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 902, Wed Aug 15 21:17:05 2001 UTC revision 1181, Wed Mar 27 21:27:27 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: 2002/03/27 16:27:00 EST
18    Tag: blume-20020327-mlrisc-divisions
19    Description:
20    
21    Added support for all four division operations (ML's div, mod, quot,
22    and rem) to MLRISC.  In the course of doing so, I also rationalized
23    the naming (no more annoying switch-around of DIV and QUOT), by
24    parameterizing the operation by div_rounding_mode (which can be either
25    DIV_TO_ZERO or DIV_TO_NEGINF).
26    
27    The generic MLTreeGen functor takes care of compiling all four
28    operations down to only round-to-zero div.
29    
30    Missing pieces:
31    
32      * Doing something smarter than relying on MLTreeGen on architectures
33        like, e.g., the x86 where hardware division delivers both quotient and
34        remainder at the same time.  With this, the implementation of the
35        round-to-neginf operations could be further streamlined.
36    
37      * Remove inlining support for div/mod/rem from the frontend and replace it
38        with primops that get carried through to the backend.  Do this for all
39        int and word types.
40    
41    ----------------------------------------------------------------------
42    Name: Matthias Blume
43    Date: 2002/03/25 17:25:00 EST
44    Tag: blume-20020325-divmod
45    Description:
46    
47    I improved (hopefully without breaking them) the implementation of Int.div,
48    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
49    advantage of the following observations:
50    
51      Let  q = x quot y      r = x rem y
52           d = x div  y      m = x mod y
53    
54    where "quot" is the round-to-zero version of integer division that
55    hardware usually provides.  Then we have:
56    
57         r = x - q * y        where neither the * nor the - will overflow
58         d = if q >= 0 orelse x = q * y then q else q - 1
59                              where neither the * nor the - will overflow
60         m = if q >= 0 orelse r = 0 then r else r + y
61                              where the + will not overflow
62    
63    This results in substantial simplification of the generated code.
64    The following table shows the number of CFG nodes and edges generated
65    for
66            fun f (x, y) = x OPER y
67            (* with OPER \in div, mod, quot, rem *)
68    
69    
70        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
71        --------------------------------------------------------
72         div |         24 |         39 |         12 |         16
73         mod |         41 |         71 |         12 |         16
74        quot |          8 |         10 |          8 |         10
75         rem |         10 |         14 |          8 |         10
76    
77    
78    ----------------------------------------------------------------------
79    Name: Matthias Blume
80    Date: 2002/03/25 22:06:00 EST
81    Tag: blume-20020325-cprotobug
82    Description:
83    
84    Fixed a bug in cproto (c prototype decoder).
85    
86    ----------------------------------------------------------------------
87    Name: Matthias Blume
88    Date: 2002/03/25 16:00:00 EST
89    Tag: blume-20020325-raw-primops
90    Description:
91    
92    I did some cleanup to Allen's new primop code and
93    replaced yesterday's bootfiles with new ones.
94    (But they are stored in the same place.)
95    
96    ----------------------------------------------------------------------
97    Name: Matthias Blume
98    Date: 2002/03/24 22:40:00 EST
99    Tag: blume-20020324-bootfiles
100    Description:
101    
102    Made the bootfiles that Allen asked for.
103    
104    ----------------------------------------------------------------------
105    Name: Allen Leung
106    Date: 2002/03/23 15:50:00 EST
107    Tag: leunga-20020323-flint-cps-rcc-primops
108    Description:
109    
110      1. Changes to FLINT primops:
111    
112        (* make a call to a C-function;
113         * The primop carries C function prototype information and specifies
114         * which of its (ML-) arguments are floating point. C prototype
115         * information is for use by the backend, ML information is for
116         * use by the CPS converter. *)
117      | RAW_CCALL of { c_proto: CTypes.c_proto,
118                       ml_args: ccall_type list,
119                       ml_res_opt: ccall_type option,
120                       reentrant : bool
121                     } option
122       (* Allocate uninitialized storage on the heap.
123        * The record is meant to hold short-lived C objects, i.e., they
124        * are not ML pointers.  With the tag, the representation is
125        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
126        *)
127      | RAW_RECORD of {tag:bool,sz:int}
128      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
129    
130      2.  These CPS primops are now overloaded:
131    
132           rawload of {kind:numkind}
133           rawstore of {kind:numkind}
134    
135          The one argument form is:
136    
137             rawload {kind} address
138    
139          The two argument form is:
140    
141             rawload {kind} [ml object, byte-offset]
142    
143      3. RAW_CCALL/RCC now takes two extra arguments:
144    
145         a. The first is whether the C call is reentrant, i.e., whether
146            ML state should be saved and restored.
147         b. The second argument is a string argument specifying the name of
148            library and the C function.
149    
150         These things are currently not handled in the code generator, yet.
151    
152      4. In CProto,
153    
154         An encoding type of "bool" means "ml object" and is mapped into
155         C prototype of PTR.  Note that "bool" is different than "string",
156         even though "string" is also mapped into PTR, because "bool"
157         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
158    
159      5. Pickler/unpicker
160    
161         Changed to handle RAW_RECORD and newest RAW_CCALL
162    
163      6. MLRiscGen,
164    
165         1. Changed to handle the new rawload/rawstore/rawrecord operators.
166         2. Code for handling C Calls has been moved to a new module CPSCCalls,
167            in the file CodeGen/cpscompile/cps-c-calls.sml
168    
169      7. Added the conditional move operator
170    
171             condmove of branch
172    
173         to cps.  Generation of this is still buggy so it is currently
174         disabled.
175    
176    ----------------------------------------------------------------------
177    Name: Lal George
178    Date: 2002/03/22 14:18:25 EST
179    Tag: george-20020322-cps-branch-prob
180    Description:
181    
182    Implemented the Ball-Larus branch prediction-heuristics, and
183    incorporated graphical viewers for control flow graphs.
184    
185    Ball-Larus Heuristics:
186    ---------------------
187    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
188    
189    By design it uses the Dempster-Shafer theory for combining
190    probabilities.  For example, in the function:
191    
192        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
193    
194    the ball-larus heuristics predicts that the n=0 is unlikely
195    (OH-heuristic), and the 'then' branch is unlikely because of the
196    RH-heuristic -- giving the 'then' branch an even lower combined
197    probability using the Dempster-Shafer theory.
198    
199    Finally, John Reppy's loop analysis in MLRISC, further lowers the
200    probability of the 'then' branch because of the loop in the else
201    branch.
202    
203    
204    Graphical Viewing:
205    ------------------
206    I merely plugged in Allen's graphical viewers into the compiler. The
207    additional code is not much. At the top level, saying:
208    
209            Control.MLRISC.getFlag "cfg-graphical-view" := true;
210    
211    will display the graphical view of the control flow graph just before
212    back-patching.  daVinci must be in your path for this to work. If
213    daVinci is not available, then the default viewer can be changed
214    using:
215    
216            Control.MLRISC.getString "viewer"
217    
218    which can be set to "dot" or "vcg" for the corresponding viewers. Of
219    course, these viewers must be in your path.
220    
221    The above will display the compilation unit at the level of clusters,
222    many of which are small, boring, and un-interesting. Also setting:
223    
224            Control.MLRISC.getInt "cfg-graphical-view_size"
225    
226    will display clusters that are larger than the value set by the above.
227    
228    
229    ----------------------------------------------------------------------
230    Name: Matthias Blume
231    Date: 2002/03/21 22:20:00 EST
232    Tag: blume-20020321-kmp-bugfix
233    Description:
234    
235    Changed the interface to the KMP routine in PreString and fixed
236    a minor bug in one place where it was used.
237    
238    ----------------------------------------------------------------------
239    Name: Allen Leung
240    Date: 2002/03/21 20:30:00 EST
241    Tag: leunga-20020321-cfg
242    Description:
243    
244      Fixed a potential problem in cfg edge splitting.
245    
246    ----------------------------------------------------------------------
247    Name: Allen Leung
248    Date: 2002/03/21 17:15:00 EST
249    Tag: leunga-20020321-x86-fp-cfg
250    Description:
251    
252      1. Recoded the buggy parts of x86-fp.
253    
254         a. All the block reordering code has been removed.
255            We now depend on the block placement phases to do this work.
256    
257         b. Critical edge splitting code has been simplified and moved into the
258            CFG modules, as where they belong.
259    
260         Both of these were quite buggy and complex.  The code is now much, much
261         simpler.
262    
263      2. X86 backend.
264    
265         a. Added instructions for 64-bit support.  Instruction selection for
266            64-bit has not been committed, however, since that
267            requires changes to MLTREE which haven't been approved by
268            Lal and John.
269    
270         b. Added support for FUCOMI and FUCOMIP when generating code for
271            PentiumPro and above.  We only generate these instructions in
272            the fast-fp mode.
