial heap image is built.
168      (Many libraries, e.g., CML, do not exist yet at this time.  The
169       only libraries that can be preloaded via config/preloads are those
170       that come bundled with the bootfiles.)
171    
172    ----------------------------------------------------------------------
173    Name: Matthias Blume
174    Date: 2002/05/20 10:59:00 EDT
175    Tag: blume-20020520-preloads
176    Description:
177    
178    Added a lot of commented-out suggestions for things to be included
179    in config/preloads.
180    
181    ----------------------------------------------------------------------
182    Name: Allen Leung
183    Date: 2002/05/18 14:20:00 EDT
184    Tag: leunga-20020518-mdl
185    Description:
186    
187    o Made the mdl tool stuff compile and run again.
188    o I've disabled all the stuff that depends on RTL specifications; they
189      are all badly broken anyway.
190    
191    ----------------------------------------------------------------------
192    Name: Matthias Blume
193    Date: 2002/05/17 16:49:00 EDT
194    Tag: blume-20020517-controls
195    Description:
196    
197    0. John Reppy made several modifications to the SML/NJ library.
198       In particular, there is a shiny new controls-lib.cm.
199    
200    1. Pushed new controls interface through compiler so that everything
201       compiles again.
202    
203    2. Added FormatComb and FORMAT_COMB to the CML version of the
204       SML/NJ library (so that CML compiles again).
205    
206    3. Modified init scripts because XXX_DEFAULT environment variables
207       are no longer with us.  (Boot-time initialization is now done
208       using the same environment variables that are also used for
209       startup-time initialization of controls.)
210    
211    ----------------------------------------------------------------------
212    Name: Lal George
213    Date: 2002/05/15 09:20:10 EDT
214    Tag: george-20020515-pseudo-op-decls
215    Description:
216    
217            All pseudo-ops emitted before the first segment declaration
218    such as TEXT, DATA, and BSS directives are assumed to be global
219    declarations and are emitted first in the assembly file. This is
220    useful in a number of situations where one has pseudo-ops that are not
221    specific to any segment, and also works around the constraint that one
222    cannot have client pseudo-ops in the TEXT segment.
223    
224            Because no segment is associated with these declarations it is
225    an error to allocate any space or objects before the first segment
226    directive and an exception will be raised. However, we cannot make
227    this check for client pseudo-ops.
228    
229            These top level declarations are a field in the CFG graph_info.
230    In theory you can continue to add to this field after the CFG has been
231    built -- provided you know what you are doing;-)
232    
233    ----------------------------------------------------------------------
234    Name: Matthias Blume
235    Date: 2002/05/13 16:40:00 EDT
236    Tag: blume-20020513-pp-etc
237    Description:
238    
239    A few minor bugfixes:
240    
241      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
242        (Bogus printouts for redefined bindings still occur.  Compiler
243        bug should no longer occur now.  We need to redo the prettyprinter
244        from scratch.)
245    
246      - CM pathname printer now also adds escape sequences for ( and )
247    
248      - commend and docu fixes for ml-nlffi
249    
250    ----------------------------------------------------------------------
251    Name: Matthias Blume
252    Date: 2002/05/10 16:40:00 EDT
253    Tag: blume-20020510-erg-textio
254    Description:
255    
256    Applied the following bugfix provided by Emden Gansner:
257    
258        Output is corrupted when outputSubstr is used rather than output.
259    
260        The problem occurs when a substring
261    
262            ss = (s, dataStart, dataLen)
263    
264        where dataStart > 0, fills a stream buffer with avail bytes left.
265        avail bytes of s, starting at index dataStart, are copied into the
266        buffer, the buffer is flushed, and then the remaining dataLen-avail
267        bytes of ss are copied into the beginning of the buffer. Instead of
268        starting this copy at index dataStart+avail in s, the current code
269        starts the copy at index avail.
270    
271        Fix:
272        In text-io-fn.sml, change line 695 from
273             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
274        to
275             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
276    
277    ----------------------------------------------------------------------
278    Name: Matthias Blume
279    Date: 2002/04/12 13:55:00 EDT
280    Tag: blume-20020412-assyntax
281    Description:
282    
283    1. Grabbed newer assyntax.h from the XFree86 project.
284    2. Fiddled with how to compile X86.prim.asm without warnings.
285    3. (Very) Minor cleanup in CM.
286    
287    ----------------------------------------------------------------------
288    Name: Matthias Blume
289    Date: 2002/04/01 (no joke!) 17:07:00 EST
290    Tag: blume-20020401-x86div
291    Description:
292    
293    Added full support for div/mod/rem/quot on the x86, using the machine
294    instruction's two results (without clumsily recomputing the remainder)
295    directly where appropriate.
296    
297    Some more extensive power-of-two support was added to the x86 instruction
298    selector (avoiding expensive divs, mods, and muls where they can be
299    replaced with cheaper shifts and masks).  However, this sort of thing
300    ought to be done earlier, e.g., within the CPS optimizer so that
301    all architectures benefit from it.
302    
303    The compiler compiles to a fixed point, but changes might be somewhat
304    fragile nevertheless.  Please, report any strange things that you might
305    see wrt. div/mod/quot/rem...
306    
307    ----------------------------------------------------------------------
308    Name: Matthias Blume
309    Date: 2002/03/29 17:22:00
310    Tag: blume-20020329-div
311    Description:
312    
313    Fixed my broken div/mod logic.  Unfortunately, this means that the
314    inline code for div/mod now has one more comparison than before.
315    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
316    The problem was with quotient = 0, because that alone does not tell
317    us which way the rounding went.  One then has to look at whether
318    remainder and divisor have the same sign...  :(
319    
320    Anyway, I replaced the bootfiles with fresh ones...
321    
322    ----------------------------------------------------------------------
323    Name: Matthias Blume
324    Date: 2002/03/29 14:10:00 EST
325    Tag: blume-20020329-inlprims
326    Description:
327    
328    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
329    
330    Primops have changed. This means that the bin/boot-file formats have
331    changed as well.
332    
333    To make sure that there is no confusion, I made a new version.
334    
335    
336    CHANGES:
337    
338    * removed REMT from mltree (remainder should never overflow).
339    
340    * added primops to deal with divisions of all flavors to the frontend
341    
342    * handled these primops all the way through so they map to their respective
343      MLRISC support
344    
345    * used these primops in the implementation of Int, Int32, Word, Word32
346    
347    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
348    
349    * parameterized INLMIN, INLMAX, and INLABS by a numkind
350    
351    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
352      floating point
353    
354    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
355      and Real (but Real.abs maps to a separate floating-point-only primop)
356    
357    
358    TODO items:
359    
360    * Hacked Alpha32 instruction selection, disabling the selection of REMx
361      instructions because the machine instruction encoder cannot handle
362      them.  (Hppa, PPC, and Sparc instruction selection did not handle
363      REM in the first place, and REM is supported by the x86 machine coder.)
364    
365    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
366      selection phase.  (The two can be streamlined because the hardware
367      delivers both quotient and remainder at the same time anyway.)
368    
369    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
370      (Currently the behavior is inconsistent both across architectures and
371      wrt. the draft Basis spec.)
372    
373    * Word8 should eventually be handled natively, too.
374    
375    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
376      as if there currently is no execution path that could trigger it in
377      SML/NJ.  (The assumptions underlying functions arith and promotable do not
378      hold for things like multiplication and division.)
379    
380    ----------------------------------------------------------------------
381    Name: Matthias Blume
382    Date: 2002/03/27 16:27:00 EST
383    Tag: blume-20020327-mlrisc-divisions
384    Description:
385    
386    Added support for all four division operations (ML's div, mod, quot,
387    and rem) to MLRISC.  In the course of doing so, I also rationalized
388    the naming (no more annoying switch-around of DIV and QUOT), by
389    parameterizing the operation by div_rounding_mode (which can be either
390    DIV_TO_ZERO or DIV_TO_NEGINF).
391    
392    The generic MLTreeGen functor takes care of compiling all four
393    operations down to only round-to-zero div.
394    
395    Missing pieces:
396    
397      * Doing something smarter than relying on MLTreeGen on architectures
398        like, e.g., the x86 where hardware division delivers both quotient and
399        remainder at the same time.  With this, the implementation of the
400        round-to-neginf operations could be further streamlined.
401    
402      * Remove inlining support for div/mod/rem from the frontend and replace it
403        with primops that get carried through to the backend.  Do this for all
404        int and word types.
405    
406    ----------------------------------------------------------------------
407    Name: Matthias Blume
408    Date: 2002/03/25 17:25:00 EST
409    Tag: blume-20020325-divmod
410    Description:
411    
412    I improved (hopefully without breaking them) the implementation of Int.div,
413    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
414    advantage of the following observations:
415    
416      Let  q = x quot y      r = x rem y
417           d = x div  y      m = x mod y
418    
419    where "quot" is the round-to-zero version of integer division that
420    hardware usually provides.  Then we have:
421    
422         r = x - q * y        where neither the * nor the - will overflow
423         d = if q >= 0 orelse x = q * y then q else q - 1
424                              where neither the * nor the - will overflow
425         m = if q >= 0 orelse r = 0 then r else r + y
426                              where the + will not overflow
427    
428    This results in substantial simplification of the generated code.
429    The following table shows the number of CFG nodes and edges generated
430    for
431            fun f (x, y) = x OPER y
432            (* with OPER \in div, mod, quot, rem *)
433    
434    
435        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
436        --------------------------------------------------------
437         div |         24 |         39 |         12 |         16
438         mod |         41 |         71 |         12 |         16
439        quot |          8 |         10 |          8 |         10
440         rem |         10 |         14 |          8 |         10
441    
442    
443    ----------------------------------------------------------------------
444    Name: Matthias Blume
445    Date: 2002/03/25 22:06:00 EST
446    Tag: blume-20020325-cprotobug
447    Description:
448    
449    Fixed a bug in cproto (c prototype decoder).
450    
451    ----------------------------------------------------------------------
452    Name: Matthias Blume
453    Date: 2002/03/25 16:00:00 EST
454    Tag: blume-20020325-raw-primops
455    Description:
456    
457    I did some cleanup to Allen's new primop code and
458    replaced yesterday's bootfiles with new ones.
459    (But they are stored in the same place.)
460    
461    ----------------------------------------------------------------------
462    Name: Matthias Blume
463    Date: 2002/03/24 22:40:00 EST
464    Tag: blume-20020324-bootfiles
465    Description:
466    
467    Made the bootfiles that Allen asked for.
468    
469    ----------------------------------------------------------------------
470    Name: Allen Leung
471    Date: 2002/03/23 15:50:00 EST
472    Tag: leunga-20020323-flint-cps-rcc-primops
473    Description:
474    
475      1. Changes to FLINT primops:
476    
477        (* make a call to a C-function;
478         * The primop carries C function prototype information and specifies
479         * which of its (ML-) arguments are floating point. C prototype
480         * information is for use by the backend, ML information is for
481         * use by the CPS converter. *)
482      | RAW_CCALL of { c_proto: CTypes.c_proto,
483                       ml_args: ccall_type list,
484                       ml_res_opt: ccall_type option,
485                       reentrant : bool
486                     } option
487       (* Allocate uninitialized storage on the heap.
488        * The record is meant to hold short-lived C objects, i.e., they
489        * are not ML pointers.  With the tag, the representation is
490        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
491        *)
492      | RAW_RECORD of {tag:bool,sz:int}
493      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
494    
495      2.  These CPS primops are now overloaded:
496    
497           rawload of {kind:numkind}
498           rawstore of {kind:numkind}
499    
500          The one argument form is:
501    
502             rawload {kind} address
503    
504          The two argument form is:
505    
506             rawload {kind} [ml object, byte-offset]
507    
508      3. RAW_CCALL/RCC now takes two extra arguments:
509    
510         a. The first is whether the C call is reentrant, i.e., whether
511            ML state should be saved and restored.
512         b. The second argument is a string argument specifying the name of
513            library and the C function.
514    
515         These things are currently not handled in the code generator, yet.
516    
517      4. In CProto,
518    
519         An encoding type of "bool" means "ml object" and is mapped into
520         C prototype of PTR.  Note that "bool" is different than "string",
521         even though "string" is also mapped into PTR, because "bool"
522         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
523    
524      5. Pickler/unpicker
525    
526         Changed to handle RAW_RECORD and newest RAW_CCALL
527    
528      6. MLRiscGen,
529    
530         1. Changed to handle the new rawload/rawstore/rawrecord operators.
531         2. Code for handling C Calls has been moved to a new module CPSCCalls,
532            in the file CodeGen/cpscompile/cps-c-calls.sml
533    
534      7. Added the conditional move operator
535    
536             condmove of branch
537    
538         to cps.  Generation of this is still buggy so it is currently
539         disabled.
540    
541    ----------------------------------------------------------------------
542    Name: Lal George
543    Date: 2002/03/22 14:18:25 EST
544    Tag: george-20020322-cps-branch-prob
545    Description:
546    
547    Implemented the Ball-Larus branch prediction-heuristics, and
548    incorporated graphical viewers for control flow graphs.
549    
550    Ball-Larus Heuristics:
551    ---------------------
552    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
553    
554    By design it uses the Dempster-Shafer theory for combining
555    probabilities.  For example, in the function:
556    
557        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
558    
559    the ball-larus heuristics predicts that the n=0 is unlikely
560    (OH-heuristic), and the 'then' branch is unlikely because of the
561    RH-heuristic -- giving the 'then' branch an even lower combined
562    probability using the Dempster-Shafer theory.
563    
564    Finally, John Reppy's loop analysis in MLRISC, further lowers the
565    probability of the 'then' branch because of the loop in the else
566    branch.
567    
568    
569    Graphical Viewing:
570    ------------------
571    I merely plugged in Allen's graphical viewers into the compiler. The
572    additional code is not much. At the top level, saying:
573    
574            Control.MLRISC.getFlag "cfg-graphical-view" := true;
575    
576    will display the graphical view of the control flow graph just before
577    back-patching.  daVinci must be in your path for this to work. If
578    daVinci is not available, then the default viewer can be changed
579    using:
580    
581            Control.MLRISC.getString "viewer"
582    
583    which can be set to "dot" or "vcg" for the corresponding viewers. Of
584    course, these viewers must be in your path.
585    
586    The above will display the compilation unit at the level of clusters,
587    many of which are small, boring, and un-interesting. Also setting:
588    
589            Control.MLRISC.getInt "cfg-graphical-view_size"
590    
591    will display clusters that are larger than the value set by the above.
592    
593    
594    ----------------------------------------------------------------------
595    Name: Matthias Blume
596    Date: 2002/03/21 22:20:00 EST
597    Tag: blume-20020321-kmp-bugfix
598    Description:
599    
600    Changed the interface to the KMP routine in PreString and fixed
601    a minor bug in one place where it was used.
602    
603    ----------------------------------------------------------------------
604    Name: Allen Leung
605    Date: 2002/03/21 20:30:00 EST
606    Tag: leunga-20020321-cfg
607    Description:
608    
609      Fixed a potential problem in cfg edge splitting.
610    
611    ----------------------------------------------------------------------
612    Name: Allen Leung
613    Date: 2002/03/21 17:15:00 EST
614    Tag: leunga-20020321-x86-fp-cfg
615    Description:
616    
617      1. Recoded the buggy parts of x86-fp.
618    
619         a. All the block reordering code has been removed.
620            We now depend on the block placement phases to do this work.
621    
622         b. Critical edge splitting code has been simplified and moved into the
623            CFG modules, as where they belong.
624    
625         Both of these were quite buggy and complex.  The code is now much, much
626         simpler.
627    
628      2. X86 backend.
629    
630         a. Added instructions for 64-bit support.  Instruction selection for
631            64-bit has not been committed, however, since that
632            requires changes to MLTREE which haven't been approved by
633            Lal and John.
634    
635         b. Added support for FUCOMI and FUCOMIP when generating code for
636            PentiumPro and above.  We only generate these instructions in
637            the fast-fp mode.
