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 601, Thu Apr 6 04:38:14 2000 UTC revision 841, Fri Jun 15 21:38:20 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: 2001/06/15 15:05:00 EDT
18    Tag: blume-20010615-moresparc
19    Description:
20    
21    Here is a short late-hour update related to Sparc c-calls:
22    
23     -- made handling of double-word arguments a bit smarter
24    
25     -- instruction selection phase tries to collapse certain clumsily
26        constructed ML-Trees; typical example:
27    
28            ADD(ty,ADD(_,e,LI d1),LI d2)  ->  ADD(ty,e,LI(d1+d2))
29    
30        This currently has no further impact on SML/NJ since mlriscGen does
31        not seem to generate such patterns in the first place, and c-calls
32        (which did generate them in the beginning) has meanwhile been fixed
33        so as to avoid them as well.
34    
35    ----------------------------------------------------------------------
36    Name: Matthias Blume
37    Date: 2001/06/15 15:05:00 EDT
38    Tag: blume-20010615-sparc
39    Description:
40    
41    The purpose of this update is to provide an implementation of NLFFI
42    on Sparc machines.
43    
44    Here are the changes in detail:
45    
46       * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
47       the Sparc implementation of the c-calls API.
48       * The Sparc backend of SML/NJ has been modified to uniformely use %fp
49       for accessing the ML frame.  Thus, we have a real frame pointer and
50       can freely modify %sp without need for an omit-frame-ptr phase.
51       The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
52       this case.
53       * ml-nlffigen has been taught to produce code for different architectures
54       and calling conventions.
55       * In a way similar to what was done in the x86 case, the Sparc
56       backend uses its own specific extension to mltree.  (For example,
57       it needs to be able to generate UNIMP instructions which are part
58       of the calling convention.)
59       * ml-nlffi-lib was reorganized to make it more modular (in particular,
60       to make it easier to plug in new machine- and os-dependent parts).
61    
62    There are some other fairly unrelated bug fixes and cleanups as well:
63    
64       * I further hacked the .cm files for MLRISC tools (like MDLGen) so
65       that they properly share their libraries with existing SML/NJ libraries.
66       * I fixed a minor cosmetic bug in CM, supressing certain spurious
67       follow-up error messages.
68       * Updates to CM/CMB documentation.
69    
70    TODO items:
71    
72       * MLRISC should use a different register as its asmTemp on the Sparc.
73         (The current %o2 is a really bad choice because it is part of the
74         calling conventions, so things might interfere in unexpected ways.)
75    
76    ----------------------------------------------------------------------
77    Name: Matthias Blume
78    Date: 2001/06/07
79    Tag: blume-20010607-calls
80    Description:
81    
82    A number of internal changes related to C calls and calling conventions:
83    
84    1. ML-Tree CALL statements now carry a "pops" field.  It indicates the
85       number of bytes popped implicitly (by the callee).  In most cases
86       this field is 0 but on x86/win32 it is some non-zero value.  This
87       is information provided for the benefit of the "omit-frameptr" pass.
88    2. The CALL instruction on the x86 carries a similar "pops" field.
89       The instruction selection phase copies its value from the ML-Tree
90       CALL statement.
91    3. On all other architectures, the instruction selection phase checks
92       whether "pops=0" and complains if not.
93    4. The c-calls implementation for x86 now accepts two calling conventions:
94       "ccall" and "stdcall".  When "ccall" is selected, the caller cleans
95       up after the call and pops is set to 0.  For "stdcall", the caller
96       does nothing, leaving the cleanup to the callee; pops is set to
97       the number of bytes that were pushed onto the stack.
98    5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
99       distinguish between "ccall" and "stdcall".
100    6. The UNIMP instruction has been added to the supported Sparc instruction
101       set. (This is needed for implementing the official C calling convention
102       on this architecture.)
103    7. I fixed some of the .cm files under src/MLRISC/Tools to make them
104       work with the latest CM.
105    
106    ----------------------------------------------------------------------
107    Name: Matthias Blume
108    Date: 2001/06/05 15:10:00 EDT
109    Tag: blume-20010605-cm-index
110    Description:
111    
112    0. The "lambdasplit" parameter for class "sml" in CM has been documented.
113    
114    1. CM can now generate "index files".  These are human-readable files
115       that list on a per-.cm-file basis each toplevel symbol defined or
116       imported.  The location of the index file for
117       <p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
118       To enable index-file generation, set CM.Control.generate_index to true
119       or export an environment-symbol: export CM_GENERATE_INDEX=true.
120    
121       The CM manual has been updated accordingly.
122    
123    2. I made some slight modifications to the c-calls API in MLRISC.
124    
125         a) There is now a callback to support saving/restoring of
126            dedicated but caller-save registers around the actual call
127            instruction.
128         b) One can optionally specify a comment-annotation for the
129            call instruction.
130    
131    3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
132       (For example, the comment annotation shows the C prototype of
133        the function being called.)
134    
135    ----------------------------------------------------------------------
136    Name: Matthias Blume
137    Date: 2001/06/01 13:30:00 EDT
138    Tag: blume-20010601-nlffi-cleanup
139    Description:
140    
141    This is mostly a cleanup of MLFFI stuff:
142    
143       - some signature files have been put into a more exposed place
144       - the ugly 'f type parameter is gone (simplifies types tremendously!)
145       - ml-nlffigen changed accordingly
146       - tutorial updated
147    
148    Other changes:
149    
150       - author's affiliation in CM manual(s) updated
151       - some more recognized keywords added to Allen's sml.sty
152    
153    ----------------------------------------------------------------------
154    Name: Matthias Blume
155    Date: 2001/05/25 15:30:00 EDT
156    Tag: blume-20010525-iptr
157    Description:
158    
159      - put the official 110.33-README (as it appears on the ftp server) under
160        CVS
161      - fixed a small bug related to incomplete pointer types in
162        ml-nlffigen
163      - small cosmetic change to the ml-nlffi-lib's "arr" type constructor
164        (it does not need the 'f type parameter)
165    
166    ----------------------------------------------------------------------
167    Name: Matthias Blume
168    Date: 2001/05/23 14:30:00 EDT
169    Tag: Release_110_33
170    Description:
171    
172    New version number (110.33).  New bootfiles.
173    
174    ----------------------------------------------------------------------
175    Name: Matthias Blume
176    Date: 2001/05/22 18:06:00 EDT
177    Tag: blume-20010522-targets
178    Description:
179    
180    Made install.sh use file config/targets.customized if it exists, falling
181    back to config/targets if it doesn't.  This way one can have a customized
182    version of the targets file without touching the "real thing", thus
183    eliminating the constant fear of accidentally checking something bogus
184    back into the CVS repository...  (File config/targets.customized must
185    not be added to the repository!)
186    
187    ----------------------------------------------------------------------
188    Name: Matthias Blume
189    Date: 2001/05/22 16:30:00 EDT
190    Tag: blume-20010522-minitut
191    Description:
192    
193    1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
194       struct returns.
195    2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt.  This is some very
196       incomplete, preliminary documentation for NLFFI.
197    
198    ----------------------------------------------------------------------
199    Name: Matthias Blume
200    Date: 2001/05/14 11:30:00 EDT
201    Tag: blume-20010514-script
202    Description:
203    
204    Some bugs in install script fixed.
205    
206    In addition to that I also made a slight change to the NLFFI API:
207    Functors generated by ml-nlffigen now take the dynamic library as a
208    straight functor argument, not as a suspended one.  (The original
209    functor code used to force the suspension right away anyway, so there
210    was nothing gained by this complication of the interface.)
211    
212    ----------------------------------------------------------------------
213    Name: Matthias Blume
214    Date: 2001/05/11 14:35:00 EDT
215    Tag: blume-20010511-ml-nlffi
216    Description:
217    
218    I finally took the plunge and added my new FFI code to the main
219    repository.  For x86-linux it is now ready for prime-time.
220    
221    There are two new subdirectories of "src":
222    
223      - ml-nlffi-lib:
224           The utility library for programs using the FFI interface.
225           Here is the implementation of $/c.cm and its associated low-level
226           partners $/c-int.cm and $/memory.cm.
227      - ml-nlffigen:
228           A stand-alone program for generating ML glue code from C source
229           code.
230    
231    Building ml-nlffigen requires $/ckit-lib.cm.
232    
233    The config/install.sh script has been updates to do the Right Thing
234    (hopefully).
235    
236    Notice that the source tree for the C-Kit will not be put under "src"
237    but directly under the installation root directory.  (This is the
238    structure that currently exists on the CVS server when you check out
239    module "sml".)  Fortunately, config/install.sh knows about this oddity.
240    
241    Bugs: No documentation yet.
242    
243    ----------------------------------------------------------------------
244    Name: Matthias Blume
245    Date: 2001/05/09 16:35:00 EDT
246    Tag: blume-20010509-cpscontract
247    Description:
248    
249    Fixed a bug in the accounting code in cpsopt/contract.sml.  (The
250    wrapper/unwrapper elimination did not decrement usage counts and some
251    dead variables got overlooked by the dead-up logic.)