273    
274         c. Added cases for JP and JNP in X86FreqProps.
275    
276      3. CFG
277    
278         CFG now has a bunch of methods for edge splitting and merging.
279    
280      4. Machine description.
281    
282         John's simplification of MLTREE_BASIS.fcond broke a few machine
283         description things:
284    
285         rtl-build.{sig,sml} and hppa.mdl fixed.
286    
287         NOTE: the machine description stuff in the repository is still broken.
288               Again, I can't put my fixes in because that involves
289               changes to MLTREE.
290    
291    ----------------------------------------------------------------------
292    Name: Matthias Blume
293    Date: 2002/03/20 15:55:00 EST
294    Tag: blume-20020320-kmp
295    Description:
296    
297    Implemented Knuth-Morris-Pratt string matching in PreString and used
298    it for String.isSubstring, Substring.isSubstring, and
299    Substring.position.
300    
301    (Might need some stress-testing.  Simple examples worked fine.)
302    
303    ----------------------------------------------------------------------
304    Name: Matthias Blume
305    Date: 2002/03/19 16:37:00 EST
306    Tag: blume-20020319-witnesses
307    Description:
308    
309    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
310    
311    This implements a generic mechanism for changing constness qualifiers
312    anywhere within big C types without resorting to outright "casts".
313    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
314    modify the constness at the outermost level.)
315    The implementation of "convert" is based on the idea of "witness"
316    values -- values that are not used by the operation but whose types
317    "testify" to their applicability.  On the implementation side, "convert"
318    is simply a projection (returning its second curried argument).  With
319    cross-module inlining, it should not result in any machine code being
320    generated.
321    
322    ----------------------------------------------------------------------
323    Name: Matthias Blume
324    Date: 2002/03/15 16:40:00 EST
325    Tag: blume-20020315-basis
326    Description:
327    
328    Provided (preliminary?) implementations for
329    
330      {String,Substring}.{concatWith,isSuffix,isSubstring}
331    
332    and
333    
334      Substring.full
335    
336    Those are in the Basis spec but they were missing in SML/NJ.
337    
338    ----------------------------------------------------------------------
339    Name: Matthias Blume
340    Date: 2002/03/14 21:30:00 EST
341    Tag: blume-20020314-controls
342    Description:
343    
344    Controls:
345    ---------
346    
347    1. Factored out the recently-added Controls : CONTROLS stuff and put
348       it into its own library $/controls-lib.cm.  The source tree for
349       this is under src/smlnj-lib/Controls.
350    
351    2. Changed the names of types and functions in this interface, so they
352       make a bit more "sense":
353    
354          module -> registry
355          'a registry -> 'a group
356    
357    3. The interface now deals in ref cells only.  The getter/setter interface
358       is (mostly) gone.
359    
360    4. Added a function that lets one register an already-existing ref cell.
361    
362    5. Made the corresponding modifications to the rest of the code so that
363       everything compiles again.
364    
365    6. Changed the implementation of Controls.MLRISC back to something closer
366       to the original.  In particular, this module (and therefore MLRISC)
367       does not depend on Controls.  There now is some link-time code in
368       int-sys.sml that registers the MLRISC controls with the Controls
369       module.
370    
371    CM:
372    ---
373    
374      * One can now specify the lambda-split aggressiveness in init.cmi.
375    
376    ----------------------------------------------------------------------
377    Name: Allen Leung
378    Date: 2002/03/13 17:30:00 EST
379    Tag: leunga-20020313-x86-fp-unary
380    Description:
381    
382    Bug fix for:
383    
384    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
385    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
386    > - fun f(x,(y,z)) = Real.~ y;
387    > [autoloading]
388    > [autoloading done]
389    >       fchsl   (%eax), 184(%esp)
390    > Error: MLRisc bug: X86MCEmitter.emitInstr
391    >
392    > uncaught exception Error
393    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
394    
395    The problem was that the code generator did not generate any fp registers
396    in this case, and the ra didn't know that it needed to run the X86FP phase to
397    translate the pseudo fp instruction.   This only happened with unary fp
398    operators in certain situations.
399    
400    ----------------------------------------------------------------------
401    Name: Matthias Blume
402    Date: 2002/03/13 14:00:00 EST
403    Tag: blume-20020313-overload-etc
404    Description:
405    
406    1. Added _overload as a synonym for overload for backward compatibility.
407       (Control.overloadKW must be true for either version to be accepted.)
408    
409    2. Fixed bug in install script that caused more things to be installed
410       than what was requested in config/targets.
411    
412    3. Made CM aware of the (_)overload construct so that autoloading
413       works.
414    
415    ----------------------------------------------------------------------
416    Name: Matthias Blume
417    Date: 2002/03/12 22:03:00 EST
418    Tag: blume-20020312-url
419    Description:
420    
421    Forgot to update BOOT and srcarchiveurl.
422    
423    ----------------------------------------------------------------------
424    Name: Matthias Blume
425    Date: 2002/03/12 17:30:00 EST
426    Tag: blume-20020312-version110392
427    Description:
428    
429    Yet another version number bump (because of small changes to the
430    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
431    
432    Changes:
433    
434      The new pid generation scheme described a few weeks ago was overly
435      complicated.  I implemented a new mechanism that is simpler and
436      provides a bit more "stability":  Once CM has seen a compilation
437      unit, it keeps its identity constant (as long as you do not delete
438      those crucial CM/GUID/* files).  This means that when you change
439      an interface, compile, then go back to the old interface, and
440      compile again, you arrive at the original pid.
441    
442      There now also is a mechanism that instructs CM to use the plain
443      environment hash as a module's pid (effectively making its GUID
444      the empty string).  For this, "noguid" must be specified as an
445      option to the .sml file in question within its .cm file.
446      This is most useful for code that is being generated by tools such
447      as ml-nlffigen (because during development programmers tend to
448      erase the tool's entire output directory tree including CM's cached
449      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
450      revert to the old, broken behavior of SML/NJ, but in specific cases
451      where there is no danger of interface confusion, its use is ok
452      (I think).
453    
454      ml-nlffigen by default generates "noguid" annotations.  They can be
455      turned off by specifying -guid in its command line.
456    
457    ----------------------------------------------------------------------
458    Name: Lal George
459    Date: 2002/03/12 12 14:42:36 EST
460    Tag: george-20020312-frequency-computation
461    Description:
462    
463    Integrated jump chaining and static block frequency into the
464    compiler. More details and numbers later.
465    
466    ----------------------------------------------------------------------
467    Name: Lal George
468    Date: 2002/03/11 11 22:38:53 EST
469    Tag: george-20020311-jump-chain-elim
470    Description:
471    
472    Tested the jump chain elimination on all architectures (except the
473    hppa).  This is on by default right now and is profitable for the
474    alpha and x86, however, it may not be profitable for the sparc and ppc
475    when compiling the compiler.
476    
477    The gc test will typically jump to a label at the end of the cluster,
478    where there is another jump to an external cluster containing the actual
479    code to invoke gc. This is to allow factoring of common gc invocation
480    sequences. That is to say, we generate:
481    
482            f:
483               testgc
484               ja   L1      % jump if above to L1
485    
486            L1:
487               jmp L2
488    
489    
490    After jump chain elimination the 'ja L1' instructions is converted to
491    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
492    up being implemented in their long form (if L2 is far away) using:
493    
494            jbe     L3      % jump if below or equal to L3
495            jmp     L2
496         L3:
497            ...
498    
499    
500    For large compilation units L2  may be far away.
501    
502    
503    ----------------------------------------------------------------------
504    Name: Matthias Blume
505    Date: 2002/03/11 13:30:00 EST
506    Tag: blume-20020311-mltreeeval
507    Description:
508    
509    A functor parameter was missing.
510    
511    ----------------------------------------------------------------------
512    Name: Allen Leung
513    Date: 2002/03/11 10:30:00 EST
514    Tag: leunga-20020311-runtime-string0
515    Description:
516    
517       The representation of the empty string now points to a
518    legal null terminated C string instead of unit.  It is now possible
519    to convert an ML string into C string with InlineT.CharVector.getData.
520    This compiles into one single machine instruction.
521    
522    ----------------------------------------------------------------------
523    Name: Allen Leung
524    Date: 2002/03/10 23:55:00 EST
525    Tag: leunga-20020310-x86-call
526    Description:
527    
528       Added machine generation for CALL instruction (relative displacement mode)
529    
530    ----------------------------------------------------------------------
531    Name: Matthias Blume
532    Date: 2002/03/08 16:05:00
533    Tag: blume-20020308-entrypoints
534    Description:
535    
536    Version number bumped to 110.39.1.  NEW BOOTFILES!
537    
538    Entrypoints: non-zero offset into a code object where execution should begin.
539    
540    - Added the notion of an entrypoint to CodeObj.
541    - Added reading/writing of entrypoint info to Binfile.