638    
639         c. Added cases for JP and JNP in X86FreqProps.
640    
641      3. CFG
642    
643         CFG now has a bunch of methods for edge splitting and merging.
644    
645      4. Machine description.
646    
647         John's simplification of MLTREE_BASIS.fcond broke a few machine
648         description things:
649    
650         rtl-build.{sig,sml} and hppa.mdl fixed.
651    
652         NOTE: the machine description stuff in the repository is still broken.
653               Again, I can't put my fixes in because that involves
654               changes to MLTREE.
655    
656    ----------------------------------------------------------------------
657    Name: Matthias Blume
658    Date: 2002/03/20 15:55:00 EST
659    Tag: blume-20020320-kmp
660    Description:
661    
662    Implemented Knuth-Morris-Pratt string matching in PreString and used
663    it for String.isSubstring, Substring.isSubstring, and
664    Substring.position.
665    
666    (Might need some stress-testing.  Simple examples worked fine.)
667    
668    ----------------------------------------------------------------------
669    Name: Matthias Blume
670    Date: 2002/03/19 16:37:00 EST
671    Tag: blume-20020319-witnesses
672    Description:
673    
674    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
675    
676    This implements a generic mechanism for changing constness qualifiers
677    anywhere within big C types without resorting to outright "casts".
678    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
679    modify the constness at the outermost level.)
680    The implementation of "convert" is based on the idea of "witness"
681    values -- values that are not used by the operation but whose types
682    "testify" to their applicability.  On the implementation side, "convert"
683    is simply a projection (returning its second curried argument).  With
684    cross-module inlining, it should not result in any machine code being
685    generated.
686    
687    ----------------------------------------------------------------------
688    Name: Matthias Blume
689    Date: 2002/03/15 16:40:00 EST
690    Tag: blume-20020315-basis
691    Description:
692    
693    Provided (preliminary?) implementations for
694    
695      {String,Substring}.{concatWith,isSuffix,isSubstring}
696    
697    and
698    
699      Substring.full
700    
701    Those are in the Basis spec but they were missing in SML/NJ.
702    
703    ----------------------------------------------------------------------
704    Name: Matthias Blume
705    Date: 2002/03/14 21:30:00 EST
706    Tag: blume-20020314-controls
707    Description:
708    
709    Controls:
710    ---------
711    
712    1. Factored out the recently-added Controls : CONTROLS stuff and put
713       it into its own library $/controls-lib.cm.  The source tree for
714       this is under src/smlnj-lib/Controls.
715    
716    2. Changed the names of types and functions in this interface, so they
717       make a bit more "sense":
718    
719          module -> registry
720          'a registry -> 'a group
721    
722    3. The interface now deals in ref cells only.  The getter/setter interface
723       is (mostly) gone.
724    
725    4. Added a function that lets one register an already-existing ref cell.
726    
727    5. Made the corresponding modifications to the rest of the code so that
728       everything compiles again.
729    
730    6. Changed the implementation of Controls.MLRISC back to something closer
731       to the original.  In particular, this module (and therefore MLRISC)
732       does not depend on Controls.  There now is some link-time code in
733       int-sys.sml that registers the MLRISC controls with the Controls
734       module.
735    
736    CM:
737    ---
738    
739      * One can now specify the lambda-split aggressiveness in init.cmi.
740    
741    ----------------------------------------------------------------------
742    Name: Allen Leung
743    Date: 2002/03/13 17:30:00 EST
744    Tag: leunga-20020313-x86-fp-unary
745    Description:
746    
747    Bug fix for:
748    
749    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
750    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
751    > - fun f(x,(y,z)) = Real.~ y;
752    > [autoloading]
753    > [autoloading done]
754    >       fchsl   (%eax), 184(%esp)
755    > Error: MLRisc bug: X86MCEmitter.emitInstr
756    >
757    > uncaught exception Error
758    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
759    
760    The problem was that the code generator did not generate any fp registers
761    in this case, and the ra didn't know that it needed to run the X86FP phase to
762    translate the pseudo fp instruction.   This only happened with unary fp
763    operators in certain situations.
764    
765    ----------------------------------------------------------------------
766    Name: Matthias Blume
767    Date: 2002/03/13 14:00:00 EST
768    Tag: blume-20020313-overload-etc
769    Description:
770    
771    1. Added _overload as a synonym for overload for backward compatibility.
772       (Control.overloadKW must be true for either version to be accepted.)
773    
774    2. Fixed bug in install script that caused more things to be installed
775       than what was requested in config/targets.
776    
777    3. Made CM aware of the (_)overload construct so that autoloading
778       works.
779    
780    ----------------------------------------------------------------------
781    Name: Matthias Blume
782    Date: 2002/03/12 22:03:00 EST
783    Tag: blume-20020312-url
784    Description:
785    
786    Forgot to update BOOT and srcarchiveurl.
787    
788    ----------------------------------------------------------------------
789    Name: Matthias Blume
790    Date: 2002/03/12 17:30:00 EST
791    Tag: blume-20020312-version110392
792    Description:
793    
794    Yet another version number bump (because of small changes to the
795    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
796    
797    Changes:
798    
799      The new pid generation scheme described a few weeks ago was overly
800      complicated.  I implemented a new mechanism that is simpler and
801      provides a bit more "stability":  Once CM has seen a compilation
802      unit, it keeps its identity constant (as long as you do not delete
803      those crucial CM/GUID/* files).  This means that when you change
804      an interface, compile, then go back to the old interface, and
805      compile again, you arrive at the original pid.
806    
807      There now also is a mechanism that instructs CM to use the plain
808      environment hash as a module's pid (effectively making its GUID
809      the empty string).  For this, "noguid" must be specified as an
810      option to the .sml file in question within its .cm file.
811      This is most useful for code that is being generated by tools such
812      as ml-nlffigen (because during development programmers tend to
813      erase the tool's entire output directory tree including CM's cached
814      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
815      revert to the old, broken behavior of SML/NJ, but in specific cases
816      where there is no danger of interface confusion, its use is ok
817      (I think).
818    
819      ml-nlffigen by default generates "noguid" annotations.  They can be
820      turned off by specifying -guid in its command line.
821    
822    ----------------------------------------------------------------------
823    Name: Lal George
824    Date: 2002/03/12 12 14:42:36 EST
825    Tag: george-20020312-frequency-computation
826    Description:
827    
828    Integrated jump chaining and static block frequency into the
829    compiler. More details and numbers later.
830    
831    ----------------------------------------------------------------------
832    Name: Lal George
833    Date: 2002/03/11 11 22:38:53 EST
834    Tag: george-20020311-jump-chain-elim
835    Description:
836    
837    Tested the jump chain elimination on all architectures (except the
838    hppa).  This is on by default right now and is profitable for the
839    alpha and x86, however, it may not be profitable for the sparc and ppc
840    when compiling the compiler.
841    
842    The gc test will typically jump to a label at the end of the cluster,
843    where there is another jump to an external cluster containing the actual
844    code to invoke gc. This is to allow factoring of common gc invocation
845    sequences. That is to say, we generate:
846    
847            f:
848               testgc
849               ja   L1      % jump if above to L1
850    
851            L1:
852               jmp L2
853    
854    
855    After jump chain elimination the 'ja L1' instructions is converted to
856    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
857    up being implemented in their long form (if L2 is far away) using:
858    
859            jbe     L3      % jump if below or equal to L3
860            jmp     L2
861         L3:
862            ...
863    
864    
865    For large compilation units L2  may be far away.
866    
867    
868    ----------------------------------------------------------------------
869    Name: Matthias Blume
870    Date: 2002/03/11 13:30:00 EST
871    Tag: blume-20020311-mltreeeval
872    Description:
873    
874    A functor parameter was missing.
875    
876    ----------------------------------------------------------------------
877    Name: Allen Leung
878    Date: 2002/03/11 10:30:00 EST
879    Tag: leunga-20020311-runtime-string0
880    Description:
881    
882       The representation of the empty string now points to a
883    legal null terminated C string instead of unit.  It is now possible
884    to convert an ML string into C string with InlineT.CharVector.getData.
885    This compiles into one single machine instruction.
886    
887    ----------------------------------------------------------------------
888    Name: Allen Leung
889    Date: 2002/03/10 23:55:00 EST
890    Tag: leunga-20020310-x86-call
891    Description:
892    
893       Added machine generation for CALL instruction (relative displacement mode)
894    
895    ----------------------------------------------------------------------
896    Name: Matthias Blume
897    Date: 2002/03/08 16:05:00
898    Tag: blume-20020308-entrypoints
899    Description:
900    
901    Version number bumped to 110.39.1.  NEW BOOTFILES!
902    
903    Entrypoints: non-zero offset into a code object where execution should begin.
904    
905    - Added the notion of an entrypoint to CodeObj.
906    - Added reading/writing of entrypoint info to Binfile.
907    - Made runtime system bootloader aware of entrypoints.
908    - Use the address of the label of the first function given to mlriscGen
909      as the entrypoint.  This address is currently always 0, but it will
910      not be 0 once we turn on block placement.
911    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
912      with entry points) from mlriscGen.
913    
914    ----------------------------------------------------------------------
915    Name: Allen Leung
916    Date: 2002/03/07 20:45:00 EST
917    Tag: leunga-20020307-x86-cmov
918    Description:
919    
920       Bug fixes for CMOVcc on x86.
921    
922       1. Added machine code generation for CMOVcc
923       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
924       3. CMOVcc cannot have an immediate operand as argument.
925    
926    ----------------------------------------------------------------------
927    Name: Matthias Blume
928    Date: 2002/03/07 16:15:00 EST
929    Tag: blume-20020307-controls
930    Description:
931    
932    This is a very large but mostly boring patch which makes (almost)
933    every tuneable compiler knob (i.e., pretty much everything under
934    Control.* plus a few other things) configurable via both the command
935    line and environment variables in the style CM did its configuration
936    until now.
937    
938    Try starting sml with '-h' (or, if you are brave, '-H')
939    
940    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
941    implements the underlying generic mechanism.
942    
943    The interface to some of the existing such facilities has changed somewhat.
944    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
945    (The getFoo interface is still there for backward-compatibility, but its
946    use is deprecated.)
947    
948    The ml-build script passes -Cxxx=yyy command-line arguments through so
949    that one can now twiddle the compiler settings when using this "batch"
950    compiler.
951    
952    TODO items:
953    
954    We should go through and throw out all controls that are no longer
955    connected to anything.  Moreover, we should go through and provide
956    meaningful (and correct!) documentation strings for those controls
957    that still are connected.
958    
959    Currently, multiple calls to Controls.new are accepted (only the first
960    has any effect).  Eventually we should make sure that every control
961    is being made (via Controls.new) exactly once.  Future access can then
962    be done using Controls.acc.
963    
964    Finally, it would probably be a good idea to use the getter-setter
965    interface to controls rather than ref cells.  For the time being, both
966    styles are provided by the Controls module, but getter-setter pairs are
967    better if thread-safety is of any concern because they can be wrapped.
968    
969    *****************************************
970    
971    One bug fix: The function blockPlacement in three of the MLRISC
972    backpatch files used to be hard-wired to one of two possibilities at
973    link time (according to the value of the placementFlag).  But (I
974    think) it should rather sense the flag every time.
975    
976    *****************************************
977    
978    Other assorted changes (by other people who did not supply a HISTORY entry):
979    
980    1. the cross-module inliner now works much better (Monnier)
981    2. representation of weights, frequencies, and probabilities in MLRISC
982       changed in preparation of using those for weighted block placement
983       (Reppy, George)
984    
985    ----------------------------------------------------------------------
986    Name: Lal George
987    Date: 2002/03/07 14:44:24 EST 2002
988    Tag: george-20020307-weighted-block-placement
989    
990    Tested the weighted block placement optimization on all architectures
991    (except the hppa) using AMPL to generate the block and edge frequencies.
992    Changes were required in the machine properties to correctly
993    categorize trap instructions. There is an MLRISC flag
994    "weighted-block-placement" that can be used to enable weighted block
995    placement, but this will be ineffective without block/edge
996    frequencies (coming soon).
997    
998    
999    ----------------------------------------------------------------------
1000    Name: Lal George
1001    Date: 2002/03/05 17:24:48 EST
1002    Tag: george-20020305-linkage-cluster
1003    
1004    In order to support the block placement optimization, a new cluster
1005    is generated as the very first cluster (called the linkage cluster).
1006    It contains a single jump to the 'real' entry point for the compilation
1007    unit. Block placement has no effect on the linkage cluster itself, but
1008    all the other clusters  have full freedom in the manner in which they
1009    reorder blocks or functions.
1010    
1011    On the x86 the typical linkage code that is generated is:
1012       ----------------------
1013            .align 2
1014       L0:
1015            addl    $L1-L0, 72(%esp)
1016            jmp     L1
1017    
1018    
1019            .align  2
1020       L1:
1021       ----------------------
1022    
1023    72(%esp) is the memory location for the stdlink register. This
1024    must contain the address of the CPS function being called. In the
1025    above example, it contains the address of  L0; before
1026    calling L1 (the real entry point for the compilation unit), it
1027    must contain the address for L1, and hence
1028    
1029            addl $L1-L0, 72(%esp)
1030    
1031    I have tested this on all architectures except the hppa.The increase
1032    in code size is of course negligible
1033    
1034    ----------------------------------------------------------------------
1035    Name: Allen Leung
1036    Date: 2002/03/03 13:20:00 EST
1037    Tag: leunga-20020303-mlrisc-tools
1038    
1039      Added #[ ... ] expressions to mlrisc tools
1040    
1041    ----------------------------------------------------------------------
1042    Name: Matthias Blume
1043    Date: 2002/02/27 12:29:00 EST
1044    Tag: blume-20020227-cdebug
1045    Description:
1046    
1047    - made types in structure C and C_Debug to be equal
1048    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
1049    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
1050    
1051    ----------------------------------------------------------------------
1052    Name: Matthias Blume
1053    Date: 2002/02/26 12:00:00 EST
1054    Tag: blume-20020226-ffi
1055    Description:
1056    
1057    1. Fixed a minor bug in CM's "noweb" tool:
1058       If numbering is turned off, then truly don't number (i.e., do not
1059       supply the -L option to noweb).  The previous behavior was to supply
1060       -L'' -- which caused noweb to use the "default" line numbering scheme.
1061       Thanks to Chris Richards for pointing this out (and supplying the fix).
1062    
1063    2. Once again, I reworked some aspects of the FFI:
1064    
1065       A. The incomplete/complete type business:
1066    
1067       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
1068         gone!
1069       - ML types representing an incomplete type are now *equal* to
1070         ML types representing their corresponding complete types (just like
1071         in C).  This is still safe because ml-nlffigen will not generate
1072         RTTI for incomplete types, nor will it generate functions that
1073         require access to such RTTI.   But when ML code generated from both
1074         incomplete and complete versions of the C type meet, the ML types
1075         are trivially interoperable.
1076    
1077         NOTE:  These changes restore the full generality of the translation
1078         (which was previously lost when I eliminated functorization)!
1079    
1080       B. Enum types:
1081    
1082       - Structure C now has a type constructor "enum" that is similar to
1083         how the "su" constructor works.  However, "enum" is not a phantom
1084         type because each "T enum" has values (and is isomorphic to
1085         MLRep.Signed.int).
1086       - There are generic access operations for enum objects (using
1087         MLRep.Signed.int).
1088       - ml-nlffigen will generate a structure E_foo for each "enum foo".
1089         * The structure contains the definition of type "mlrep" (the ML-side
1090         representation type of the enum).  Normally, mlrep is the same
1091         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
1092         then mlrep will be defined as a datatype -- thus facilitating
1093         pattern matching on mlrep values.
1094         ("-ec" will be suppressed if there are duplicate values in an
1095          enumeration.)
1096         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
1097         will be generated for each C enum constant xxx.
1098         * Conversion functions m2i and i2m convert between mlrep and
1099         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
1100         * Coversion functions c and ml convert between mlrep and "tag enum".