252    
253    ----------------------------------------------------------------------
254    Name: Lal George
255    Date: 2001/05/08  17:26:09 EDT
256    Tag: george-20010508-omit-frameptr
257    Description:
258    
259    Changes to implement the omit-frame-pointer optimization to support
260    raw C calls. For now, there is only support on the Intel x86, but
261    other architectures will follow as more experience is gained with this.
262    
263    
264    ----------------------------------------------------------------------
265    Name: Matthias Blume
266    Date: 2001/05/07 14:40:00 EDT
267    Tag: blume-20010507-proxies
268    Description:
269    
270    I made into "proxy libraries" all libraries that qualify for such a
271    change.  (A qualifying library is a library that has another library or
272    groups as its sole member and repeats that member's export list
273    verbatim.  A proxy library avoids this repetition by omitting its export
274    list, effectively inheriting the list that its (only) member exports.
275    See the CM manual for more explanation.)
276    The main effect is that explicit export lists for these libraries
277    do not have to be kepts in sync, making maintenance a bit easier.
278    
279    I also added copyright notices to many .cm-files.
280    
281    Last but not least, I made a new set of bootfiles.
282    
283    ----------------------------------------------------------------------
284    Name: Matthias Blume
285    Date: 2001/05/04 17:00:00 EDT
286    Tag: blume-20010504-cm-lsplit
287    Description:
288    
289    0. John merged pending changes to $/smlnj-lib.cm
290    
291    1. Allen's previous change accidentally backed out of one of Lal's
292       earlier changes.  I undid this mistake (re-introducing Lal's change).
293    
294    2. I used the new topOrder' function from graph-scc.sml (from $/smlnj-lib.cm)
295       within the compiler where applicable.  There is some code simplification
296       because of that.
297    
298    3. The "split" phase (in FLINT) is now part of the default list of phases.
299       Compiler.Control.LambdaSplitting.* can be used to globally control the
300       lambda-splitting (cross-module-inlining) engine.  In addition to that,
301       it can now also be controlled on a per-source basis: CM has been taught
302       a new tool parameter applicable to ML source files.
303    
304       - To turn lambda-splitting off completely:
305            local open Compiler.Control.LambdaSplitting in
306                val _ = set Off
307            end
308       - To make "no lambda-splitting" the global default (but allow per-source
309         overriding); this is the initial setting:
310            local open Compiler.Control.LambdaSplitting in
311                val _ = set (Default NONE)
312            end
313       - To make "lambda-splitting with aggressiveness a" the global default
314         (and allow per-source overriding):
315            local open Compiler.Control.LambdaSplitting in
316                val _ = set (Default (SOME a))
317            end
318    
319       - To turn lambda-splitting off for a given ML souce file (say: a.sml)
320         write (in the respective .cm-file):
321            a.sml (lambdasplitting:off)
322       - To turn lambda-splitting for a.sml on with minimal aggressiveness:
323            a.sml (lambdasplitting:on)
324       - To turn lambda-splitting for a.sml on with aggressiveness <a> (where
325         <a> is a decimal non-negative integer):
326            a.sml (lambdasplitting:<a>)
327       - To turn lambda-splitting for a.sml on with maximal aggressiveness:
328            a.sml (lambdasplitting:infinity)
329       - To use the global default for a.sml:
330            a.sml (lambdasplitting:default)
331         or simply
332            a.sml
333    
334    ----------------------------------------------------------------------
335    Name: Allen Leung
336    Date: 2001/05/04 01:57:00 EDT
337    Tag: leunga-20010504-sync
338    Description:
339    
340      MLRISC features.
341    
342      1. Fix to CMPXCHG instructions.
343      2. Changed RA interface to allow annotations in callbacks.
344      3. Added a new method to the stream interface to allow annotations updates.
345    
346    ----------------------------------------------------------------------
347    Name: Matthias Blume
348    Date: 2001/05/01 11:45:00 EDT
349    Tag: blume-20010501-pcedittmp
350    Description:
351    
352    Changed install.sh to use the current working directory instead of
353    /usr/tmp for a temporary file (pcedittmp).  The previous choice
354    of /usr/tmp caused trouble with MacOS X because of file premission
355    problems.
356    
357    ----------------------------------------------------------------------
358    Name: Matthias Blume
359    Date: 2001/04/20 11:10:00 EDT
360    Tag: blume-20010420-inMLflag
361    Description:
362    
363     - added vp_limitPtrMask to vproc-state.h
364       (for use by the raw-C-calls mechanism to implement proper interrupt
365        handling)
366     - made the ML compiler aware of various data-structure offsets so it
367       can generate code for accessing the vp_inML flag and vp_limitPtrMask
368     - tweaked mlriscGen.sml to have it emit interrupt-handling code for
369       raw C-calls
370    
371    ----------------------------------------------------------------------
372    Name: Lal George
373    Date: 2001/04/20 09:15:28 EDT
374    Tag: george-20010420-macosX
375    Description:
376    
377     - Changes to port to Mac OS X; Darwin.
378    
379     - In the process I found that sqrt was broken on the PPC, because the
380       fsqrt instruction is not implemented.
381    
382    ----------------------------------------------------------------------
383    Name: Matthias Blume
384    Date: 2001/04/18 12:45:00 EDT
385    Tag: blume-20010418-ccalls
386    Description:
387    
388     - fixed two off-by-4 errors in the x86-specific c-calls implementation
389       (this bug prevented structure arguments containing pointers from being
390        passed correctly)
391     - changed the raw-C-call code in mlriscGen.sml in such a way that
392       structure arguments are represented as a pointer to the beginning
393       of the structure (instead of having a series of synthesized arguments,
394       one for each structure member)
395    
396     - made makeml script's verbosity level configurable via environment
397       variable (MAKEML_VERBOSITY)
398    
399     - eliminated placeholder implementations for f32l, w16s, i16s, and f32s
400       in rawmem-x86.sml; we are now using the real thing
401    
402    ----------------------------------------------------------------------
403    Name: Matthias Blume
404    Date: 2001/03/22 16:25:00 EST
405    Tag: blume-20010322-bootfiles
406    Description:
407    
408    Created a new set of bootfiles (for your automatic installation convenience).
409    
410    ----------------------------------------------------------------------
411    Name: Matthias Blume
412    Date: 2001/03/22 15:10:00 EST
413    Tag: blume-20010322-rawmem-parcm
414    Description:
415    
416    1. All "raw memory access" primitives for the new FFI are implemented now
417       (at least on the x86).
418    2. Some further cleanup of CM's parallel make mechanism.
419    
420    ----------------------------------------------------------------------
421    Name: Matthias Blume
422    Date: 2001/03/19 17:53:00 EST
423    Tag: blume-20010319-parallel
424    Description:
425    
426    Parallel make (using compile servers) now works again.
427    
428    To this end, CM.stabilize and CMB.make have been modified to work in
429    two passes when compile servers are attached:
430       1. Compile everything, do not perform stabilization; this pass
431          uses compile servers
432       2. Stabilize everything; this pass does not use compile servers
433    If there are no compile servers, the two passes are combined into one
434    (as before).  Splitting the passes increases the inherent parallelism
435    in the dependency graph because the entire graph including all
436    libraries is available at the same time.  This, in turn, improves
437    server utilization.  The downside is that the master process will
438    have to do some extra work after compilation is done (because for
439    technical reasons it must re-read all the binfiles during stabilization).
440    
441    ----------------------------------------------------------------------
442    Name: Matthias Blume
443    Date: 2001/03/16 12:22:00 EST
444    Tag: blume-20010316-bootfiles
445    Description:
446    
447    Created a new set of bootfiles (for your automatic installation convenience).
448    
449    ----------------------------------------------------------------------
450    Name: Matthias Blume
451    Date: 2001/03/16 11:00:00 EST
452    Tag: blume-20010316-MLTREE-fixup
453    Description:
454    
455    This is a minor fixup for an (untagged) earlier commit by Allen.
456    (A file was missing).
457    
458    ----------------------------------------------------------------------
459    Name: Allen Leung
460    Date: Mon Mar  5 18:54:57 EST 2001
461    Tag: leunga-20010305-cut-support
462    
463    1. New support for alternative control-flow in MLTREE.
464       Currently we support
465    
466          FLOW_TO(CALL ...., [k1,...,kn])
467    
468       This is needed for 'cuts to' in C-- and try/handle-like constructs
469       in Moby
470    
471       New assembler flag "asm-show-cutsto" to turn on control-flow debugging.
472    
473    2. Register Allocator
474    
475       Changes in interface [from Fermin, John]
476    
477    3. Alpha 8-bit SLL support [Fermin]
478    
479    4. All architectures
480    
481       A new module (ClusterExpandCopies) for expanding parallel copies.
482    
483    ----------------------------------------------------------------------
484    Name: Allen Leung
485    Date: 2001/02/27 23:07:00 EST
486    Tag: leunga-20010227-minor-stuff
487    
488    1. Alpha bug fix for CMOVNE
489    2. Handle mltree COND(..,FCMP ...,...)