542    - Made runtime system bootloader aware of entrypoints.
543    - Use the address of the label of the first function given to mlriscGen
544      as the entrypoint.  This address is currently always 0, but it will
545      not be 0 once we turn on block placement.
546    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
547      with entry points) from mlriscGen.
548    
549    ----------------------------------------------------------------------
550    Name: Allen Leung
551    Date: 2002/03/07 20:45:00 EST
552    Tag: leunga-20020307-x86-cmov
553    Description:
554    
555       Bug fixes for CMOVcc on x86.
556    
557       1. Added machine code generation for CMOVcc
558       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
559       3. CMOVcc cannot have an immediate operand as argument.
560    
561    ----------------------------------------------------------------------
562    Name: Matthias Blume
563    Date: 2002/03/07 16:15:00 EST
564    Tag: blume-20020307-controls
565    Description:
566    
567    This is a very large but mostly boring patch which makes (almost)
568    every tuneable compiler knob (i.e., pretty much everything under
569    Control.* plus a few other things) configurable via both the command
570    line and environment variables in the style CM did its configuration
571    until now.
572    
573    Try starting sml with '-h' (or, if you are brave, '-H')
574    
575    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
576    implements the underlying generic mechanism.
577    
578    The interface to some of the existing such facilities has changed somewhat.
579    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
580    (The getFoo interface is still there for backward-compatibility, but its
581    use is deprecated.)
582    
583    The ml-build script passes -Cxxx=yyy command-line arguments through so
584    that one can now twiddle the compiler settings when using this "batch"
585    compiler.
586    
587    TODO items:
588    
589    We should go through and throw out all controls that are no longer
590    connected to anything.  Moreover, we should go through and provide
591    meaningful (and correct!) documentation strings for those controls
592    that still are connected.
593    
594    Currently, multiple calls to Controls.new are accepted (only the first
595    has any effect).  Eventually we should make sure that every control
596    is being made (via Controls.new) exactly once.  Future access can then
597    be done using Controls.acc.
598    
599    Finally, it would probably be a good idea to use the getter-setter
600    interface to controls rather than ref cells.  For the time being, both
601    styles are provided by the Controls module, but getter-setter pairs are
602    better if thread-safety is of any concern because they can be wrapped.
603    
604    *****************************************
605    
606    One bug fix: The function blockPlacement in three of the MLRISC
607    backpatch files used to be hard-wired to one of two possibilities at
608    link time (according to the value of the placementFlag).  But (I
609    think) it should rather sense the flag every time.
610    
611    *****************************************
612    
613    Other assorted changes (by other people who did not supply a HISTORY entry):
614    
615    1. the cross-module inliner now works much better (Monnier)
616    2. representation of weights, frequencies, and probabilities in MLRISC
617       changed in preparation of using those for weighted block placement
618       (Reppy, George)
619    
620    ----------------------------------------------------------------------
621    Name: Lal George
622    Date: 2002/03/07 14:44:24 EST 2002
623    Tag: george-20020307-weighted-block-placement
624    
625    Tested the weighted block placement optimization on all architectures
626    (except the hppa) using AMPL to generate the block and edge frequencies.
627    Changes were required in the machine properties to correctly
628    categorize trap instructions. There is an MLRISC flag
629    "weighted-block-placement" that can be used to enable weighted block
630    placement, but this will be ineffective without block/edge
631    frequencies (coming soon).
632    
633    
634    ----------------------------------------------------------------------
635    Name: Lal George
636    Date: 2002/03/05 17:24:48 EST
637    Tag: george-20020305-linkage-cluster
638    
639    In order to support the block placement optimization, a new cluster
640    is generated as the very first cluster (called the linkage cluster).
641    It contains a single jump to the 'real' entry point for the compilation
642    unit. Block placement has no effect on the linkage cluster itself, but
643    all the other clusters  have full freedom in the manner in which they
644    reorder blocks or functions.
645    
646    On the x86 the typical linkage code that is generated is:
647       ----------------------
648            .align 2
649       L0:
650            addl    $L1-L0, 72(%esp)
651            jmp     L1
652    
653    
654            .align  2
655       L1:
656       ----------------------
657    
658    72(%esp) is the memory location for the stdlink register. This
659    must contain the address of the CPS function being called. In the
660    above example, it contains the address of  L0; before
661    calling L1 (the real entry point for the compilation unit), it
662    must contain the address for L1, and hence
663    
664            addl $L1-L0, 72(%esp)
665    
666    I have tested this on all architectures except the hppa.The increase
667    in code size is of course negligible
668    
669    ----------------------------------------------------------------------
670    Name: Allen Leung
671    Date: 2002/03/03 13:20:00 EST
672    Tag: leunga-20020303-mlrisc-tools
673    
674      Added #[ ... ] expressions to mlrisc tools
675    
676    ----------------------------------------------------------------------
677    Name: Matthias Blume
678    Date: 2002/02/27 12:29:00 EST
679    Tag: blume-20020227-cdebug
680    Description:
681    
682    - made types in structure C and C_Debug to be equal
683    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
684    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
685    
686    ----------------------------------------------------------------------
687    Name: Matthias Blume
688    Date: 2002/02/26 12:00:00 EST
689    Tag: blume-20020226-ffi
690    Description:
691    
692    1. Fixed a minor bug in CM's "noweb" tool:
693       If numbering is turned off, then truly don't number (i.e., do not
694       supply the -L option to noweb).  The previous behavior was to supply
695       -L'' -- which caused noweb to use the "default" line numbering scheme.
696       Thanks to Chris Richards for pointing this out (and supplying the fix).
697    
698    2. Once again, I reworked some aspects of the FFI:
699    
700       A. The incomplete/complete type business:
701    
702       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
703         gone!
704       - ML types representing an incomplete type are now *equal* to
705         ML types representing their corresponding complete types (just like
706         in C).  This is still safe because ml-nlffigen will not generate
707         RTTI for incomplete types, nor will it generate functions that
708         require access to such RTTI.   But when ML code generated from both
709         incomplete and complete versions of the C type meet, the ML types
710         are trivially interoperable.
711    
712         NOTE:  These changes restore the full generality of the translation
713         (which was previously lost when I eliminated functorization)!
714    
715       B. Enum types:
716    
717       - Structure C now has a type constructor "enum" that is similar to
718         how the "su" constructor works.  However, "enum" is not a phantom
719         type because each "T enum" has values (and is isomorphic to
720         MLRep.Signed.int).
721       - There are generic access operations for enum objects (using
722         MLRep.Signed.int).
723       - ml-nlffigen will generate a structure E_foo for each "enum foo".
724         * The structure contains the definition of type "mlrep" (the ML-side
725         representation type of the enum).  Normally, mlrep is the same
726         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
727         then mlrep will be defined as a datatype -- thus facilitating
728         pattern matching on mlrep values.
729         ("-ec" will be suppressed if there are duplicate values in an
730          enumeration.)
731         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
732         will be generated for each C enum constant xxx.
733         * Conversion functions m2i and i2m convert between mlrep and
734         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
735         * Coversion functions c and ml convert between mlrep and "tag enum".
736         * Access functions (get/set) fetch and store mlrep values.
737       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
738         enumerations are merged into one single enumeration represented by
739         structure E_'.
740    
741    ----------------------------------------------------------------------
742    Name: Allen Leung
743    Date: 2002/02/25 04:45:00 EST
744    Tag: leunga-20020225-cps-spill
745    
746    This is a new implementation of the CPS spill phase.
747    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
748    In case of problems, replace it with the old file spill.sml
749    
750    The current compiler runs into some serious performance problems when
751    constructing a large record.  This can happen when we try to compile a
752    structure with many items.  Even a very simple structure like the following
753    makes the compiler slow down.
754    
755        structure Foo = struct
756           val x_1 = 0w1 : Word32.int
757           val x_2 = 0w2 : Word32.int
758           val x_3 = 0w3 : Word32.int
759           ...