1101         * Access functions (get/set) fetch and store mlrep values.
1102       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
1103         enumerations are merged into one single enumeration represented by
1104         structure E_'.
1105    
1106    ----------------------------------------------------------------------
1107    Name: Allen Leung
1108    Date: 2002/02/25 04:45:00 EST
1109    Tag: leunga-20020225-cps-spill
1110    
1111    This is a new implementation of the CPS spill phase.
1112    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
1113    In case of problems, replace it with the old file spill.sml
1114    
1115    The current compiler runs into some serious performance problems when
1116    constructing a large record.  This can happen when we try to compile a
1117    structure with many items.  Even a very simple structure like the following
1118    makes the compiler slow down.
1119    
1120        structure Foo = struct
1121           val x_1 = 0w1 : Word32.int
1122           val x_2 = 0w2 : Word32.int
1123           val x_3 = 0w3 : Word32.int
1124           ...
1125           val x_N = 0wN : Word32.int
1126        end
1127    
1128    The following table shows the compile time, from N=1000 to N=4000,
1129    with the old compiler:
1130    
1131    N
1132    1000   CPS 100 spill                           0.04u  0.00s  0.00g
1133           MLRISC ra                               0.06u  0.00s  0.05g
1134              (spills = 0 reloads = 0)
1135           TOTAL                                   0.63u  0.07s  0.21g
1136    
1137    1100   CPS 100 spill                           8.25u  0.32s  0.64g
1138           MLRISC ra                               5.68u  0.59s  3.93g
1139              (spills = 0 reloads = 0)
1140           TOTAL                                   14.71u  0.99s  4.81g
1141    
1142    1500   CPS 100 spill                           58.55u  2.34s  1.74g
1143           MLRISC ra                               5.54u  0.65s  3.91g
1144              (spills = 543 reloads = 1082)
1145           TOTAL                                   65.40u  3.13s  6.00g
1146    
1147    2000   CPS 100 spill                           126.69u  4.84s  3.08g
1148           MLRISC ra                               0.80u  0.10s  0.55g
1149              (spills = 42 reloads = 84)
1150           TOTAL                                   129.42u  5.10s  4.13g
1151    
1152    3000   CPS 100 spill                           675.59u  19.03s  11.64g
1153           MLRISC ra                               2.69u  0.27s  1.38g
1154              (spills = 62 reloads = 124)
1155           TOTAL                                   682.48u  19.61s  13.99g
1156    
1157    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
1158           MLRISC ra                               4.96u  0.27s  2.72g
1159              (spills = 85 reloads = 170)
1160           TOTAL                                   2375.26u  57.21s  48.00g
1161    
1162    As you can see the old cps spill module suffers from some serious
1163    performance problem.  But since I cannot decipher the old code fully,
1164    instead of patching the problems up, I'm reimplementing it
1165    with a different algorithm.  The new code is more modular,
1166    smaller when compiled, and substantially faster
1167    (O(n log n) time and O(n) space).  Timing of the new spill module:
1168    
1169    4000  CPS 100 spill                           0.02u  0.00s  0.00g
1170          MLRISC ra                               0.25u  0.02s  0.15g
1171             (spills=1 reloads=3)
1172          TOTAL                                   7.74u  0.34s  1.62g
1173    
1174    Implementation details:
1175    
1176    As far as I can tell, the purpose of the CPS spill module is to make sure the
1177    number of live variables at any program point (the bandwidth)
1178    does not exceed a certain limit, which is determined by the
1179    size of the spill area.
1180    
1181    When the bandwidth is too large, we decrease the register pressure by
1182    packing live variables into spill records.  How we achieve this is
1183    completely different than what we did in the old code.
1184    
1185    First, there is something about the MLRiscGen code generator
1186    that we should be aware of:
1187    
1188    o MLRiscGen performs code motion!
1189    
1190       In particular, it will move floating point computations and
1191       address computations involving only the heap pointer to
1192       their use sites (if there is only a single use).
1193       What this means is that if we have a CPS record construction
1194       statement
1195    
1196           RECORD(k,vl,w,e)
1197    
1198       we should never count the new record address w as live if w
1199       has only one use (which is often the case).
1200    
1201       We should do something similar to floating point, but the transformation
1202       there is much more complex, so I won't deal with that.
1203    
1204    Secondly, there are now two new cps primops at our disposal:
1205    
1206     1. rawrecord of record_kind option
1207        This pure operator allocates some uninitialized storage from the heap.
1208        There are two forms:
1209    
1210         rawrecord NONE [INT n]  allocates a tagless record of length n
1211         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
1212                                     and initializes the tag.
1213    
1214     2. rawupdate of cty
1215          rawupdate cty (v,i,x)
1216          Assigns to x to the ith component of record v.
1217          The storelist is not updated.
1218    
1219    We use these new primops for both spilling and increment record construction.
1220    
1221     1. Spilling.
1222    
1223        This is implemented with a linear scan algorithm (but generalized
1224        to trees).  The algorithm will create a single spill record at the
1225        beginning of the cps function and use rawupdate to spill to it,
1226        and SELECT or SELp to reload from it.  So both spills and reloads
1227        are fine-grain operations.  In contrast, in the old algorithm
1228        "spills" have to be bundled together in records.
1229    
1230        Ideally, we should sink the spill record construction to where
1231        it is needed.  We can even split the spill record into multiple ones
1232        at the places where they are needed.  But CPS is not a good
1233        representation for global code motion, so I'll keep it simple and
1234        am not attempting this.
1235    
1236     2. Incremental record construction (aka record splitting).
1237    
1238        Long records with many component values which are simulatenously live
1239        (recall that single use record addresses are not considered to
1240         be live) are constructed with rawrecord and rawupdate.
1241        We allocate space on the heap with rawrecord first, then gradually
1242        fill it in with rawupdate.  This is the technique suggested to me
1243        by Matthias.
1244    
1245        Some restrictions on when this is applicable:
1246        1. It is not a VECTOR record.  The code generator currently does not handle
1247           this case. VECTOR record uses double indirection like arrays.
1248        2. All the record component values are defined in the same "basic block"
1249           as the record constructor.  This is to prevent speculative
1250           record construction.
1251    
1252    ----------------------------------------------------------------------
1253    Name: Allen Leung
1254    Date: 2002/02/22 01:02:00 EST
1255    Tag: leunga-20020222-mlrisc-tools
1256    
1257    Minor bug fixes in the parser and rewriter
1258    
1259    ----------------------------------------------------------------------
1260    Name: Allen Leung
1261    Date: 2002/02/21 20:20:00 EST
1262    Tag: leunga-20020221-peephole
1263    
1264    Regenerated the peephole files.  Some contained typos in the specification
1265    and some didn't compile because of pretty printing bugs in the old version
1266    of 'nowhere'.
1267    
1268    ----------------------------------------------------------------------
1269    Name: Allen Leung
1270    Date: 2002/02/19 20:20:00 EST
1271    Tag: leunga-20020219-mlrisc-tools
1272    Description:
1273    
1274       Minor bug fixes to the mlrisc-tools library:
1275    
1276       1.  Fixed up parsing colon suffixed keywords
1277       2.  Added the ability to shut the error messages up
1278       3.  Reimplemented the pretty printer and fixed up/improved
1279           the pretty printing of handle and -> types.
1280       4.  Fixed up generation of literal symbols in the nowhere tool.
1281       5.  Added some SML keywords to to sml.sty
1282    
1283    ----------------------------------------------------------------------
1284    Name: Matthias Blume
1285    Date: 2002/02/19 16:20:00 EST
1286    Tag: blume-20020219-cmffi
1287    Description:
1288    
1289    A wild mix of changes, some minor, some major:
1290    
1291    * All C FFI-related libraries are now anchored under $c:
1292        $/c.cm      --> $c/c.cm
1293        $/c-int.cm  --> $c/internals/c-int.cm
1294        $/memory.cm --> $c/memory/memory.cm
1295    
1296    * "make" tool (in CM) now treats its argument pathname slightly
1297      differently:
1298        1. If the native expansion is an absolute name, then before invoking
1299           the "make" command on it, CM will apply OS.Path.mkRelative
1300           (with relativeTo = OS.FileSys.getDir()) to it.
1301        2. The argument will be passed through to subsequent phases of CM
1302           processing without "going native".  In particular, if the argument
1303           was an anchored path, then "make" will not lose track of that anchor.
1304    
1305    * Compiler backends now "know" their respective C calling conventions
1306      instead of having to be told about it by ml-nlffigen.  This relieves
1307      ml-nlffigen from one of its burdens.
1308    
1309    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
1310    
1311    * Export C_DEBUG and C_Debug from $c/c.cm.
1312    
1313    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
1314      subtyping relationship between incomplete pointers and their complete
1315      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
1316      with the convention of instantiating 'o with ('t, 'c) obj whenever
1317      the pointer target type is complete.  In the incomplete case, 'o
1318      will be instantiated with some "'c iobj" -- a type obtained by
1319      using one of the functors PointerToIncompleteType or PointerToCompleteType.
1320    
1321      Operations that work on both incomplete and complete pointer types are
1322      typed as taking an 'o ptr while operations that require the target to
1323      be known are typed as taking some ('t, 'c) obj ptr.
1324    
1325      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
1326      where void is an eqtype without any values.  This makes it possible
1327      to work on voidptr values using functions meant to operate on light
1328      incomplete pointers.
1329    
1330    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
1331      been vastly simplified.
1332    
1333    ----------------------------------------------------------------------
1334    Name: Matthias Blume
1335    Date: 2002/02/19 10:48:00 EST
1336    Tag: blume-20020219-pqfix
1337    Description:
1338    
1339    Applied Chris Okasaki's bug fix for priority queues.
1340    
1341    ----------------------------------------------------------------------
1342    Name: Matthias Blume
1343    Date: 2002/02/15 17:05:00
1344    Tag: Release_110_39
1345    Description:
1346    
1347    Last-minute retagging is becoming a tradition... :-(
1348    
1349    This is the working release 110.39.
1350    
1351    ----------------------------------------------------------------------
1352    Name: Matthias Blume
1353    Date: 2002/02/15 16:00:00 EST
1354    Tag: Release_110_39-orig
1355    Description:
1356    
1357    Working release 110.39.  New bootfiles.
1358    
1359    (Update: There was a small bug in the installer so it wouldn't work
1360    with all shells.  So I retagged. -Matthias)
1361    
1362    ----------------------------------------------------------------------
1363    Name: Matthias Blume
1364    Date: 2002/02/15 14:17:00 EST
1365    Tag: blume-20020215-showbindings
1366    Description:
1367    
1368    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
1369    the latter can be useful for exploring what bindings are available at
1370    the interactive prompt.  (The first function returns only the list
1371    of symbols that are really bound, the second prints those but also the
1372    ones that CM's autoloading mechanism knows about.)
1373    
1374    ----------------------------------------------------------------------
1375    Name: Matthias Blume
1376    Date: 2002/02/15 12:08:00 EST
1377    Tag: blume-20020215-iptrs
1378    Description:
1379    
1380    Two improvements to ml-nlffigen:
1381    
1382      1. Write files only if they do not exist or if their current contents
1383         do not coincide with what's being written.  (That is, avoid messing
1384         with the time stamps unless absolutely necessary.)
1385    
1386      2. Implement a "repository" mechanism for generated files related
1387         to "incomplete pointer types".   See the README file for details.
1388    
1389    ----------------------------------------------------------------------
1390    Name: Matthias Blume
1391    Date: 2002/02/14 11:50:00 EST
1392    Tag: blume-20020214-quote
1393    Description:
1394    
1395    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
1396    because of the new and improved tag generation scheme.  (Thanks to Allen
1397    Leung for pointing it out.)
1398    
1399    ----------------------------------------------------------------------
1400    Name: Lal George
1401    Date: 2002/02/14 09:55:27 EST 2002
1402    Tag: george-20020214-isabelle-bug
1403    Description:
1404    
1405    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
1406    of Isabelle on the x86.
1407    
1408    From Allen:
1409    -----------
1410     I've found the problem:
1411    
1412         in ra-core.sml, I use the counter "blocked" to keep track of the
1413         true number of elements in the freeze queue.  When the counter goes
1414         to zero, I skip examining the queue.  But I've messed up the
1415         bookkeeping in combine():
1416    
1417             else ();
1418             case !ucol of
1419               PSEUDO => (if !cntv > 0 then
1420                     (if !cntu > 0 then blocked := !blocked - 1 else ();
1421                                        ^^^^^^^^^^^^^^^^^^^^^^^
1422                      moveu := mergeMoveList(!movev, !moveu)
1423                     )
1424                  else ();
1425    
1426         combine() is called to coalesce two nodes u and v.
1427         I think I was thinking that if the move counts of u and v are both
1428         greater than zero then after they are coalesced then one node is
1429         removed from the freeze queue.  Apparently I was thinking that
1430         both u and v are of low degree, but that's clearly not necessarily true.
1431    
1432    
1433    02/12/2002:
1434        Here's the patch.  HOL now compiles.
1435    
1436        I don't know how this impact on performance (compile
1437        time or runtime).  This bug caused the RA (especially on the x86)
1438        to go thru the potential spill phase when there are still nodes on the
1439        freeze queue.
1440    
1441    
1442    
1443    
1444    ----------------------------------------------------------------------
1445    Name: Matthias Blume
1446    Date: 2002/02/13 22:40:00 EST
1447    Tag: blume-20020213-fptr-rtti
1448    Description:
1449    
1450    Fixed a bug in ml-nlffigen that was introduced with one of the previous
1451    updates.
1452    
1453    ----------------------------------------------------------------------
1454    Name: Matthias Blume
1455    Date: 2002/02/13 16:41:00 EST
1456    Tag: blume-20020213-cmlpq
1457    Description:
1458    
1459    Added new priority queue export symbols (which have just been added to
1460    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
1461    not compile and the installer would choke.)
1462    
1463    ----------------------------------------------------------------------
1464    Name: Matthias Blume
1465    Date: 2002/02/13 16:15:00 EST
1466    Tag: blume-20020213-various
1467    Description:
1468    
1469    1. More tweaks to ml-nlffigen:
1470    
1471       - better internal datastructures (resulting in slight speedup)
1472       - "-match" option requires exact match
1473       - "localized" gensym counters (untagged structs/unions nested within
1474         other structs/unions or within typedefs get a fresh counter; their
1475         tag will be prefixed by a concatenation of their parents' tags)
1476       - bug fixes (related to calculation of transitive closure of types
1477         to be included in the output)
1478    
1479    2. Minor Basis updates:
1480    
1481       - added implementations for List.collate and Option.app
1482    
1483    ----------------------------------------------------------------------
1484    Name: Matthias Blume
1485    Date: 2002/02/11 15:55:00 EST
1486    Tag: blume-20020211-gensym
1487    Description:
1488    
1489    Added a "-gensym" option to command line of ml-nlffigen.  This can be
1490    used to specify a "stem" -- a string that is inserted in all "gensym'd"
1491    names (ML structure names that correspond to unnamed C structs, unions,
1492    and enums), so that separate runs of ml-nlffigen do not clash.
1493    
1494    ----------------------------------------------------------------------
1495    Name: Matthias Blume
1496    Date: 2002/02/11 12:05:00 EST
1497    Tag: blume-20020211-gensml
1498    Description:
1499    
1500    A quick fix for a problem with GenSML (in the pgraph-util library):
1501    Make generation of toplevel "local" optional.  (Strictly speaking,
1502    signature definitions within "local" are not legal SML.)
1503    
1504    Other than that: updates to INSTALL and cm/TODO.
1505    
1506    ----------------------------------------------------------------------
1507    Name: Matthias Blume
1508    Date: 2002/02/08 15:00:00 EST
1509    Tag: blume-20020208-uniquepid
1510    Description:
1511    
1512    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
1513    
1514    1. The installer (config/install.sh) has gotten smarter:
1515    
1516         - Configuration options are a bit easier to specify now
1517           (in config/targets).
1518         - Bug in recognizing .tar.bz2 files fixed.