490    3. Bug fix in simplifier
491    
492    ----------------------------------------------------------------------
493    Name: Matthias Blume
494    Date: 2001/01/30 17:50:00 EST
495    Tag: blume-20010130-sync
496    Description:
497    
498    This is just a minor update to sync my devel branch with the main brach.
499    The only visible change is the addition of some README files.
500    
501    ----------------------------------------------------------------------
502    Name: Matthias Blume
503    Date: 2001/01/12 23:30:00 JST
504    Tag: blume-20010112-bootfiles
505    Description:
506    
507    Made a new set of bootfiles that goes with the current state of the
508    repository.
509    
510    ----------------------------------------------------------------------
511    Name: Matthias Blume
512    Date: 2001/01/12 21:20:00 JST
513    Tag: blume-20010112-sync
514    Description:
515    
516    I am just flushing out some minor changes that had accumulated in
517    my private branch in order to sync with the main tree.  (This is
518    mainly because I had CVS trouble when trying to merge _into_ my
519    private branch.)
520    
521    Most people should be completely unaffected by this.
522    
523    ----------------------------------------------------------------------
524    Name: Allen Leung
525    Date: Thu Jan 11 21:03:00 EST 2001
526    Tag: leunga-20010111-labexp=mltree
527    Description:
528    
529    1.  Removed the type LabelExp and replace it by MLTree.
530    2.  Rewritten mltree-simplify with the pattern matcher tool.
531    3.  There were some bugs in alpha code generator which would break
532        64-bit code generation.
533    4.  Redo the tools to generate code with the
534    5.  The CM files in MLRISC (and in src/system/smlnj/MLRISC)
535        are now generated by perl scripts.
536    
537    ----------------------------------------------------------------------
538    Name: Matthias Blume
539    Date: 2001/01/10 21:55:00 JST
540    Tag: blume-20010110-rcc
541    Description:
542    
543    The RCC stuff now seems to work (but only on the x86).
544    This required hacking of the c-calls interface (and -implementation) in
545    MLRISC.
546    
547    Normal compiler users should be unaffected.
548    
549    ----------------------------------------------------------------------
550    Name: Matthias Blume
551    Date: 2001/01/09 01:20:00 JST
552    Tag: blume-20010109-rcc
553    Description:
554    
555    This is a fairly big patch, flushing out a large number of pending
556    changes that I made to my development copy over the last couple of days.
557    
558    Of practical relevance at this moment is a workaround for a pickling
559    bug that Allen ran into the other day.  The cause of the bug itself is
560    still unknown and it might be hard to fix it properly, but the
561    workaround has some merits of its own (namely somewhat reducing pickling
562    overhead for certain libraries).  Therefore, I think this solution should
563    be satisfactory at this time.
564    
565    The rest of the changes (i.e., the vast majority) has to do with my
566    ongoing efforts of providing direct support for C function calls from
567    ML.  At the moment there is a new primop "RAW_CCALL", typing magic
568    in types/cproto.sml (invoked from FLINT/trans/translate.sml), a new
569    case in the FLINT CPS datatype (RCC), changes to cps/convert.sml to
570    translate uses of RAW_CCALL into RCC, and changes to mlriscGen.sml to
571    handle RCC.
572    
573    The last part (the changes to mlriscGen.sml) are still known to be
574    wrong on the x86 and not implemented on all other architectures.  But
575    the infrastructure is in place. I had to change a few functor
576    signatures in the backend to be able to route the CCalls interface
577    from MLRISC there, and I had to specialize the mltree type (on the
578    x86) to include the necessary extensions. (The extensions themselves
579    were already there and redy to go in MLRISC/x86).
580    
581    Everything should be very happy as soon as someone helps me with
582    mlriscGen.sml...
583    
584    In any case, nothing of this should matter to anyone as long as the
585    new primop is not being used (which is going to be the case unless you
586    find it where I hid it :). The rest of the compiler is completely
587    unaffected.
588    
589    ----------------------------------------------------------------------
590    Name: Matthias Blume
591    Date: 2001/01/05 00:30:00 JST
592    Tag: blume-20010105-primops
593    Description:
594    
595    Added some experimental support for work that I am doing right now.
596    These changes mostly concern added primops, but there is also a new
597    experimental C library in the runtime system (but currently not enabled
598    anywhere except on Linux/X86).
599    
600    In the course of adding primops (and playing with them), I discovered that
601    Zhong's INL_PRIM hack (no type info for certain primops) was, in fact, badly
602    broken.  (Zhong was very right he labeled this stuff as "major gross hack".)
603    To recover, I made type information in INL_PRIM mandatory and changed
604    prim.sml as well as built-in.sml accordingly.  The InLine structure now
605    has complete, correct type information (i.e., no bottom types).
606    
607    Since all these changes mean that we need new binfiles, I also bumped the
608    version number to 110.32.1.
609    
610    ----------------------------------------------------------------------
611    Name: Matthias Blume
612    Date: 2000/12/30 22:10:00 JST
613    Tag: blume-20001230-various
614    Description:
615    
616    Added proxy libraries for MLRISC and let MLRISC libraries refer
617    to each other using path anchors.  (See CM manual for explanation.)
618    
619    Updated CM documentation.
620    
621    Fixed some bugs in CM.
622    
623    Implemented "proxy" libraries (= syntactic sugar for CM).
624    
625    Added "-quiet" option to makeml and changed runtime system accordingly.
626    
627    Added cleanup handler for exportML to reset timers and compiler stats.
628    
629    ----------------------------------------------------------------------
630    Name: Lal George
631    Date: 2000/12/22 22:22:58 EST 2000
632    Tag: Release_110_32
633    Description:
634    
635            Infinite precision used throughout MLRISC.
636            see MLRISC/mltree/machine-int.sig
637    
638    ----------------------------------------------------------------------
639    Name: Matthias Blume
640    Date: 2000/12/22 23:16:00 JST
641    Tag: blume-20001222-warn
642    Description:
643    
644    Corrected wording and formatting of some CM warning message which I
645    broke in my previous patch.
646    
647    ----------------------------------------------------------------------
648    Name: Matthias Blume
649    Date: 2000/12/22 21:20:00 JST
650    Tag: blume-20001222-anchorenv
651    Description:
652    
653    Fixed CM's handling of anchor environments in connection with CMB.make.
654    
655    ----------------------------------------------------------------------
656    Name: Matthias Blume
657    Date: 2000/12/22 13:15:00 JST
658    Tag: blume-20001222-cleanup
659    Description:
660    
661    Removed src/cm/ffi which does not (and did not) belong here.
662    
663    ----------------------------------------------------------------------
664    Name: Matthias Blume
665    Date: 2000/12/21 23:55:00 JST
666    Tag: blume-20001221-exn
667    Description:
668    
669    Probably most important: CM no longer silently swallows all exceptions
670    in the compiler.
671    Plus: some other minor CM changes.  For example, CM now reports some
672    sizes for generated binfiles (code, data, envpickle, lambdapickle).
673    
674    ----------------------------------------------------------------------
675    Name: Matthias Blume
676    Date: 2000/12/15 00:01:05 JST
677    Tag: blume-20001215-dirtool
678    Description:
679    
680    - "dir" tool added.
681    - improvements and cleanup to Tools structure
682    - documentation updates
683    
684    ----------------------------------------------------------------------
685    Name: Allen Leung
686    Date: Thu Dec 14 03:45:24 EST 2000
687    Description:
688    Tag:  leunga-20001214-int-inf
689    Description:
690    
691       In IntInf, added these standard functions, which are missing from our
692    implementation:
693    
694        andb : int * int -> int
695        xorb : int * int -> int
696        orb  : int * int -> int
697        notb : int -> int
698         <<   : int * word -> int
699        ~>>  : int * word -> int
700    
701       Not tested, I hope they are correct.
702    
703    ----------------------------------------------------------------------
704    Name: Allen Leung
705    Date: Fri Dec  8 19:23:26 EST 2000
706    Description:
707    Tag:  leunga-20001208-nowhere
708    Description:
709    
710      Slight improvements to the 'nowhere' tool to handle OR-patterns,
711    to generate better error messages etc.  Plus a brief manual.
712    
713    ----------------------------------------------------------------------
714    Name: Lal George
715    Date: 2000/12/08 09:54:02 EST 2000
716    Tag: Release_110_31
717    Description:
718    
719    - Version 110.31
720    ----------------------------------------------------------------------
721    Name: Allen Leung
722    Date: Thu Dec  7 22:01:04 EST 2000
723    Tag:  leunga-20001207-cell-monster-hack
724    Description:
725    
726    Major MLRISC internal changes.  Affect all clients.
727    Summary:
728    
729    1.  Type CELLS.cell = int is now replaced by a datatype.
730        As a result, the old regmap is now gone.  Almost all interfaces
731        in MLRISC change as a consequence.
732    
733    2.  A new brand version of machine description tool (v3.0) that generates
734        modules expecting the new interface.  The old version is removed.
735    
736    3.  The RA interface has been further abstracted into two new functors.
737        RISC_RA and X86RA.  These functors have much simpler interfaces.
738        [See also directory MLRISC/demo.]
