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 595, Wed Apr 5 03:30:17 2000 UTC revision 873, Mon Jul 9 15:26:20 2001 UTC
# Line 11  Line 11 
11  Date:  Date:
12  Tag: <post-commit CVS tag>  Tag: <post-commit CVS tag>
13  Description:  Description:
14    
15    ----------------------------------------------------------------------
16    Name: Matthias Blume
17    Date: 2001/07/09 11:25:00 EDT
18    Tag: blume-20010709-varargs
19    Description:
20    
21    1. ckit-lib.cm now exports structure Error
22    2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
23       with a warning accompanied by a source location.  Moreover, it
24       merely skips the offending function or type and proceeds with the
25       rest of its work.u  As a result, one can safely feed C code containing
26       "..." to ml-nlffigen.
27    3. There are some internal improvements to CM, providing slightly
28       more general string substitutions in the tools subsystem.
29    
30    ----------------------------------------------------------------------
31    Name: Matthias Blume
32    Date: 2001/06/27 15:10:00 EDT
33    Tag: blume-20010627-concur
34    Description:
35    
36    Fixed a small bug in CM's handling of parallel compilation.
37    (You could observe the bug by Control-C-interrupting an ordinary
38    CMB.make or CM.stabilize and then attaching some compile servers.
39    The result was that all of a sudden the previously interrupted
40    compilation would continue on its own.  This was because of
41    an over-optimization: CM did not bother to clean out certain queues
42    when no servers were attached "anyway", resulting in the contents
43    of these queues to grab control when new servers did get attached.)
44    
45    There is also another minor update to the CM manual.
46    
47    ----------------------------------------------------------------------
48    Name: Matthias Blume
49    Date: 2001/06/26 16:15:00 EDT
50    Tag: blume-20010626-cmdoc
51    Description:
52    
53    Minor typo fixed in CM manual (syntax diagram for libraries).
54    
55    ----------------------------------------------------------------------
56    Name: Matthias Blume
57    Date: 2001/06/25 22:55:00 EDT
58    Tag: blume-20010625-x86pc
59    Description:
60    
61    Fixed a nasty bug in the X86 assembly code that caused signal
62    handlers to fail (crash) randomly.
63    
64    ----------------------------------------------------------------------
65    Name: Matthias Blume
66    Date: 2001/06/25 12:05:00 EDT
67    Tag: blume-20010625-nlffigen
68    Description:
69    
70    This update fixes a number of minor bugs in ml-nlffigen as reported by
71    Nick Carter <nbc@andrew.cmu.edu>.
72    
73      1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
74      2. Default names for generated files are now derived from the name of
75         the C file *without its directory*.  In particular, this causes generated
76         files to be placed locally even if the C file is in some system directory.
77      3. Default names for generated signatures and structures are also derived
78         from the C file name without its directory.  This avoids silly things
79         like "structure GL/GL".
80         (Other silly names are still possible because ml-nlffigen does not do
81          a thorough check of whether generated names are legal ML identifiers.
82          When in doubt, use command line arguments to force particular names.)
83    
84    ----------------------------------------------------------------------
85    Name: Matthias Blume
86    Date: 2001/06/21 12:25:00 EDT
87    Tag: blume-20010621-eXene
88    Description:
89    
90    eXene now compiles and (sort of) works again.
91    
92    The library name (for version > 110.33) is $/eXene.cm.
93    
94    I also added an new example in src/eXene/examples/nbody.  See the
95    README file there for details.
96    
97    ----------------------------------------------------------------------
98    Name: Matthias Blume
99    Date: 2001/06/20 16:40:00 EDT
100    Tag: blume-20010620-cml
101    Description:
102    
103    CML now compiles and works again.
104    
105    Libraries (for version > 110.33):
106    
107      $cml/cml.cm            Main CML library.
108      $cml/basis.cm          CML's version of $/basis.cm.
109      $cml/cml-internal.cm   Internal helper library.
110      $cml/core-cml.cm       Internal helper library.
111      $cml-lib/trace-cml.cm  Tracing facility.
112      $cml-lib/smlnj-lib.cm  CML's version of $/smlnj-lib.cm
113    
114    The installer (config/install.sh) has been taught how to properly
115    install this stuff.
116    
117    ----------------------------------------------------------------------
118    Name: Matthias Blume
119    Date: 2001/06/19 17:55:00 EDT
120    Tag: blume-20010619-instantiate
121    Description:
122    
123    This un-breaks the fix for bug 1432.
124    (The bug was originally fixed in 110.9 but I broke it again some
125    time after that.)
126    
127    ----------------------------------------------------------------------
128    Name: Matthias Blume
129    Date: 2001/06/19 17:25:00 EDT
130    Tag: blume-20010619-signals
131    Description:
132    
133    This should (hopefully) fix the long-standing signal handling bug.
134    (The runtime system was constructing a continuation record with an
135    incorrect descriptor which would cause the GC to drop data on the floor...)
136    
137    ----------------------------------------------------------------------
138    Name: Matthias Blume
139    Date: 2001/06/15 15:05:00 EDT
140    Tag: blume-20010615-moresparc
141    Description:
142    
143    Here is a short late-hour update related to Sparc c-calls:
144    
145     -- made handling of double-word arguments a bit smarter
146    
147     -- instruction selection phase tries to collapse certain clumsily
148        constructed ML-Trees; typical example:
149    
150            ADD(ty,ADD(_,e,LI d1),LI d2)  ->  ADD(ty,e,LI(d1+d2))
151    
152        This currently has no further impact on SML/NJ since mlriscGen does
153        not seem to generate such patterns in the first place, and c-calls
154        (which did generate them in the beginning) has meanwhile been fixed
155        so as to avoid them as well.
156    
157    ----------------------------------------------------------------------
158    Name: Matthias Blume
159    Date: 2001/06/15 15:05:00 EDT
160    Tag: blume-20010615-sparc
161    Description:
162    
163    The purpose of this update is to provide an implementation of NLFFI
164    on Sparc machines.
165    
166    Here are the changes in detail:
167    
168       * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
169       the Sparc implementation of the c-calls API.
170       * The Sparc backend of SML/NJ has been modified to uniformely use %fp
171       for accessing the ML frame.  Thus, we have a real frame pointer and
172       can freely modify %sp without need for an omit-frame-ptr phase.
173       The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
174       this case.
175       * ml-nlffigen has been taught to produce code for different architectures
176       and calling conventions.
177       * In a way similar to what was done in the x86 case, the Sparc
178       backend uses its own specific extension to mltree.  (For example,
179       it needs to be able to generate UNIMP instructions which are part
180       of the calling convention.)
181       * ml-nlffi-lib was reorganized to make it more modular (in particular,
182       to make it easier to plug in new machine- and os-dependent parts).
183    
184    There are some other fairly unrelated bug fixes and cleanups as well:
185    
186       * I further hacked the .cm files for MLRISC tools (like MDLGen) so
187       that they properly share their libraries with existing SML/NJ libraries.
188       * I fixed a minor cosmetic bug in CM, supressing certain spurious
189       follow-up error messages.
190       * Updates to CM/CMB documentation.
191    
192    TODO items:
193    
194       * MLRISC should use a different register as its asmTemp on the Sparc.
195         (The current %o2 is a really bad choice because it is part of the
196         calling conventions, so things might interfere in unexpected ways.)
197    
198    ----------------------------------------------------------------------
199    Name: Matthias Blume
200    Date: 2001/06/07
201    Tag: blume-20010607-calls
202    Description:
203    
204    A number of internal changes related to C calls and calling conventions:
205    
206    1. ML-Tree CALL statements now carry a "pops" field.  It indicates the
207       number of bytes popped implicitly (by the callee).  In most cases
208       this field is 0 but on x86/win32 it is some non-zero value.  This
209       is information provided for the benefit of the "omit-frameptr" pass.
210    2. The CALL instruction on the x86 carries a similar "pops" field.
211       The instruction selection phase copies its value from the ML-Tree
212       CALL statement.
213    3. On all other architectures, the instruction selection phase checks
214       whether "pops=0" and complains if not.
215    4. The c-calls implementation for x86 now accepts two calling conventions:
216       "ccall" and "stdcall".  When "ccall" is selected, the caller cleans
217       up after the call and pops is set to 0.  For "stdcall", the caller
218       does nothing, leaving the cleanup to the callee; pops is set to
219       the number of bytes that were pushed onto the stack.
220    5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
221       distinguish between "ccall" and "stdcall".
222    6. The UNIMP instruction has been added to the supported Sparc instruction
223       set. (This is needed for implementing the official C calling convention
224       on this architecture.)
225    7. I fixed some of the .cm files under src/MLRISC/Tools to make them
226       work with the latest CM.
227    
228    ----------------------------------------------------------------------
229    Name: Matthias Blume
230    Date: 2001/06/05 15:10:00 EDT
231    Tag: blume-20010605-cm-index
232    Description:
233    
234    0. The "lambdasplit" parameter for class "sml" in CM has been documented.
235    
236    1. CM can now generate "index files".  These are human-readable files
237       that list on a per-.cm-file basis each toplevel symbol defined or
238       imported.  The location of the index file for
239       <p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
240       To enable index-file generation, set CM.Control.generate_index to true
241       or export an environment-symbol: export CM_GENERATE_INDEX=true.
242    
243       The CM manual has been updated accordingly.
244    
245    2. I made some slight modifications to the c-calls API in MLRISC.
246    
247         a) There is now a callback to support saving/restoring of
248            dedicated but caller-save registers around the actual call
249            instruction.
250         b) One can optionally specify a comment-annotation for the
251            call instruction.
252    
253    3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
254       (For example, the comment annotation shows the C prototype of
255        the function being called.)
256    
257    ----------------------------------------------------------------------
258    Name: Matthias Blume
259    Date: 2001/06/01 13:30:00 EDT
260    Tag: blume-20010601-nlffi-cleanup
261    Description:
262    
263    This is mostly a cleanup of MLFFI stuff:
264    
265       - some signature files have been put into a more exposed place
266       - the ugly 'f type parameter is gone (simplifies types tremendously!)
267       - ml-nlffigen changed accordingly
268       - tutorial updated
269    
270    Other changes:
271    
272       - author's affiliation in CM manual(s) updated
273       - some more recognized keywords added to Allen's sml.sty
274    
275    ----------------------------------------------------------------------
276    Name: Matthias Blume
277    Date: 2001/05/25 15:30:00 EDT
278    Tag: blume-20010525-iptr
279    Description:
280    
281      - put the official 110.33-README (as it appears on the ftp server) under
282        CVS
283      - fixed a small bug related to incomplete pointer types in
284        ml-nlffigen
285      - small cosmetic change to the ml-nlffi-lib's "arr" type constructor
286        (it does not need the 'f type parameter)
287    
288    ----------------------------------------------------------------------
289    Name: Matthias Blume
290    Date: 2001/05/23 14:30:00 EDT
291    Tag: Release_110_33
292    Description:
293    
294    New version number (110.33).  New bootfiles.
295    
296    ----------------------------------------------------------------------
297    Name: Matthias Blume
298    Date: 2001/05/22 18:06:00 EDT
299    Tag: blume-20010522-targets
300    Description:
301    
302    Made install.sh use file config/targets.customized if it exists, falling
303    back to config/targets if it doesn't.  This way one can have a customized
304    version of the targets file without touching the "real thing", thus
305    eliminating the constant fear of accidentally checking something bogus
306    back into the CVS repository...  (File config/targets.customized must
307    not be added to the repository!)
308    
309    ----------------------------------------------------------------------
310    Name: Matthias Blume
311    Date: 2001/05/22 16:30:00 EDT
312    Tag: blume-20010522-minitut
313    Description:
314    
315    1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
316       struct returns.
317    2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt.  This is some very
318       incomplete, preliminary documentation for NLFFI.