760           val x_N = 0wN : Word32.int
761        end
762    
763    The following table shows the compile time, from N=1000 to N=4000,
764    with the old compiler:
765    
766    N
767    1000   CPS 100 spill                           0.04u  0.00s  0.00g
768           MLRISC ra                               0.06u  0.00s  0.05g
769              (spills = 0 reloads = 0)
770           TOTAL                                   0.63u  0.07s  0.21g
771    
772    1100   CPS 100 spill                           8.25u  0.32s  0.64g
773           MLRISC ra                               5.68u  0.59s  3.93g
774              (spills = 0 reloads = 0)
775           TOTAL                                   14.71u  0.99s  4.81g
776    
777    1500   CPS 100 spill                           58.55u  2.34s  1.74g
778           MLRISC ra                               5.54u  0.65s  3.91g
779              (spills = 543 reloads = 1082)
780           TOTAL                                   65.40u  3.13s  6.00g
781    
782    2000   CPS 100 spill                           126.69u  4.84s  3.08g
783           MLRISC ra                               0.80u  0.10s  0.55g
784              (spills = 42 reloads = 84)
785           TOTAL                                   129.42u  5.10s  4.13g
786    
787    3000   CPS 100 spill                           675.59u  19.03s  11.64g
788           MLRISC ra                               2.69u  0.27s  1.38g
789              (spills = 62 reloads = 124)
790           TOTAL                                   682.48u  19.61s  13.99g
791    
792    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
793           MLRISC ra                               4.96u  0.27s  2.72g
794              (spills = 85 reloads = 170)
795           TOTAL                                   2375.26u  57.21s  48.00g
796    
797    As you can see the old cps spill module suffers from some serious
798    performance problem.  But since I cannot decipher the old code fully,
799    instead of patching the problems up, I'm reimplementing it
800    with a different algorithm.  The new code is more modular,
801    smaller when compiled, and substantially faster
802    (O(n log n) time and O(n) space).  Timing of the new spill module:
803    
804    4000  CPS 100 spill                           0.02u  0.00s  0.00g
805          MLRISC ra                               0.25u  0.02s  0.15g
806             (spills=1 reloads=3)
807          TOTAL                                   7.74u  0.34s  1.62g
808    
809    Implementation details:
810    
811    As far as I can tell, the purpose of the CPS spill module is to make sure the
812    number of live variables at any program point (the bandwidth)
813    does not exceed a certain limit, which is determined by the
814    size of the spill area.
815    
816    When the bandwidth is too large, we decrease the register pressure by
817    packing live variables into spill records.  How we achieve this is
818    completely different than what we did in the old code.
819    
820    First, there is something about the MLRiscGen code generator
821    that we should be aware of:
822    
823    o MLRiscGen performs code motion!
824    
825       In particular, it will move floating point computations and
826       address computations involving only the heap pointer to
827       their use sites (if there is only a single use).
828       What this means is that if we have a CPS record construction
829       statement
830    
831           RECORD(k,vl,w,e)
832    
833       we should never count the new record address w as live if w
834       has only one use (which is often the case).
835    
836       We should do something similar to floating point, but the transformation
837       there is much more complex, so I won't deal with that.
838    
839    Secondly, there are now two new cps primops at our disposal:
840    
841     1. rawrecord of record_kind option
842        This pure operator allocates some uninitialized storage from the heap.
843        There are two forms:
844    
845         rawrecord NONE [INT n]  allocates a tagless record of length n
846         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
847                                     and initializes the tag.
848    
849     2. rawupdate of cty
850          rawupdate cty (v,i,x)
851          Assigns to x to the ith component of record v.
852          The storelist is not updated.
853    
854    We use these new primops for both spilling and increment record construction.
855    
856     1. Spilling.
857    
858        This is implemented with a linear scan algorithm (but generalized
859        to trees).  The algorithm will create a single spill record at the
860        beginning of the cps function and use rawupdate to spill to it,
861        and SELECT or SELp to reload from it.  So both spills and reloads
862        are fine-grain operations.  In contrast, in the old algorithm
863        "spills" have to be bundled together in records.
864    
865        Ideally, we should sink the spill record construction to where
866        it is needed.  We can even split the spill record into multiple ones
867        at the places where they are needed.  But CPS is not a good
868        representation for global code motion, so I'll keep it simple and
869        am not attempting this.
870    
871     2. Incremental record construction (aka record splitting).
872    
873        Long records with many component values which are simulatenously live
874        (recall that single use record addresses are not considered to
875         be live) are constructed with rawrecord and rawupdate.
876        We allocate space on the heap with rawrecord first, then gradually
877        fill it in with rawupdate.  This is the technique suggested to me
878        by Matthias.
879    
880        Some restrictions on when this is applicable:
881        1. It is not a VECTOR record.  The code generator currently does not handle
882           this case. VECTOR record uses double indirection like arrays.
883        2. All the record component values are defined in the same "basic block"
884           as the record constructor.  This is to prevent speculative
885           record construction.
886    
887    ----------------------------------------------------------------------
888    Name: Allen Leung
889    Date: 2002/02/22 01:02:00 EST
890    Tag: leunga-20020222-mlrisc-tools
891    
892    Minor bug fixes in the parser and rewriter
893    
894    ----------------------------------------------------------------------
895    Name: Allen Leung
896    Date: 2002/02/21 20:20:00 EST
897    Tag: leunga-20020221-peephole
898    
899    Regenerated the peephole files.  Some contained typos in the specification
900    and some didn't compile because of pretty printing bugs in the old version
901    of 'nowhere'.
902    
903    ----------------------------------------------------------------------
904    Name: Allen Leung
905    Date: 2002/02/19 20:20:00 EST
906    Tag: leunga-20020219-mlrisc-tools
907    Description:
908    
909       Minor bug fixes to the mlrisc-tools library:
910    
911       1.  Fixed up parsing colon suffixed keywords
912       2.  Added the ability to shut the error messages up
913       3.  Reimplemented the pretty printer and fixed up/improved
914           the pretty printing of handle and -> types.
915       4.  Fixed up generation of literal symbols in the nowhere tool.
916       5.  Added some SML keywords to to sml.sty
917    
918    ----------------------------------------------------------------------
919    Name: Matthias Blume
920    Date: 2002/02/19 16:20:00 EST
921    Tag: blume-20020219-cmffi
922    Description:
923    
924    A wild mix of changes, some minor, some major:
925    
926    * All C FFI-related libraries are now anchored under $c:
927        $/c.cm      --> $c/c.cm
928        $/c-int.cm  --> $c/internals/c-int.cm
929        $/memory.cm --> $c/memory/memory.cm
930    
931    * "make" tool (in CM) now treats its argument pathname slightly
932      differently:
933        1. If the native expansion is an absolute name, then before invoking
934           the "make" command on it, CM will apply OS.Path.mkRelative
935           (with relativeTo = OS.FileSys.getDir()) to it.
936        2. The argument will be passed through to subsequent phases of CM
937           processing without "going native".  In particular, if the argument
938           was an anchored path, then "make" will not lose track of that anchor.
939    
940    * Compiler backends now "know" their respective C calling conventions
941      instead of having to be told about it by ml-nlffigen.  This relieves
942      ml-nlffigen from one of its burdens.
943    
944    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
945    
946    * Export C_DEBUG and C_Debug from $c/c.cm.
947    
948    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
949      subtyping relationship between incomplete pointers and their complete
950      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
951      with the convention of instantiating 'o with ('t, 'c) obj whenever
952      the pointer target type is complete.  In the incomplete case, 'o
953      will be instantiated with some "'c iobj" -- a type obtained by
954      using one of the functors PointerToIncompleteType or PointerToCompleteType.
955    
956      Operations that work on both incomplete and complete pointer types are
957      typed as taking an 'o ptr while operations that require the target to
958      be known are typed as taking some ('t, 'c) obj ptr.
959    
960      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
961      where void is an eqtype without any values.  This makes it possible
962      to work on voidptr values using functions meant to operate on light
963      incomplete pointers.
964    
965    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
966      been vastly simplified.
967    
968    ----------------------------------------------------------------------
969    Name: Matthias Blume
970    Date: 2002/02/19 10:48:00 EST
971    Tag: blume-20020219-pqfix
972    Description:
973    
974    Applied Chris Okasaki's bug fix for priority queues.
975    
976    ----------------------------------------------------------------------
977    Name: Matthias Blume
978    Date: 2002/02/15 17:05:00
979    Tag: Release_110_39
980    Description:
981    
982    Last-minute retagging is becoming a tradition... :-(
983    
984    This is the working release 110.39.
985    
986    ----------------------------------------------------------------------
987    Name: Matthias Blume
988    Date: 2002/02/15 16:00:00 EST
989    Tag: Release_110_39-orig
990    Description:
991    
992    Working release 110.39.  New bootfiles.
993    
994    (Update: There was a small bug in the installer so it wouldn't work
995    with all shells.  So I retagged. -Matthias)
996    
997    ----------------------------------------------------------------------
998    Name: Matthias Blume
999    Date: 2002/02/15 14:17:00 EST
1000    Tag: blume-20020215-showbindings
1001    Description:
1002    
1003    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
1004    the latter can be useful for exploring what bindings are available at
1005    the interactive prompt.  (The first function returns only the list
1006    of symbols that are really bound, the second prints those but also the
1007    ones that CM's autoloading mechanism knows about.)
1008    
1009    ----------------------------------------------------------------------
1010    Name: Matthias Blume
1011    Date: 2002/02/15 12:08:00 EST
1012    Tag: blume-20020215-iptrs
1013    Description:
1014    
1015    Two improvements to ml-nlffigen:
1016    
1017      1. Write files only if they do not exist or if their current contents
1018         do not coincide with what's being written.  (That is, avoid messing
1019         with the time stamps unless absolutely necessary.)