739    
740    4.  Some other new source->source code generation tools are available:
741    
742        a. MLRISC/Tools/RewriteGen -- generate rewriters from rules.
743        b. MLRISC/Tools/WhereGen -- expands conditional pattern matching rules.
744           I use this tool to generate the peephole optimizers---with the new
745           cell type changes, peephole rules are becoming difficult to write
746           without conditional pattern matching.
747    
748    5.  More Intmap -> IntHashTable change.  Previous changes by Matthias didn't
749        cover the entire MLRISC source tree so many things broke.
750    
751    6.  CM files have been moved to the subdirectory MLRISC/cm.
752        They are moved because there are a lot of them and they clutter up the
753        root dir.
754    
755    7.  More detailed documentation to come...
756    
757        NOTE: To rebuild from 110.30 (ftp distribution), you'll have to do
758        a makeml -rebuild first.  This is because of other other
759        changes that Matthias has made (see below).
760    
761    
762    ----------------------------------------------------------------------
763    Name: Matthias Blume
764    Date: 2000/11/30 23:12:00 JST
765    Tag: blume-20001130-filereorg
766    Description:
767    
768    Some manual updates and some file reorganizations in CM.
769    
770    ----------------------------------------------------------------------
771    Name: Matthias Blume
772    Date: 2000/11/24 17:45:00 JST
773    Tag: blume-20001124-link
774    Description:
775    
776    Drastically improved link traversal code for the case that the dynamic
777    value was already loaded at bootstrap time.  As a result, CM and CMB
778    now both load blazingly fast -- even on a very slow machine.  Also,
779    memory consumption has been further reduced by this.
780    
781    Warning: The format of the PIDMAP file has changed.  THerefore, to
782    bootstrap you have to do this:
783    
784    1. Run CMB.make
785    2. Make a symbolic link for the boot directory:
786         ln -s sml.boot.ARCH-OS xxx
787    3. "Rebuild" the boot directory:
788         ./makeml -boot xxx -rebuild sml ; rm xxx
789    4. Boot normally:
790          ./makeml
791    
792    ----------------------------------------------------------------------
793    Name: Matthias Blume
794    Date: 2000/11/21 21:20:00 JST
795    Tag: blume-20001121-tools
796    Description:
797    
798    Continued hacking on autoloading problem -- with success this time.
799    Also changed tool-plugin mechanism.  See new CM manual.
800    
801    ----------------------------------------------------------------------
802    Name: Matthias Blume
803    Date: 2000/11/19 14:30:00 JST
804    Tag:  blume-20001119-autoload
805    Description:
806    
807    Some hacking to make autoloading faster.  Success for CMB, no success
808    so far for CM.  There is a reduced structure CM' that autoloads faster.
809    (This is a temporary, non-documented hack to be eliminated again when
810    the general problem is solved.)
811    
812    ----------------------------------------------------------------------
813    Name: Matthias Blume
814    Date: 2000/11/17 14:10:00 JST
815    Tag: blume-20001117-pickle-lib
816    Description:
817    
818    1. Eliminated comp-lib.cm
819    2. Made pickle-lib.cm
820    3. Eliminated all uses of intset.sml (from comp-lib.cm)
821    4. Replaced all uses of intmap.{sig,sml} (from comp-lib.cm) with
822       equivalent constructs from smlnj-lib.cm (INtHashTable).
823    5. Point 4. also goes for those uses of intmap.* in MLRISC.
824       Duplicated intmap modules thrown out.
825    6. Hunted down all duplicated SCC code and replaced it with
826       equivalent stuff (GraphSCCFn from smlnj-lib.cm).
827    7. Rewrote Feedback module.
828    8. Moved sortedlist.sml into viscomp-lib.cm.  Eventually it
829       should be thrown out and equivalent modules from smlnj-lib.cm
830       should be used (IntRedBlackSet, IntListSet, ...).
831    
832    Confirmed that compiler compiles to fixpoint.
833    
834    ----------------------------------------------------------------------
835    Name: Allen Leung
836    Date: 2000/11/10 18:00:00
837    Tag: leunga-20001110-new-x86-fp
838    
839    A new x86 floating point code generator has been added.
840    By default this is turned off.  To turn this on, do:
841    
842        CM.autoload "$smlnj/compiler.cm";
843        Compiler.Control.MLRISC.getFlag "x86-fast-fp" := true;
844    
845    Changes:
846    
847    1.  Changed FTAN to FPTAN so that the assembly output is correct.
848    2.  Changed the extension callback for FTANGENT to generate:
849    
850              fptan
851              fstp  %st(0)
852        instead of
853              fptan
854              fstpl ftempmem
855    
856    3.  Numerous assembly fixes for x86.
857    
858    5.  Cleaned up the machine code output module x86/x86MC.sml and added
859        support for a whole bunch of instructions and addressing modes:
860    
861          fadd/fsub/fsubr/fmul/fdiv/fdivr  %st, %st(n)
862          faddp/fsubp/fsubrp/fmulp/fdivp/fdivrp  %st, %st(n)
863          fadd/fsub/fsubr/fmul/fdiv/fdivr  %st(n), %st
864          fiadd/fisub/fisubr/fimul/fidiv/fidivr mem
865          fxch %st(n)
866          fld %st(n)
867          fst %st(n)
868          fst mem
869          fstp %st(n)
870          fucom %st(n)
871          fucomp %st(n)
872    
873        All these are now generated when the fast fp mode is turned on.
874    
875    6.  Removed the dedicated registers %st(0), ..., %st(7) from X86CpsRegs
876    
877    ----------------------------------------------------------------------
878    Name: Matthias Blume
879    Date: 2000/11/09 11:20:00 JST
880    Tag: blume-20001109-scc
881    Description:
882    
883    Eliminated some code duplication:
884    
885    1. Added "where" clause to GraphSCCFn in SML/NJ Library.
886       (Otherwise the functor is useless.)
887    2. Used GraphSCCFn where SCCUtilFun was used previously.
888    3. Got rid of SCCUtilFun (in comp-lib.cm).
889    
890    ----------------------------------------------------------------------
891    Name: Lal George
892    Date: 2000/11/06 09:02:21 EST 2000
893    Tag: Release_110_30
894    Description:
895    
896    - Version 110.30
897    ----------------------------------------------------------------------
898    Name: Matthias Blume
899    Date: 2000/11/04 14:45:00
900    Tag: blume-20001104-mlbuild
901    Description:
902    
903    - Made ml-build faster on startup.
904    - Documentation fixes.
905    
906    ----------------------------------------------------------------------
907    Name: Matthias Blume
908    Date: 2000/11/02 17:00:00 JST
909    Tag: blume-20001102-condcomp
910    Description:
911    
912    - Small tweaks to pickler -- new BOOTFILES!
913    - Version bumped to 110.29.2.
914    - Added conditional compilation facility to init.cmi (see comment there).
915    ----------------------------------------------------------------------
916    Name: Allen Leung
917    Date: 2000/10/23 19:31:00
918    Tag: leunga-20001023-demo-ra
919    
920    1. Minor RA changes that improves spilling on x86 (affects Moby and C-- only)
921    2. Test programs for the graph library updated
922    3. Some new MLRISC demo programs added
923    
924    ----------------------------------------------------------------------
925    Name: Matthias Blume
926    Date: 2000/08/31 22:15:00 JST
927    Tag: blume-20001017-errmsg
928    Description:
929    
930    More error message grief: Where there used to be no messages, there
931    now were some that had bogus error regions.  Fixed.
932    
933    ----------------------------------------------------------------------
934    Name: Matthias Blume
935    Date: 2000/08/31 17:30:00 JST
936    Tag: blume-20001017-v110p29p1
937    Description:
938    
939    I made a version 110.29.1 with new bootfiles.
940    
941    Changes:  Modified pickler/unpickler for faster and leaner unpickling.
942              CM documentation changes and a small bugfix in CM's error reporting.
943    
944    ----------------------------------------------------------------------
945    Name: Lal George
946    Date: 2000/09/27 14:42:35 EDT
947    Tag: george-20000927-nodestatus
948    Description:
949    
950    Changed the type of the nodestatus, so that:
951    
952            SPILLED(~1)             is now SPILLED
953            SPILLED(m) where m>=0   is now MEMREG(m)
954            SPILLED(s) where s<~1   is now SPILL_LOC(~s)
955    
956    ----------------------------------------------------------------------
957    Name: Matthias Blume
958    Date: 2000/09/07 14:45:00 JST
959    Tag: blume-20000907-cmerrmsg
960    Description:
961    
962    Small tweak to CM to avoid getting ML syntax error messages twice.
963    
964    ----------------------------------------------------------------------
965    Name: Matthias Blume
966    Date: 2000/08/31 18:00:00 JST
967    Tag: blume-20000831-cvsbootfiles
968    Description:
969    
970    New URL for boot files (because the 110.29 files on the BL server do
971    now work correctly with my updated install scripts for yacc and lex).