319    
320    ----------------------------------------------------------------------
321    Name: Matthias Blume
322    Date: 2001/05/14 11:30:00 EDT
323    Tag: blume-20010514-script
324    Description:
325    
326    Some bugs in install script fixed.
327    
328    In addition to that I also made a slight change to the NLFFI API:
329    Functors generated by ml-nlffigen now take the dynamic library as a
330    straight functor argument, not as a suspended one.  (The original
331    functor code used to force the suspension right away anyway, so there
332    was nothing gained by this complication of the interface.)
333    
334    ----------------------------------------------------------------------
335    Name: Matthias Blume
336    Date: 2001/05/11 14:35:00 EDT
337    Tag: blume-20010511-ml-nlffi
338    Description:
339    
340    I finally took the plunge and added my new FFI code to the main
341    repository.  For x86-linux it is now ready for prime-time.
342    
343    There are two new subdirectories of "src":
344    
345      - ml-nlffi-lib:
346           The utility library for programs using the FFI interface.
347           Here is the implementation of $/c.cm and its associated low-level
348           partners $/c-int.cm and $/memory.cm.
349      - ml-nlffigen:
350           A stand-alone program for generating ML glue code from C source
351           code.
352    
353    Building ml-nlffigen requires $/ckit-lib.cm.
354    
355    The config/install.sh script has been updates to do the Right Thing
356    (hopefully).
357    
358    Notice that the source tree for the C-Kit will not be put under "src"
359    but directly under the installation root directory.  (This is the
360    structure that currently exists on the CVS server when you check out
361    module "sml".)  Fortunately, config/install.sh knows about this oddity.
362    
363    Bugs: No documentation yet.
364    
365    ----------------------------------------------------------------------
366    Name: Matthias Blume
367    Date: 2001/05/09 16:35:00 EDT
368    Tag: blume-20010509-cpscontract
369    Description:
370    
371    Fixed a bug in the accounting code in cpsopt/contract.sml.  (The
372    wrapper/unwrapper elimination did not decrement usage counts and some
373    dead variables got overlooked by the dead-up logic.)
374    
375    ----------------------------------------------------------------------
376    Name: Lal George
377    Date: 2001/05/08  17:26:09 EDT
378    Tag: george-20010508-omit-frameptr
379    Description:
380    
381    Changes to implement the omit-frame-pointer optimization to support
382    raw C calls. For now, there is only support on the Intel x86, but
383    other architectures will follow as more experience is gained with this.
384    
385    
386    ----------------------------------------------------------------------
387    Name: Matthias Blume
388    Date: 2001/05/07 14:40:00 EDT
389    Tag: blume-20010507-proxies
390    Description:
391    
392    I made into "proxy libraries" all libraries that qualify for such a
393    change.  (A qualifying library is a library that has another library or
394    groups as its sole member and repeats that member's export list
395    verbatim.  A proxy library avoids this repetition by omitting its export
396    list, effectively inheriting the list that its (only) member exports.
397    See the CM manual for more explanation.)
398    The main effect is that explicit export lists for these libraries
399    do not have to be kepts in sync, making maintenance a bit easier.
400    
401    I also added copyright notices to many .cm-files.
402    
403    Last but not least, I made a new set of bootfiles.
404    
405    ----------------------------------------------------------------------
406    Name: Matthias Blume
407    Date: 2001/05/04 17:00:00 EDT
408    Tag: blume-20010504-cm-lsplit
409    Description:
410    
411    0. John merged pending changes to $/smlnj-lib.cm
412    
413    1. Allen's previous change accidentally backed out of one of Lal's
414       earlier changes.  I undid this mistake (re-introducing Lal's change).
415    
416    2. I used the new topOrder' function from graph-scc.sml (from $/smlnj-lib.cm)
417       within the compiler where applicable.  There is some code simplification
418       because of that.
419    
420    3. The "split" phase (in FLINT) is now part of the default list of phases.
421       Compiler.Control.LambdaSplitting.* can be used to globally control the
422       lambda-splitting (cross-module-inlining) engine.  In addition to that,
423       it can now also be controlled on a per-source basis: CM has been taught
424       a new tool parameter applicable to ML source files.
425    
426       - To turn lambda-splitting off completely:
427            local open Compiler.Control.LambdaSplitting in
428                val _ = set Off
429            end
430       - To make "no lambda-splitting" the global default (but allow per-source
431         overriding); this is the initial setting:
432            local open Compiler.Control.LambdaSplitting in
433                val _ = set (Default NONE)
434            end
435       - To make "lambda-splitting with aggressiveness a" the global default
436         (and allow per-source overriding):
437            local open Compiler.Control.LambdaSplitting in
438                val _ = set (Default (SOME a))
439            end
440    
441       - To turn lambda-splitting off for a given ML souce file (say: a.sml)
442         write (in the respective .cm-file):
443            a.sml (lambdasplitting:off)
444       - To turn lambda-splitting for a.sml on with minimal aggressiveness:
445            a.sml (lambdasplitting:on)
446       - To turn lambda-splitting for a.sml on with aggressiveness <a> (where
447         <a> is a decimal non-negative integer):
448            a.sml (lambdasplitting:<a>)
449       - To turn lambda-splitting for a.sml on with maximal aggressiveness:
450            a.sml (lambdasplitting:infinity)
451       - To use the global default for a.sml:
452            a.sml (lambdasplitting:default)
453         or simply
454            a.sml
455    
456    ----------------------------------------------------------------------
457    Name: Allen Leung
458    Date: 2001/05/04 01:57:00 EDT
459    Tag: leunga-20010504-sync
460    Description:
461    
462      MLRISC features.
463    
464      1. Fix to CMPXCHG instructions.
465      2. Changed RA interface to allow annotations in callbacks.
466      3. Added a new method to the stream interface to allow annotations updates.
467    
468    ----------------------------------------------------------------------
469    Name: Matthias Blume
470    Date: 2001/05/01 11:45:00 EDT
471    Tag: blume-20010501-pcedittmp
472    Description:
473    
474    Changed install.sh to use the current working directory instead of
475    /usr/tmp for a temporary file (pcedittmp).  The previous choice
476    of /usr/tmp caused trouble with MacOS X because of file premission
477    problems.
478    
479    ----------------------------------------------------------------------
480    Name: Matthias Blume
481    Date: 2001/04/20 11:10:00 EDT
482    Tag: blume-20010420-inMLflag
483    Description:
484    
485     - added vp_limitPtrMask to vproc-state.h
486       (for use by the raw-C-calls mechanism to implement proper interrupt
487        handling)
488     - made the ML compiler aware of various data-structure offsets so it
489       can generate code for accessing the vp_inML flag and vp_limitPtrMask
490     - tweaked mlriscGen.sml to have it emit interrupt-handling code for
491       raw C-calls
492    
493    ----------------------------------------------------------------------
494    Name: Lal George
495    Date: 2001/04/20 09:15:28 EDT
496    Tag: george-20010420-macosX
497    Description:
498    
499     - Changes to port to Mac OS X; Darwin.
500    
501     - In the process I found that sqrt was broken on the PPC, because the
502       fsqrt instruction is not implemented.
503    
504    ----------------------------------------------------------------------
505    Name: Matthias Blume
506    Date: 2001/04/18 12:45:00 EDT
507    Tag: blume-20010418-ccalls
508    Description:
509    
510     - fixed two off-by-4 errors in the x86-specific c-calls implementation
511       (this bug prevented structure arguments containing pointers from being
512        passed correctly)
513     - changed the raw-C-call code in mlriscGen.sml in such a way that
514       structure arguments are represented as a pointer to the beginning
515       of the structure (instead of having a series of synthesized arguments,
516       one for each structure member)
517    
518     - made makeml script's verbosity level configurable via environment
519       variable (MAKEML_VERBOSITY)
520    
521     - eliminated placeholder implementations for f32l, w16s, i16s, and f32s
522       in rawmem-x86.sml; we are now using the real thing
523    
524    ----------------------------------------------------------------------
525    Name: Matthias Blume
526    Date: 2001/03/22 16:25:00 EST
527    Tag: blume-20010322-bootfiles
528    Description:
529    
530    Created a new set of bootfiles (for your automatic installation convenience).
531    
532    ----------------------------------------------------------------------
533    Name: Matthias Blume
534    Date: 2001/03/22 15:10:00 EST
535    Tag: blume-20010322-rawmem-parcm
536    Description:
537    
538    1. All "raw memory access" primitives for the new FFI are implemented now
539       (at least on the x86).
540    2. Some further cleanup of CM's parallel make mechanism.
541    
542    ----------------------------------------------------------------------
543    Name: Matthias Blume
544    Date: 2001/03/19 17:53:00 EST
545    Tag: blume-20010319-parallel
546    Description:
547    
548    Parallel make (using compile servers) now works again.
549    
550    To this end, CM.stabilize and CMB.make have been modified to work in
551    two passes when compile servers are attached:
552       1. Compile everything, do not perform stabilization; this pass
553          uses compile servers
554       2. Stabilize everything; this pass does not use compile servers
555    If there are no compile servers, the two passes are combined into one
556    (as before).  Splitting the passes increases the inherent parallelism
557    in the dependency graph because the entire graph including all
558    libraries is available at the same time.  This, in turn, improves
559    server utilization.  The downside is that the master process will
560    have to do some extra work after compilation is done (because for
561    technical reasons it must re-read all the binfiles during stabilization).
562    
563    ----------------------------------------------------------------------
564    Name: Matthias Blume
565    Date: 2001/03/16 12:22:00 EST
566    Tag: blume-20010316-bootfiles
567    Description:
568    
569    Created a new set of bootfiles (for your automatic installation convenience).
570    
571    ----------------------------------------------------------------------
572    Name: Matthias Blume
573    Date: 2001/03/16 11:00:00 EST
574    Tag: blume-20010316-MLTREE-fixup
575    Description:
576    
577    This is a minor fixup for an (untagged) earlier commit by Allen.
578    (A file was missing).
579    
580    ----------------------------------------------------------------------
581    Name: Allen Leung
582    Date: Mon Mar  5 18:54:57 EST 2001
583    Tag: leunga-20010305-cut-support
584    
585    1. New support for alternative control-flow in MLTREE.
586       Currently we support
587    
588          FLOW_TO(CALL ...., [k1,...,kn])
589    
590       This is needed for 'cuts to' in C-- and try/handle-like constructs
591       in Moby
592    
593       New assembler flag "asm-show-cutsto" to turn on control-flow debugging.
594    
595    2. Register Allocator
596    
597       Changes in interface [from Fermin, John]
598    
599    3. Alpha 8-bit SLL support [Fermin]
600    
601    4. All architectures
602    
603       A new module (ClusterExpandCopies) for expanding parallel copies.
604    
605    ----------------------------------------------------------------------
606    Name: Allen Leung
607    Date: 2001/02/27 23:07:00 EST
608    Tag: leunga-20010227-minor-stuff
609    
610    1. Alpha bug fix for CMOVNE
611    2. Handle mltree COND(..,FCMP ...,...)
612    3. Bug fix in simplifier
613    
614    ----------------------------------------------------------------------
615    Name: Matthias Blume
616    Date: 2001/01/30 17:50:00 EST
617    Tag: blume-20010130-sync
618    Description:
619    
620    This is just a minor update to sync my devel branch with the main brach.
621    The only visible change is the addition of some README files.
622    
623    ----------------------------------------------------------------------
624    Name: Matthias Blume
625    Date: 2001/01/12 23:30:00 JST
626    Tag: blume-20010112-bootfiles
627    Description:
628    
629    Made a new set of bootfiles that goes with the current state of the
630    repository.
631    
632    ----------------------------------------------------------------------
633    Name: Matthias Blume
634    Date: 2001/01/12 21:20:00 JST
635    Tag: blume-20010112-sync
636    Description:
637    
638    I am just flushing out some minor changes that had accumulated in
639    my private branch in order to sync with the main tree.  (This is
640    mainly because I had CVS trouble when trying to merge _into_ my
641    private branch.)