1020    
1021      2. Implement a "repository" mechanism for generated files related
1022         to "incomplete pointer types".   See the README file for details.
1023    
1024    ----------------------------------------------------------------------
1025    Name: Matthias Blume
1026    Date: 2002/02/14 11:50:00 EST
1027    Tag: blume-20020214-quote
1028    Description:
1029    
1030    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
1031    because of the new and improved tag generation scheme.  (Thanks to Allen
1032    Leung for pointing it out.)
1033    
1034    ----------------------------------------------------------------------
1035    Name: Lal George
1036    Date: 2002/02/14 09:55:27 EST 2002
1037    Tag: george-20020214-isabelle-bug
1038    Description:
1039    
1040    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
1041    of Isabelle on the x86.
1042    
1043    From Allen:
1044    -----------
1045     I've found the problem:
1046    
1047         in ra-core.sml, I use the counter "blocked" to keep track of the
1048         true number of elements in the freeze queue.  When the counter goes
1049         to zero, I skip examining the queue.  But I've messed up the
1050         bookkeeping in combine():
1051    
1052             else ();
1053             case !ucol of
1054               PSEUDO => (if !cntv > 0 then
1055                     (if !cntu > 0 then blocked := !blocked - 1 else ();
1056                                        ^^^^^^^^^^^^^^^^^^^^^^^
1057                      moveu := mergeMoveList(!movev, !moveu)
1058                     )
1059                  else ();
1060    
1061         combine() is called to coalesce two nodes u and v.
1062         I think I was thinking that if the move counts of u and v are both
1063         greater than zero then after they are coalesced then one node is
1064         removed from the freeze queue.  Apparently I was thinking that
1065         both u and v are of low degree, but that's clearly not necessarily true.
1066    
1067    
1068    02/12/2002:
1069        Here's the patch.  HOL now compiles.
1070    
1071        I don't know how this impact on performance (compile
1072        time or runtime).  This bug caused the RA (especially on the x86)
1073        to go thru the potential spill phase when there are still nodes on the
1074        freeze queue.
1075    
1076    
1077    
1078    
1079    ----------------------------------------------------------------------
1080    Name: Matthias Blume
1081    Date: 2002/02/13 22:40:00 EST
1082    Tag: blume-20020213-fptr-rtti
1083    Description:
1084    
1085    Fixed a bug in ml-nlffigen that was introduced with one of the previous
1086    updates.
1087    
1088    ----------------------------------------------------------------------
1089    Name: Matthias Blume
1090    Date: 2002/02/13 16:41:00 EST
1091    Tag: blume-20020213-cmlpq
1092    Description:
1093    
1094    Added new priority queue export symbols (which have just been added to
1095    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
1096    not compile and the installer would choke.)
1097    
1098    ----------------------------------------------------------------------
1099    Name: Matthias Blume
1100    Date: 2002/02/13 16:15:00 EST
1101    Tag: blume-20020213-various
1102    Description:
1103    
1104    1. More tweaks to ml-nlffigen:
1105    
1106       - better internal datastructures (resulting in slight speedup)
1107       - "-match" option requires exact match
1108       - "localized" gensym counters (untagged structs/unions nested within
1109         other structs/unions or within typedefs get a fresh counter; their
1110         tag will be prefixed by a concatenation of their parents' tags)
1111       - bug fixes (related to calculation of transitive closure of types
1112         to be included in the output)
1113    
1114    2. Minor Basis updates:
1115    
1116       - added implementations for List.collate and Option.app
1117    
1118    ----------------------------------------------------------------------
1119    Name: Matthias Blume
1120    Date: 2002/02/11 15:55:00 EST
1121    Tag: blume-20020211-gensym
1122    Description:
1123    
1124    Added a "-gensym" option to command line of ml-nlffigen.  This can be
1125    used to specify a "stem" -- a string that is inserted in all "gensym'd"
1126    names (ML structure names that correspond to unnamed C structs, unions,
1127    and enums), so that separate runs of ml-nlffigen do not clash.
1128    
1129    ----------------------------------------------------------------------
1130    Name: Matthias Blume
1131    Date: 2002/02/11 12:05:00 EST
1132    Tag: blume-20020211-gensml
1133    Description:
1134    
1135    A quick fix for a problem with GenSML (in the pgraph-util library):
1136    Make generation of toplevel "local" optional.  (Strictly speaking,
1137    signature definitions within "local" are not legal SML.)
1138    
1139    Other than that: updates to INSTALL and cm/TODO.
1140    
1141    ----------------------------------------------------------------------
1142    Name: Matthias Blume
1143    Date: 2002/02/08 15:00:00 EST
1144    Tag: blume-20020208-uniquepid
1145    Description:
1146    
1147    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
1148    
1149    1. The installer (config/install.sh) has gotten smarter:
1150    
1151         - Configuration options are a bit easier to specify now
1152           (in config/targets).
1153         - Bug in recognizing .tar.bz2 files fixed.
1154         - Installer automatically resolves dependencies between
1155           configuration options (e.g., if you ask for eXene, you will
1156           also get cml -- regardless whether you asked for it or not).
1157         - Installer can run in "quieter mode" by setting the environment
1158           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
1159           "completely silent", though.
1160         - Build HashCons library as part of smlnj-lib.
1161    
1162    2. A new scheme for assigning persistent identifiers to compilation
1163       units (and, by extension, to types etc.) has been put into place.
1164       This fixes a long-standing bug where types and even dynamic values
1165       can get internally confused, thereby compromising type safety
1166       (abstraction) and dynamic correctness.  See
1167    
1168         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
1169    
1170       for an example of how things could go wrong until now.
1171    
1172       The downside of the new scheme is that pids are not quite as
1173       persistent as they used to be: CM will generate a fresh pid
1174       for every compilation unit that it thinks it sees for the first
1175       time.  That means that if you compile starting from a clean, fresh
1176       source tree at two different times, you end up with different
1177       binaries.
1178    
1179       Cutoff recompilation, however, has not been compromised because
1180       CM keeps pid information in special caches between runs.
1181    
1182    ----------------------------------------------------------------------
1183    Name: Lal George
1184    Date: 2002/02/07 15:34:13 EST 2002
1185    Tag: <none>
1186    Description:
1187    
1188    Compilers that generate assembly code may produce  global labels
1189    whose value is resolved at link time. The various peephole optimization
1190    modules did not take this in account.
1191    
1192    TODO. The Labels.addrOf function should really return an option
1193    type so that clients are forced to deal with this issue, rather
1194    than an exception being raised.
1195    
1196    ----------------------------------------------------------------------
1197    Name: Lal George
1198    Date: 2002/02/06 13:55:02 EST
1199    Tag: george-20020206-ra-breakup
1200    Description:
1201    
1202    1. A bug fix from Allen.
1203    
1204        A typo causes extra fstp %st(0)'s to be generated at compensation
1205        edges, which might cause stack underflow traps at runtime.  This
1206        occurs in fft where there are extraneous fstps right before the 'into'
1207        trap instruction (in this case they are harmless since none of the
1208        integers overflow.)
1209    
1210    2. Pulled out various utility modules that were embedded in the modules
1211       of the register allocator. I need these modules for other purposes, but
1212       they are not complete enough to put into a library (just yet).
1213    ----------------------------------------------------------------------
1214    Name: Matthias Blume
1215    Date: 2002/01/31 16:05:00 EST
1216    Tag: blume-20020131-sparc-ccalls
1217    Description:
1218    
1219    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
1220       of extra stack space by mistake.  Fixed.
1221    
1222    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
1223    
1224    ----------------------------------------------------------------------
1225    Name: Allen Leung
1226    Date: 2002/01/30
1227    Tag: leunga-20020130-nowhere-bug-fix
1228    Description:
1229    
1230       MLRISC bug fixes:
1231       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
1232       2. MachineInt.fromString was negating its value.
1233    
1234    ----------------------------------------------------------------------
1235    Name: Matthias Blume
1236    Date: 2002/01/29
1237    Tag: blume-20020129-INSTALL
1238    Description:
1239    
1240    - Added somewhat detailed installation instructions (file INSTALL).
1241    - Fixed curl-detection bug in config/install.sh.
1242    - It is now possible to select the URL getter using the URLGETTER
1243      environment variable:
1244    
1245          not set / "unknown"      --> automatic detection (script tries wget,
1246                                       curl, and lynx)
1247          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
1248                                       how to properly invoke them)
1249          other                    --> use $URLGETTER directly, it must take
1250                                       precisely two command-line arguments
1251                                       (source URL and destination file name)
1252    
1253    ----------------------------------------------------------------------
1254    Name: Matthias Blume
1255    Date: 2002/01/28
1256    Tag: blume-20020128-sparc-ccalls
1257    Description:
1258    
1259    - Fixed problem with calculation of "used" registers in sparc-c-calls.