1519         - Installer automatically resolves dependencies between
1520           configuration options (e.g., if you ask for eXene, you will
1521           also get cml -- regardless whether you asked for it or not).
1522         - Installer can run in "quieter mode" by setting the environment
1523           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
1524           "completely silent", though.
1525         - Build HashCons library as part of smlnj-lib.
1526    
1527    2. A new scheme for assigning persistent identifiers to compilation
1528       units (and, by extension, to types etc.) has been put into place.
1529       This fixes a long-standing bug where types and even dynamic values
1530       can get internally confused, thereby compromising type safety
1531       (abstraction) and dynamic correctness.  See
1532    
1533         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
1534    
1535       for an example of how things could go wrong until now.
1536    
1537       The downside of the new scheme is that pids are not quite as
1538       persistent as they used to be: CM will generate a fresh pid
1539       for every compilation unit that it thinks it sees for the first
1540       time.  That means that if you compile starting from a clean, fresh
1541       source tree at two different times, you end up with different
1542       binaries.
1543    
1544       Cutoff recompilation, however, has not been compromised because
1545       CM keeps pid information in special caches between runs.
1546    
1547    ----------------------------------------------------------------------
1548    Name: Lal George
1549    Date: 2002/02/07 15:34:13 EST 2002
1550    Tag: <none>
1551    Description:
1552    
1553    Compilers that generate assembly code may produce  global labels
1554    whose value is resolved at link time. The various peephole optimization
1555    modules did not take this in account.
1556    
1557    TODO. The Labels.addrOf function should really return an option
1558    type so that clients are forced to deal with this issue, rather
1559    than an exception being raised.
1560    
1561    ----------------------------------------------------------------------
1562    Name: Lal George
1563    Date: 2002/02/06 13:55:02 EST
1564    Tag: george-20020206-ra-breakup
1565    Description:
1566    
1567    1. A bug fix from Allen.
1568    
1569        A typo causes extra fstp %st(0)'s to be generated at compensation
1570        edges, which might cause stack underflow traps at runtime.  This
1571        occurs in fft where there are extraneous fstps right before the 'into'
1572        trap instruction (in this case they are harmless since none of the
1573        integers overflow.)
1574    
1575    2. Pulled out various utility modules that were embedded in the modules
1576       of the register allocator. I need these modules for other purposes, but
1577       they are not complete enough to put into a library (just yet).
1578    ----------------------------------------------------------------------
1579    Name: Matthias Blume
1580    Date: 2002/01/31 16:05:00 EST
1581    Tag: blume-20020131-sparc-ccalls
1582    Description:
1583    
1584    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
1585       of extra stack space by mistake.  Fixed.
1586    
1587    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
1588    
1589    ----------------------------------------------------------------------
1590    Name: Allen Leung
1591    Date: 2002/01/30
1592    Tag: leunga-20020130-nowhere-bug-fix
1593    Description:
1594    
1595       MLRISC bug fixes:
1596       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
1597       2. MachineInt.fromString was negating its value.
1598    
1599    ----------------------------------------------------------------------
1600    Name: Matthias Blume
1601    Date: 2002/01/29
1602    Tag: blume-20020129-INSTALL
1603    Description:
1604    
1605    - Added somewhat detailed installation instructions (file INSTALL).
1606    - Fixed curl-detection bug in config/install.sh.
1607    - It is now possible to select the URL getter using the URLGETTER
1608      environment variable:
1609    
1610          not set / "unknown"      --> automatic detection (script tries wget,
1611                                       curl, and lynx)
1612          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
1613                                       how to properly invoke them)
1614          other                    --> use $URLGETTER directly, it must take
1615                                       precisely two command-line arguments
1616                                       (source URL and destination file name)
1617    
1618    ----------------------------------------------------------------------
1619    Name: Matthias Blume
1620    Date: 2002/01/28
1621    Tag: blume-20020128-sparc-ccalls
1622    Description:
1623    
1624    - Fixed problem with calculation of "used" registers in sparc-c-calls.
1625    - Make use of the allocParam argument in sparc-c-calls.
1626    
1627    ----------------------------------------------------------------------
1628    Name: Matthias Blume
1629    Date: 2002/01/28
1630    Tag: blume-20020128-allocParam
1631    Description:
1632    
1633    John Reppy:  Changes c-calls API to accept client-callback for
1634    allocating extra stack space.
1635    me: Corresponding changes to mlriscGen (using a dummy argument that
1636        does not change the current behavior).
1637    
1638    ----------------------------------------------------------------------
1639    Name: Matthias Blume
1640    Date: 2002/01/28 12:00:00
1641    Tag: Release_110_38
1642    Description:
1643    
1644    This time for real!!!
1645    
1646    ----------------------------------------------------------------------
1647    Name: Matthias Blume
1648    Date: 2002/01/28 10:56:00 EST
1649    Tag: blume-20020128-retraction
1650    Description:
1651    
1652    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
1653       with blume-Release_110_38-retracted.)
1654    
1655    1. Fixed a problem with incorrect rounding modes in real64.sml.
1656       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
1657    
1658    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
1659       and enums fixed.  The naming of corresponding ML identifiers should
1660       now be consistent again.
1661    
1662    ----------------------------------------------------------------------
1663    Name: Allen Leung
1664    Date: 2002/01/27
1665    Tag: leunga-20020127-nowhere
1666    Description:
1667    
1668       Added a target called nowhere in the configuration scripts.
1669       Enabling this will build the MLRISC 'nowhere' tool (for translating
1670       programs with where-clauses into legal SML code) during installation.
1671    
1672    ----------------------------------------------------------------------
1673    Name: Matthias Blume
1674    Date: 2002/01/25 21:27:00 EST
1675    Tag: blume-Release_110_38-retracted
1676    Description:
1677    
1678    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
1679    
1680    README will be added later.
1681    
1682    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
1683    (Release_110_38) removed.  Reason: Last-minute bug fixes.
1684    
1685    ----------------------------------------------------------------------
1686    Name: Matthias Blume
1687    Date: 2002/01/25
1688    Tag: blume-20020125-ffi
1689    Description:
1690    
1691    A large number of tweaks and improvements to ml-nlffi-lib and
1692    ml-nlffigen:
1693    
1694       - ML represenation types have been streamlined
1695       - getter and setter functions work with concrete values, not abstract
1696         ones where possible
1697       - ml-nlffigen command line more flexible (see README file there)
1698       - some bugs have been fixed (hopefully)
1699    
1700    ----------------------------------------------------------------------
1701    Name: Lal George
1702    Date: 2002/01/24
1703    Tag: george-20020124-risc-ra-interface
1704    Description:
1705    
1706       There is a dramatic simplification in the interface to the
1707       register allocator for RISC architectures as a result of making
1708       parallel copy instructions explicit.
1709    
1710    ----------------------------------------------------------------------
1711    Name: Matthias Blume
1712    Date: 2002/01/22
1713    Tag: blume-20020122-x86-ccalls
1714    Description:
1715    
1716    Bug fix for c-calls on x86 (having to do with how char- and
1717    short-arguments are being handled).
1718    
1719    ----------------------------------------------------------------------
1720    Name: Matthias Blume
1721    Date: 2002/01/21
1722    Tag: blume-20020121-ff
1723    Description:
1724    
1725    Another day of fiddling with the FFI...
1726    
1727    1. Bug fix/workaround:  CKIT does not complain about negative array
1728       dimensions, so ml-nlffigen has to guard itself against this possibility.
1729       (Otherwise a negative dimension would send it into an infinite loop.)
1730    
1731    2. Some of the abstract types (light objects, light pointers, most "base"
1732       types) in structure C are now eqtypes.
1733    
1734    3. Added constructors and test functions for NULL function pointers.
1735    
1736    ----------------------------------------------------------------------
1737    Name: Matthias Blume
1738    Date: 2002/01/18
1739    Tag: blume-20020118-ready-for-new-release
1740    Description:
1741    
1742    Made config/srcarchiveurl point to a new place.  (Will provide boot
1743    files shortly.)
1744    
1745    Maybe we christen this to be 110.38?
1746    
1747    ----------------------------------------------------------------------
1748    Name: Matthias Blume
1749    Date: 2002/01/18
1750    Tag: blume-20020118-more-ffifiddle
1751    Description:
1752    
1753    Today's FFI fiddling:
1754    
1755      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
1756        C.Set.* that use concrete (MLRep.*) arguments and results instead
1757        of abstract ones.
1758    
1759      - Provided word-style bit operations etc. for "int" representation
1760        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
1761        and Long.
1762    
1763    ----------------------------------------------------------------------
1764    Name: Matthias Blume
1765    Date: 2002/01/18
1766    Tag: blume-20020118-use-x86-fp
1767    Description:
1768    
1769    Now that x86-fast-fp seems to be working, I turned it back on again
1770    by default.  (Seems to work fine now, even with the FFI.)
1771    
1772    Other than that, I added some documentation about the FFI to
1773    src/ml-nlffigen/README and updated the FFI test examples in
1774    src/ml-nlffi-lib/Tests/*.
1775    
1776    ----------------------------------------------------------------------
1777    Name: Allen Leung
1778    Date: 2002/01/17
1779    Tag: leunga-20020117-x86-fast-fp-call
1780    Description:
1781    
1782       1. Fixed a problem with handling return fp values when x86's fast fp
1783          mode is turned on.
1784    
1785       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
1786          of %f32.
1787    
1788       3. Added a constructor INT32lit to the ast of MLRISC tools.
1789    
1790    ----------------------------------------------------------------------
1791    Name: Matthias Blume
1792    Date: 2002/01/16
1793    Tag: blume-20020116-ffifiddle
1794    Description:
1795    
1796    More fiddling with the FFI interface:
1797    
1798     - Make constness 'c instead of rw wherever possible.  This eliminates
1799       the need for certain explicit coercions.  (However, due to ML's
1800       value polymorphism, there will still be many cases where explicit
1801       coercions are necessary.  Phantom types are not the whole answer
1802       to modeling a subtyping relationship in ML.)
1803    
1804     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
1805    
1806     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
1807       clumsy workaround for varargs problem.)
1808    
1809    ----------------------------------------------------------------------
1810    Name: Lal George
1811    Date: 2002/01/15
1812    Tag: <none>
1813    Description:
1814    
1815    1. Since COPY instructions are no longer native to the architecture,
1816       a generic functor can be used to implement the expandCopies function.
1817    
1818    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
1819       TEXT segment.
1820    
1821    ----------------------------------------------------------------------
1822    Name: Matthias Blume
1823    Date: 2002/01/15
1824    Tag: blume-20020115-ffiupdates
1825    Description:
1826    
1827    1. Fix for bug resulting in single-precision float values being returned
1828       incorrectly from FFI calls.
1829    
1830    2. Small modifications to C FFI API:
1831    
1832        - memory-allocation routines return straight objects (no options)
1833          and raise an exception in out-of-memory situations
1834        - unsafe extensions to cast between function pointers and pointers
1835          from/to ints
1836        - added structure C_Debug as an alternative to structure C where
1837          pointer-dereferencing (|*| and |*!) always check for null-pointers
1838        - added open_lib' to DynLinkage;  open_lib' works like open_lib
1839          but also takes a (possibly empty) list of existing library handles
1840          that the current library depends on
1841    
1842    ----------------------------------------------------------------------
1843    Name: Matthias Blume
1844    Date: 2002/01/10
1845    Tag: blume-20020110-newffigen
1846    Description:
1847    
1848    1. Updates to portable graph code.
1849    
1850    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
1851       more scalable now so that even huge interfaces such as the one
1852       for GTK compile in finite time and space. :-)
1853       See src/ml-nlffigen/README for details on what's new.
1854    
1855    ----------------------------------------------------------------------
1856    Name: Lal George
1857    Date: 2001/01/09 14:31:35 EST 2002
1858    Tag: george-20011206-rm-native-copy
1859    Description:
1860    
1861            Removed the native COPY and FCOPY instructions
1862            from all the architectures and replaced it with the
1863            explicit COPY instruction from the previous commit.
1864    
1865            It is now possible to simplify many of the optimizations
1866            modules that manipulate copies. This has not been
1867            done in this change.
1868    
1869    ----------------------------------------------------------------------
1870    Name: Lal George
1871    Date: 2001/12/06 16:50:13 EST 2001
1872    Tag: george-20011206-mlrisc-instruction
1873    Description:
1874    
1875    Changed the representation of instructions from being fully abstract
1876    to being partially concrete. That is to say:
1877    
1878      from
1879            type instruction
1880    
1881      to
1882            type instr                              (* machine instruction *)
1883    
1884            datatype instruction =
1885                LIVE of {regs: C.cellset, spilled: C.cellset}
1886              | KILL of {regs: C.cellset, spilled: C.cellset}
1887              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
1888              | ANNOTATION of {i: instruction, a: Annotations.annotation}
1889              | INSTR of instr
1890    
1891    This makes the handling of certain special instructions that appear on
1892    all architectures easier and uniform.
1893    
1894    LIVE and KILL say that a list of registers are live or killed at the
1895    program point where they appear. No spill code is generated when an
1896    element of the 'regs' field is spilled, but the register is moved to
1897    the 'spilled' (which is present, more for debugging than anything else).
1898    
1899    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
1900    We used to generate:
1901    
1902            DEFFREG f1
1903            f1 := f2 + f3
1904            trapb
1905    
1906    but now generate:
1907    
1908            f1 := f2 + f3
1909            trapb
1910            LIVE {regs=[f1,f2,f3], spilled=[]}
1911    
1912    Furthermore, the DEFFREG (hack) required that all floating point instruction
1913    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
1914    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
1915    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
1916    this architecture).
1917    
1918    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
1919    all the architectures. This will result in further simplification of the
1920    register allocator that must be aware of them for coalescing purposes, and
1921    will also simplify certain aspects of the machine description that provides
1922    callbacks related to parallel copies.
1923    
1924    ANNOTATION should be obvious, and now INSTR represents the honest to God
1925    machine instruction set!
1926    
1927    The <arch>/instructions/<arch>Instr.sml files define certain utility
1928    functions for making porting easier -- essentially converting upper case
1929    to lower case. All machine instructions (of type instr) are in upper case,
1930    and the lower case form generates an MLRISC instruction. For example on
1931    the alpha we have:
1932    
1933      datatype instr =
1934         LDA of {r:cell, b:cell, d:operand}
1935       | ...
1936    
1937      val lda : {r:cell, b:cell, d:operand} -> instruction
1938        ...
1939    
1940    where lda is just (INSTR o LDA), etc.
1941    
1942    ----------------------------------------------------------------------
1943    Name: Matthias Blume
1944    Date: 2001/11/22 21:40:00 EST
1945    Tag: Release_110_37
1946    Description:
1947    
1948    Release 110.37.  This time for real.
1949    
1950    ----------------------------------------------------------------------
1951    Name: Matthias Blume
1952    Date: 2001/11/21 16:35:00 EST
1953    Tag: blume-20011121-foot-in-mouth
1954    Description:
1955    
1956    Removed the "Release_110_37" tag because of a serious bug.
1957    This will be re-tagged once the bug is fixed.
1958    
1959    ----------------------------------------------------------------------
1960    Name: Matthias Blume
1961    Date: 2001/11/21 16:14:00 EST
1962    Tag: blume-20011121-forgottenfile
1963    Description:
1964    
1965    Forgot to add a file.  (Just a .tex-file -- part of
1966    the CM manual source.)
1967    
1968    ----------------------------------------------------------------------
1969    Name: Matthias Blume
1970    Date: 2001/11/21 16:10:00 EST
1971    Tag: blume-20011121-invalid_110_37
1972    Description:
1973    
1974    Note: I removed the original tag "Release_110_37" from this commit
1975          because we found a serious bug in all non-x86 backends.
1976          - Matthias
1977    
1978    1. Modifications to the SML/NJ code generator and to the runtime system
1979       so that code object name strings are directly inserted into code
1980       objects at code generation time.  The only business the runtime system
1981       has with this is now to read the name strings on occasions.