642    
643    Most people should be completely unaffected by this.
644    
645    ----------------------------------------------------------------------
646    Name: Allen Leung
647    Date: Thu Jan 11 21:03:00 EST 2001
648    Tag: leunga-20010111-labexp=mltree
649    Description:
650    
651    1.  Removed the type LabelExp and replace it by MLTree.
652    2.  Rewritten mltree-simplify with the pattern matcher tool.
653    3.  There were some bugs in alpha code generator which would break
654        64-bit code generation.
655    4.  Redo the tools to generate code with the
656    5.  The CM files in MLRISC (and in src/system/smlnj/MLRISC)
657        are now generated by perl scripts.
658    
659    ----------------------------------------------------------------------
660    Name: Matthias Blume
661    Date: 2001/01/10 21:55:00 JST
662    Tag: blume-20010110-rcc
663    Description:
664    
665    The RCC stuff now seems to work (but only on the x86).
666    This required hacking of the c-calls interface (and -implementation) in
667    MLRISC.
668    
669    Normal compiler users should be unaffected.
670    
671    ----------------------------------------------------------------------
672    Name: Matthias Blume
673    Date: 2001/01/09 01:20:00 JST
674    Tag: blume-20010109-rcc
675    Description:
676    
677    This is a fairly big patch, flushing out a large number of pending
678    changes that I made to my development copy over the last couple of days.
679    
680    Of practical relevance at this moment is a workaround for a pickling
681    bug that Allen ran into the other day.  The cause of the bug itself is
682    still unknown and it might be hard to fix it properly, but the
683    workaround has some merits of its own (namely somewhat reducing pickling
684    overhead for certain libraries).  Therefore, I think this solution should
685    be satisfactory at this time.
686    
687    The rest of the changes (i.e., the vast majority) has to do with my
688    ongoing efforts of providing direct support for C function calls from
689    ML.  At the moment there is a new primop "RAW_CCALL", typing magic
690    in types/cproto.sml (invoked from FLINT/trans/translate.sml), a new
691    case in the FLINT CPS datatype (RCC), changes to cps/convert.sml to
692    translate uses of RAW_CCALL into RCC, and changes to mlriscGen.sml to
693    handle RCC.
694    
695    The last part (the changes to mlriscGen.sml) are still known to be
696    wrong on the x86 and not implemented on all other architectures.  But
697    the infrastructure is in place. I had to change a few functor
698    signatures in the backend to be able to route the CCalls interface
699    from MLRISC there, and I had to specialize the mltree type (on the
700    x86) to include the necessary extensions. (The extensions themselves
701    were already there and redy to go in MLRISC/x86).
702    
703    Everything should be very happy as soon as someone helps me with
704    mlriscGen.sml...
705    
706    In any case, nothing of this should matter to anyone as long as the
707    new primop is not being used (which is going to be the case unless you
708    find it where I hid it :). The rest of the compiler is completely
709    unaffected.
710    
711    ----------------------------------------------------------------------
712    Name: Matthias Blume
713    Date: 2001/01/05 00:30:00 JST
714    Tag: blume-20010105-primops
715    Description:
716    
717    Added some experimental support for work that I am doing right now.
718    These changes mostly concern added primops, but there is also a new
719    experimental C library in the runtime system (but currently not enabled
720    anywhere except on Linux/X86).
721    
722    In the course of adding primops (and playing with them), I discovered that
723    Zhong's INL_PRIM hack (no type info for certain primops) was, in fact, badly
724    broken.  (Zhong was very right he labeled this stuff as "major gross hack".)
725    To recover, I made type information in INL_PRIM mandatory and changed
726    prim.sml as well as built-in.sml accordingly.  The InLine structure now
727    has complete, correct type information (i.e., no bottom types).
728    
729    Since all these changes mean that we need new binfiles, I also bumped the
730    version number to 110.32.1.
731    
732    ----------------------------------------------------------------------
733    Name: Matthias Blume
734    Date: 2000/12/30 22:10:00 JST
735    Tag: blume-20001230-various
736    Description:
737    
738    Added proxy libraries for MLRISC and let MLRISC libraries refer
739    to each other using path anchors.  (See CM manual for explanation.)
740    
741    Updated CM documentation.
742    
743    Fixed some bugs in CM.
744    
745    Implemented "proxy" libraries (= syntactic sugar for CM).
746    
747    Added "-quiet" option to makeml and changed runtime system accordingly.
748    
749    Added cleanup handler for exportML to reset timers and compiler stats.
750    
751    ----------------------------------------------------------------------
752    Name: Lal George
753    Date: 2000/12/22 22:22:58 EST 2000
754    Tag: Release_110_32
755    Description:
756    
757            Infinite precision used throughout MLRISC.
758            see MLRISC/mltree/machine-int.sig
759    
760    ----------------------------------------------------------------------
761    Name: Matthias Blume
762    Date: 2000/12/22 23:16:00 JST
763    Tag: blume-20001222-warn
764    Description:
765    
766    Corrected wording and formatting of some CM warning message which I
767    broke in my previous patch.
768    
769    ----------------------------------------------------------------------
770    Name: Matthias Blume
771    Date: 2000/12/22 21:20:00 JST
772    Tag: blume-20001222-anchorenv
773    Description:
774    
775    Fixed CM's handling of anchor environments in connection with CMB.make.
776    
777    ----------------------------------------------------------------------
778    Name: Matthias Blume
779    Date: 2000/12/22 13:15:00 JST
780    Tag: blume-20001222-cleanup
781    Description:
782    
783    Removed src/cm/ffi which does not (and did not) belong here.
784    
785    ----------------------------------------------------------------------
786    Name: Matthias Blume
787    Date: 2000/12/21 23:55:00 JST
788    Tag: blume-20001221-exn
789    Description:
790    
791    Probably most important: CM no longer silently swallows all exceptions
792    in the compiler.
793    Plus: some other minor CM changes.  For example, CM now reports some
794    sizes for generated binfiles (code, data, envpickle, lambdapickle).
795    
796    ----------------------------------------------------------------------
797    Name: Matthias Blume
798    Date: 2000/12/15 00:01:05 JST
799    Tag: blume-20001215-dirtool
800    Description:
801    
802    - "dir" tool added.
803    - improvements and cleanup to Tools structure
804    - documentation updates
805    
806    ----------------------------------------------------------------------
807    Name: Allen Leung
808    Date: Thu Dec 14 03:45:24 EST 2000
809    Description:
810    Tag:  leunga-20001214-int-inf
811    Description:
812    
813       In IntInf, added these standard functions, which are missing from our
814    implementation:
815    
816        andb : int * int -> int
817        xorb : int * int -> int
818        orb  : int * int -> int
819        notb : int -> int
820         <<   : int * word -> int
821        ~>>  : int * word -> int
822    
823       Not tested, I hope they are correct.
824    
825    ----------------------------------------------------------------------
826    Name: Allen Leung
827    Date: Fri Dec  8 19:23:26 EST 2000
828    Description:
829    Tag:  leunga-20001208-nowhere
830    Description:
831    
832      Slight improvements to the 'nowhere' tool to handle OR-patterns,
833    to generate better error messages etc.  Plus a brief manual.
834    
835    ----------------------------------------------------------------------
836    Name: Lal George
837    Date: 2000/12/08 09:54:02 EST 2000
838    Tag: Release_110_31
839    Description:
840    
841    - Version 110.31
842    ----------------------------------------------------------------------
843    Name: Allen Leung
844    Date: Thu Dec  7 22:01:04 EST 2000
845    Tag:  leunga-20001207-cell-monster-hack
846    Description:
847    
848    Major MLRISC internal changes.  Affect all clients.
849    Summary:
850    
851    1.  Type CELLS.cell = int is now replaced by a datatype.
852        As a result, the old regmap is now gone.  Almost all interfaces
853        in MLRISC change as a consequence.
854    
855    2.  A new brand version of machine description tool (v3.0) that generates
856        modules expecting the new interface.  The old version is removed.
857    
858    3.  The RA interface has been further abstracted into two new functors.
859        RISC_RA and X86RA.  These functors have much simpler interfaces.
860        [See also directory MLRISC/demo.]
861    
862    4.  Some other new source->source code generation tools are available:
863    
864        a. MLRISC/Tools/RewriteGen -- generate rewriters from rules.
865        b. MLRISC/Tools/WhereGen -- expands conditional pattern matching rules.
866           I use this tool to generate the peephole optimizers---with the new
867           cell type changes, peephole rules are becoming difficult to write
868           without conditional pattern matching.
869    
870    5.  More Intmap -> IntHashTable change.  Previous changes by Matthias didn't
871        cover the entire MLRISC source tree so many things broke.
872    
873    6.  CM files have been moved to the subdirectory MLRISC/cm.
874        They are moved because there are a lot of them and they clutter up the
875        root dir.
876    
877    7.  More detailed documentation to come...
878    
879        NOTE: To rebuild from 110.30 (ftp distribution), you'll have to do
880        a makeml -rebuild first.  This is because of other other
881        changes that Matthias has made (see below).
882    
883    
884    ----------------------------------------------------------------------
885    Name: Matthias Blume
886    Date: 2000/11/30 23:12:00 JST
887    Tag: blume-20001130-filereorg
888    Description:
889    
890    Some manual updates and some file reorganizations in CM.
891    
892    ----------------------------------------------------------------------
893    Name: Matthias Blume
894    Date: 2000/11/24 17:45:00 JST
895    Tag: blume-20001124-link
896    Description:
897    
898    Drastically improved link traversal code for the case that the dynamic
899    value was already loaded at bootstrap time.  As a result, CM and CMB
900    now both load blazingly fast -- even on a very slow machine.  Also,
901    memory consumption has been further reduced by this.
902    
903    Warning: The format of the PIDMAP file has changed.  THerefore, to
904    bootstrap you have to do this:
905    
906    1. Run CMB.make
907    2. Make a symbolic link for the boot directory:
908         ln -s sml.boot.ARCH-OS xxx
909    3. "Rebuild" the boot directory:
910         ./makeml -boot xxx -rebuild sml ; rm xxx
911    4. Boot normally:
912          ./makeml
913    
914    ----------------------------------------------------------------------
915    Name: Matthias Blume
916    Date: 2000/11/21 21:20:00 JST
917    Tag: blume-20001121-tools
918    Description:
919    
920    Continued hacking on autoloading problem -- with success this time.
921    Also changed tool-plugin mechanism.  See new CM manual.
922    
923    ----------------------------------------------------------------------
924    Name: Matthias Blume
925    Date: 2000/11/19 14:30:00 JST
926    Tag:  blume-20001119-autoload
927    Description:
928    
929    Some hacking to make autoloading faster.  Success for CMB, no success
930    so far for CM.  There is a reduced structure CM' that autoloads faster.
931    (This is a temporary, non-documented hack to be eliminated again when
932    the general problem is solved.)
933    
934    ----------------------------------------------------------------------
935    Name: Matthias Blume
936    Date: 2000/11/17 14:10:00 JST
937    Tag: blume-20001117-pickle-lib
938    Description:
939    
940    1. Eliminated comp-lib.cm
941    2. Made pickle-lib.cm
942    3. Eliminated all uses of intset.sml (from comp-lib.cm)
943    4. Replaced all uses of intmap.{sig,sml} (from comp-lib.cm) with
944       equivalent constructs from smlnj-lib.cm (INtHashTable).
945    5. Point 4. also goes for those uses of intmap.* in MLRISC.
946       Duplicated intmap modules thrown out.
947    6. Hunted down all duplicated SCC code and replaced it with
948       equivalent stuff (GraphSCCFn from smlnj-lib.cm).
949    7. Rewrote Feedback module.
950    8. Moved sortedlist.sml into viscomp-lib.cm.  Eventually it
951       should be thrown out and equivalent modules from smlnj-lib.cm
952       should be used (IntRedBlackSet, IntListSet, ...).