1260    - Make use of the allocParam argument in sparc-c-calls.
1261    
1262    ----------------------------------------------------------------------
1263    Name: Matthias Blume
1264    Date: 2002/01/28
1265    Tag: blume-20020128-allocParam
1266    Description:
1267    
1268    John Reppy:  Changes c-calls API to accept client-callback for
1269    allocating extra stack space.
1270    me: Corresponding changes to mlriscGen (using a dummy argument that
1271        does not change the current behavior).
1272    
1273    ----------------------------------------------------------------------
1274    Name: Matthias Blume
1275    Date: 2002/01/28 12:00:00
1276    Tag: Release_110_38
1277    Description:
1278    
1279    This time for real!!!
1280    
1281    ----------------------------------------------------------------------
1282    Name: Matthias Blume
1283    Date: 2002/01/28 10:56:00 EST
1284    Tag: blume-20020128-retraction
1285    Description:
1286    
1287    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
1288       with blume-Release_110_38-retracted.)
1289    
1290    1. Fixed a problem with incorrect rounding modes in real64.sml.
1291       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
1292    
1293    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
1294       and enums fixed.  The naming of corresponding ML identifiers should
1295       now be consistent again.
1296    
1297    ----------------------------------------------------------------------
1298    Name: Allen Leung
1299    Date: 2002/01/27
1300    Tag: leunga-20020127-nowhere
1301    Description:
1302    
1303       Added a target called nowhere in the configuration scripts.
1304       Enabling this will build the MLRISC 'nowhere' tool (for translating
1305       programs with where-clauses into legal SML code) during installation.
1306    
1307    ----------------------------------------------------------------------
1308    Name: Matthias Blume
1309    Date: 2002/01/25 21:27:00 EST
1310    Tag: blume-Release_110_38-retracted
1311    Description:
1312    
1313    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
1314    
1315    README will be added later.
1316    
1317    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
1318    (Release_110_38) removed.  Reason: Last-minute bug fixes.
1319    
1320    ----------------------------------------------------------------------
1321    Name: Matthias Blume
1322    Date: 2002/01/25
1323    Tag: blume-20020125-ffi
1324    Description:
1325    
1326    A large number of tweaks and improvements to ml-nlffi-lib and
1327    ml-nlffigen:
1328    
1329       - ML represenation types have been streamlined
1330       - getter and setter functions work with concrete values, not abstract
1331         ones where possible
1332       - ml-nlffigen command line more flexible (see README file there)
1333       - some bugs have been fixed (hopefully)
1334    
1335    ----------------------------------------------------------------------
1336    Name: Lal George
1337    Date: 2002/01/24
1338    Tag: george-20020124-risc-ra-interface
1339    Description:
1340    
1341       There is a dramatic simplification in the interface to the
1342       register allocator for RISC architectures as a result of making
1343       parallel copy instructions explicit.
1344    
1345    ----------------------------------------------------------------------
1346    Name: Matthias Blume
1347    Date: 2002/01/22
1348    Tag: blume-20020122-x86-ccalls
1349    Description:
1350    
1351    Bug fix for c-calls on x86 (having to do with how char- and
1352    short-arguments are being handled).
1353    
1354    ----------------------------------------------------------------------
1355    Name: Matthias Blume
1356    Date: 2002/01/21
1357    Tag: blume-20020121-ff
1358    Description:
1359    
1360    Another day of fiddling with the FFI...
1361    
1362    1. Bug fix/workaround:  CKIT does not complain about negative array
1363       dimensions, so ml-nlffigen has to guard itself against this possibility.
1364       (Otherwise a negative dimension would send it into an infinite loop.)
1365    
1366    2. Some of the abstract types (light objects, light pointers, most "base"
1367       types) in structure C are now eqtypes.
1368    
1369    3. Added constructors and test functions for NULL function pointers.
1370    
1371    ----------------------------------------------------------------------
1372    Name: Matthias Blume
1373    Date: 2002/01/18
1374    Tag: blume-20020118-ready-for-new-release
1375    Description:
1376    
1377    Made config/srcarchiveurl point to a new place.  (Will provide boot
1378    files shortly.)
1379    
1380    Maybe we christen this to be 110.38?
1381    
1382    ----------------------------------------------------------------------
1383    Name: Matthias Blume
1384    Date: 2002/01/18
1385    Tag: blume-20020118-more-ffifiddle
1386    Description:
1387    
1388    Today's FFI fiddling:
1389    
1390      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
1391        C.Set.* that use concrete (MLRep.*) arguments and results instead
1392        of abstract ones.
1393    
1394      - Provided word-style bit operations etc. for "int" representation
1395        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
1396        and Long.
1397    
1398    ----------------------------------------------------------------------
1399    Name: Matthias Blume
1400    Date: 2002/01/18
1401    Tag: blume-20020118-use-x86-fp
1402    Description:
1403    
1404    Now that x86-fast-fp seems to be working, I turned it back on again
1405    by default.  (Seems to work fine now, even with the FFI.)
1406    
1407    Other than that, I added some documentation about the FFI to
1408    src/ml-nlffigen/README and updated the FFI test examples in
1409    src/ml-nlffi-lib/Tests/*.
1410    
1411    ----------------------------------------------------------------------
1412    Name: Allen Leung
1413    Date: 2002/01/17
1414    Tag: leunga-20020117-x86-fast-fp-call
1415    Description:
1416    
1417       1. Fixed a problem with handling return fp values when x86's fast fp
1418          mode is turned on.
1419    
1420       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
1421          of %f32.
1422    
1423       3. Added a constructor INT32lit to the ast of MLRISC tools.
1424    
1425    ----------------------------------------------------------------------
1426    Name: Matthias Blume
1427    Date: 2002/01/16
1428    Tag: blume-20020116-ffifiddle
1429    Description:
1430    
1431    More fiddling with the FFI interface:
1432    
1433     - Make constness 'c instead of rw wherever possible.  This eliminates
1434       the need for certain explicit coercions.  (However, due to ML's
1435       value polymorphism, there will still be many cases where explicit
1436       coercions are necessary.  Phantom types are not the whole answer
1437       to modeling a subtyping relationship in ML.)
1438    
1439     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
1440    
1441     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
1442       clumsy workaround for varargs problem.)
1443    
1444    ----------------------------------------------------------------------
1445    Name: Lal George
1446    Date: 2002/01/15
1447    Tag: <none>
1448    Description:
1449    
1450    1. Since COPY instructions are no longer native to the architecture,
1451       a generic functor can be used to implement the expandCopies function.
1452    
1453    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
1454       TEXT segment.
1455    
1456    ----------------------------------------------------------------------
1457    Name: Matthias Blume
1458    Date: 2002/01/15
1459    Tag: blume-20020115-ffiupdates
1460    Description:
1461    
1462    1. Fix for bug resulting in single-precision float values being returned
1463       incorrectly from FFI calls.
1464    
1465    2. Small modifications to C FFI API:
1466    
1467        - memory-allocation routines return straight objects (no options)
1468          and raise an exception in out-of-memory situations
1469        - unsafe extensions to cast between function pointers and pointers
1470          from/to ints
1471        - added structure C_Debug as an alternative to structure C where
1472          pointer-dereferencing (|*| and |*!) always check for null-pointers
1473        - added open_lib' to DynLinkage;  open_lib' works like open_lib
1474          but also takes a (possibly empty) list of existing library handles
1475          that the current library depends on
1476    
1477    ----------------------------------------------------------------------
1478    Name: Matthias Blume
1479    Date: 2002/01/10
1480    Tag: blume-20020110-newffigen
1481    Description:
1482    
1483    1. Updates to portable graph code.
1484    
1485    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
1486       more scalable now so that even huge interfaces such as the one
1487       for GTK compile in finite time and space. :-)
1488       See src/ml-nlffigen/README for details on what's new.
1489    
1490    ----------------------------------------------------------------------
1491    Name: Lal George
1492    Date: 2001/01/09 14:31:35 EST 2002
1493    Tag: george-20011206-rm-native-copy
1494    Description:
1495    
1496            Removed the native COPY and FCOPY instructions
1497            from all the architectures and replaced it with the
1498            explicit COPY instruction from the previous commit.
1499    
1500            It is now possible to simplify many of the optimizations
1501            modules that manipulate copies. This has not been
1502            done in this change.
1503    
1504    ----------------------------------------------------------------------
1505    Name: Lal George
1506    Date: 2001/12/06 16:50:13 EST 2001
1507    Tag: george-20011206-mlrisc-instruction
1508    Description:
1509    
1510    Changed the representation of instructions from being fully abstract
1511    to being partially concrete. That is to say:
1512    
1513      from
1514            type instruction
1515    
1516      to
1517            type instr                              (* machine instruction *)
1518    
1519            datatype instruction =
1520                LIVE of {regs: C.cellset, spilled: C.cellset}
1521              | KILL of {regs: C.cellset, spilled: C.cellset}
1522              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
1523              | ANNOTATION of {i: instruction, a: Annotations.annotation}
1524              | INSTR of instr
1525    
1526    This makes the handling of certain special instructions that appear on
1527    all architectures easier and uniform.