1982       (The encoding of the name string has also changed somewhat.)
1983    
1984    2. CM now implements a simple "set calculus" for specifying export lists.
1985       In particular, it is now possible to refer to the export lists of
1986       other libraries/groups/sources and form unions as well as differences.
1987       See the latest CM manual for details.
1988    
1989    3. An separate notion of "proxy" libraries has again be eliminated from
1990       CM's model.  (Proxy libraries are now simply a special case of using
1991       the export list calculus.)
1992    
1993    4. Some of the existing libraries now take advantage of the new set
1994       calculus.
1995       (Notice that not all libraries have been converted because some
1996       of the existing .cm-files are supposed to be backward compatible
1997       with 110.0.x.)
1998    
1999    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
2000       "General.exnMessage"!  The former relies on a certain hook to be
2001       initialized, and that often does not happen in the stand-alone case.)
2002    
2003    ----------------------------------------------------------------------
2004    Name: Lal George
2005    Date: 2001/11/21  13:56:18 EST
2006    Tag: george-2001121-pseudo-ops
2007    Description:
2008    
2009      Implemented a complete redesign of MLRISC pseudo-ops. Now there
2010      ought to never be any question of incompatabilities with
2011      pseudo-op syntax expected by host assemblers.
2012    
2013      For now, only modules supporting GAS syntax are implemented
2014      but more should follow, such as MASM, and vendor assembler
2015      syntax, e.g. IBM as, Sun as, etc.
2016    
2017    ----------------------------------------------------------------------
2018    Name: Matthias Blume
2019    Date: 2001/11/14 11:52:00 EST
2020    Tag: blume-20011114-srcname
2021    Description:
2022    
2023    1. Routed the name of the current source file to mlriscgen where it
2024       should be directly emitted into the code object.  (This last part
2025       is yet to be done.)
2026    
2027    2. Some cleanup of the pgraph code to make it match the proposal that
2028       I put out the other day.  (The proposal notwithstanding, things are
2029       still in flux here.)
2030    
2031    ----------------------------------------------------------------------
2032    Name: Lal George
2033    Date: 2001/11/14 09:44:04 EST
2034    Tag:
2035    Description:
2036    
2037      Fix for a backpatching bug reported by Allen.
2038    
2039      Because the boundary between short and long span-dependent
2040      instructions is +/- 128, there are an astounding number of
2041      span-dependent instructions whose size is over estimated.
2042    
2043      Allen came up with the idea of letting the size of span
2044      dependent instructions be non-monotonic, for a maxIter
2045      number of times, after which the size must be monotonically
2046      increasing.
2047    
2048      This table shows the number of span-dependent instructions
2049      whose size was over-estimated as a function of maxIter, for the
2050      file Parse/parse/ml.grm.sml:
2051    
2052         maxIter            # of instructions:
2053            10                      687
2054            20                      438
2055            30                      198
2056            40                        0
2057    
2058      In compiling the compiler, there is no significant difference in
2059      compilation speed between maxIter=10 and maxIter=40. Actually,
2060      my measurements showed that maxIter=40 was a tad faster than
2061      maxIter=10! Also 96% of the  files in the compiler reach a fix
2062      point within 13 iterations, so fixing maxIter at 40, while high,
2063      is okay.
2064    
2065    ----------------------------------------------------------------------
2066    Name: Matthias Blume
2067    Date: 2001/10/31 15:25:00 EST
2068    Tag: blume-20011031-pgraph
2069    Description:
2070    
2071    CKIT:
2072    * Changed the "Function" constructor of type Ast.ctype to carry optional
2073      argument identifiers.
2074    * Changed the return type of TypeUtil.getFunction accordingly.
2075    * Type equality ignores the argument names.
2076    * TypeUtil.composite tries to preserve argument names but gives up quickly
2077      if there is a mismatch.
2078    
2079    installation script:
2080    * attempts to use "curl" if available (unless "wget" is available as well)
2081    
2082    CM:
2083    * has an experimental implementation of "portable graphs" which I will
2084      soon propose as an implementation-independent library format
2085    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
2086    
2087    NLFFI-LIB:
2088    * some cleanup (all cosmetic)
2089    
2090    NLFFIGEN:
2091    * temporarily disabled the mechanism that suppresses ML output for
2092      C definitions whose identifiers start with an underscore character
2093    * generate val bindings for enum constants
2094    * user can request that only one style (light or heavy) is being used;
2095      default is to use both (command-line arguments: -heavy and -light)
2096    * fixed bug in handling of function types involving incomplete pointers
2097    * generate ML entry points that take record arguments (i.e., using
2098      named arguments) for C functions that have a prototype with named
2099      arguments
2100      (see changes to CKIT)
2101    
2102    ----------------------------------------------------------------------
2103    Name: Allen Leung
2104    Date: 2001/10/27 20:34:00 EDT
2105    Tag: leunga-20011027-x86-fast-fp-call
2106    Description:
2107    
2108       Fixed the bug described in blume-20010920-slowfp.
2109    
2110       The fix involves
2111          1. generating FCOPYs in FSTP in ia32-svid
2112          2. marking a CALL with the appropriate annotation
2113    
2114    ----------------------------------------------------------------------
2115    Name: Matthias Blume
2116    Date: 2001/10/16 11:32:00 EDT
2117    Tag: blume-20011016-netbsd
2118    Description:
2119    
2120    Underscore patch from Chris Richards (fixing problem with compiling
2121    runtime system under recent NetBSD).
2122    
2123    ----------------------------------------------------------------------
2124    Name: Allen Leung
2125    Date: 2001/10/12 17:18:32 EDT 2001
2126    Tag: leung-20011012-x86-printflowgraph
2127    Description:
2128    
2129    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
2130    
2131    ----------------------------------------------------------------------
2132    Name: Lal George
2133    Date: 2001/10/11 23:51:34 EDT
2134    Tag: george-20011011-too-many-instrs
2135    Description:
2136    
2137    The representation of a program point never expected to see more
2138    than 65536 instructions in a basic block!
2139    
2140    ----------------------------------------------------------------------
2141    Name: Lal George
2142    Date: 2001/10/09 09:41:37 EDT
2143    Tag: george-20011008-mlrisc-labels
2144    Description:
2145    
2146    Changed the machine description files to support printing of
2147    local and global labels in assembly code, based on host assembler
2148    conventions.
2149    
2150    ----------------------------------------------------------------------
2151    Name: Matthias Blume
2152    Date: 2001/09/25 15:25:00 EDT
2153    Tag: blume-20010925-exninfo
2154    Description:
2155    
2156    I provided a non-hook implementation of exnName (at the toplevel) and
2157    made the "dummy" implementation of exnMessage (at the toplevel) more
2158    useful: if nothing gets "hooked in", then at least you are going to
2159    see the exception name and a message indicating why you don't see more.
2160    
2161    [For the time being, programs that need exnMessage and want to use
2162    ml-build should either use General.exnMessage (strongly recommended) or
2163    refer to structure General at some other point so that CM sees a
2164    static dependency.]
2165    
2166    [Similar remarks go for "print" and "use":  If you want to use their
2167    functionality in stand-alone programs generated by ml-build, then use
2168    TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
2169    
2170    ----------------------------------------------------------------------
2171    Name: Matthias Blume
2172    Date: 2001/09/20 17:28:00 EDT
2173    Tag: blume-20010920-slowfp
2174    Description:
2175    
2176    Allen says that x86-fast-fp is not safe yet, so I turned it off again...
2177    
2178    ----------------------------------------------------------------------
2179    Name: Matthias Blume
2180    Date: 2001/09/20 17:20:00 EDT
2181    Tag: blume-20010920-canonicalpaths
2182    Description:
2183    
2184    0. Updated the BOOT file (something that I forgot to do earlier).
2185    
2186    1. Small internal change to CM so that it avoids "/../" in filenames
2187       as much as possible (but only where it is safe).
2188    
2189    2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
2190       that arguments that contain delimiters are passed through correctly.
2191       This change also means that all "special" arguments of the form
2192       @SMLxxx... must come first.
2193    
2194    3. Changed install script to put relative anchor names for tool commands
2195       into pathconfig.
2196    
2197    ----------------------------------------------------------------------
2198    Name: Matthias Blume
2199    Date: 2001/09/18 15:35:00 EDT
2200    Tag: blume-20010918-readme11036
2201    Description:
2202    
2203    Added README files.
2204    
2205    ----------------------------------------------------------------------
2206    Name: Matthias Blume
2207    Date: 2001/09/18 11:45:00 EDT
2208    Tag: Release_110_36 (retag)
2209    Description:
2210    
2211    Fixed mistake in config/preloads. Retagged as 110.36.
2212    
2213    ----------------------------------------------------------------------
2214    Name: Matthias Blume
2215    Date: 2001/09/18 09:40:00 EDT
2216    Tag: Release_110_36_orig (tag changed)
2217    Description:
2218    
2219    New version (110.36).  New bootfiles.
2220    
2221    ----------------------------------------------------------------------
2222    Name: Matthias Blume
2223    Date: 2001/09/14 16:15:00 EDT
2224    Tag: blume-20010914-x86fastfp
2225    Description:
2226    
2227    John committed some changes that Allen made, in particular a (hopefully)
2228    correctly working version of the x86-fp module.
2229    
2230    I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
2231    flag to "true".  Everything seems to compile to a fixpoint ok, and
2232    "mandelbrot" speeds up by about 15%.
2233    
2234    ----------------------------------------------------------------------
2235    Name: Matthias Blume
2236    Date: 2001/09/13 11:20:00 EDT
2237    Tag: blume-20010913-minimal
2238    Description:
2239    
2240    1. Stefan Monnier's patch to fix a miscompilation problem that
2241       was brought to light by John Reppy's work on Moby.
2242    
2243    2. Implemented a minimal "structure Compiler" that contains just
2244       "version" and "architecture".  The minimal version will be
2245       available when the full version is not.  This is for backward-
2246       compatibility with code that wants to test Compiler.version.
2247    
2248    ----------------------------------------------------------------------
2249    Name: Matthias Blume
2250    Date: 2001/08/28 14:03:00 EDT
2251    Tag: blume-20010828-ml-lex
2252    Description:
2253    
2254    Fix for bug 1581, received from Neophytos Michael.
2255    
2256    ----------------------------------------------------------------------
2257    Name: Matthias Blume
2258    Date: 2001/08/27 11:20:00 EDT
2259    Tag: blume-20010827-readme11035
2260    Description:
2261    
2262    Fleshed out the README file for 110.35.
2263    
2264    ----------------------------------------------------------------------
2265    Name: Matthias Blume
2266    Date: 2001/08/24 17:10:00 EDT
2267    Tag: Release_110_35
2268    Description:
2269    
2270    New version number (110.35).  New bootfiles.
2271    
2272    ----------------------------------------------------------------------
2273    Name: Lal George
2274    Date: 2001/08/24 13:47:18 EDT 2001
2275    Tag: george-20010824-MLRISC-graphs
2276    Description:
2277    
2278     removed clusters from MLRISC completely and replaced with graphs.
2279    
2280    ----------------------------------------------------------------------
2281    Name: Matthias Blume
2282    Date: 2001/08/23 17:50:00 EDT
2283    Tag: blume-20010823-toplevel
2284    Description:
2285    
2286    - some reorganization of the code that implements various kinds of
2287      environments in the compiler (static, dynamic, symbolic, combined)
2288    - re-implemented the EnvRef module so that evalStream works properly
2289      (if the stream contains references to "use", "CM.make", etc.)
2290    - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
2291    
2292    ----------------------------------------------------------------------
2293    Name: Matthias Blume
2294    Date: 2001/08/20 15:50 EDT
2295    Tag: blume20010820-slipup
2296    Description:
2297    
2298    I forgot to commit a few files.  Here they are...
2299    
2300    ----------------------------------------------------------------------
2301    Name: Matthias Blume
2302    Date: 2001/08/20 15:35:00 EDT
2303    Tag: blume-20010820-debugprof
2304    Description:
2305    
2306    !!!! NEW BOOTFILES !!!!
2307    
2308    This is another round of reorganizing the compiler sources.  This
2309    time the main goal was to factor out all the "instrumentation"
2310    passes (for profiling and backtracing) into their own library.
2311    The difficulty was to do it in such a way that it does not depend
2312    on elaborate.cm but only on elabdata.cm.
2313    
2314    Therefore there have been further changes to both elaborate.cm and
2315    elabdata.cm -- more "generic" things have been moved from the former
2316    to the latter.  As a result, I was forced to split the assignment
2317    of numbers indicating "primtyc"s into two portions: SML-generic and
2318    SML/NJ-specific.  Since it would have been awkward to maintain,
2319    I bit the bullet and actually _changed_ the mapping between these
2320    numbers and primtycs.  The bottom line of this is that you need
2321    a new set of bin- and bootfiles.
2322    
2323    I have built new bootfiles for all architectures, so doing a fresh
2324    checkout and config/install.sh should be all you need.
2325    
2326    The newly created library's name is
2327    
2328        $smlnj/viscomp/debugprof.cm
2329    
2330    and its sources live under
2331    
2332        src/compiler/DebugProf
2333    
2334    ----------------------------------------------------------------------
2335    Name: Matthias Blume
2336    Date: 2001/08/15 17:15:00 EDT
2337    Tag: blume-20010815-compreorg
2338    Description:
2339    
2340    This is a first cut at reorganizing the CM libraries that make up the
2341    core of the compiler.  The idea is to separate out pieces that could
2342    be used independently by tools, e.g., the parser, the typechecker, etc.
2343    
2344    The current status is a step in this direction, but it is not quite
2345    satisfactory yet.  Expect more changes in the future.
2346    
2347    Here is the current (new) organization...
2348    
2349        What used to be $smlnj/viscomp/core.cm is now divided into
2350        six CM libraries:
2351    
2352             $smlnj/viscomp/basics.cm
2353                           /parser.cm
2354                           /elabdata.cm
2355                           /elaborate.cm
2356                           /execute.cm
2357                           /core.cm
2358    
2359        The CM files for these libraries live under src/system/smlnj/viscomp.
2360        All these libraries are proxy libraries that contain precisely
2361        one CM library component.  Here are the locations of the components
2362        (all within the src/compiler tree):
2363    
2364             Basics/basics.cm
2365             Parse/parser.cm
2366             ElabData/elabdata.cm
2367             Elaborator/elaborate.cm
2368             Execution/execute.cm
2369             core.cm
2370    
2371         [This organization is the same that has been used already
2372         for a while for the architecture-specific parts of the visible
2373         compiler and for the old version of core.cm.]
2374    
2375         As you will notice, many source files have been moved from their
2376         respective original locations to a new home in one of the above
2377         subtrees.
2378    
2379         The division of labor between the new libraries is the following:
2380    
2381             basics.cm:
2382                - Simple, basic definitions that pertain to many (or all) of
2383                  the other libraries.
2384             parser.cm:
2385                - The SML parser, producing output of type Ast.dec.
2386                - The type family for Ast is also defined and exported here.
2387             elabdata.cm:
2388                - The datatypes that describe input and output of the elaborator.
2389                  This includes types, absyn, and static environments.
2390             elaborator.cm:
2391                - The SML/NJ type checker and elaborator.
2392                  This maps an Ast.dec (with a given static environment) to
2393                  an Absyn.dec (with a new static environment).
2394                - This libraries implements certain modules that used to be
2395                  structures as functors (to remove dependencies on FLINT).
2396             execute.cm:
2397                - Everything having to do with executing binary code objects.
2398                - Dynamic environments.
2399             core.cm:
2400                - SML/NJ-specific instantiations of the elaborator and MLRISC.
2401                - Top-level modules.
2402                - FLINT (this should eventually become its own library)
2403    
2404    Notes:
2405    
2406    I am not 100% happy with the way I separated the elaborator (and its
2407    data structures) from FLINT.  Two instances of the same problem:
2408    
2409        1. Data structures contain certain fields that carry FLINT-specific
2410           information.  I hacked around this using exn and the property list
2411           module from smlnj-lib.  But the fact that there are middle-end
2412           specific fields around at all is a bit annoying.