953    
954    Confirmed that compiler compiles to fixpoint.
955    
956    ----------------------------------------------------------------------
957    Name: Allen Leung
958    Date: 2000/11/10 18:00:00
959    Tag: leunga-20001110-new-x86-fp
960    
961    A new x86 floating point code generator has been added.
962    By default this is turned off.  To turn this on, do:
963    
964        CM.autoload "$smlnj/compiler.cm";
965        Compiler.Control.MLRISC.getFlag "x86-fast-fp" := true;
966    
967    Changes:
968    
969    1.  Changed FTAN to FPTAN so that the assembly output is correct.
970    2.  Changed the extension callback for FTANGENT to generate:
971    
972              fptan
973              fstp  %st(0)
974        instead of
975              fptan
976              fstpl ftempmem
977    
978    3.  Numerous assembly fixes for x86.
979    
980    5.  Cleaned up the machine code output module x86/x86MC.sml and added
981        support for a whole bunch of instructions and addressing modes:
982    
983          fadd/fsub/fsubr/fmul/fdiv/fdivr  %st, %st(n)
984          faddp/fsubp/fsubrp/fmulp/fdivp/fdivrp  %st, %st(n)
985          fadd/fsub/fsubr/fmul/fdiv/fdivr  %st(n), %st
986          fiadd/fisub/fisubr/fimul/fidiv/fidivr mem
987          fxch %st(n)
988          fld %st(n)
989          fst %st(n)
990          fst mem
991          fstp %st(n)
992          fucom %st(n)
993          fucomp %st(n)
994    
995        All these are now generated when the fast fp mode is turned on.
996    
997    6.  Removed the dedicated registers %st(0), ..., %st(7) from X86CpsRegs
998    
999    ----------------------------------------------------------------------
1000    Name: Matthias Blume
1001    Date: 2000/11/09 11:20:00 JST
1002    Tag: blume-20001109-scc
1003    Description:
1004    
1005    Eliminated some code duplication:
1006    
1007    1. Added "where" clause to GraphSCCFn in SML/NJ Library.
1008       (Otherwise the functor is useless.)
1009    2. Used GraphSCCFn where SCCUtilFun was used previously.
1010    3. Got rid of SCCUtilFun (in comp-lib.cm).
1011    
1012    ----------------------------------------------------------------------
1013    Name: Lal George
1014    Date: 2000/11/06 09:02:21 EST 2000
1015    Tag: Release_110_30
1016    Description:
1017    
1018    - Version 110.30
1019    ----------------------------------------------------------------------
1020    Name: Matthias Blume
1021    Date: 2000/11/04 14:45:00
1022    Tag: blume-20001104-mlbuild
1023    Description:
1024    
1025    - Made ml-build faster on startup.
1026    - Documentation fixes.
1027    
1028    ----------------------------------------------------------------------
1029    Name: Matthias Blume
1030    Date: 2000/11/02 17:00:00 JST
1031    Tag: blume-20001102-condcomp
1032    Description:
1033    
1034    - Small tweaks to pickler -- new BOOTFILES!
1035    - Version bumped to 110.29.2.
1036    - Added conditional compilation facility to init.cmi (see comment there).
1037    ----------------------------------------------------------------------
1038    Name: Allen Leung
1039    Date: 2000/10/23 19:31:00
1040    Tag: leunga-20001023-demo-ra
1041    
1042    1. Minor RA changes that improves spilling on x86 (affects Moby and C-- only)
1043    2. Test programs for the graph library updated
1044    3. Some new MLRISC demo programs added
1045    
1046    ----------------------------------------------------------------------
1047    Name: Matthias Blume
1048    Date: 2000/08/31 22:15:00 JST
1049    Tag: blume-20001017-errmsg
1050    Description:
1051    
1052    More error message grief: Where there used to be no messages, there
1053    now were some that had bogus error regions.  Fixed.
1054    
1055    ----------------------------------------------------------------------
1056    Name: Matthias Blume
1057    Date: 2000/08/31 17:30:00 JST
1058    Tag: blume-20001017-v110p29p1
1059    Description:
1060    
1061    I made a version 110.29.1 with new bootfiles.
1062    
1063    Changes:  Modified pickler/unpickler for faster and leaner unpickling.
1064              CM documentation changes and a small bugfix in CM's error reporting.
1065    
1066    ----------------------------------------------------------------------
1067    Name: Lal George
1068    Date: 2000/09/27 14:42:35 EDT
1069    Tag: george-20000927-nodestatus
1070    Description:
1071    
1072    Changed the type of the nodestatus, so that:
1073    
1074            SPILLED(~1)             is now SPILLED
1075            SPILLED(m) where m>=0   is now MEMREG(m)
1076            SPILLED(s) where s<~1   is now SPILL_LOC(~s)
1077    
1078    ----------------------------------------------------------------------
1079    Name: Matthias Blume
1080    Date: 2000/09/07 14:45:00 JST
1081    Tag: blume-20000907-cmerrmsg
1082    Description:
1083    
1084    Small tweak to CM to avoid getting ML syntax error messages twice.
1085    
1086    ----------------------------------------------------------------------
1087    Name: Matthias Blume
1088    Date: 2000/08/31 18:00:00 JST
1089    Tag: blume-20000831-cvsbootfiles
1090    Description:
1091    
1092    New URL for boot files (because the 110.29 files on the BL server do
1093    now work correctly with my updated install scripts for yacc and lex).
1094    
1095    ----------------------------------------------------------------------
1096    Name: Matthias Blume
1097    Date: 2000/08/08 12:33:00 JST
1098    Tag: blume-20000808-manual
1099    Description:
1100    
1101    Tiny update to CM manual.
1102    
1103    ----------------------------------------------------------------------
1104    Name: Allen Leung
1105    Date: 2000/08/7 19:31:00
1106    Tag: leunga-20000807-a-whole-bunch-of-stuff
1107    
1108      Moby, C--, SSA, x86, machine descriptions etc.  Should only affect C--
1109    and Mobdy.
1110    
1111    1.  x86
1112    
1113       a.  Fixes to peephole module by John and Dan.
1114       b.  Assembly fix to SETcc by Allen.
1115       c.  Fix to c-call by John.
1116       d.  Fix to spilling by John.  (This one deals with the missing FSTPT case)
1117       e.  Instruction selection optimization to SETcc as suggested by John.
1118    
1119           For example,
1120    
1121            MV(32, x, COND(32, CMP(32, LT, a, b), LI 1, LI 0))
1122    
1123           should generate:
1124    
1125            MOVL a, x
1126            SUBL b, x
1127            SHRL 31, x
1128    
1129    2.  IR stuff
1130    
1131         A bunch of new DJ-graph related algorithms added.  These
1132         speed up SSA construction.
1133    
1134    3.  SSA + Scheduling
1135    
1136         Added code for SSA and scheduling to the repository
1137    
1138    ----------------------------------------------------------------------
1139    Name: Lal George
1140    Date: 2000/07/27 11:53:14 EDT
1141    
1142    Tag: lal-20000727-linux-ppc
1143    Description:
1144    
1145     Made changes to support Linux PPC.
1146     p.s. I have confirmation that the 110.29 boot files work fine.
1147    
1148    ----------------------------------------------------------------------
1149    Name: Matthias Blume
1150    Date: 2000/07/27 17:40:00 JST
1151    Tag: blume-20000727-scripts
1152    Description:
1153    
1154    !!!! WARNING !!!!
1155    You must recompile the runtime system!
1156    !!!! WARNING !!!!
1157    
1158    This is basically another round of script-enhancements:
1159    
1160    1. sml, ml-build, and ml-makedepend accept options -D and -U to define
1161       and undefine CM preprocessor symbols.
1162    
1163    2. ml-build avoids generating a new heap image if it finds that the
1164       existing one is still ok.  (The condition is that no ML file had to
1165       be recompiled and all ML files are found to be older that the heap
1166       file.)
1167    
1168       To make this work smoothly, I also hacked the runtime system as
1169       well as SMLofNJ.SysInfo to get access to the heap image suffix
1170       (.sparc-solaris, ...) that is currently being used.
1171    
1172       Moreover, the signature of CM.mk_standalone has changed.  See the
1173       CM manual.
1174    
1175    3. ml-makedepend accepts additional options -n, -a, and -o.  (See the
1176       CM manual for details.)
1177    
1178    4. More CM manual updates:
1179        - all of the above has been documented.
1180        - there is now a section describing the (CM-related) command line
1181          arguments that are accepted by the "sml" command
1182    
1183    ----------------------------------------------------------------------
1184    Name: Matthias Blume
1185    Date: 2000/07/25 16:20:00 JST
1186    Tag: blume-20000725-makedepend
1187    Description:
1188    
1189    Added a script called ml-makedepend.  This can be used in makefiles
1190    for Unix' make in a way very similar to the "makedepend" command for
1191    C.
1192    
1193    The script internally uses function CM.sources.
1194    
1195    Synopsis:
1196    
1197        ml-makedepend [-f makefile] cmfile targetname
1198    
1199    The default for the makefile is "makefile" (or "Makefile" should
1200    "makefile" not exist).
1201    
1202    ml-makedepend adds a cmfile/targetname-specific section to this
1203    makefile (after removing the previous version of this section).  The
1204    section contains a single dependency specification with targetname on
1205    the LHS (targetname is an arbitrary name), and a list of files derived
1206    from the cmfile on the RHS.  Some of the files on the RHS are
1207    ARCH/OPSYS-specific.  Therefore, ml-makedepend inserts references to
1208    "make" variables $(ARCH) and $(OPSYS) in place of the corresponding
1209    path names.  The makefile writer is responsible for making sure that
1210    these variables have correct at the time "make" is invoked.
1211    
1212    ----------------------------------------------------------------------
1213    Name: Matthias Blume
1214    Date: 2000/07/22 23:30:00 JST
1215    Tag: blume-20000722-urlupdate
1216    Description:
1217    
1218    Changed BOOT and config/srcarchiveurl to point to BL server:
1219    
1220        ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.29/
1221    
1222    ----------------------------------------------------------------------
1223    Name: Matthias Blume
1224    Date: 2000/07/18 18:00:00 JST
1225    Tag: blume-20000718-Version_110_29
1226    Description:
1227    
1228    1. Updated src/compiler/TopLevel/main/version.sml to version 110.29
1229    
1230    2. Updated config/version to 110.29
1231    
1232    3. Updated config/srcarchiveurl
1233    
1234    3. New boot files!
1235       ftp://ftp.cs.princeton.edu/pub/people/blume/sml/110.29-autofetch
1236    
1237    ----------------------------------------------------------------------
1238    Name: Matthias Blume
1239    Date: 2000/07/11 13:58:00 JST
1240    Tag: blume-20000711-doctypo
1241    Description:
1242    
1243    Fixed a few typos in CM manual.
1244    
1245    ----------------------------------------------------------------------
1246    Name: Allen Leung
1247    Date: 2000/06/15 00:38:00
1248    Tag: leunga-20000704-sparc-x86
1249    
1250    1. x86 peephole improvement sp += k; sp -= k => nop  [from John]
1251    2. fix to x86 RET bug [found by Dan Grossman]
1252    3. sparc assembly bug fix for ticc instructions [found by Fermin]
1253    
1254       Affects c-- and moby only
1255    
1256    ----------------------------------------------------------------------
1257    Name: Matthias Blume
1258    Date: 2000/07/04 15:26:00
1259    Tag: blume-20000704-trigger
1260    Description:
1261    
1262    1. Improvements to CM manual.
1263    2. SMLofNJ.Internals.BTrace.trigger reinstated as an alternative way
1264       of getting a back-trace.  The function, when called, raises an
1265       internal exception which explicitly carries the full back-trace history,
1266       so it is unaffected by any intervening handle-raise pairs ("trivial"
1267       or not).  The interactive loop will print that history once it arrives
1268       at top level.