972    
973    ----------------------------------------------------------------------
974    Name: Matthias Blume
975    Date: 2000/08/08 12:33:00 JST
976    Tag: blume-20000808-manual
977    Description:
978    
979    Tiny update to CM manual.
980    
981    ----------------------------------------------------------------------
982    Name: Allen Leung
983    Date: 2000/08/7 19:31:00
984    Tag: leunga-20000807-a-whole-bunch-of-stuff
985    
986      Moby, C--, SSA, x86, machine descriptions etc.  Should only affect C--
987    and Mobdy.
988    
989    1.  x86
990    
991       a.  Fixes to peephole module by John and Dan.
992       b.  Assembly fix to SETcc by Allen.
993       c.  Fix to c-call by John.
994       d.  Fix to spilling by John.  (This one deals with the missing FSTPT case)
995       e.  Instruction selection optimization to SETcc as suggested by John.
996    
997           For example,
998    
999            MV(32, x, COND(32, CMP(32, LT, a, b), LI 1, LI 0))
1000    
1001           should generate:
1002    
1003            MOVL a, x
1004            SUBL b, x
1005            SHRL 31, x
1006    
1007    2.  IR stuff
1008    
1009         A bunch of new DJ-graph related algorithms added.  These
1010         speed up SSA construction.
1011    
1012    3.  SSA + Scheduling
1013    
1014         Added code for SSA and scheduling to the repository
1015    
1016    ----------------------------------------------------------------------
1017    Name: Lal George
1018    Date: 2000/07/27 11:53:14 EDT
1019    
1020    Tag: lal-20000727-linux-ppc
1021    Description:
1022    
1023     Made changes to support Linux PPC.
1024     p.s. I have confirmation that the 110.29 boot files work fine.
1025    
1026    ----------------------------------------------------------------------
1027    Name: Matthias Blume
1028    Date: 2000/07/27 17:40:00 JST
1029    Tag: blume-20000727-scripts
1030    Description:
1031    
1032    !!!! WARNING !!!!
1033    You must recompile the runtime system!
1034    !!!! WARNING !!!!
1035    
1036    This is basically another round of script-enhancements:
1037    
1038    1. sml, ml-build, and ml-makedepend accept options -D and -U to define
1039       and undefine CM preprocessor symbols.
1040    
1041    2. ml-build avoids generating a new heap image if it finds that the
1042       existing one is still ok.  (The condition is that no ML file had to
1043       be recompiled and all ML files are found to be older that the heap
1044       file.)
1045    
1046       To make this work smoothly, I also hacked the runtime system as
1047       well as SMLofNJ.SysInfo to get access to the heap image suffix
1048       (.sparc-solaris, ...) that is currently being used.
1049    
1050       Moreover, the signature of CM.mk_standalone has changed.  See the
1051       CM manual.
1052    
1053    3. ml-makedepend accepts additional options -n, -a, and -o.  (See the
1054       CM manual for details.)
1055    
1056    4. More CM manual updates:
1057        - all of the above has been documented.
1058        - there is now a section describing the (CM-related) command line
1059          arguments that are accepted by the "sml" command
1060    
1061    ----------------------------------------------------------------------
1062    Name: Matthias Blume
1063    Date: 2000/07/25 16:20:00 JST
1064    Tag: blume-20000725-makedepend
1065    Description:
1066    
1067    Added a script called ml-makedepend.  This can be used in makefiles
1068    for Unix' make in a way very similar to the "makedepend" command for
1069    C.
1070    
1071    The script internally uses function CM.sources.
1072    
1073    Synopsis:
1074    
1075        ml-makedepend [-f makefile] cmfile targetname
1076    
1077    The default for the makefile is "makefile" (or "Makefile" should
1078    "makefile" not exist).
1079    
1080    ml-makedepend adds a cmfile/targetname-specific section to this
1081    makefile (after removing the previous version of this section).  The
1082    section contains a single dependency specification with targetname on
1083    the LHS (targetname is an arbitrary name), and a list of files derived
1084    from the cmfile on the RHS.  Some of the files on the RHS are
1085    ARCH/OPSYS-specific.  Therefore, ml-makedepend inserts references to
1086    "make" variables $(ARCH) and $(OPSYS) in place of the corresponding
1087    path names.  The makefile writer is responsible for making sure that
1088    these variables have correct at the time "make" is invoked.
1089    
1090    ----------------------------------------------------------------------
1091    Name: Matthias Blume
1092    Date: 2000/07/22 23:30:00 JST
1093    Tag: blume-20000722-urlupdate
1094    Description:
1095    
1096    Changed BOOT and config/srcarchiveurl to point to BL server:
1097    
1098        ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.29/
1099    
1100    ----------------------------------------------------------------------
1101    Name: Matthias Blume
1102    Date: 2000/07/18 18:00:00 JST
1103    Tag: blume-20000718-Version_110_29
1104    Description:
1105    
1106    1. Updated src/compiler/TopLevel/main/version.sml to version 110.29
1107    
1108    2. Updated config/version to 110.29
1109    
1110    3. Updated config/srcarchiveurl
1111    
1112    3. New boot files!
1113       ftp://ftp.cs.princeton.edu/pub/people/blume/sml/110.29-autofetch
1114    
1115    ----------------------------------------------------------------------
1116    Name: Matthias Blume
1117    Date: 2000/07/11 13:58:00 JST
1118    Tag: blume-20000711-doctypo
1119    Description:
1120    
1121    Fixed a few typos in CM manual.
1122    
1123    ----------------------------------------------------------------------
1124    Name: Allen Leung
1125    Date: 2000/06/15 00:38:00
1126    Tag: leunga-20000704-sparc-x86
1127    
1128    1. x86 peephole improvement sp += k; sp -= k => nop  [from John]
1129    2. fix to x86 RET bug [found by Dan Grossman]
1130    3. sparc assembly bug fix for ticc instructions [found by Fermin]
1131    
1132       Affects c-- and moby only
1133    
1134    ----------------------------------------------------------------------
1135    Name: Matthias Blume
1136    Date: 2000/07/04 15:26:00
1137    Tag: blume-20000704-trigger
1138    Description:
1139    
1140    1. Improvements to CM manual.
1141    2. SMLofNJ.Internals.BTrace.trigger reinstated as an alternative way
1142       of getting a back-trace.  The function, when called, raises an
1143       internal exception which explicitly carries the full back-trace history,
1144       so it is unaffected by any intervening handle-raise pairs ("trivial"
1145       or not).  The interactive loop will print that history once it arrives
1146       at top level.
1147       Short of having all exceptions implicitly carry the full history, the
1148       recommended way of using this facility is:
1149         - compile your program with instrumentation "on"
1150         - run it, when it raises an exception, look at the history
1151         - if the history is "cut off" because of some handler, go and modify
1152           your program so that it explicitly calls BTrace.trigger
1153         - recompile (still instrumented), and rerun; look at the full history
1154    
1155    ----------------------------------------------------------------------
1156    Name: Matthias Blume
1157    Date: 2000/07/03 15:36:00 JST
1158    Tag: blume-20000702-manual
1159    Description:
1160    
1161    Small corrections and updates to CM manual.
1162    
1163    ----------------------------------------------------------------------
1164    Name: Matthias Blume
1165    Date: 2000/06/29 16:04:00 JST
1166    Tag: blume-20000629-yacctool
1167    Description:
1168    
1169    Changes:
1170    
1171    1. Class "mlyacc" now takes separate arguments to pass options to
1172       generated .sml- and .sig-files independently.
1173    2. Corresponding CM manual updates.
1174    3. BTrace module now also reports call sites.  (However, for loop clusters
1175       it only shows from where the cluster was entered.)  There are associated
1176       modifications to core.sml, internals.{sig,sml}, btrace.sml, and btimp.sml.
1177    
1178    ----------------------------------------------------------------------
1179    Name: Matthias Blume
1180    Date: 2000/06/27 16:51:00 JST
1181    Tag: blume-20000627-noweb
1182    Description:
1183    
1184    Changes:
1185    
1186     1. Implemented "subdir" and "witness" options for noweb tool.
1187        This caused some slight internal changes in CM's tool implementation.
1188     2. Fixed bug in "tool plugin" mechanism.  This is essentially cleaning
1189        some remaining issues from earlier path anchor changes.
1190     3. Updated CM manual accordingly.
1191    
1192     4. Changed implementation of back-tracing so that I now consider it
1193        ready for prime-time.
1194    
1195        In particular, you don't have to explicitly trigger the back-trace
1196        anymore.  Instead, if you are running BTrace-instrumented code and
1197        there is an uncaught exception (regardless of whether or not it was
1198        raised in instrumented code), the top-level evalloop will print
1199        the back-trace.
1200    
1201        Features:
1202    
1203          - Instrumented and uninstrumented code work together seemlessly.
1204            (Of course, uninstrumented code is never mentioned in actual
1205             back-traces.)
1206    
1207          - Asymptotic time- and space-complexity of instrumented code is
1208            equal to that of uninstrumented code.  (This means that
1209            tail-recursion is preserved by the instrumentation phase.)
1210    
1211          - Modules whose code has been instrumented in different sessions
1212            work together without problem.