1528    
1529    LIVE and KILL say that a list of registers are live or killed at the
1530    program point where they appear. No spill code is generated when an
1531    element of the 'regs' field is spilled, but the register is moved to
1532    the 'spilled' (which is present, more for debugging than anything else).
1533    
1534    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
1535    We used to generate:
1536    
1537            DEFFREG f1
1538            f1 := f2 + f3
1539            trapb
1540    
1541    but now generate:
1542    
1543            f1 := f2 + f3
1544            trapb
1545            LIVE {regs=[f1,f2,f3], spilled=[]}
1546    
1547    Furthermore, the DEFFREG (hack) required that all floating point instruction
1548    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
1549    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
1550    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
1551    this architecture).
1552    
1553    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
1554    all the architectures. This will result in further simplification of the
1555    register allocator that must be aware of them for coalescing purposes, and
1556    will also simplify certain aspects of the machine description that provides
1557    callbacks related to parallel copies.
1558    
1559    ANNOTATION should be obvious, and now INSTR represents the honest to God
1560    machine instruction set!
1561    
1562    The <arch>/instructions/<arch>Instr.sml files define certain utility
1563    functions for making porting easier -- essentially converting upper case
1564    to lower case. All machine instructions (of type instr) are in upper case,
1565    and the lower case form generates an MLRISC instruction. For example on
1566    the alpha we have:
1567    
1568      datatype instr =
1569         LDA of {r:cell, b:cell, d:operand}
1570       | ...
1571    
1572      val lda : {r:cell, b:cell, d:operand} -> instruction
1573        ...
1574    
1575    where lda is just (INSTR o LDA), etc.
1576    
1577    ----------------------------------------------------------------------
1578    Name: Matthias Blume
1579    Date: 2001/11/22 21:40:00 EST
1580    Tag: Release_110_37
1581    Description:
1582    
1583    Release 110.37.  This time for real.
1584    
1585    ----------------------------------------------------------------------
1586    Name: Matthias Blume
1587    Date: 2001/11/21 16:35:00 EST
1588    Tag: blume-20011121-foot-in-mouth
1589    Description:
1590    
1591    Removed the "Release_110_37" tag because of a serious bug.
1592    This will be re-tagged once the bug is fixed.
1593    
1594    ----------------------------------------------------------------------
1595    Name: Matthias Blume
1596    Date: 2001/11/21 16:14:00 EST
1597    Tag: blume-20011121-forgottenfile
1598    Description:
1599    
1600    Forgot to add a file.  (Just a .tex-file -- part of
1601    the CM manual source.)
1602    
1603    ----------------------------------------------------------------------
1604    Name: Matthias Blume
1605    Date: 2001/11/21 16:10:00 EST
1606    Tag: blume-20011121-invalid_110_37
1607    Description:
1608    
1609    Note: I removed the original tag "Release_110_37" from this commit
1610          because we found a serious bug in all non-x86 backends.
1611          - Matthias
1612    
1613    1. Modifications to the SML/NJ code generator and to the runtime system
1614       so that code object name strings are directly inserted into code
1615       objects at code generation time.  The only business the runtime system
1616       has with this is now to read the name strings on occasions.
1617       (The encoding of the name string has also changed somewhat.)
1618    
1619    2. CM now implements a simple "set calculus" for specifying export lists.
1620       In particular, it is now possible to refer to the export lists of
1621       other libraries/groups/sources and form unions as well as differences.
1622       See the latest CM manual for details.
1623    
1624    3. An separate notion of "proxy" libraries has again be eliminated from
1625       CM's model.  (Proxy libraries are now simply a special case of using
1626       the export list calculus.)
1627    
1628    4. Some of the existing libraries now take advantage of the new set
1629       calculus.
1630       (Notice that not all libraries have been converted because some
1631       of the existing .cm-files are supposed to be backward compatible
1632       with 110.0.x.)
1633    
1634    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
1635       "General.exnMessage"!  The former relies on a certain hook to be
1636       initialized, and that often does not happen in the stand-alone case.)
1637    
1638    ----------------------------------------------------------------------
1639    Name: Lal George
1640    Date: 2001/11/21  13:56:18 EST
1641    Tag: george-2001121-pseudo-ops
1642    Description:
1643    
1644      Implemented a complete redesign of MLRISC pseudo-ops. Now there
1645      ought to never be any question of incompatabilities with
1646      pseudo-op syntax expected by host assemblers.
1647    
1648      For now, only modules supporting GAS syntax are implemented
1649      but more should follow, such as MASM, and vendor assembler
1650      syntax, e.g. IBM as, Sun as, etc.
1651    
1652    ----------------------------------------------------------------------
1653    Name: Matthias Blume
1654    Date: 2001/11/14 11:52:00 EST
1655    Tag: blume-20011114-srcname
1656    Description:
1657    
1658    1. Routed the name of the current source file to mlriscgen where it
1659       should be directly emitted into the code object.  (This last part
1660       is yet to be done.)
1661    
1662    2. Some cleanup of the pgraph code to make it match the proposal that
1663       I put out the other day.  (The proposal notwithstanding, things are
1664       still in flux here.)
1665    
1666    ----------------------------------------------------------------------
1667    Name: Lal George
1668    Date: 2001/11/14 09:44:04 EST
1669    Tag:
1670    Description:
1671    
1672      Fix for a backpatching bug reported by Allen.
1673    
1674      Because the boundary between short and long span-dependent
1675      instructions is +/- 128, there are an astounding number of
1676      span-dependent instructions whose size is over estimated.
1677    
1678      Allen came up with the idea of letting the size of span
1679      dependent instructions be non-monotonic, for a maxIter
1680      number of times, after which the size must be monotonically
1681      increasing.
1682    
1683      This table shows the number of span-dependent instructions
1684      whose size was over-estimated as a function of maxIter, for the
1685      file Parse/parse/ml.grm.sml:
1686    
1687         maxIter            # of instructions:
1688            10                      687
1689            20                      438
1690            30                      198
1691            40                        0
1692    
1693      In compiling the compiler, there is no significant difference in
1694      compilation speed between maxIter=10 and maxIter=40. Actually,
1695      my measurements showed that maxIter=40 was a tad faster than
1696      maxIter=10! Also 96% of the  files in the compiler reach a fix
1697      point within 13 iterations, so fixing maxIter at 40, while high,
1698      is okay.
1699    
1700    ----------------------------------------------------------------------
1701    Name: Matthias Blume
1702    Date: 2001/10/31 15:25:00 EST
1703    Tag: blume-20011031-pgraph
1704    Description:
1705    
1706    CKIT:
1707    * Changed the "Function" constructor of type Ast.ctype to carry optional
1708      argument identifiers.
1709    * Changed the return type of TypeUtil.getFunction accordingly.
1710    * Type equality ignores the argument names.
1711    * TypeUtil.composite tries to preserve argument names but gives up quickly
1712      if there is a mismatch.
1713    
1714    installation script:
1715    * attempts to use "curl" if available (unless "wget" is available as well)
1716    
1717    CM:
1718    * has an experimental implementation of "portable graphs" which I will
1719      soon propose as an implementation-independent library format
1720    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
1721    
1722    NLFFI-LIB:
1723    * some cleanup (all cosmetic)
1724    
1725    NLFFIGEN:
1726    * temporarily disabled the mechanism that suppresses ML output for
1727      C definitions whose identifiers start with an underscore character
1728    * generate val bindings for enum constants
1729    * user can request that only one style (light or heavy) is being used;
1730      default is to use both (command-line arguments: -heavy and -light)
1731    * fixed bug in handling of function types involving incomplete pointers
1732    * generate ML entry points that take record arguments (i.e., using
1733      named arguments) for C functions that have a prototype with named
1734      arguments
1735      (see changes to CKIT)
1736    
1737    ----------------------------------------------------------------------
1738    Name: Allen Leung
1739    Date: 2001/10/27 20:34:00 EDT
1740    Tag: leunga-20011027-x86-fast-fp-call
1741    Description:
1742    
1743       Fixed the bug described in blume-20010920-slowfp.
1744    
1745       The fix involves
1746          1. generating FCOPYs in FSTP in ia32-svid
1747          2. marking a CALL with the appropriate annotation
1748    
1749    ----------------------------------------------------------------------
1750    Name: Matthias Blume
1751    Date: 2001/10/16 11:32:00 EDT
1752    Tag: blume-20011016-netbsd
1753    Description:
1754    
1755    Underscore patch from Chris Richards (fixing problem with compiling
1756    runtime system under recent NetBSD).