1269       Short of having all exceptions implicitly carry the full history, the
1270       recommended way of using this facility is:
1271         - compile your program with instrumentation "on"
1272         - run it, when it raises an exception, look at the history
1273         - if the history is "cut off" because of some handler, go and modify
1274           your program so that it explicitly calls BTrace.trigger
1275         - recompile (still instrumented), and rerun; look at the full history
1276    
1277    ----------------------------------------------------------------------
1278    Name: Matthias Blume
1279    Date: 2000/07/03 15:36:00 JST
1280    Tag: blume-20000702-manual
1281    Description:
1282    
1283    Small corrections and updates to CM manual.
1284    
1285    ----------------------------------------------------------------------
1286    Name: Matthias Blume
1287    Date: 2000/06/29 16:04:00 JST
1288    Tag: blume-20000629-yacctool
1289    Description:
1290    
1291    Changes:
1292    
1293    1. Class "mlyacc" now takes separate arguments to pass options to
1294       generated .sml- and .sig-files independently.
1295    2. Corresponding CM manual updates.
1296    3. BTrace module now also reports call sites.  (However, for loop clusters
1297       it only shows from where the cluster was entered.)  There are associated
1298       modifications to core.sml, internals.{sig,sml}, btrace.sml, and btimp.sml.
1299    
1300    ----------------------------------------------------------------------
1301    Name: Matthias Blume
1302    Date: 2000/06/27 16:51:00 JST
1303    Tag: blume-20000627-noweb
1304    Description:
1305    
1306    Changes:
1307    
1308     1. Implemented "subdir" and "witness" options for noweb tool.
1309        This caused some slight internal changes in CM's tool implementation.
1310     2. Fixed bug in "tool plugin" mechanism.  This is essentially cleaning
1311        some remaining issues from earlier path anchor changes.
1312     3. Updated CM manual accordingly.
1313    
1314     4. Changed implementation of back-tracing so that I now consider it
1315        ready for prime-time.
1316    
1317        In particular, you don't have to explicitly trigger the back-trace
1318        anymore.  Instead, if you are running BTrace-instrumented code and
1319        there is an uncaught exception (regardless of whether or not it was
1320        raised in instrumented code), the top-level evalloop will print
1321        the back-trace.
1322    
1323        Features:
1324    
1325          - Instrumented and uninstrumented code work together seemlessly.
1326            (Of course, uninstrumented code is never mentioned in actual
1327             back-traces.)
1328    
1329          - Asymptotic time- and space-complexity of instrumented code is
1330            equal to that of uninstrumented code.  (This means that
1331            tail-recursion is preserved by the instrumentation phase.)
1332    
1333          - Modules whose code has been instrumented in different sessions
1334            work together without problem.
1335    
1336          - There is no penalty whatsoever on uninstrumented code.
1337    
1338          - There is no penalty on "raise" expressions, even in
1339            instrumented code.
1340    
1341        A potential bug (or perhaps it is a feature, too):
1342    
1343          A back-trace reaches no further than the outermost instrumented
1344          non-trivial "raise".  Here, a "trivial" raise is one that is the
1345          sole RHS of a "handle" rule.  Thus, back-traces reach trough
1346    
1347               <exp> handle e => raise e
1348    
1349          and even
1350    
1351               <exp> handle Foo => raise Bar
1352    
1353          and, of course, through
1354    
1355               <exp> handle Foo => ...
1356    
1357         if the exception was not Foo.
1358    
1359         Back-traces always reach right through any un-instrumented code
1360         including any of its "handle" expressions, trivial or not.
1361    
1362       To try this out, do the following:
1363    
1364         - Erase all existing binfiles for your program.
1365           (You may keep binfiles for those modules where you think you
1366            definitely don't need back-tracing.)
1367         - Turn on back-trace instrumentation:
1368              SMLofNJ.Internals.BTrace.mode (SOME true);
1369         - Recompile your program.  (I.e., run "CM.make" or "use".)
1370         - You may now turn instrumentation off again (if you want):
1371              SMLofNJ.Internals.BTrace.mode (SOME false);
1372         - Run your program as usual.  If it raises an exception that
1373           reaches the interactive toplevel, then a back-trace will
1374           automatically be printed.  After that, the toplevel loop
1375           will print the exception history as usual.
1376    
1377    ----------------------------------------------------------------------
1378    Name: Matthias Blume
1379    Date: 2000/06/26 09:56:46 JST
1380    Tag: blume-20000626-setup
1381    Description:
1382    
1383    CM: - setup-parameter to "sml" added; this can be used to run arbitrary
1384          ML code before and after compiling a file (e.g., to set compiler
1385          flags)
1386    
1387    Compiler: - improved btrace API (in core.sml, internals.{sig,sml})
1388              - associated changes to btrace.sml (BTrace instrumentation pass)
1389              - cleaner implementation of btimp.sml (BTrace tracing and report
1390                module)
1391    
1392    CM manual: * new path encoding documented
1393               * description of setup-parameter to "sml" added
1394    
1395    The biggest user-visible change to back-tracing is that it is no
1396    longer necessary to compile all traced modules within the same
1397    session.  (This was a real limitation.)
1398    
1399    ----------------------------------------------------------------------
1400    Name: Matthias Blume
1401    Date: 2000/06/24 12:40:00 JST
1402    Tag: blume-20000624-startup
1403    Description:
1404    
1405    Fixes startup slowdown problem.  (I was calling SrcPath.sync a _tad_
1406    bit too often -- to put it mildly. :)
1407    
1408    ----------------------------------------------------------------------
1409    Name: Matthias Blume
1410    Date: 2000/06/23 18:20:00 JST
1411    Tag: blume-20000623-btrace
1412    Description:
1413    
1414    This updates adds a backtrace facility to aid programmers in debugging
1415    their programs.  This involves the following changes:
1416    
1417    1. Module system/smlnj/init/core.sml (structure _Core) now has hooks for
1418       keeping track of the current call stack.  When programs are compiled
1419       in a special mode, the compiler will insert calls to these hooks
1420       into the user program.
1421       "Hook" means that it is possible for different implementations of
1422       back-tracing to register themselves (at different times).
1423    
1424    2. compiler/MiscUtil/profile/btrace.sml implements the annotation phase
1425       as an Absyn.dec->Absyn.dec rewrite.  Normally this phase is turned off.
1426       It can be turned on using this call:
1427         SMLofNJ.Internals.BTrace.mode (SOME true);
1428       Turning it off again:
1429         SMLofNJ.Internals.BTrace.mode (SOME false);
1430       Querying the current status:
1431         SMLofNJ.Internals.BTrace.mode NONE;
1432       Annotated programs are about twice as big as normal ones, and they
1433       run a factor of 2 to 4 slower with a dummy back-trace plugin (one
1434       where all hooks do nothing).  The slowdown with a plugin that is
1435       actually useful (such as the one supplied by default) is even greater,
1436       but in the case of the default plugin it is still only an constant
1437       factor (amortized).
1438    
1439    3. system/Basis/Implementation/NJ/internals.{sig,sml} have been augmented
1440       with a sub-structure BTrace for controlling back-tracing.  In particular,
1441       the above-mentioned function "mode" controls whether the annotation
1442       phase is invoked by the compiler.  Another important function is
1443       "trigger": when called it aborts the current execution and causes
1444       the top-level loop to print a full back-trace.
1445    
1446    4. compiler/MiscUtil/profile/btimp.sml is the current default plugin
1447       for back-tracing.  It keeps track of the dynamic call stack and in
1448       addition to that it keeps a partial history at each "level" of that
1449       stack.  For example, if a tail-calls b, b tail-calls c, and c tail-calls
1450       d and b (at separate times, dynamically), then the report will show:
1451    
1452       GOTO   d
1453             /c
1454       GOTO  \b
1455       CALL   a
1456    
1457       This shows that there was an initial non-tail call of a, then a
1458       tail-call to b or c, looping behavior in a cluster of functions that
1459       consist of b and c, and then a goto from that cluster (i.e., either from
1460       b or from c) to d.
1461    
1462       Note that (depending on the user program) the amount of information
1463       that the back-trace module has to keep track of at each level is bounded
1464       by a constant.  Thus, the whole implementation has the same asymptotical
1465       complexity as the original program (both in space and in time).
1466    
1467    5. compiler/TopLevel/interact/evalloop.sml has been modified to
1468       handle the special exception SMLofNJ.Internals.BTrace.BTrace
1469       which is raised by the "trigger" function mentioned above.
1470    
1471    Notes on usage:
1472    
1473    - Annotated code works well together with unannotated code:
1474    Unannotated calls simply do not show up at all in the backtrace.
1475    
1476    - It is not a good idea to let modules that were annotated during
1477    different sessions run at the same time.  This is because the compiler
1478    chooses small integers to identify individual functions, and there
1479    will be clashes if different modules were compiled in separate sessions.
1480    (Nothing will crash, and you will even be told about the clashes, but
1481    back-trace information will in general not be useful.)
1482    
1483    - Back-tracing can be confused by callcc and capture.
1484    
1485    - The only way of getting a back-trace right now is to explicitly
1486    invoke the "trigger" function from your user program.  Eventually, we
1487    should make every exception carry back-trace information (if
1488    available).  But since this creates more overhead at "raise"-time
1489    (similar to the current exnHistory overhead), I have not yet
1490    implemented this.  (The implementation will be rather easy.)  With
1491    exceptions carrying back-trace information, this facility will be even
1492    more useful because users don't need to modify their programs...
1493    
1494    - While it is possible to compile the compiler with back-trace
1495    annotations turned on (I did it to get some confidence in
1496    correctness), you must make absolutely sure that core.sml and
1497    btimp.sml are compiled WITHOUT annotation!  (core.sml cannot actually
1498    be compiled with annotation because there is no core access yet, but
1499    if you compile btimp.sml with annotation, then the system will go into
1500    an infinite recursion and crash.)
1501    Since CM currently does not know about BTrace, the only way to turn
1502    annotations on and off for different modules of the compiler is to
1503    interrupt CMB.make, change the settings, and re-invoke it.  Of course,
1504    this is awkward and clumsy.
1505    
1506    Sample sessions:
1507    
1508    Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
1509    - SMLofNJ.Internals.BTrace.mode (SOME true);
1510    [autoloading]
1511    [autoloading done]
1512    val it = false : bool
1513    - structure X = struct
1514    -     fun main n = let
1515    -         fun a (x, 0) = d x
1516    -           | a (x, n) = b (x, n - 1)
1517    -         and b (x, n) = c (x, n)
1518    -         and c (x, n) = a (x, n)
1519    -         and d x = e (x, 3)
1520    -         and e (x, 0) = f x
1521    -           | e (x, n) = e (x, n - 1)
1522    -         and f 0 = SMLofNJ.Internals.BTrace.trigger ()
1523    -           | f n = n * g (n - 1)
1524    -         and g n = a (n, 3)
1525    -     in
1526    -         f n
1527    -     end
1528    - end;
1529    structure X : sig val main : int -> int end
1530    - X.main 3;
1531    *** BACK-TRACE ***
1532    GOTO   stdIn:4.2-13.20: X.main[2].f
1533    GOTO-( stdIn:4.2-13.20: X.main[2].e
1534    GOTO   stdIn:4.2-13.20: X.main[2].d
1535         / stdIn:4.2-13.20: X.main[2].a
1536         | stdIn:4.2-13.20: X.main[2].b
1537    GOTO-\ stdIn:4.2-13.20: X.main[2].c
1538    CALL   stdIn:4.2-13.20: X.main[2].g
1539    GOTO   stdIn:4.2-13.20: X.main[2].f
1540    GOTO-( stdIn:4.2-13.20: X.main[2].e
1541    GOTO   stdIn:4.2-13.20: X.main[2].d
1542         / stdIn:4.2-13.20: X.main[2].a
1543         | stdIn:4.2-13.20: X.main[2].b
1544    GOTO-\ stdIn:4.2-13.20: X.main[2].c
1545    CALL   stdIn:4.2-13.20: X.main[2].g
1546    GOTO   stdIn:4.2-13.20: X.main[2].f
1547    GOTO-( stdIn:4.2-13.20: X.main[2].e
1548    GOTO   stdIn:4.2-13.20: X.main[2].d
1549         / stdIn:4.2-13.20: X.main[2].a
1550         | stdIn:4.2-13.20: X.main[2].b
1551    GOTO-\ stdIn:4.2-13.20: X.main[2].c
1552    CALL   stdIn:4.2-13.20: X.main[2].g
1553    GOTO   stdIn:4.2-13.20: X.main[2].f
1554    CALL   stdIn:2.15-17.4: X.main[2]
1555    -
1556    
1557    (Note that because of a FLINt bug the above code currently does not
1558    compile without BTrace turned on.)