2413    
2414        2. The elaborator calculates certain FLINT-related information.  I tried
2415           to make this as abstract as I could using functorization, but, again,
2416           the fact that the elaborator has to perform calculations on behalf
2417           of the middle-end at all is not nice.
2418    
2419        3. Having to used exn and property lists is unfortunate because it
2420           weakens type checking.  The other alternative (parameterizing
2421           nearly *everything*) is not appealing, though.
2422    
2423    I removed the "rebinding =" warning hack because due to the new organization
2424    it was awkward to maintain it.  As a result, the compiler now issues some of
2425    these warnings when compiling init.cmi during bootstrap compilation. On
2426    the plus side, you also get a warning when you do, for example:
2427       val op = = Int32.+
2428    which was not the case up to now.
2429    
2430    I placed "assign" and "deref" into the _Core structure so that the
2431    code that deals with the "lazy" keyword can find them there.  This
2432    removes the need for having access to the primitive environment
2433    during elaboration.
2434    
2435    ----------------------------------------------------------------------
2436    Name: Matthias Blume
2437    Date: 2001/08/13
2438    Tag: blume-20010813-closures
2439    Description:
2440    
2441    This fix was sent to us by Zhong Shao.  It is supposed to improve the
2442    performance of certain loops by avoiding needless closure allocation.
2443    
2444    ----------------------------------------------------------------------
2445    Name: Lal George
2446    Date: 2001/07/31 10:03:23 EDT 2001
2447    Tag: george-20010731-x86-fmalloc
2448    Description: Fixed bug in x86 calls
2449    
2450        There was a bug where call instructions would mysteriously
2451        vanish. The call instruction had to be one that returned
2452        a floating point value.
2453    
2454    ----------------------------------------------------------------------
2455    Name: Lal George
2456    Date: 2001/07/19 16:36:29 EDT 2001
2457    Tag: george-20010719-simple-cells
2458    Description:
2459    
2460    I have dramatically simplified the interface for CELLS in MLRISC.
2461    
2462    In summary, the cells interface is broken up into three parts:
2463    
2464      1. CellsBasis : CELLS_BASIS
2465    
2466            CellsBasis is a top level structure and common for all
2467            architectures.  it contains the definitions of basic datatypes
2468            and utility  functions over these types.
2469    
2470      2. functor Cells() : CELLS
2471    
2472            Cells generates an interface for CELLS that incorporates the
2473            specific resources on the target architecture, such as the
2474            presence of special register classes, their number and size,
2475            and various useful substructures.
2476    
2477      3. <ARCH>CELLS
2478    
2479            e.g. SparcCells: SPARCCELLS
2480    
2481            <ARCH>CELLS usually contains additional bindings for special
2482            registers  on the architecture, such as:
2483    
2484                    val r0 : cell           (* register zero *)
2485                    val y : cell            (* Y register *)
2486                    val psr : cell          (* processor status register *)
2487                    ...
2488    
2489            The structure returned by applying the Cells functor is opened
2490            in this interface.
2491    
2492    The main implication of all this is that the datatypes for cells is
2493    split between CellsBasis and CELLS -- a fairly simple change for user
2494    code.
2495    
2496    In the old scheme the CELLS interface had a definitional binding of
2497    the form:
2498    
2499            signature CELLS = sig
2500    
2501               structure CellsBasis = CellsBasis
2502    
2503               ...
2504    
2505            end
2506    
2507    With all the sharing constraints that goes on in MLRISC, this old
2508    design  quickly leads to errors such as:
2509    
2510            "structure definition spec inside of sharing ... "
2511    
2512    
2513    and appears to require an unacceptable amount of sharing and where
2514    constraint hackery.
2515    
2516    I think this error message (the interaction of definitional specs and
2517    sharing) requires more explanation on our web page.
2518    
2519    ----------------------------------------------------------------------
2520    Name: Matthias Blume
2521    Date: 2001/07/19 15:00:00 EDT
2522    Tag: blume-20010719-libreorg
2523    Description:
2524    
2525    This update puts together a fairly extensive but straightforward change
2526    to the way the libraries that implement the interactive system are
2527    organized:
2528    
2529       The biggest change is the elimination of structure Compiler.  As a
2530       replacement for this structure, there is now a CM library
2531       (known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
2532       that exports all the substructures of the original structure Compiler
2533       directly.  So instead of saying Compiler.Foo.bar one now simply
2534       says Foo.bar.  (The CM libraries actually export a collection of
2535       structures that is richer than the collection of substructures of
2536       structure Compiler.)
2537    
2538       To make the transition smooth, there is a separate library called
2539       $smlnj/compiler/compiler.cm which puts together and exports the
2540       original structure Compiler (or at least something very close to it).
2541    
2542       There are five members of the original structure Compiler
2543       that are not exported directly but which instead became members
2544       of a new structure Backend (described by signature BACKEND).  These are:
2545       structure Profile (: PROFILE), structure Compile (: COMPILE), structure
2546       Interact (: INTERACT), structure Machine (: MACHINE), and val
2547       architecture (: string).
2548    
2549       Structure Compiler.Version has become structure CompilerVersion.
2550    
2551       Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
2552       by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
2553       or x86, respectively.
2554       Each of these exports the same frontend structures that
2555       $smlnj/compiler.cm exports.  But they do not have a structure Backend
2556       and instead export some structure <Arch>Backend where <Arch> is Alpha32,
2557       Hppa, PPC, Sparc, or X86, respectively.
2558    
2559       Library $smlnj/compiler/all.cm exports the union of the exports of
2560       $smlnj/compiler/<arch>.cm
2561    
2562       There are no structures <Arch>Compiler anymore, use
2563       $smlnj/compiler/<arch>.cm instead.
2564    
2565       Library host-compiler-0.cm is gone.  Instead, the internal library
2566       that instantiates CM is now called cm0.cm.  Selection of the host
2567       compiler (backend) is no longer done here but. (Responsibility for it
2568       now lies with $smlnj/compiler/current.cm.  This seems to be more
2569       logical.)
2570    
2571       Many individual files have been moved or renamed.  Some files have
2572       been split into multiple files, and some "dead" files have been deleted.
2573    
2574    Aside from these changes to library organization, there are also changes
2575    to the way the code itself is organized:
2576    
2577       Structure Binfile has been re-implemented in such a way that it no
2578       longer needs any knowledge of the compiler.  It exclusively deals
2579       with the details of binfile layout.  It no longer invokes the
2580       compiler (for the purpose of creating new prospective binfile
2581       content), and it no longer has any knowledge of how to interpret
2582       pickles.
2583    
2584       Structure Compile (: COMPILE) has been stripped down to the bare
2585       essentials of compilation.  It no longer deals with linking/execution.
2586       The interface has been cleaned up considerably.
2587    
2588       Utility routines for dealing with linking and execution have been
2589       moved into their own substructures.
2590    
2591       (The ultimate goal of these changes is to provide a light-weight
2592       binfile loader/linker (at least for, e.g., stable libraries) that
2593       does not require CM or the compiler to be present.)
2594    
2595    CM documentation has been updated to reflect the changes to library
2596    organization.
2597    
2598    ----------------------------------------------------------------------
2599    Name: Matthias Blume
2600    Date: 2001/07/10 17:30:00 EDT
2601    Tag: Release_110_34
2602    Description:
2603    
2604    Minor tweak to 110.34 (re-tagged):
2605    
2606      - README.html file added to CVS repository
2607      - runtime compiles properly under FreeBSD 3.X and 4.X
2608    
2609    ----------------------------------------------------------------------
2610    Name: Matthias Blume
2611    Date: 2001/07/10 17:30:00 EDT
2612    Tag: Release_110_34
2613    Description:
2614    
2615    New version number (110.34). New bootfiles.
2616    
2617    ----------------------------------------------------------------------
2618    Name: Matthias Blume
2619    Date: 2001/07/09 16:00:00 EDT
2620    Tag: blume-20010709-more-varargs
2621    Description:
2622    
2623    I changed the handling of varargs in ml-nlffigen again:
2624    The ellipsis ... will now simply be ignored (with an accompanying warning).
2625    
2626    The immediate effect is that you can actually call a varargs function
2627    from ML -- but you can't actually supply any arguments beyond the ones
2628    specified explicitly.  (For example, you can call printf with its format
2629    string, but you cannot pass additional arguments.)
2630    
2631    This behavior is only marginally more useful than the one before, but
2632    it has the advantage that a function or, more importantly, a function
2633    type never gets dropped on the floor, thus avoiding follow-up problems with
2634    other types that refer to the offending one.
2635    
2636    ----------------------------------------------------------------------
2637    Name: Matthias Blume
2638    Date: 2001/07/09 11:25:00 EDT
2639    Tag: blume-20010709-varargs
2640    Description:
2641    
2642    1. ckit-lib.cm now exports structure Error
2643    2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
2644       with a warning accompanied by a source location.  Moreover, it
2645       merely skips the offending function or type and proceeds with the
2646       rest of its work.u  As a result, one can safely feed C code containing
2647       "..." to ml-nlffigen.
2648    3. There are some internal improvements to CM, providing slightly
2649       more general string substitutions in the tools subsystem.
2650    
2651    ----------------------------------------------------------------------
2652    Name: Matthias Blume
2653    Date: 2001/06/27 15:10:00 EDT
2654    Tag: blume-20010627-concur
2655    Description:
2656    
2657    Fixed a small bug in CM's handling of parallel compilation.
2658    (You could observe the bug by Control-C-interrupting an ordinary
2659    CMB.make or CM.stabilize and then attaching some compile servers.
2660    The result was that all of a sudden the previously interrupted
2661    compilation would continue on its own.  This was because of
2662    an over-optimization: CM did not bother to clean out certain queues
2663    when no servers were attached "anyway", resulting in the contents
2664    of these queues to grab control when new servers did get attached.)
2665    
2666    There is also another minor update to the CM manual.
2667    
2668    ----------------------------------------------------------------------
2669    Name: Matthias Blume
2670    Date: 2001/06/26 16:15:00 EDT
2671    Tag: blume-20010626-cmdoc
2672    Description:
2673    
2674    Minor typo fixed in CM manual (syntax diagram for libraries).
2675    
2676    ----------------------------------------------------------------------
2677    Name: Matthias Blume
2678    Date: 2001/06/25 22:55:00 EDT
2679    Tag: blume-20010625-x86pc
2680    Description:
2681    
2682    Fixed a nasty bug in the X86 assembly code that caused signal
2683    handlers to fail (crash) randomly.
2684    
2685    ----------------------------------------------------------------------
2686    Name: Matthias Blume
2687    Date: 2001/06/25 12:05:00 EDT
2688    Tag: blume-20010625-nlffigen
2689    Description:
2690    
2691    This update fixes a number of minor bugs in ml-nlffigen as reported by
2692    Nick Carter <nbc@andrew.cmu.edu>.
2693    
2694      1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
2695      2. Default names for generated files are now derived from the name of
2696         the C file *without its directory*.  In particular, this causes generated
2697         files to be placed locally even if the C file is in some system directory.
2698      3. Default names for generated signatures and structures are also derived
2699         from the C file name without its directory.  This avoids silly things
2700         like "structure GL/GL".
2701         (Other silly names are still possible because ml-nlffigen does not do
2702          a thorough check of whether generated names are legal ML identifiers.
2703          When in doubt, use command line arguments to force particular names.)
2704    
2705    ----------------------------------------------------------------------
2706    Name: Matthias Blume
2707    Date: 2001/06/21 12:25:00 EDT
2708    Tag: blume-20010621-eXene
2709    Description:
2710    
2711    eXene now compiles and (sort of) works again.
2712    
2713    The library name (for version > 110.33) is $/eXene.cm.
2714    
2715    I also added an new example in src/eXene/examples/nbody.  See the
2716    README file there for details.
2717    
2718    ----------------------------------------------------------------------
2719    Name: Matthias Blume
2720    Date: 2001/06/20 16:40:00 EDT
2721    Tag: blume-20010620-cml
2722    Description:
2723    
2724    CML now compiles and works again.
2725    
2726    Libraries (for version > 110.33):
2727    
2728      $cml/cml.cm            Main CML library.
2729      $cml/basis.cm          CML's version of $/basis.cm.
2730      $cml/cml-internal.cm   Internal helper library.
2731      $cml/core-cml.cm       Internal helper library.
2732      $cml-lib/trace-cml.cm  Tracing facility.
2733      $cml-lib/smlnj-lib.cm  CML's version of $/smlnj-lib.cm
2734    
2735    The installer (config/install.sh) has been taught how to properly
2736    install this stuff.
2737    
2738    ----------------------------------------------------------------------
2739    Name: Matthias Blume
2740    Date: 2001/06/19 17:55:00 EDT
2741    Tag: blume-20010619-instantiate
2742    Description:
2743    
2744    This un-breaks the fix for bug 1432.
2745    (The bug was originally fixed in 110.9 but I broke it again some
2746    time after that.)
2747    
2748    ----------------------------------------------------------------------
2749    Name: Matthias Blume
2750    Date: 2001/06/19 17:25:00 EDT
2751    Tag: blume-20010619-signals
2752    Description:
2753    
2754    This should (hopefully) fix the long-standing signal handling bug.
2755    (The runtime system was constructing a continuation record with an
2756    incorrect descriptor which would cause the GC to drop data on the floor...)
2757    
2758    ----------------------------------------------------------------------
2759    Name: Matthias Blume
2760    Date: 2001/06/15 15:05:00 EDT
2761    Tag: blume-20010615-moresparc
2762    Description:
2763    
2764    Here is a short late-hour update related to Sparc c-calls:
2765    
2766     -- made handling of double-word arguments a bit smarter
2767    
2768     -- instruction selection phase tries to collapse certain clumsily
2769        constructed ML-Trees; typical example:
2770    
2771            ADD(ty,ADD(_,e,LI d1),LI d2)  ->  ADD(ty,e,LI(d1+d2))
2772    
2773        This currently has no further impact on SML/NJ since mlriscGen does
2774        not seem to generate such patterns in the first place, and c-calls
2775        (which did generate them in the beginning) has meanwhile been fixed
2776        so as to avoid them as well.
2777    
2778    ----------------------------------------------------------------------
2779    Name: Matthias Blume
2780    Date: 2001/06/15 15:05:00 EDT
2781    Tag: blume-20010615-sparc
2782    Description:
2783    
2784    The purpose of this update is to provide an implementation of NLFFI
2785    on Sparc machines.
2786    
2787    Here are the changes in detail:
2788    
2789       * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
2790       the Sparc implementation of the c-calls API.
2791       * The Sparc backend of SML/NJ has been modified to uniformely use %fp
2792       for accessing the ML frame.  Thus, we have a real frame pointer and
2793       can freely modify %sp without need for an omit-frame-ptr phase.
2794       The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
2795       this case.
2796       * ml-nlffigen has been taught to produce code for different architectures
2797       and calling conventions.
2798       * In a way similar to what was done in the x86 case, the Sparc
2799       backend uses its own specific extension to mltree.  (For example,
2800       it needs to be able to generate UNIMP instructions which are part
2801       of the calling convention.)
2802       * ml-nlffi-lib was reorganized to make it more modular (in particular,
2803       to make it easier to plug in new machine- and os-dependent parts).
2804    
2805    There are some other fairly unrelated bug fixes and cleanups as well:
2806    
2807       * I further hacked the .cm files for MLRISC tools (like MDLGen) so
2808       that they properly share their libraries with existing SML/NJ libraries.
2809       * I fixed a minor cosmetic bug in CM, supressing certain spurious
2810       follow-up error messages.
2811       * Updates to CM/CMB documentation.
2812    
2813    TODO items:
2814    
2815       * MLRISC should use a different register as its asmTemp on the Sparc.
2816         (The current %o2 is a really bad choice because it is part of the
2817         calling conventions, so things might interfere in unexpected ways.)