1757    
1758    ----------------------------------------------------------------------
1759    Name: Allen Leung
1760    Date: 2001/10/12 17:18:32 EDT 2001
1761    Tag: leung-20011012-x86-printflowgraph
1762    Description:
1763    
1764    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
1765    
1766    ----------------------------------------------------------------------
1767    Name: Lal George
1768    Date: 2001/10/11 23:51:34 EDT
1769    Tag: george-20011011-too-many-instrs
1770    Description:
1771    
1772    The representation of a program point never expected to see more
1773    than 65536 instructions in a basic block!
1774    
1775    ----------------------------------------------------------------------
1776    Name: Lal George
1777    Date: 2001/10/09 09:41:37 EDT
1778    Tag: george-20011008-mlrisc-labels
1779    Description:
1780    
1781    Changed the machine description files to support printing of
1782    local and global labels in assembly code, based on host assembler
1783    conventions.
1784    
1785    ----------------------------------------------------------------------
1786    Name: Matthias Blume
1787    Date: 2001/09/25 15:25:00 EDT
1788    Tag: blume-20010925-exninfo
1789    Description:
1790    
1791    I provided a non-hook implementation of exnName (at the toplevel) and
1792    made the "dummy" implementation of exnMessage (at the toplevel) more
1793    useful: if nothing gets "hooked in", then at least you are going to
1794    see the exception name and a message indicating why you don't see more.
1795    
1796    [For the time being, programs that need exnMessage and want to use
1797    ml-build should either use General.exnMessage (strongly recommended) or
1798    refer to structure General at some other point so that CM sees a
1799    static dependency.]
1800    
1801    [Similar remarks go for "print" and "use":  If you want to use their
1802    functionality in stand-alone programs generated by ml-build, then use
1803    TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
1804    
1805    ----------------------------------------------------------------------
1806    Name: Matthias Blume
1807    Date: 2001/09/20 17:28:00 EDT
1808    Tag: blume-20010920-slowfp
1809    Description:
1810    
1811    Allen says that x86-fast-fp is not safe yet, so I turned it off again...
1812    
1813    ----------------------------------------------------------------------
1814    Name: Matthias Blume
1815    Date: 2001/09/20 17:20:00 EDT
1816    Tag: blume-20010920-canonicalpaths
1817    Description:
1818    
1819    0. Updated the BOOT file (something that I forgot to do earlier).
1820    
1821    1. Small internal change to CM so that it avoids "/../" in filenames
1822       as much as possible (but only where it is safe).
1823    
1824    2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
1825       that arguments that contain delimiters are passed through correctly.
1826       This change also means that all "special" arguments of the form
1827       @SMLxxx... must come first.
1828    
1829    3. Changed install script to put relative anchor names for tool commands
1830       into pathconfig.
1831    
1832    ----------------------------------------------------------------------
1833    Name: Matthias Blume
1834    Date: 2001/09/18 15:35:00 EDT
1835    Tag: blume-20010918-readme11036
1836    Description:
1837    
1838    Added README files.
1839    
1840    ----------------------------------------------------------------------
1841    Name: Matthias Blume
1842    Date: 2001/09/18 11:45:00 EDT
1843    Tag: Release_110_36 (retag)
1844    Description:
1845    
1846    Fixed mistake in config/preloads. Retagged as 110.36.
1847    
1848    ----------------------------------------------------------------------
1849    Name: Matthias Blume
1850    Date: 2001/09/18 09:40:00 EDT
1851    Tag: Release_110_36_orig (tag changed)
1852    Description:
1853    
1854    New version (110.36).  New bootfiles.
1855    
1856    ----------------------------------------------------------------------
1857    Name: Matthias Blume
1858    Date: 2001/09/14 16:15:00 EDT
1859    Tag: blume-20010914-x86fastfp
1860    Description:
1861    
1862    John committed some changes that Allen made, in particular a (hopefully)
1863    correctly working version of the x86-fp module.
1864    
1865    I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
1866    flag to "true".  Everything seems to compile to a fixpoint ok, and
1867    "mandelbrot" speeds up by about 15%.
1868    
1869    ----------------------------------------------------------------------
1870    Name: Matthias Blume
1871    Date: 2001/09/13 11:20:00 EDT
1872    Tag: blume-20010913-minimal
1873    Description:
1874    
1875    1. Stefan Monnier's patch to fix a miscompilation problem that
1876       was brought to light by John Reppy's work on Moby.
1877    
1878    2. Implemented a minimal "structure Compiler" that contains just
1879       "version" and "architecture".  The minimal version will be
1880       available when the full version is not.  This is for backward-
1881       compatibility with code that wants to test Compiler.version.
1882    
1883    ----------------------------------------------------------------------
1884    Name: Matthias Blume
1885    Date: 2001/08/28 14:03:00 EDT
1886    Tag: blume-20010828-ml-lex
1887    Description:
1888    
1889    Fix for bug 1581, received from Neophytos Michael.
1890    
1891    ----------------------------------------------------------------------
1892    Name: Matthias Blume
1893    Date: 2001/08/27 11:20:00 EDT
1894    Tag: blume-20010827-readme11035
1895    Description:
1896    
1897    Fleshed out the README file for 110.35.
1898    
1899    ----------------------------------------------------------------------
1900    Name: Matthias Blume
1901    Date: 2001/08/24 17:10:00 EDT
1902    Tag: Release_110_35
1903    Description:
1904    
1905    New version number (110.35).  New bootfiles.
1906    
1907    ----------------------------------------------------------------------
1908    Name: Lal George
1909    Date: 2001/08/24 13:47:18 EDT 2001
1910    Tag: george-20010824-MLRISC-graphs
1911    Description:
1912    
1913     removed clusters from MLRISC completely and replaced with graphs.
1914    
1915    ----------------------------------------------------------------------
1916    Name: Matthias Blume
1917    Date: 2001/08/23 17:50:00 EDT
1918    Tag: blume-20010823-toplevel
1919    Description:
1920    
1921    - some reorganization of the code that implements various kinds of
1922      environments in the compiler (static, dynamic, symbolic, combined)
1923    - re-implemented the EnvRef module so that evalStream works properly
1924      (if the stream contains references to "use", "CM.make", etc.)
1925    - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
1926    
1927    ----------------------------------------------------------------------
1928    Name: Matthias Blume
1929    Date: 2001/08/20 15:50 EDT
1930    Tag: blume20010820-slipup
1931    Description:
1932    
1933    I forgot to commit a few files.  Here they are...
1934    
1935    ----------------------------------------------------------------------
1936    Name: Matthias Blume
1937    Date: 2001/08/20 15:35:00 EDT
1938    Tag: blume-20010820-debugprof
1939    Description:
1940    
1941    !!!! NEW BOOTFILES !!!!
1942    
1943    This is another round of reorganizing the compiler sources.  This
1944    time the main goal was to factor out all the "instrumentation"
1945    passes (for profiling and backtracing) into their own library.
1946    The difficulty was to do it in such a way that it does not depend
1947    on elaborate.cm but only on elabdata.cm.
1948    
1949    Therefore there have been further changes to both elaborate.cm and
1950    elabdata.cm -- more "generic" things have been moved from the former
1951    to the latter.  As a result, I was forced to split the assignment
1952    of numbers indicating "primtyc"s into two portions: SML-generic and
1953    SML/NJ-specific.  Since it would have been awkward to maintain,
1954    I bit the bullet and actually _changed_ the mapping between these
1955    numbers and primtycs.  The bottom line of this is that you need
1956    a new set of bin- and bootfiles.
1957    
1958    I have built new bootfiles for all architectures, so doing a fresh
1959    checkout and config/install.sh should be all you need.
1960    
1961    The newly created library's name is
1962    
1963        $smlnj/viscomp/debugprof.cm
1964    
1965    and its sources live under
1966    
1967        src/compiler/DebugProf
1968    
1969    ----------------------------------------------------------------------
1970    Name: Matthias Blume
1971  Date: 2001/08/15 17:15:00 EDT  Date: 2001/08/15 17:15:00 EDT
1972  Tag: blume-20010815-compreorg  Tag: blume-20010815-compreorg
1973  Description:  Description:
# Line 2853  Line 4807 
4807       elaborator). There were a lot of changes during my "linkpath" trials       elaborator). There were a lot of changes during my "linkpath" trials
4808       that could have been reverted to their original state but weren't.       that could have been reverted to their original state but weren't.
4809       Please, don't be too harsh on me for messing with this code a bit more       Please, don't be too harsh on me for messing with this code a bit more
4810       than what was strictly necessary...  (I _did_ resist the tempation       than what was strictly necessary...  (I _did_ resist the temptation
4811       of doing any "global reformatting" to avoid an untimely death at       of doing any "global reformatting" to avoid an untimely death at
4812       Dave's hands. :)       Dave's hands. :)
4813    

Legend:
Removed from v.902  
changed lines
  Added in v.1181

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