1559    
1560    Here is another example, using my modified Tiger compiler:
1561    
1562    Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
1563    - SMLofNJ.Internals.BTrace.mode (SOME true);
1564    [autoloading]
1565    [autoloading done]
1566    val it = false : bool
1567    - CM.make "sources.cm";
1568    [autoloading]
1569    ...
1570    [autoloading done]
1571    [scanning sources.cm]
1572    [parsing (sources.cm):parse.sml]
1573    [creating directory CM/SKEL ...]
1574    [parsing (sources.cm):tiger.lex.sml]
1575    ...
1576    [wrote CM/sparc-unix/semant.sml]
1577    [compiling (sources.cm):main.sml]
1578    [wrote CM/sparc-unix/main.sml]
1579    [New bindings added.]
1580    val it = true : bool
1581    - Main.compile ("../testcases/merge.tig", "foo.out");
1582    *** BACK-TRACE ***
1583    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trvar
1584    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1585    CALL   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
1586    GOTO   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
1587    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1588    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1589    CALL   lib/semant.sml:488.3-505.6: SemantFun[2].transDec.trdec[2].transBody[2]
1590         / lib/semant.sml:411.65-543.8: SemantFun[2].transDec
1591    CALL-\ lib/semant.sml:413.2-540.9: SemantFun[2].transDec.trdec[2]
1592    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1593    CALL   lib/semant.sml:8.52-558.4: SemantFun[2].transProg[2]
1594    CALL   main.sml:1.18-118.4: Main.compile[2]
1595    -
1596    
1597    ----------------------------------------------------------------------
1598    Name: Matthias Blumen
1599    Date: 2000/06/21 18:00:00 JST
1600    Tag: blume-20000621-manual
1601    Description:
1602    
1603    CM manual update: Path environments documented.
1604    
1605    ----------------------------------------------------------------------
1606    Name: Matthias Blume
1607    Date: 2000/06/19 13:40:00
1608    Tag: blume-20000619-manual
1609    Description:
1610    
1611    CM manual and system/README update.  This only covers the fact that
1612    there are no more implicit anchors.  (Path environments and the "bind"
1613    option to "cm" have yet to be documented.)
1614    
1615    ----------------------------------------------------------------------
1616    Name: Matthias Blume
1617    Date: 2000/06/19 11:05:00 JST
1618    Tag: blume-20000619-chdir-bugfix
1619    Description:
1620    
1621    Fixed a bug in new SrcPath module that sometimes led to a bad chDir call.
1622    
1623    ----------------------------------------------------------------------
1624    Name: Matthias Blume
1625    Date: 2000/06/18 22:00:10 JST
1626    Tag: blume-20000618-implicit-anchors-really-gone
1627    Description:
1628    
1629    I updates the previous HISTORY entry where I forgot to mention that
1630    implicit anchors are no longer with us.
1631    
1632    The current update also gets rid of the (now useless) controller
1633    CM.Control.implicit_anchors.
1634    
1635    ----------------------------------------------------------------------
1636    Name: Matthias Blume
1637    Date: 2000/06/16 17:30:00 JST
1638    Tag: blume-20000616-anchorenv
1639    Description:
1640    
1641    This patch implements the long anticipated (just kidding :) "anchor
1642    environment" mechanism.  In the course of doing this, I also
1643    re-implemented CM's internal "SrcPath" module from scratch.  The new
1644    one should be more robust in certain boundary cases.  In any case, it
1645    is a lot cleaner than its predecessor (IMHO).
1646    
1647    This time, although there is yet another boot file format change, I
1648    kept the unpickler backward-compatible.  As a result, no new bootfiles
1649    are necessary and bootstrapping is straightforward.  (You cannot read
1650    new bootfiles into an old system, but the other way around is no
1651    problem.)
1652    
1653    Visible changes:
1654    
1655    ** 0. Implicit path anchors (without the leading $-symbol) are no
1656    longer recognized at all. This means that such path names are not
1657    illegal either.  For example, the name basis.cm simply refers to a
1658    local file called "basis.cm" (i.e, the name is an ordinary path
1659    relative to .cm-files directory).  Or, to put it differently, only
1660    names that start with $ are anchored paths.
1661    
1662    ** 1. The $<singlearc> abbreviation for $/<singlearc> has finally
1663    vanished.
1664    
1665    John (Reppy) had critizised this as soon as I originally proposed and
1666    implemented it, but at that time I did not really deeply believe
1667    him. :) Now I came full-circle because I need the $<singlearc> syntax
1668    in another place where it cannot be seen as an abbreviation for
1669    $/<singlearc>.  To avoid the confusion, $<singlearc> now means what it
1670    seems to mean (i.e., it "expands" into the corresponding anchor
1671    value).
1672    
1673    However, when paths are used as members in CM description files, it
1674    continues to be true that there must be at least another arc after the
1675    anchor.  This is now enforced separately during semantic analysis
1676    (i.e., from a lexical/syntactical point of view, the notation is ok.)
1677    
1678    ** 2. The "cm" class now accepts an option "bind".  The option's value
1679    is a sub-option list of precisely two items -- one labeled "anchor"
1680    and the other one labeled "value".  As you might expect, "anchor" is
1681    used to specify an anchor name to be bound, and "value" specifies what
1682    the anchor is being bound to.
1683    
1684    The value must be a directory name and can be given in either standard
1685    syntax (including the possibility that it is itself an anchored path)
1686    or native syntax.
1687    
1688    Examples:
1689    
1690       foo.cm (bind:(anchor:bar value:$mystuff/bar))
1691       lib.cm (bind:(anchor:a value:"H:\\x\\y\\z"))  (* only works under windows *)
1692    
1693    and so on.
1694    
1695    The meaning of this is that the .cm-file will be processed with an
1696    augmented anchor environment where the given anchor(s) is/are bound to
1697    the given values(s).
1698    
1699    The rationale for having this feature is this: Suppose you are trying
1700    to use two different (already stable) libraries a.cm and b.cm (that
1701    you perhaps didn't write yourself).  Further, suppose each of these
1702    two libraries internally uses its own auxiliary library $aux/lib.cm.
1703    Normally you would now have a problem because the anchor "lib" can not
1704    be bound to more than one value globally.  Therefore, the project that
1705    uses both a.cm and b.cm must locally redirect the anchor to some other
1706    place:
1707    
1708       a.cm (bind:(anchor:lib value:/usr/lib/smlnj/a-stuff))
1709       b.cm (bind:(anchor:lib value:/usr/lib/smlnj/b-stuff))
1710    
1711    This hard-wires $lib/aux.cm to /usr/lib/smlnj/a-stuff/aux.cm or
1712    /usr/lib/smlnj/b-stuff/aux.cm, respectively.
1713    
1714    Hard-wiring path names is a bit inflexible (and CM will verbosely warn
1715    you when you do so at the time of CM.stabilize).  Therefore, you can
1716    also use an anchored path as the value:
1717    
1718      a.cm (bind:(anchor:lib value:$a-lib))
1719      b.cm (bind:(anchor:lib value:$b-lib))
1720    
1721    Now you can globally configure (using the usual CM.Anchor.anchor or
1722    pathconfig machinery) bindings for "a-lib" and "b-lib".  Since "lib"
1723    itself is always locally bound, setting it globally is no longer
1724    meaningful or necessary (but it does not hurt either).  In fact, "lib"
1725    can still be used as a global anchor for separate purposes.  As a
1726    matter of fact, one can locally define "lib" in terms of a global
1727    "lib":
1728    
1729      a.cm (bind:(anchor:lib value:$lib/a))
1730      b.cm (bind:(anchor:lib value:$lib/b))
1731    
1732    ** 3: The encoding of path names has changed.  This affects the way
1733    path names are shown in CM's progress report and also the internal
1734    protocol encoding used for parallel make.
1735    
1736    The encoding now uses one or more ':'-separated segments.  Each
1737    segments corresponds to a file that has been specified relative to the
1738    file given by its preceding segment.  The first segment is either
1739    relative to the CWD, absolute, or anchored.  Each segment itself is
1740    basically a Unix pathname; all segments but the first are relative.
1741    
1742    Example:
1743    
1744       $foo/bar/baz.cm:a/b/c.sml
1745    
1746    This path denotes the file bar/a/b/c.sml relative to the directory
1747    denoted by anchor "foo".  Notice that the encoding also includes
1748    baz.cm which is the .cm-file that listed a/b/c.sml.  As usual, such
1749    paths are resolved relative to the .cm-files directory, so baz.cm must
1750    be ignored to get the "real" pathname.
1751    
1752    To make this fact more obvious, CM puts the names of such "virtual
1753    arcs" into parentheses when they appear in progress reports. (No
1754    parentheses will appear in the internal protocol encoding.)  Thus,
1755    what you really see is:
1756    
1757      $foo/bar/(baz.cm):a/b/c.sml
1758    
1759    I find this notation to be much more informative than before.
1760    
1761    Another new feature of the encoding is that special characters
1762    including parentheses, colons, (back)slashes, and white space are
1763    written as \ddd (where ddd is the decimal encoding of the character).
1764    
1765    *** The CM manual still needs to be updated.
1766    
1767    ----------------------------------------------------------------------
1768    Name: Allen Leung
1769    Date: 2000/06/15 00:38:00
1770    Tag: leunga-20000615-x86-peephole
1771    
1772    x86 Peephole fix by Fermin.  Affects c-- and moby only.
1773    
1774    ----------------------------------------------------------------------
1775    Name: Matthias Blume
1776    Date: 2000/06/12 11:40:00
1777    Tag: blume-20000612-parmakefix
1778    Description:
1779    
1780    More cleanup after changing the file naming scheme: This time I
1781    repaired the parallel make mechanism for CMB.make which I broke earlier.
1782    
1783    ----------------------------------------------------------------------
1784    Name: Allen Leung
1785    Date: 2000/06/09 01:25:00
1786    Tag: leunga-20000609-various
1787    
1788    None of these things should affect normal SML/NJ operations
1789    
1790    1. Peephole improvements provided by Fermin (c--)
1791    2. New annotation DEFUSE for adding extra dependence (moby)
1792    3. New X86 LOCK instructions (moby)
1793    4. New machine description language for reservation tables (scheduling)
1794    5. Fixes to various optimization/analysis modules (branch chaining, dominator
1795       trees etc.)
1796    6. I've changed the CM files so that they can work with versions
1797       110.0.6, 110.25 and 110.28
1798    
1799    ----------------------------------------------------------------------
1800    Name: Matthias Blume
1801    Date: 2000/06/09 12:40:00
1802    Tag: blume-20000609-log
1803    Description:
1804    
1805    - Removed all(?) remaining RCS Log entries from sources.
1806    
1807    - Fixed bug in ml-yacc and ml-lex sources (use explicit anchors for
1808      anchored paths).
1809    
1810    ----------------------------------------------------------------------
1811    Name: Matthias Blume
1812    Date: 2000/06/07 17:00:00 JST
1813    Tag: blume-20000607-no-implicit-anchors
1814    Description:
1815    
1816    1. This update changes the default setting for
1817    CM.Control.implicit_anchors from true to false.  This means that
1818    implicit anchors are no longer permitted by default.  I also tried to
1819    make sure that nothing else still relies on implicit anchors.
1820    (This is the next step on the schedule towards a CM that does not even
1821    have the notion of implicit anchors anymore.)
1822    
1823    2. More CM manual updates.