2818    
2819    ----------------------------------------------------------------------
2820    Name: Matthias Blume
2821    Date: 2001/06/07
2822    Tag: blume-20010607-calls
2823    Description:
2824    
2825    A number of internal changes related to C calls and calling conventions:
2826    
2827    1. ML-Tree CALL statements now carry a "pops" field.  It indicates the
2828       number of bytes popped implicitly (by the callee).  In most cases
2829       this field is 0 but on x86/win32 it is some non-zero value.  This
2830       is information provided for the benefit of the "omit-frameptr" pass.
2831    2. The CALL instruction on the x86 carries a similar "pops" field.
2832       The instruction selection phase copies its value from the ML-Tree
2833       CALL statement.
2834    3. On all other architectures, the instruction selection phase checks
2835       whether "pops=0" and complains if not.
2836    4. The c-calls implementation for x86 now accepts two calling conventions:
2837       "ccall" and "stdcall".  When "ccall" is selected, the caller cleans
2838       up after the call and pops is set to 0.  For "stdcall", the caller
2839       does nothing, leaving the cleanup to the callee; pops is set to
2840       the number of bytes that were pushed onto the stack.
2841    5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
2842       distinguish between "ccall" and "stdcall".
2843    6. The UNIMP instruction has been added to the supported Sparc instruction
2844       set. (This is needed for implementing the official C calling convention
2845       on this architecture.)
2846    7. I fixed some of the .cm files under src/MLRISC/Tools to make them
2847       work with the latest CM.
2848    
2849    ----------------------------------------------------------------------
2850    Name: Matthias Blume
2851    Date: 2001/06/05 15:10:00 EDT
2852    Tag: blume-20010605-cm-index
2853    Description:
2854    
2855    0. The "lambdasplit" parameter for class "sml" in CM has been documented.
2856    
2857    1. CM can now generate "index files".  These are human-readable files
2858       that list on a per-.cm-file basis each toplevel symbol defined or
2859       imported.  The location of the index file for
2860       <p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
2861       To enable index-file generation, set CM.Control.generate_index to true
2862       or export an environment-symbol: export CM_GENERATE_INDEX=true.
2863    
2864       The CM manual has been updated accordingly.
2865    
2866    2. I made some slight modifications to the c-calls API in MLRISC.
2867    
2868         a) There is now a callback to support saving/restoring of
2869            dedicated but caller-save registers around the actual call
2870            instruction.
2871         b) One can optionally specify a comment-annotation for the
2872            call instruction.
2873    
2874    3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
2875       (For example, the comment annotation shows the C prototype of
2876        the function being called.)
2877    
2878    ----------------------------------------------------------------------
2879    Name: Matthias Blume
2880    Date: 2001/06/01 13:30:00 EDT
2881    Tag: blume-20010601-nlffi-cleanup
2882    Description:
2883    
2884    This is mostly a cleanup of MLFFI stuff:
2885    
2886       - some signature files have been put into a more exposed place
2887       - the ugly 'f type parameter is gone (simplifies types tremendously!)
2888       - ml-nlffigen changed accordingly
2889       - tutorial updated
2890    
2891    Other changes:
2892    
2893       - author's affiliation in CM manual(s) updated
2894       - some more recognized keywords added to Allen's sml.sty
2895    
2896    ----------------------------------------------------------------------
2897    Name: Matthias Blume
2898    Date: 2001/05/25 15:30:00 EDT
2899    Tag: blume-20010525-iptr
2900    Description:
2901    
2902      - put the official 110.33-README (as it appears on the ftp server) under
2903        CVS
2904      - fixed a small bug related to incomplete pointer types in
2905        ml-nlffigen
2906      - small cosmetic change to the ml-nlffi-lib's "arr" type constructor
2907        (it does not need the 'f type parameter)
2908    
2909    ----------------------------------------------------------------------
2910    Name: Matthias Blume
2911    Date: 2001/05/23 14:30:00 EDT
2912    Tag: Release_110_33
2913    Description:
2914    
2915    New version number (110.33).  New bootfiles.
2916    
2917    ----------------------------------------------------------------------
2918    Name: Matthias Blume
2919    Date: 2001/05/22 18:06:00 EDT
2920    Tag: blume-20010522-targets
2921    Description:
2922    
2923    Made install.sh use file config/targets.customized if it exists, falling
2924    back to config/targets if it doesn't.  This way one can have a customized
2925    version of the targets file without touching the "real thing", thus
2926    eliminating the constant fear of accidentally checking something bogus
2927    back into the CVS repository...  (File config/targets.customized must
2928    not be added to the repository!)
2929    
2930    ----------------------------------------------------------------------
2931    Name: Matthias Blume
2932    Date: 2001/05/22 16:30:00 EDT
2933    Tag: blume-20010522-minitut
2934    Description:
2935    
2936    1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
2937       struct returns.
2938    2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt.  This is some very
2939       incomplete, preliminary documentation for NLFFI.
2940    
2941    ----------------------------------------------------------------------
2942    Name: Matthias Blume
2943    Date: 2001/05/14 11:30:00 EDT
2944    Tag: blume-20010514-script
2945    Description:
2946    
2947    Some bugs in install script fixed.
2948    
2949    In addition to that I also made a slight change to the NLFFI API:
2950    Functors generated by ml-nlffigen now take the dynamic library as a
2951    straight functor argument, not as a suspended one.  (The original
2952    functor code used to force the suspension right away anyway, so there
2953    was nothing gained by this complication of the interface.)
2954    
2955    ----------------------------------------------------------------------
2956    Name: Matthias Blume
2957    Date: 2001/05/11 14:35:00 EDT
2958    Tag: blume-20010511-ml-nlffi
2959    Description:
2960    
2961    I finally took the plunge and added my new FFI code to the main
2962    repository.  For x86-linux it is now ready for prime-time.
2963    
2964    There are two new subdirectories of "src":
2965    
2966      - ml-nlffi-lib:
2967           The utility library for programs using the FFI interface.
2968           Here is the implementation of $/c.cm and its associated low-level
2969           partners $/c-int.cm and $/memory.cm.
2970      - ml-nlffigen:
2971           A stand-alone program for generating ML glue code from C source
2972           code.
2973    
2974    Building ml-nlffigen requires $/ckit-lib.cm.
2975    
2976    The config/install.sh script has been updates to do the Right Thing
2977    (hopefully).
2978    
2979    Notice that the source tree for the C-Kit will not be put under "src"
2980    but directly under the installation root directory.  (This is the
2981    structure that currently exists on the CVS server when you check out
2982    module "sml".)  Fortunately, config/install.sh knows about this oddity.
2983    
2984    Bugs: No documentation yet.
2985    
2986    ----------------------------------------------------------------------
2987    Name: Matthias Blume
2988    Date: 2001/05/09 16:35:00 EDT
2989    Tag: blume-20010509-cpscontract
2990    Description:
2991    
2992    Fixed a bug in the accounting code in cpsopt/contract.sml.  (The
2993    wrapper/unwrapper elimination did not decrement usage counts and some
2994    dead variables got overlooked by the dead-up logic.)
2995    
2996    ----------------------------------------------------------------------
2997    Name: Lal George
2998    Date: 2001/05/08  17:26:09 EDT
2999    Tag: george-20010508-omit-frameptr
3000    Description:
3001    
3002    Changes to implement the omit-frame-pointer optimization to support
3003    raw C calls. For now, there is only support on the Intel x86, but
3004    other architectures will follow as more experience is gained with this.
3005    
3006    
3007    ----------------------------------------------------------------------
3008    Name: Matthias Blume
3009    Date: 2001/05/07 14:40:00 EDT
3010    Tag: blume-20010507-proxies
3011    Description:
3012    
3013    I made into "proxy libraries" all libraries that qualify for such a
3014    change.  (A qualifying library is a library that has another library or
3015    groups as its sole member and repeats that member's export list
3016    verbatim.  A proxy library avoids this repetition by omitting its export
3017    list, effectively inheriting the list that its (only) member exports.
3018    See the CM manual for more explanation.)
3019    The main effect is that explicit export lists for these libraries
3020    do not have to be kepts in sync, making maintenance a bit easier.
3021    
3022    I also added copyright notices to many .cm-files.
3023    
3024    Last but not least, I made a new set of bootfiles.
3025    
3026    ----------------------------------------------------------------------
3027    Name: Matthias Blume
3028    Date: 2001/05/04 17:00:00 EDT
3029    Tag: blume-20010504-cm-lsplit
3030    Description:
3031    
3032    0. John merged pending changes to $/smlnj-lib.cm
3033    
3034    1. Allen's previous change accidentally backed out of one of Lal's
3035       earlier changes.  I undid this mistake (re-introducing Lal's change).
3036    
3037    2. I used the new topOrder' function from graph-scc.sml (from $/smlnj-lib.cm)
3038       within the compiler where applicable.  There is some code simplification
3039       because of that.
3040    
3041    3. The "split" phase (in FLINT) is now part of the default list of phases.
3042       Compiler.Control.LambdaSplitting.* can be used to globally control the
3043       lambda-splitting (cross-module-inlining) engine.  In addition to that,
3044       it can now also be controlled on a per-source basis: CM has been taught
3045       a new tool parameter applicable to ML source files.
3046    
3047       - To turn lambda-splitting off completely:
3048            local open Compiler.Control.LambdaSplitting in
3049                val _ = set Off
3050            end
3051       - To make "no lambda-splitting" the global default (but allow per-source
3052         overriding); this is the initial setting:
3053            local open Compiler.Control.LambdaSplitting in
3054                val _ = set (Default NONE)
3055            end
3056       - To make "lambda-splitting with aggressiveness a" the global default
3057         (and allow per-source overriding):
3058            local open Compiler.Control.LambdaSplitting in
3059                val _ = set (Default (SOME a))
3060            end
3061    
3062       - To turn lambda-splitting off for a given ML souce file (say: a.sml)
3063         write (in the respective .cm-file):
3064            a.sml (lambdasplitting:off)
3065       - To turn lambda-splitting for a.sml on with minimal aggressiveness:
3066            a.sml (lambdasplitting:on)
3067       - To turn lambda-splitting for a.sml on with aggressiveness <a> (where
3068         <a> is a decimal non-negative integer):
3069            a.sml (lambdasplitting:<a>)
3070       - To turn lambda-splitting for a.sml on with maximal aggressiveness:
3071            a.sml (lambdasplitting:infinity)
3072       - To use the global default for a.sml:
3073            a.sml (lambdasplitting:default)
3074         or simply
3075            a.sml
3076    
3077    ----------------------------------------------------------------------
3078    Name: Allen Leung
3079    Date: 2001/05/04 01:57:00 EDT
3080    Tag: leunga-20010504-sync
3081    Description:
3082    
3083      MLRISC features.
3084    
3085      1. Fix to CMPXCHG instructions.
3086      2. Changed RA interface to allow annotations in callbacks.
3087      3. Added a new method to the stream interface to allow annotations updates.
3088    
3089    ----------------------------------------------------------------------
3090    Name: Matthias Blume
3091    Date: 2001/05/01 11:45:00 EDT
3092    Tag: blume-20010501-pcedittmp
3093    Description:
3094    
3095    Changed install.sh to use the current working directory instead of
3096    /usr/tmp for a temporary file (pcedittmp).  The previous choice
3097    of /usr/tmp caused trouble with MacOS X because of file premission
3098    problems.
3099    
3100    ----------------------------------------------------------------------
3101    Name: Matthias Blume
3102    Date: 2001/04/20 11:10:00 EDT
3103    Tag: blume-20010420-inMLflag
3104    Description:
3105    
3106     - added vp_limitPtrMask to vproc-state.h
3107       (for use by the raw-C-calls mechanism to implement proper interrupt
3108        handling)
3109     - made the ML compiler aware of various data-structure offsets so it
3110       can generate code for accessing the vp_inML flag and vp_limitPtrMask
3111     - tweaked mlriscGen.sml to have it emit interrupt-handling code for
3112       raw C-calls
3113    
3114    ----------------------------------------------------------------------
3115    Name: Lal George
3116    Date: 2001/04/20 09:15:28 EDT
3117    Tag: george-20010420-macosX
3118    Description:
3119    
3120     - Changes to port to Mac OS X; Darwin.
3121    
3122     - In the process I found that sqrt was broken on the PPC, because the
3123       fsqrt instruction is not implemented.
3124    
3125    ----------------------------------------------------------------------
3126    Name: Matthias Blume
3127    Date: 2001/04/18 12:45:00 EDT
3128    Tag: blume-20010418-ccalls
3129    Description:
3130    
3131     - fixed two off-by-4 errors in the x86-specific c-calls implementation
3132       (this bug prevented structure arguments containing pointers from being
3133        passed correctly)
3134     - changed the raw-C-call code in mlriscGen.sml in such a way that
3135       structure arguments are represented as a pointer to the beginning
3136       of the structure (instead of having a series of synthesized arguments,
3137       one for each structure member)
3138    
3139     - made makeml script's verbosity level configurable via environment
3140       variable (MAKEML_VERBOSITY)
3141    
3142     - eliminated placeholder implementations for f32l, w16s, i16s, and f32s
3143       in rawmem-x86.sml; we are now using the real thing
3144    
3145    ----------------------------------------------------------------------
3146    Name: Matthias Blume
3147    Date: 2001/03/22 16:25:00 EST
3148    Tag: blume-20010322-bootfiles
3149    Description:
3150    
3151    Created a new set of bootfiles (for your automatic installation convenience).
3152    
3153    ----------------------------------------------------------------------
3154    Name: Matthias Blume
3155    Date: 2001/03/22 15:10:00 EST
3156    Tag: blume-20010322-rawmem-parcm
3157    Description:
3158    
3159    1. All "raw memory access" primitives for the new FFI are implemented now
3160       (at least on the x86).
3161    2. Some further cleanup of CM's parallel make mechanism.
3162    
3163    ----------------------------------------------------------------------
3164    Name: Matthias Blume
3165    Date: 2001/03/19 17:53:00 EST
3166    Tag: blume-20010319-parallel
3167    Description:
3168    
3169    Parallel make (using compile servers) now works again.
3170    
3171    To this end, CM.stabilize and CMB.make have been modified to work in
3172    two passes when compile servers are attached:
3173       1. Compile everything, do not perform stabilization; this pass
3174          uses compile servers
3175       2. Stabilize everything; this pass does not use compile servers
3176    If there are no compile servers, the two passes are combined into one
3177    (as before).  Splitting the passes increases the inherent parallelism
3178    in the dependency graph because the entire graph including all
3179    libraries is available at the same time.  This, in turn, improves
3180    server utilization.  The downside is that the master process will
3181    have to do some extra work after compilation is done (because for
3182    technical reasons it must re-read all the binfiles during stabilization).
3183    
3184    ----------------------------------------------------------------------
3185    Name: Matthias Blume
3186    Date: 2001/03/16 12:22:00 EST
3187    Tag: blume-20010316-bootfiles
3188    Description:
3189    
3190    Created a new set of bootfiles (for your automatic installation convenience).
3191    
3192    ----------------------------------------------------------------------
3193    Name: Matthias Blume
3194    Date: 2001/03/16 11:00:00 EST
3195    Tag: blume-20010316-MLTREE-fixup
3196    Description:
3197    
3198    This is a minor fixup for an (untagged) earlier commit by Allen.
3199    (A file was missing).
3200    
3201    ----------------------------------------------------------------------
3202    Name: Allen Leung
3203    Date: Mon Mar  5 18:54:57 EST 2001
3204    Tag: leunga-20010305-cut-support
3205    
3206    1. New support for alternative control-flow in MLTREE.