1213    
1214          - There is no penalty whatsoever on uninstrumented code.
1215    
1216          - There is no penalty on "raise" expressions, even in
1217            instrumented code.
1218    
1219        A potential bug (or perhaps it is a feature, too):
1220    
1221          A back-trace reaches no further than the outermost instrumented
1222          non-trivial "raise".  Here, a "trivial" raise is one that is the
1223          sole RHS of a "handle" rule.  Thus, back-traces reach trough
1224    
1225               <exp> handle e => raise e
1226    
1227          and even
1228    
1229               <exp> handle Foo => raise Bar
1230    
1231          and, of course, through
1232    
1233               <exp> handle Foo => ...
1234    
1235         if the exception was not Foo.
1236    
1237         Back-traces always reach right through any un-instrumented code
1238         including any of its "handle" expressions, trivial or not.
1239    
1240       To try this out, do the following:
1241    
1242         - Erase all existing binfiles for your program.
1243           (You may keep binfiles for those modules where you think you
1244            definitely don't need back-tracing.)
1245         - Turn on back-trace instrumentation:
1246              SMLofNJ.Internals.BTrace.mode (SOME true);
1247         - Recompile your program.  (I.e., run "CM.make" or "use".)
1248         - You may now turn instrumentation off again (if you want):
1249              SMLofNJ.Internals.BTrace.mode (SOME false);
1250         - Run your program as usual.  If it raises an exception that
1251           reaches the interactive toplevel, then a back-trace will
1252           automatically be printed.  After that, the toplevel loop
1253           will print the exception history as usual.
1254    
1255    ----------------------------------------------------------------------
1256    Name: Matthias Blume
1257    Date: 2000/06/26 09:56:46 JST
1258    Tag: blume-20000626-setup
1259    Description:
1260    
1261    CM: - setup-parameter to "sml" added; this can be used to run arbitrary
1262          ML code before and after compiling a file (e.g., to set compiler
1263          flags)
1264    
1265    Compiler: - improved btrace API (in core.sml, internals.{sig,sml})
1266              - associated changes to btrace.sml (BTrace instrumentation pass)
1267              - cleaner implementation of btimp.sml (BTrace tracing and report
1268                module)
1269    
1270    CM manual: * new path encoding documented
1271               * description of setup-parameter to "sml" added
1272    
1273    The biggest user-visible change to back-tracing is that it is no
1274    longer necessary to compile all traced modules within the same
1275    session.  (This was a real limitation.)
1276    
1277    ----------------------------------------------------------------------
1278    Name: Matthias Blume
1279    Date: 2000/06/24 12:40:00 JST
1280    Tag: blume-20000624-startup
1281    Description:
1282    
1283    Fixes startup slowdown problem.  (I was calling SrcPath.sync a _tad_
1284    bit too often -- to put it mildly. :)
1285    
1286    ----------------------------------------------------------------------
1287    Name: Matthias Blume
1288    Date: 2000/06/23 18:20:00 JST
1289    Tag: blume-20000623-btrace
1290    Description:
1291    
1292    This updates adds a backtrace facility to aid programmers in debugging
1293    their programs.  This involves the following changes:
1294    
1295    1. Module system/smlnj/init/core.sml (structure _Core) now has hooks for
1296       keeping track of the current call stack.  When programs are compiled
1297       in a special mode, the compiler will insert calls to these hooks
1298       into the user program.
1299       "Hook" means that it is possible for different implementations of
1300       back-tracing to register themselves (at different times).
1301    
1302    2. compiler/MiscUtil/profile/btrace.sml implements the annotation phase
1303       as an Absyn.dec->Absyn.dec rewrite.  Normally this phase is turned off.
1304       It can be turned on using this call:
1305         SMLofNJ.Internals.BTrace.mode (SOME true);
1306       Turning it off again:
1307         SMLofNJ.Internals.BTrace.mode (SOME false);
1308       Querying the current status:
1309         SMLofNJ.Internals.BTrace.mode NONE;
1310       Annotated programs are about twice as big as normal ones, and they
1311       run a factor of 2 to 4 slower with a dummy back-trace plugin (one
1312       where all hooks do nothing).  The slowdown with a plugin that is
1313       actually useful (such as the one supplied by default) is even greater,
1314       but in the case of the default plugin it is still only an constant
1315       factor (amortized).
1316    
1317    3. system/Basis/Implementation/NJ/internals.{sig,sml} have been augmented
1318       with a sub-structure BTrace for controlling back-tracing.  In particular,
1319       the above-mentioned function "mode" controls whether the annotation
1320       phase is invoked by the compiler.  Another important function is
1321       "trigger": when called it aborts the current execution and causes
1322       the top-level loop to print a full back-trace.
1323    
1324    4. compiler/MiscUtil/profile/btimp.sml is the current default plugin
1325       for back-tracing.  It keeps track of the dynamic call stack and in
1326       addition to that it keeps a partial history at each "level" of that
1327       stack.  For example, if a tail-calls b, b tail-calls c, and c tail-calls
1328       d and b (at separate times, dynamically), then the report will show:
1329    
1330       GOTO   d
1331             /c
1332       GOTO  \b
1333       CALL   a
1334    
1335       This shows that there was an initial non-tail call of a, then a
1336       tail-call to b or c, looping behavior in a cluster of functions that
1337       consist of b and c, and then a goto from that cluster (i.e., either from
1338       b or from c) to d.
1339    
1340       Note that (depending on the user program) the amount of information
1341       that the back-trace module has to keep track of at each level is bounded
1342       by a constant.  Thus, the whole implementation has the same asymptotical
1343       complexity as the original program (both in space and in time).
1344    
1345    5. compiler/TopLevel/interact/evalloop.sml has been modified to
1346       handle the special exception SMLofNJ.Internals.BTrace.BTrace
1347       which is raised by the "trigger" function mentioned above.
1348    
1349    Notes on usage:
1350    
1351    - Annotated code works well together with unannotated code:
1352    Unannotated calls simply do not show up at all in the backtrace.
1353    
1354    - It is not a good idea to let modules that were annotated during
1355    different sessions run at the same time.  This is because the compiler
1356    chooses small integers to identify individual functions, and there
1357    will be clashes if different modules were compiled in separate sessions.
1358    (Nothing will crash, and you will even be told about the clashes, but
1359    back-trace information will in general not be useful.)
1360    
1361    - Back-tracing can be confused by callcc and capture.
1362    
1363    - The only way of getting a back-trace right now is to explicitly
1364    invoke the "trigger" function from your user program.  Eventually, we
1365    should make every exception carry back-trace information (if
1366    available).  But since this creates more overhead at "raise"-time
1367    (similar to the current exnHistory overhead), I have not yet
1368    implemented this.  (The implementation will be rather easy.)  With
1369    exceptions carrying back-trace information, this facility will be even
1370    more useful because users don't need to modify their programs...
1371    
1372    - While it is possible to compile the compiler with back-trace
1373    annotations turned on (I did it to get some confidence in
1374    correctness), you must make absolutely sure that core.sml and
1375    btimp.sml are compiled WITHOUT annotation!  (core.sml cannot actually
1376    be compiled with annotation because there is no core access yet, but
1377    if you compile btimp.sml with annotation, then the system will go into
1378    an infinite recursion and crash.)
1379    Since CM currently does not know about BTrace, the only way to turn
1380    annotations on and off for different modules of the compiler is to
1381    interrupt CMB.make, change the settings, and re-invoke it.  Of course,
1382    this is awkward and clumsy.
1383    
1384    Sample sessions:
1385    
1386    Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
1387    - SMLofNJ.Internals.BTrace.mode (SOME true);
1388    [autoloading]
1389    [autoloading done]
1390    val it = false : bool
1391    - structure X = struct
1392    -     fun main n = let
1393    -         fun a (x, 0) = d x
1394    -           | a (x, n) = b (x, n - 1)
1395    -         and b (x, n) = c (x, n)
1396    -         and c (x, n) = a (x, n)
1397    -         and d x = e (x, 3)
1398    -         and e (x, 0) = f x
1399    -           | e (x, n) = e (x, n - 1)
1400    -         and f 0 = SMLofNJ.Internals.BTrace.trigger ()
1401    -           | f n = n * g (n - 1)
1402    -         and g n = a (n, 3)
1403    -     in
1404    -         f n
1405    -     end
1406    - end;
1407    structure X : sig val main : int -> int end
1408    - X.main 3;
1409    *** BACK-TRACE ***
1410    GOTO   stdIn:4.2-13.20: X.main[2].f
1411    GOTO-( stdIn:4.2-13.20: X.main[2].e
1412    GOTO   stdIn:4.2-13.20: X.main[2].d
1413         / stdIn:4.2-13.20: X.main[2].a
1414         | stdIn:4.2-13.20: X.main[2].b
1415    GOTO-\ stdIn:4.2-13.20: X.main[2].c
1416    CALL   stdIn:4.2-13.20: X.main[2].g
1417    GOTO   stdIn:4.2-13.20: X.main[2].f
1418    GOTO-( stdIn:4.2-13.20: X.main[2].e
1419    GOTO   stdIn:4.2-13.20: X.main[2].d
1420         / stdIn:4.2-13.20: X.main[2].a
1421         | stdIn:4.2-13.20: X.main[2].b
1422    GOTO-\ stdIn:4.2-13.20: X.main[2].c
1423    CALL   stdIn:4.2-13.20: X.main[2].g
1424    GOTO   stdIn:4.2-13.20: X.main[2].f
1425    GOTO-( stdIn:4.2-13.20: X.main[2].e
1426    GOTO   stdIn:4.2-13.20: X.main[2].d
1427         / stdIn:4.2-13.20: X.main[2].a
1428         | stdIn:4.2-13.20: X.main[2].b
1429    GOTO-\ stdIn:4.2-13.20: X.main[2].c
1430    CALL   stdIn:4.2-13.20: X.main[2].g
1431    GOTO   stdIn:4.2-13.20: X.main[2].f
1432    CALL   stdIn:2.15-17.4: X.main[2]
1433    -
1434    
1435    (Note that because of a FLINt bug the above code currently does not
1436    compile without BTrace turned on.)