1824    
1825    3. I managed to track down and fix the pickling bug I mentioned last
1826    time.  Because of the previously existing workaround, this entails no
1827    immediate practical changes.
1828    
1829    ----------------------------------------------------------------------
1830    Name: Matthias Blume
1831    Date: 2000/06/06 11:15:00 JST
1832    Tag: blume-20000606-lazierpickle
1833    Description:
1834    
1835    !!!! NEW BOOT FILES !!!!
1836    
1837    * The main purpose of this update is to make library pickles lazier in
1838    order to reduce the initial space penalty for autoloading a library.
1839    As a result, it is now possible to have $smlnj/compiler.cm
1840    pre-registered.  This should take care of the many complaints or
1841    inquiries about missing structure Compiler.  This required changes to
1842    CM's internal data structures and small tweaks to some algorithms.
1843    
1844    As a neat additional effect, it is no longer necessary (for the sake
1845    of lean heap image files) to distinguish between a "minimal" CM and a
1846    "full" CM.  Now, there is only one CM (i.e., the "full" version:
1847    $smlnj/cm.cm aka $smlnj/cm/full.cm), and it is always available at the
1848    interactive top level. ($smlnj/cm/minimal.cm is gone.)
1849    
1850    To make the life of compiler-hackers easier, "makeml" now also
1851    pre-registers $smlnj/cmb.cm (aka $smlnj/cmb/current.cm).  In other
1852    words, after you bootstrap a new sml for the first time, you will not
1853    have to autoload $smlnj/cmb.cm again afterwards.  (The first time
1854    around you will still have to do it, though.)
1855    
1856    * A second change consists of major updates to the CM manual.  There
1857    are now several appendices with summary information and also a full
1858    specification of the CM description file syntax.
1859    
1860    * In directory src/system I added the script "allcross".  This script
1861    invokes sml and cross-compiles the compiler for all supported
1862    architectures.  (Useful when providing a new set of boot files.)
1863    
1864    * There seems to be a latent bug in my "lazy pickles" mechanism.  I
1865    added a small tweak to pickle-util.sml to work around this problem,
1866    but it is not a proper fix yet.  I will investigate further.  (The
1867    effect of the bug was an inflation of library pickle size.)
1868    
1869    * Version number increased to 110.28.1 (to avoid compatibility problems).
1870    
1871    ----------------------------------------------------------------------
1872    Name: Allen Leung
1873    Date: 2000/05/25 17:28 EDT
1874    Tag: leunga-20000525-ra
1875    Description:
1876    
1877      Fixed a bug in freezing phase of the register allocator.
1878    
1879    ----------------------------------------------------------------------
1880    Name: Allen Leung
1881    Date: 2000/05/15 22:53 EDT
1882    Tag: leunga-20000515-alpha-x86-ra
1883    Description:
1884    
1885      1. Alpha
1886    
1887          Slight cleanup.  Removed the instruction SGNXL
1888    
1889      2. X86
1890    
1891          Added the following instructions to the instruction set:
1892    
1893            ROLx, RORx,
1894            BTx, BTSx, BTLx, BTRx,
1895            XCHGx, and variants with the LOCK prefix
1896    
1897      3. Register Allocation
1898    
1899          The module ra-rewrite-with-renaming has been improved.
1900    
1901      These have no effect on SML/NJ.
1902    
1903    ----------------------------------------------------------------------
1904    Name: Matthias Blume
1905    Date: 2000/05/15 16:20:00 JST
1906    Tag: blume-20000515-lightrebuild
1907    Description:
1908    
1909    1. I added an alternative to "-rebuild" to "makeml".  The difference is
1910       that prior to calling CMB.make' the CM-variable "LIGHT" will be
1911       defined.  In effect, the command will not build any cross-compiler
1912       backends and therefore finish more quickly.
1913    
1914       The "fixpt" script also takes a "-light" switch to be able to use
1915       this new facility while compiling for a fixpoint.
1916    
1917    2. I replaced all mentions of anchored paths in group owner specifications
1918       with simple relative paths (usually starting with "..").
1919       The rationale is that a library's internal workings should not be
1920       compromised by the lack of some anchor.  (An anchor is necessary
1921       for someone who wants to refer to the library by an anchored path,
1922       but it should not be necessary to build the same library in the first
1923       place.)
1924    
1925    3. I changed the way CM's tool mechanism determines the shell command
1926       string used for things like ml-yacc etc. so that it does not break
1927       when CM.Control.implicit_anchors is turned off.
1928    
1929    ----------------------------------------------------------------------
1930    Name: Matthias Blume
1931    Date: 2000/05/12 18:20:00 JST
1932    Tag: blume-20000512-ml-build
1933    Description:
1934    
1935    Fixed a bug in config/_ml-build that prevented ml-yacc and ml-lex from
1936    getting installed properly (by config/install.sh).
1937    
1938    ----------------------------------------------------------------------
1939    Name: Matthias Blume
1940    Date: 2000/05/12 17:30:00 JST
1941    Tag: blume-20000512-anchors
1942    Description:
1943    
1944    !!! NEW BOOT FILES !!!
1945    
1946    This change is in preparation of fading out support for "implicitly
1947    anchored path names".  I went through all sources and used the
1948    explicit (and relatively new) $-notation.  See system/README and the
1949    CM manual for more info on this.
1950    
1951    I also modified the anchoring scheme for some things such as "smlnj",
1952    "MLRISC", "cm", etc. to take advantage of the fact that explicit
1953    anchors are more expressive: anchor name and first arc do not have to
1954    coincide.  This entails the following user-visible change:
1955    
1956    You have to write $smlnj/foo/bar instead of smlnj/foo/bar.  In
1957    particular, when you fire up sml with a command-line argument, say,
1958    e.g.:
1959    
1960       sml '$smlnj/cmb.cm'
1961    
1962    At the ML toplevel prompt:
1963    
1964       CM.autoload "$smlnj/cmb.cm";
1965    
1966    There is also a new controller in CM.Control that can be used to turn
1967    off all remaining support for implicit anchors by saying:
1968    
1969        CM.autoload "$smlnj/
1970        #set CM.Control.implicit_anchors false;
1971    
1972    This causes CM to reject implicitly anchored paths.  This is (for the
1973    time being) less permissive than the "final" version where there will
1974    be no more such implicit anchors and relative paths will be just that:
1975    relative.
1976    
1977    The next step (version after next version?) will be to make the
1978    default for CM.Control.implicit_anchors false.  After the dust has
1979    settled, I can then produce the "final" version of this...
1980    
1981    Note: Since bootstrapping is a bit tricky, I provided new boot files.
1982    
1983    ----------------------------------------------------------------------
1984    Name: Matthias Blume
1985    Date: 2000/05/11 16:30:00 JST
1986    Tag: blume-20000511-sources
1987    Description:
1988    
1989    The main change is that I added function CM.sources as a generalized
1990    version of the earlier CM.makedepend.  This entails the following
1991    additional changes:
1992    
1993      - CM.makedepend has been dropped.
1994    
1995      - CM manual has been updated.
1996    
1997      - TOOLS signature and API have been changed.
1998    
1999    ----------------------------------------------------------------------
2000    Name: Allen Leung
2001    Date: 2000/05/10 21:17 EDT
2002    Tag: leunga-20000510-moby-c--ssa
2003    Description:
2004    
2005      Various bug fixes and new features for C--, Moby and MLRISC optimizations.
2006    None of these affect SML/NJ.
2007    
2008    1. Register Allocation
2009    
2010        a. A new ra spilling module (ra/ra-spill-with-renaming) is implemented.
2011           This module tries to remove local (i.e. basic block level) redundancies
2012           during spilling.
2013    
2014        b. A new framework for performing region based register allocation.
2015           Not yet entirely functional.
2016    
2017    2. X86
2018    
2019       a. DefUse for POP was missing the stack pointer [found by Lal]
2020       b. Reload for CALL was incorrect in X86Spill [found by John]
2021       c. Various fixes in X86Spill so that it can be used correctly for
2022          the new spilling module.
2023    
2024    3. SSA/IR
2025    
2026       a. New module ir/dj-dataflow.sml implements elimination based
2027          data flow analysis.
2028    
2029    4. MLRiscGen
2030    
2031       a. Fix for gc type annotation
2032    
2033    5. MDGen
2034    
2035       Various fixes for machine description -> ml code translation.  For ssa
2036       only.
2037    
2038    ----------------------------------------------------------------------
2039    Name: Allen Leung
2040    Date: 2000/05/08 22:17 EDT
2041    Tag: leunga-20000508-labexp
2042    Description:
2043    
2044      Fermin has found a few assembly problems with constant expressions
2045      generated in LabelExp.  Mostly, the problems involve extra parentheses,
2046      which choke on dumb assemblers.  This is his fix.
2047    
2048    ----------------------------------------------------------------------
2049    Name: Dave MacQueen
2050    Date: 2000/04/09 14:00 EDT
2051    Tag: dbm-20000502-Version_110_28
2052    Description:
2053    
2054    1. Updated src/compiler/TopLevel/main/version.sml to version 110.28
2055    
2056    2. Updated config/version to 110.28
2057    
2058    3. Updated config/srcarchiveurl
2059    
2060    3. New boot files!
2061       ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.28/
2062    
2063    ----------------------------------------------------------------------
2064    Name: Matthias Blume
2065    Date: 2000/05/01 19:05:00 JST
2066    Tag: blume-20000501-noweb
2067    Description:
2068    
2069    A new noweb tool has been added.  The existing system is entirely
2070    unaffected by this, but some CM users have asked for renewed noweb
2071    support.  Everything is documented in the CM manual.
2072    
2073    New (plugin) libraries:
2074    
2075       noweb-tool.cm
2076       nw-ext.cm
2077    
2078    ----------------------------------------------------------------------
2079    Name: Dave MacQueen
2080    Date: 2000/04/30 12:40PM EDT
2081    Tag: dbm-20000430-bug_fixes
2082    Description:
2083    
2084    1. Fix for bug 1498
2085       smlnj/src/system/Basis/Implementation/Unsafe/object.sig
2086       smlnj/src/system/Basis/Implementation/Unsafe/object.sml
2087         added toRealArray function
2088       smlnj/src/compiler/MiscUtil/print/ppobj.sml
2089         added check for tag Obj.RealArray to array printing case in ppObj
2090    
2091    2. Fix for bug 1510
2092       smlnj/src/compiler/Semant/types/typesutil.sml
2093         fixed definition of dummyargs (used by equalTycon) so that
2094         dummy args are distinct types
2095    
2096    ----------------------------------------------------------------------
2097    Name: Matthias Blume
2098    Date: 2000/04/30 01:00:00 JST
2099    Tag: blume-20000430-versions
2100    Description:
2101    
2102    1. CM version numbering added.  This is an implementation of Lal's
2103       proposal for adding version numbers and version checking to .cm
2104       files.  Lal said that his proposal was just that -- a proposal.
2105       For the time being I went ahead and implemented it so that people
2106       can comment on it.  Everything is completely backward-compatible
2107       (except for the stable library format, i.e., new bootfiles!).
2108    
2109       As usual, see the CM manual for details.
2110    
2111    2. An alternative syntax for anchored paths has been implemented.
2112       Dave has recently voiced the same concerns that I had when I did
2113       this, so there should be some support.  My take is that eventually
2114       I will let support for the current syntax (where anchors are
2115       "implicit") fade out in favor of the new, explicit syntax.
2116       In order to be backward-compatible, both old and new syntax are
2117       currently supported.
2118    
2119       Again, see the CM manual for details.
2120    
2121    3. Parallel make is trying to be slightly smarter:  When the master
2122       process finds a "bottleneck", i.e., when there is only one
2123       compilation unit that can be compiled and everybody else is
2124       waiting on it, then it will simply compile it directly instead
2125       of clumsily telling one of the slaves to do it.
2126    
2127    4. Support for "unsharing" added.  This is necessary in order to be
2128       able to have two different versions of the same library running
2129       at the same time (e.g., for trying out a new MLRISC while still
2130       having the old MLRISC linked into the current compiler, etc.)