3207       Currently we support
3208    
3209          FLOW_TO(CALL ...., [k1,...,kn])
3210    
3211       This is needed for 'cuts to' in C-- and try/handle-like constructs
3212       in Moby
3213    
3214       New assembler flag "asm-show-cutsto" to turn on control-flow debugging.
3215    
3216    2. Register Allocator
3217    
3218       Changes in interface [from Fermin, John]
3219    
3220    3. Alpha 8-bit SLL support [Fermin]
3221    
3222    4. All architectures
3223    
3224       A new module (ClusterExpandCopies) for expanding parallel copies.
3225    
3226    ----------------------------------------------------------------------
3227    Name: Allen Leung
3228    Date: 2001/02/27 23:07:00 EST
3229    Tag: leunga-20010227-minor-stuff
3230    
3231    1. Alpha bug fix for CMOVNE
3232    2. Handle mltree COND(..,FCMP ...,...)
3233    3. Bug fix in simplifier
3234    
3235    ----------------------------------------------------------------------
3236    Name: Matthias Blume
3237    Date: 2001/01/30 17:50:00 EST
3238    Tag: blume-20010130-sync
3239    Description:
3240    
3241    This is just a minor update to sync my devel branch with the main brach.
3242    The only visible change is the addition of some README files.
3243    
3244    ----------------------------------------------------------------------
3245    Name: Matthias Blume
3246    Date: 2001/01/12 23:30:00 JST
3247    Tag: blume-20010112-bootfiles
3248    Description:
3249    
3250    Made a new set of bootfiles that goes with the current state of the
3251    repository.
3252    
3253    ----------------------------------------------------------------------
3254    Name: Matthias Blume
3255    Date: 2001/01/12 21:20:00 JST
3256    Tag: blume-20010112-sync
3257    Description:
3258    
3259    I am just flushing out some minor changes that had accumulated in
3260    my private branch in order to sync with the main tree.  (This is
3261    mainly because I had CVS trouble when trying to merge _into_ my
3262    private branch.)
3263    
3264    Most people should be completely unaffected by this.
3265    
3266    ----------------------------------------------------------------------
3267    Name: Allen Leung
3268    Date: Thu Jan 11 21:03:00 EST 2001
3269    Tag: leunga-20010111-labexp=mltree
3270    Description:
3271    
3272    1.  Removed the type LabelExp and replace it by MLTree.
3273    2.  Rewritten mltree-simplify with the pattern matcher tool.
3274    3.  There were some bugs in alpha code generator which would break
3275        64-bit code generation.
3276    4.  Redo the tools to generate code with the
3277    5.  The CM files in MLRISC (and in src/system/smlnj/MLRISC)
3278        are now generated by perl scripts.
3279    
3280    ----------------------------------------------------------------------
3281    Name: Matthias Blume
3282    Date: 2001/01/10 21:55:00 JST
3283    Tag: blume-20010110-rcc
3284    Description:
3285    
3286    The RCC stuff now seems to work (but only on the x86).
3287    This required hacking of the c-calls interface (and -implementation) in
3288    MLRISC.
3289    
3290    Normal compiler users should be unaffected.
3291    
3292    ----------------------------------------------------------------------
3293    Name: Matthias Blume
3294    Date: 2001/01/09 01:20:00 JST
3295    Tag: blume-20010109-rcc
3296    Description:
3297    
3298    This is a fairly big patch, flushing out a large number of pending
3299    changes that I made to my development copy over the last couple of days.
3300    
3301    Of practical relevance at this moment is a workaround for a pickling
3302    bug that Allen ran into the other day.  The cause of the bug itself is
3303    still unknown and it might be hard to fix it properly, but the
3304    workaround has some merits of its own (namely somewhat reducing pickling
3305    overhead for certain libraries).  Therefore, I think this solution should
3306    be satisfactory at this time.
3307    
3308    The rest of the changes (i.e., the vast majority) has to do with my
3309    ongoing efforts of providing direct support for C function calls from
3310    ML.  At the moment there is a new primop "RAW_CCALL", typing magic
3311    in types/cproto.sml (invoked from FLINT/trans/translate.sml), a new
3312    case in the FLINT CPS datatype (RCC), changes to cps/convert.sml to
3313    translate uses of RAW_CCALL into RCC, and changes to mlriscGen.sml to
3314    handle RCC.
3315    
3316    The last part (the changes to mlriscGen.sml) are still known to be
3317    wrong on the x86 and not implemented on all other architectures.  But
3318    the infrastructure is in place. I had to change a few functor
3319    signatures in the backend to be able to route the CCalls interface
3320    from MLRISC there, and I had to specialize the mltree type (on the
3321    x86) to include the necessary extensions. (The extensions themselves
3322    were already there and redy to go in MLRISC/x86).
3323    
3324    Everything should be very happy as soon as someone helps me with
3325    mlriscGen.sml...
3326    
3327    In any case, nothing of this should matter to anyone as long as the
3328    new primop is not being used (which is going to be the case unless you
3329    find it where I hid it :). The rest of the compiler is completely
3330    unaffected.
3331    
3332    ----------------------------------------------------------------------
3333    Name: Matthias Blume
3334    Date: 2001/01/05 00:30:00 JST
3335    Tag: blume-20010105-primops
3336    Description:
3337    
3338    Added some experimental support for work that I am doing right now.
3339    These changes mostly concern added primops, but there is also a new
3340    experimental C library in the runtime system (but currently not enabled
3341    anywhere except on Linux/X86).
3342    
3343    In the course of adding primops (and playing with them), I discovered that
3344    Zhong's INL_PRIM hack (no type info for certain primops) was, in fact, badly
3345    broken.  (Zhong was very right he labeled this stuff as "major gross hack".)
3346    To recover, I made type information in INL_PRIM mandatory and changed
3347    prim.sml as well as built-in.sml accordingly.  The InLine structure now
3348    has complete, correct type information (i.e., no bottom types).
3349    
3350    Since all these changes mean that we need new binfiles, I also bumped the
3351    version number to 110.32.1.
3352    
3353    ----------------------------------------------------------------------
3354    Name: Matthias Blume
3355    Date: 2000/12/30 22:10:00 JST
3356    Tag: blume-20001230-various
3357    Description:
3358    
3359    Added proxy libraries for MLRISC and let MLRISC libraries refer
3360    to each other using path anchors.  (See CM manual for explanation.)
3361    
3362    Updated CM documentation.
3363    
3364    Fixed some bugs in CM.
3365    
3366    Implemented "proxy" libraries (= syntactic sugar for CM).
3367    
3368    Added "-quiet" option to makeml and changed runtime system accordingly.
3369    
3370    Added cleanup handler for exportML to reset timers and compiler stats.
3371    
3372    ----------------------------------------------------------------------
3373    Name: Lal George
3374    Date: 2000/12/22 22:22:58 EST 2000
3375    Tag: Release_110_32
3376    Description:
3377    
3378            Infinite precision used throughout MLRISC.
3379            see MLRISC/mltree/machine-int.sig
3380    
3381    ----------------------------------------------------------------------
3382    Name: Matthias Blume
3383    Date: 2000/12/22 23:16:00 JST
3384    Tag: blume-20001222-warn
3385    Description:
3386    
3387    Corrected wording and formatting of some CM warning message which I
3388    broke in my previous patch.
3389    
3390    ----------------------------------------------------------------------
3391    Name: Matthias Blume
3392    Date: 2000/12/22 21:20:00 JST
3393    Tag: blume-20001222-anchorenv
3394    Description:
3395    
3396    Fixed CM's handling of anchor environments in connection with CMB.make.
3397    
3398    ----------------------------------------------------------------------
3399    Name: Matthias Blume
3400    Date: 2000/12/22 13:15:00 JST
3401    Tag: blume-20001222-cleanup
3402    Description:
3403    
3404    Removed src/cm/ffi which does not (and did not) belong here.
3405    
3406    ----------------------------------------------------------------------
3407    Name: Matthias Blume
3408    Date: 2000/12/21 23:55:00 JST
3409    Tag: blume-20001221-exn
3410    Description:
3411    
3412    Probably most important: CM no longer silently swallows all exceptions
3413    in the compiler.
3414    Plus: some other minor CM changes.  For example, CM now reports some
3415    sizes for generated binfiles (code, data, envpickle, lambdapickle).
3416    
3417    ----------------------------------------------------------------------
3418    Name: Matthias Blume
3419    Date: 2000/12/15 00:01:05 JST
3420    Tag: blume-20001215-dirtool
3421    Description:
3422    
3423    - "dir" tool added.
3424    - improvements and cleanup to Tools structure
3425    - documentation updates
3426    
3427    ----------------------------------------------------------------------
3428    Name: Allen Leung
3429    Date: Thu Dec 14 03:45:24 EST 2000
3430    Description:
3431    Tag:  leunga-20001214-int-inf
3432    Description:
3433    
3434       In IntInf, added these standard functions, which are missing from our
3435    implementation:
3436    
3437        andb : int * int -> int
3438        xorb : int * int -> int
3439        orb  : int * int -> int
3440        notb : int -> int
3441         <<   : int * word -> int
3442        ~>>  : int * word -> int
3443    
3444       Not tested, I hope they are correct.
3445    
3446    ----------------------------------------------------------------------
3447    Name: Allen Leung
3448    Date: Fri Dec  8 19:23:26 EST 2000
3449    Description:
3450    Tag:  leunga-20001208-nowhere
3451    Description:
3452    
3453      Slight improvements to the 'nowhere' tool to handle OR-patterns,
3454    to generate better error messages etc.  Plus a brief manual.
3455    
3456    ----------------------------------------------------------------------
3457    Name: Lal George
3458    Date: 2000/12/08 09:54:02 EST 2000
3459    Tag: Release_110_31
3460    Description:
3461    
3462    - Version 110.31
3463    ----------------------------------------------------------------------
3464    Name: Allen Leung
3465    Date: Thu Dec  7 22:01:04 EST 2000
3466    Tag:  leunga-20001207-cell-monster-hack
3467    Description:
3468    
3469    Major MLRISC internal changes.  Affect all clients.
3470    Summary:
3471    
3472    1.  Type CELLS.cell = int is now replaced by a datatype.
3473        As a result, the old regmap is now gone.  Almost all interfaces
3474        in MLRISC change as a consequence.
3475    
3476    2.  A new brand version of machine description tool (v3.0) that generates
3477        modules expecting the new interface.  The old version is removed.
3478    
3479    3.  The RA interface has been further abstracted into two new functors.
3480        RISC_RA and X86RA.  These functors have much simpler interfaces.
3481        [See also directory MLRISC/demo.]
3482    
3483    4.  Some other new source->source code generation tools are available:
3484    
3485        a. MLRISC/Tools/RewriteGen -- generate rewriters from rules.
3486        b. MLRISC/Tools/WhereGen -- expands conditional pattern matching rules.
3487           I use this tool to generate the peephole optimizers---with the new
3488           cell type changes, peephole rules are becoming difficult to write
3489           without conditional pattern matching.
3490    
3491    5.  More Intmap -> IntHashTable change.  Previous changes by Matthias didn't
3492        cover the entire MLRISC source tree so many things broke.
3493    
3494    6.  CM files have been moved to the subdirectory MLRISC/cm.
3495        They are moved because there are a lot of them and they clutter up the
3496        root dir.
3497    
3498    7.  More detailed documentation to come...
3499    
3500        NOTE: To rebuild from 110.30 (ftp distribution), you'll have to do
3501        a makeml -rebuild first.  This is because of other other
3502        changes that Matthias has made (see below).
3503    
3504    
3505    ----------------------------------------------------------------------
3506    Name: Matthias Blume
3507    Date: 2000/11/30 23:12:00 JST
3508    Tag: blume-20001130-filereorg
3509    Description:
3510    
3511    Some manual updates and some file reorganizations in CM.
3512    
3513    ----------------------------------------------------------------------
3514    Name: Matthias Blume
3515    Date: 2000/11/24 17:45:00 JST
3516    Tag: blume-20001124-link
3517    Description:
3518    
3519    Drastically improved link traversal code for the case that the dynamic
3520    value was already loaded at bootstrap time.  As a result, CM and CMB
3521    now both load blazingly fast -- even on a very slow machine.  Also,
3522    memory consumption has been further reduced by this.
3523    
3524    Warning: The format of the PIDMAP file has changed.  THerefore, to
3525    bootstrap you have to do this:
3526    
3527    1. Run CMB.make
3528    2. Make a symbolic link for the boot directory:
3529         ln -s sml.boot.ARCH-OS xxx
3530    3. "Rebuild" the boot directory:
3531         ./makeml -boot xxx -rebuild sml ; rm xxx
3532    4. Boot normally:
3533          ./makeml
3534    
3535    ----------------------------------------------------------------------
3536    Name: Matthias Blume
3537    Date: 2000/11/21 21:20:00 JST
3538    Tag: blume-20001121-tools
3539    Description:
3540    
3541    Continued hacking on autoloading problem -- with success this time.
3542    Also changed tool-plugin mechanism.  See new CM manual.
3543    
3544    ----------------------------------------------------------------------
3545    Name: Matthias Blume
3546    Date: 2000/11/19 14:30:00 JST
3547    Tag:  blume-20001119-autoload
3548    Description:
3549    
3550    Some hacking to make autoloading faster.  Success for CMB, no success
3551    so far for CM.  There is a reduced structure CM' that autoloads faster.
3552    (This is a temporary, non-documented hack to be eliminated again when
3553    the general problem is solved.)
3554    
3555    ----------------------------------------------------------------------
3556    Name: Matthias Blume
3557    Date: 2000/11/17 14:10:00 JST
3558    Tag: blume-20001117-pickle-lib
3559    Description:
3560    
3561    1. Eliminated comp-lib.cm
3562    2. Made pickle-lib.cm
3563    3. Eliminated all uses of intset.sml (from comp-lib.cm)
3564    4. Replaced all uses of intmap.{sig,sml} (from comp-lib.cm) with
3565       equivalent constructs from smlnj-lib.cm (INtHashTable).
3566    5. Point 4. also goes for those uses of intmap.* in MLRISC.
3567       Duplicated intmap modules thrown out.
3568    6. Hunted down all duplicated SCC code and replaced it with
3569       equivalent stuff (GraphSCCFn from smlnj-lib.cm).
3570    7. Rewrote Feedback module.
3571    8. Moved sortedlist.sml into viscomp-lib.cm.  Eventually it
3572       should be thrown out and equivalent modules from smlnj-lib.cm
3573       should be used (IntRedBlackSet, IntListSet, ...).
3574    
3575    Confirmed that compiler compiles to fixpoint.
3576    
3577    ----------------------------------------------------------------------
3578    Name: Allen Leung
3579    Date: 2000/11/10 18:00:00
3580    Tag: leunga-20001110-new-x86-fp
3581    
3582    A new x86 floating point code generator has been added.
3583    By default this is turned off.  To turn this on, do:
3584    
3585        CM.autoload "$smlnj/compiler.cm";
3586        Compiler.Control.MLRISC.getFlag "x86-fast-fp" := true;
3587    
3588    Changes:
3589    
3590    1.  Changed FTAN to FPTAN so that the assembly output is correct.
3591    2.  Changed the extension callback for FTANGENT to generate:
3592    
3593              fptan
3594              fstp  %st(0)
3595        instead of
3596              fptan
3597              fstpl ftempmem
3598    
3599    3.  Numerous assembly fixes for x86.
3600    
3601    5.  Cleaned up the machine code output module x86/x86MC.sml and added
3602        support for a whole bunch of instructions and addressing modes:
3603    
3604          fadd/fsub/fsubr/fmul/fdiv/fdivr  %st, %st(n)
3605          faddp/fsubp/fsubrp/fmulp/fdivp/fdivrp  %st, %st(n)
3606          fadd/fsub/fsubr/fmul/fdiv/fdivr  %st(n), %st
3607          fiadd/fisub/fisubr/fimul/fidiv/fidivr mem
3608          fxch %st(n)
3609          fld %st(n)
3610          fst %st(n)
3611          fst mem
3612          fstp %st(n)
3613          fucom %st(n)
3614          fucomp %st(n)
3615    
3616        All these are now generated when the fast fp mode is turned on.
3617    
3618    6.  Removed the dedicated registers %st(0), ..., %st(7) from X86CpsRegs
3619    
3620    ----------------------------------------------------------------------
3621    Nam