1437    
1438    Here is another example, using my modified Tiger compiler:
1439    
1440    Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
1441    - SMLofNJ.Internals.BTrace.mode (SOME true);
1442    [autoloading]
1443    [autoloading done]
1444    val it = false : bool
1445    - CM.make "sources.cm";
1446    [autoloading]
1447    ...
1448    [autoloading done]
1449    [scanning sources.cm]
1450    [parsing (sources.cm):parse.sml]
1451    [creating directory CM/SKEL ...]
1452    [parsing (sources.cm):tiger.lex.sml]
1453    ...
1454    [wrote CM/sparc-unix/semant.sml]
1455    [compiling (sources.cm):main.sml]
1456    [wrote CM/sparc-unix/main.sml]
1457    [New bindings added.]
1458    val it = true : bool
1459    - Main.compile ("../testcases/merge.tig", "foo.out");
1460    *** BACK-TRACE ***
1461    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trvar
1462    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1463    CALL   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
1464    GOTO   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
1465    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1466    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1467    CALL   lib/semant.sml:488.3-505.6: SemantFun[2].transDec.trdec[2].transBody[2]
1468         / lib/semant.sml:411.65-543.8: SemantFun[2].transDec
1469    CALL-\ lib/semant.sml:413.2-540.9: SemantFun[2].transDec.trdec[2]
1470    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1471    CALL   lib/semant.sml:8.52-558.4: SemantFun[2].transProg[2]
1472    CALL   main.sml:1.18-118.4: Main.compile[2]
1473    -
1474    
1475    ----------------------------------------------------------------------
1476    Name: Matthias Blumen
1477    Date: 2000/06/21 18:00:00 JST
1478    Tag: blume-20000621-manual
1479    Description:
1480    
1481    CM manual update: Path environments documented.
1482    
1483    ----------------------------------------------------------------------
1484    Name: Matthias Blume
1485    Date: 2000/06/19 13:40:00
1486    Tag: blume-20000619-manual
1487    Description:
1488    
1489    CM manual and system/README update.  This only covers the fact that
1490    there are no more implicit anchors.  (Path environments and the "bind"
1491    option to "cm" have yet to be documented.)
1492    
1493    ----------------------------------------------------------------------
1494    Name: Matthias Blume
1495    Date: 2000/06/19 11:05:00 JST
1496    Tag: blume-20000619-chdir-bugfix
1497    Description:
1498    
1499    Fixed a bug in new SrcPath module that sometimes led to a bad chDir call.
1500    
1501    ----------------------------------------------------------------------
1502    Name: Matthias Blume
1503    Date: 2000/06/18 22:00:10 JST
1504    Tag: blume-20000618-implicit-anchors-really-gone
1505    Description:
1506    
1507    I updates the previous HISTORY entry where I forgot to mention that
1508    implicit anchors are no longer with us.
1509    
1510    The current update also gets rid of the (now useless) controller
1511    CM.Control.implicit_anchors.
1512    
1513    ----------------------------------------------------------------------
1514    Name: Matthias Blume
1515    Date: 2000/06/16 17:30:00 JST
1516    Tag: blume-20000616-anchorenv
1517    Description:
1518    
1519    This patch implements the long anticipated (just kidding :) "anchor
1520    environment" mechanism.  In the course of doing this, I also
1521    re-implemented CM's internal "SrcPath" module from scratch.  The new
1522    one should be more robust in certain boundary cases.  In any case, it
1523    is a lot cleaner than its predecessor (IMHO).
1524    
1525    This time, although there is yet another boot file format change, I
1526    kept the unpickler backward-compatible.  As a result, no new bootfiles
1527    are necessary and bootstrapping is straightforward.  (You cannot read
1528    new bootfiles into an old system, but the other way around is no
1529    problem.)
1530    
1531    Visible changes:
1532    
1533    ** 0. Implicit path anchors (without the leading $-symbol) are no
1534    longer recognized at all. This means that such path names are not
1535    illegal either.  For example, the name basis.cm simply refers to a
1536    local file called "basis.cm" (i.e, the name is an ordinary path
1537    relative to .cm-files directory).  Or, to put it differently, only
1538    names that start with $ are anchored paths.
1539    
1540    ** 1. The $<singlearc> abbreviation for $/<singlearc> has finally
1541    vanished.
1542    
1543    John (Reppy) had critizised this as soon as I originally proposed and
1544    implemented it, but at that time I did not really deeply believe
1545    him. :) Now I came full-circle because I need the $<singlearc> syntax
1546    in another place where it cannot be seen as an abbreviation for
1547    $/<singlearc>.  To avoid the confusion, $<singlearc> now means what it
1548    seems to mean (i.e., it "expands" into the corresponding anchor
1549    value).
1550    
1551    However, when paths are used as members in CM description files, it
1552    continues to be true that there must be at least another arc after the
1553    anchor.  This is now enforced separately during semantic analysis
1554    (i.e., from a lexical/syntactical point of view, the notation is ok.)
1555    
1556    ** 2. The "cm" class now accepts an option "bind".  The option's value
1557    is a sub-option list of precisely two items -- one labeled "anchor"
1558    and the other one labeled "value".  As you might expect, "anchor" is
1559    used to specify an anchor name to be bound, and "value" specifies what
1560    the anchor is being bound to.
1561    
1562    The value must be a directory name and can be given in either standard
1563    syntax (including the possibility that it is itself an anchored path)
1564    or native syntax.
1565    
1566    Examples:
1567    
1568       foo.cm (bind:(anchor:bar value:$mystuff/bar))
1569       lib.cm (bind:(anchor:a value:"H:\\x\\y\\z"))  (* only works under windows *)
1570    
1571    and so on.
1572    
1573    The meaning of this is that the .cm-file will be processed with an
1574    augmented anchor environment where the given anchor(s) is/are bound to
1575    the given values(s).
1576    
1577    The rationale for having this feature is this: Suppose you are trying
1578    to use two different (already stable) libraries a.cm and b.cm (that
1579    you perhaps didn't write yourself).  Further, suppose each of these
1580    two libraries internally uses its own auxiliary library $aux/lib.cm.
1581    Normally you would now have a problem because the anchor "lib" can not
1582    be bound to more than one value globally.  Therefore, the project that
1583    uses both a.cm and b.cm must locally redirect the anchor to some other
1584    place:
1585    
1586       a.cm (bind:(anchor:lib value:/usr/lib/smlnj/a-stuff))
1587       b.cm (bind:(anchor:lib value:/usr/lib/smlnj/b-stuff))
1588    
1589    This hard-wires $lib/aux.cm to /usr/lib/smlnj/a-stuff/aux.cm or
1590    /usr/lib/smlnj/b-stuff/aux.cm, respectively.
1591    
1592    Hard-wiring path names is a bit inflexible (and CM will verbosely warn
1593    you when you do so at the time of CM.stabilize).  Therefore, you can
1594    also use an anchored path as the value:
1595    
1596      a.cm (bind:(anchor:lib value:$a-lib))
1597      b.cm (bind:(anchor:lib value:$b-lib))
1598    
1599    Now you can globally configure (using the usual CM.Anchor.anchor or
1600    pathconfig machinery) bindings for "a-lib" and "b-lib".  Since "lib"
1601    itself is always locally bound, setting it globally is no longer
1602    meaningful or necessary (but it does not hurt either).  In fact, "lib"
1603    can still be used as a global anchor for separate purposes.  As a
1604    matter of fact, one can locally define "lib" in terms of a global
1605    "lib":
1606    
1607      a.cm (bind:(anchor:lib value:$lib/a))
1608      b.cm (bind:(anchor:lib value:$lib/b))
1609    
1610    ** 3: The encoding of path names has changed.  This affects the way
1611    path names are shown in CM's progress report and also the internal
1612    protocol encoding used for parallel make.
1613    
1614    The encoding now uses one or more ':'-separated segments.  Each
1615    segments corresponds to a file that has been specified relative to the
1616    file given by its preceding segment.  The first segment is either
1617    relative to the CWD, absolute, or anchored.  Each segment itself is
1618    basically a Unix pathname; all segments but the first are relative.