2131       See the CM manual.
2132    
2133    5. Simple "makedepend" functionality added for generating Makefile
2134       dependency information.  (This is rather crude at the moment.
2135       Expect some changes here in the future.)
2136    
2137    6. ".fun" added as a recognized suffix for ML files. Also documented
2138       explicitly in the manual that the fallback behavior (unknown suffix
2139       -> ML file) is not an official feature!
2140    
2141    7. Small changes to the pickler for stable libraries.
2142    
2143    8. Several internal changes to CM (for cleanup/improvement).
2144    
2145    
2146    !!!! NEW BINFILES !!!!
2147    
2148    ----------------------------------------------------------------------
2149    Name: Matthias Blume
2150    Date: 2000/04/28 17:30:00 JST
2151    Tag: blume-20000428-pathconfig
2152    Description:
2153    
2154    1. I changed config/install.sh to remove duplicate entries from the
2155       lib/pathconfig file at the end.  Moreover, the final version of
2156       lib/pathconfig is sorted alphabetically.  The same (sorting) is done
2157       in src/system/installml.
2158    
2159    2. The config/install.sh script now consistently uses relative
2160       pathnames in lib/pathconfig whenever the anchor is in the lib
2161       directory.  (So far this was true for the libraries that come
2162       pre-compiled and bundled as part of the bootfiles but not for
2163       libraries that are compiled by the script itself.)
2164    
2165    ----------------------------------------------------------------------
2166    Name: Matthias Blume
2167    Date: 2000/04/26 13:10:00 JST
2168    Tag: blume-20000426-fun_suffix
2169    Description:
2170    
2171    Added ".fun" as a recognized file name suffix (for ML code).
2172    
2173    ----------------------------------------------------------------------
2174    Name: Allen Leung
2175    Date: 2000/04/25 17:00:00 EST
2176    Tag: leunga-20000425-alpha-ra
2177    Description:
2178    
2179    1. Alpha
2180    
2181        PSEUDOARITH was missing in AlphaRewrite.  This causes an endless loop
2182    in C--.
2183    
2184    2. RA
2185    
2186       Added a flag "ra-dump-size" to print out the size of the flowgraph
2187       and the interference graph.
2188    
2189    ----------------------------------------------------------------------
2190    Name: Dave MacQueen
2191    Date: 2000/04/25/
2192    Tag: dbm-20000425-mlyacc_doc_examples
2193    Description:
2194      Updated mlyacc.tex sections 5 and 7 for SML '97 and CM.
2195      Updated all three examples in src/ml-yacc/examples to run
2196      under 110.* using CM.make.
2197    
2198    ----------------------------------------------------------------------
2199    Name: Allen Leung
2200    Date: 2000/04/20 23:04:00 EST
2201    Tag: leunga-20000420-ssa-c---stuff
2202    Description:
2203    
2204      This update synchronizes my repository with Yale's.  Most of these
2205    changes, however, do not affect SML/NJ at all (the RA is an exception).
2206    
2207    1. Register Allocator
2208    
2209       a. An improvement in the interference graph construction:
2210          Given a copy
2211    
2212                s <- t
2213    
2214          no interference edge between s and t is added for this definition of s.
2215    
2216       b. I've added two new spill heuristic modules that Fermin and I developed
2217          (in the new library RA.cm). These are unused in SML/NJ but maybe
2218          useful for others (Moby?)
2219    
2220    2. X86
2221    
2222       a. Various fixes in the backend provided by Fermin [C--] and Lal.
2223    
2224    3. Alpha
2225    
2226       a. Added the BSR instruction and code generation that goes with it [C--]
2227       b. Other fixes too numerous to recount provided by Fermin [C--]
2228    
2229    4. Regmaps
2230    
2231       a. The regmaps are not initialized with the identity physical bindings
2232          at creation time.  This is unneeded.
2233    
2234    5. MLRISC Optimizations
2235    
2236       a. The DJ-Graph module can now compute the iterated dominance frontiers
2237          intersects with liveness incrementally in linear time! Woohoo!
2238          This is now used in my new SSA construction algorithm.
2239    
2240       b. THe branch reorganization module is now smarter about linear chains of
2241          basic blocks.
2242    
2243    
2244    ----------------------------------------------------------------------
2245    Name: Matthias Blume
2246    Date: 2000/04/12 13:52:00 JST
2247    Tag: blume_main_v110p27_1
2248    Description:
2249    
2250    Changed install.sh script to handle archive files without version number
2251    and to use "boot.<arch>-<os>" instead of "sml.boot.<arch>-<os>" for the
2252    name of the boot file archive.
2253    
2254    ----------------------------------------------------------------------
2255    Name: Dave MacQueen
2256    Date: 2000/04/09 14:00 EDT
2257    Tag: dbm-20000410-Version_110_27
2258    Description:
2259    
2260    1. Updated src/compiler/TopLevel/main/version.sml to version 110.27
2261    
2262    2. Updated src/config/version to 110.27
2263    
2264    3. New boot files!
2265    
2266    ----------------------------------------------------------------------
2267    Name: Allen Leung
2268    Date: 2000/04/09 19:09:00 EST
2269    Tag: leunga-20000409-misc
2270    Description:
2271    
2272    1.  Yet another fix for x86 assembly for idivl, imull, mull and friends.
2273    
2274    2.  Miscellaneous improvements to MLRISC (unused in sml/nj)
2275    
2276    ----------------------------------------------------------------------
2277    Name: Stefan
2278    Date: 2000/04/07 10:00:00 EDT
2279    Tag: monnier-20000406-branch-handling
2280    Description:
2281    
2282    Improved handling of branches (mostly those generated from
2283    polymorphic equality), removed switchoff and changed the
2284    default optimization settings (more cpsopt and less flintopt).
2285    
2286    ----------------------------------------------------------------------
2287    Name: Allen Leung
2288    Date: 2000/04/06 01:30:00 EST
2289    Tag: leunga-20000406-peephole-x86-SSA-2
2290    Description:
2291    
2292       Forgot a few files.
2293    
2294    ----------------------------------------------------------------------
2295    Name: Allen Leung
2296    Date: 2000/04/06 00:36:00 EST
2297    Tag: leunga-20000406-peephole-x86-SSA
2298    Description:
2299    
2300    1.  New Peephole code
2301    
2302    2.  Minor improvement to X86 instruction selection
2303    
2304    3.  Various fixes to SSA and machine description -> code translator
2305    
2306  ----------------------------------------------------------------------  ----------------------------------------------------------------------
2307  Name: Matthias Blume  Name: Matthias Blume
2308  Date: 2000/04/05 12:30:00 JST  Date: 2000/04/05 12:30:00 JST
# Line 67  Line 2359 
2359    
2360  3.  Assembly  3.  Assembly
2361    
2362      When generating assemby, resolve the value of client defined constants,      When generating assembly, resolve the value of client defined constants,
2363      instead of generating symbolic values.  This is controlled by the      instead of generating symbolic values.  This is controlled by the
2364      new flag "asm-resolve-constants", which is default to true.      new flag "asm-resolve-constants", which is default to true.
2365    
# Line 90  Line 2382 
2382    
2383  To this end, I arranged that instead of "structure Core" as "structure  To this end, I arranged that instead of "structure Core" as "structure
2384  _Core" is bound in the pervasive environment.  Core access is done via  _Core" is bound in the pervasive environment.  Core access is done via
2385  _Core (which can never be accidentially rebound because _Core is not a  _Core (which can never be accidentally rebound because _Core is not a
2386  legal surface-syntax symbol).  legal surface-syntax symbol).
2387    
2388  The current solution is much cleaner because the core environment is  The current solution is much cleaner because the core environment is
# Line 100  Line 2392 
2392  with dynamic and symbolic parts of the core environment.  with dynamic and symbolic parts of the core environment.
2393    
2394  Remaining hackery (to bind the "magic" symbol _Core) is localized in the  Remaining hackery (to bind the "magic" symbol _Core) is localized in the
2395  compilation mananger's bootstrap compiler (actually: in the "init group"  compilation manager's bootstrap compiler (actually: in the "init group"
2396  handling).  See the comments in src/system/smlnj/init/init.cmi for  handling).  See the comments in src/system/smlnj/init/init.cmi for
2397  more details.  more details.
2398    
# Line 215  Line 2507 
2507     (specified in the .cm file at each instance where the tool's class is     (specified in the .cm file at each instance where the tool's class is
2508     used).     used).
2509    
2510     This was done to accomodate the new "make" and "shell" tools which     This was done to accommodate the new "make" and "shell" tools which
2511     facilitate fairly seemless hookup to portions of code managed using     facilitate fairly seamless hookup to portions of code managed using
2512     Makefiles or Shell scripts.     Makefiles or Shell scripts.
2513    
2514     There are no classes "shared" or "private" anymore.  Instead, the     There are no classes "shared" or "private" anymore.  Instead, the
# Line 228  Line 2520 
2520    
2521     All existing tools are described in the CM manual.     All existing tools are described in the CM manual.
2522    
2523     - Slightly better error handling.  (CM now surpresses many followup     - Slightly better error handling.  (CM now suppresses many followup
2524     error messages that tended to be more annoying than helpful.)     error messages that tended to be more annoying than helpful.)
2525    
2526  2. Major changes to the compiler's static environment data structures.  2. Major changes to the compiler's static environment data structures.
# Line 362  Line 2654 
2654    
2655       I've changed andl to testl in the floating point test sequence       I've changed andl to testl in the floating point test sequence
2656       whenever appropriate.  The Intel optimization guide states that       whenever appropriate.  The Intel optimization guide states that
2657       testl is perferable to andl.       testl is preferable to andl.
2658    
2659  7. RA (x86 only)  7. RA (x86 only)
2660    
# Line 544  Line 2836 
2836    
2837  1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an  1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an
2838  additional argument called "template" which is an optional string that  additional argument called "template" which is an optional string that
2839  specifiel the layout of the tool command line.  See the CM manual for  specifies the layout of the tool command line.  See the CM manual for
2840  explanation.  explanation.
2841    
2842  2. A special-purpose tool can be "regisitered" by simply dropping the  2. A special-purpose tool can be "registered" by simply dropping the
2843  corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same  corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same
2844  directory where the .cm file lives that uses this tool.  (The  directory where the .cm file lives that uses this tool.  (The
2845  behavior/misfeature until now was to look for the tool description  behavior/misfeature until now was to look for the tool description
# Line 591  Line 2883 
2883  rebuild.  Having sets of the form <base><k>.{bin,boot}.<arch>-unix for  rebuild.  Having sets of the form <base><k>.{bin,boot}.<arch>-unix for
2884  <k>=1,2,... is normally not a good idea when invoking fixpt.  However,  <k>=1,2,... is normally not a good idea when invoking fixpt.  However,
2885  they might be the result of an earlier partial run of fixpt (which  they might be the result of an earlier partial run of fixpt (which
2886  perhaps got accidentially killed).  In this case, fixpt will quickly  perhaps got accidentally killed).  In this case, fixpt will quickly
2887  move through what exists before continuing where it left off earlier,  move through what exists before continuing where it left off earlier,
2888  and, thus, saves a lot of time.  and, thus, saves a lot of time.
2889    
# Line 641  Line 2933 
2933    it from that remote directory.    it from that remote directory.
2934    This should simplify installation further:  For machines that have    This should simplify installation further:  For machines that have
2935    access to the internet, just fetch <version>-config.tgz, unpack it,    access to the internet, just fetch <version>-config.tgz, unpack it,
2936    edit config/targets, and go (run config/install.sh).  The scipt will    edit config/targets, and go (run config/install.sh).  The script will
2937    fetch everything else that it might need all by itself.    fetch everything else that it might need all by itself.
2938    
2939    For CVS users, this mechanism is not relevant for source archives, but    For CVS users, this mechanism is not relevant for source archives, but

Legend:
Removed from v.595  
changed lines
  Added in v.873

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