1619    
1620    Example:
1621    
1622       $foo/bar/baz.cm:a/b/c.sml
1623    
1624    This path denotes the file bar/a/b/c.sml relative to the directory
1625    denoted by anchor "foo".  Notice that the encoding also includes
1626    baz.cm which is the .cm-file that listed a/b/c.sml.  As usual, such
1627    paths are resolved relative to the .cm-files directory, so baz.cm must
1628    be ignored to get the "real" pathname.
1629    
1630    To make this fact more obvious, CM puts the names of such "virtual
1631    arcs" into parentheses when they appear in progress reports. (No
1632    parentheses will appear in the internal protocol encoding.)  Thus,
1633    what you really see is:
1634    
1635      $foo/bar/(baz.cm):a/b/c.sml
1636    
1637    I find this notation to be much more informative than before.
1638    
1639    Another new feature of the encoding is that special characters
1640    including parentheses, colons, (back)slashes, and white space are
1641    written as \ddd (where ddd is the decimal encoding of the character).
1642    
1643    *** The CM manual still needs to be updated.
1644    
1645    ----------------------------------------------------------------------
1646    Name: Allen Leung
1647    Date: 2000/06/15 00:38:00
1648    Tag: leunga-20000615-x86-peephole
1649    
1650    x86 Peephole fix by Fermin.  Affects c-- and moby only.
1651    
1652    ----------------------------------------------------------------------
1653    Name: Matthias Blume
1654    Date: 2000/06/12 11:40:00
1655    Tag: blume-20000612-parmakefix
1656    Description:
1657    
1658    More cleanup after changing the file naming scheme: This time I
1659    repaired the parallel make mechanism for CMB.make which I broke earlier.
1660    
1661    ----------------------------------------------------------------------
1662    Name: Allen Leung
1663    Date: 2000/06/09 01:25:00
1664    Tag: leunga-20000609-various
1665    
1666    None of these things should affect normal SML/NJ operations
1667    
1668    1. Peephole improvements provided by Fermin (c--)
1669    2. New annotation DEFUSE for adding extra dependence (moby)
1670    3. New X86 LOCK instructions (moby)
1671    4. New machine description language for reservation tables (scheduling)
1672    5. Fixes to various optimization/analysis modules (branch chaining, dominator
1673       trees etc.)
1674    6. I've changed the CM files so that they can work with versions
1675       110.0.6, 110.25 and 110.28
1676    
1677    ----------------------------------------------------------------------
1678    Name: Matthias Blume
1679    Date: 2000/06/09 12:40:00
1680    Tag: blume-20000609-log
1681    Description:
1682    
1683    - Removed all(?) remaining RCS Log entries from sources.
1684    
1685    - Fixed bug in ml-yacc and ml-lex sources (use explicit anchors for
1686      anchored paths).
1687    
1688    ----------------------------------------------------------------------
1689    Name: Matthias Blume
1690    Date: 2000/06/07 17:00:00 JST
1691    Tag: blume-20000607-no-implicit-anchors
1692    Description:
1693    
1694    1. This update changes the default setting for
1695    CM.Control.implicit_anchors from true to false.  This means that
1696    implicit anchors are no longer permitted by default.  I also tried to
1697    make sure that nothing else still relies on implicit anchors.
1698    (This is the next step on the schedule towards a CM that does not even
1699    have the notion of implicit anchors anymore.)
1700    
1701    2. More CM manual updates.
1702    
1703    3. I managed to track down and fix the pickling bug I mentioned last
1704    time.  Because of the previously existing workaround, this entails no
1705    immediate practical changes.
1706    
1707    ----------------------------------------------------------------------
1708    Name: Matthias Blume
1709    Date: 2000/06/06 11:15:00 JST
1710    Tag: blume-20000606-lazierpickle
1711    Description:
1712    
1713    !!!! NEW BOOT FILES !!!!
1714    
1715    * The main purpose of this update is to make library pickles lazier in
1716    order to reduce the initial space penalty for autoloading a library.
1717    As a result, it is now possible to have $smlnj/compiler.cm
1718    pre-registered.  This should take care of the many complaints or
1719    inquiries about missing structure Compiler.  This required changes to
1720    CM's internal data structures and small tweaks to some algorithms.
1721    
1722    As a neat additional effect, it is no longer necessary (for the sake
1723    of lean heap image files) to distinguish between a "minimal" CM and a
1724    "full" CM.  Now, there is only one CM (i.e., the "full" version:
1725    $smlnj/cm.cm aka $smlnj/cm/full.cm), and it is always available at the
1726    interactive top level. ($smlnj/cm/minimal.cm is gone.)
1727    
1728    To make the life of compiler-hackers easier, "makeml" now also
1729    pre-registers $smlnj/cmb.cm (aka $smlnj/cmb/current.cm).  In other
1730    words, after you bootstrap a new sml for the first time, you will not
1731    have to autoload $smlnj/cmb.cm again afterwards.  (The first time
1732    around you will still have to do it, though.)
1733    
1734    * A second change consists of major updates to the CM manual.  There
1735    are now several appendices with summary information and also a full
1736    specification of the CM description file syntax.
1737    
1738    * In directory src/system I added the script "allcross".  This script
1739    invokes sml and cross-compiles the compiler for all supported
1740    architectures.  (Useful when providing a new set of boot files.)
1741    
1742    * There seems to be a latent bug in my "lazy pickles" mechanism.  I
1743    added a small tweak to pickle-util.sml to work around this problem,
1744    but it is not a proper fix yet.  I will investigate further.  (The
1745    effect of the bug was an inflation of library pickle size.)
1746    
1747    * Version number increased to 110.28.1 (to avoid compatibility problems).
1748    
1749    ----------------------------------------------------------------------
1750    Name: Allen Leung
1751    Date: 2000/05/25 17:28 EDT
1752    Tag: leunga-20000525-ra
1753    Description:
1754    
1755      Fixed a bug in freezing phase of the register allocator.
1756    
1757    ----------------------------------------------------------------------
1758    Name: Allen Leung
1759    Date: 2000/05/15 22:53 EDT
1760    Tag: leunga-20000515-alpha-x86-ra
1761    Description:
1762    
1763      1. Alpha
1764    
1765          Slight cleanup.  Removed the instruction SGNXL
1766    
1767      2. X86
1768    
1769          Added the following instructions to the instruction set:
1770    
1771            ROLx, RORx,
1772            BTx, BTSx, BTLx, BTRx,
1773            XCHGx, and variants with the LOCK prefix
1774    
1775      3. Register Allocation
1776    
1777          The module ra-rewrite-with-renaming has been improved.
1778    
1779      These have no effect on SML/NJ.
1780    
1781    ----------------------------------------------------------------------
1782    Name: Matthias Blume
1783    Date: 2000/05/15 16:20:00 JST
1784    Tag: blume-20000515-lightrebuild
1785    Description:
1786    
1787    1. I added an alternative to "-rebuild" to "makeml".  The difference is
1788       that prior to calling CMB.make' the CM-variable "LIGHT" will be
1789       defined.  In effect, the command will not build any cross-compiler
1790       backends and therefore finish more quickly.
1791    
1792       The "fixpt" script also takes a "-light" switch to be able to use
1793       this new facility while compiling for a fixpoint.
1794    
1795    2. I replaced all mentions of anchored paths in group owner specifications
1796       with simple relative paths (usually starting with "..").
1797       The rationale is that a library's internal workings should not be
1798       compromised by the lack of some anchor.  (An anchor is necessary
1799       for someone who wants to refer to the library by an anchored path,
1800       but it should not be necessary to build the same library in the first
1801       place.)
1802    
1803    3. I changed the way CM's tool mechanism determines the shell command
1804       string used for things like ml-yacc etc. so that it does not break
1805       when CM.Control.implicit_anchors is turned off.
1806    
1807    ----------------------------------------------------------------------
1808    Name: Matthias Blume
1809    Date: 2000/05/12 18:20:00 JST
1810    Tag: blume-20000512-ml-build
1811    Description:
1812    
1813    Fixed a bug in config/_ml-build that prevented ml-yacc and ml-lex from
1814    getting installed properly (by config/install.sh).
1815    
1816    ----------------------------------------------------------------------
1817    Name: Matthias Blume
1818    Date: 2000/05/12 17:30:00 JST
1819    Tag: blume-20000512-anchors
1820    Description:
1821    
1822    !!! NEW BOOT FILES !!!
1823    
1824    This change is in preparation of fading out support for "implicitly
1825    anchored path names".  I went through all sources and used the
1826    explicit (and relatively new) $-notation.  See system/README and the
1827    CM manual for more info on this.
1828    
1829    I also modified the anchoring scheme for some things such as "smlnj",
1830    "MLRISC", "cm", etc. to take advantage of the fact that explicit
1831    anchors are more expressive: anchor name and first arc do not have to
1832    coincide.  This entails the following user-visible change:
1833    
1834    You have to write $smlnj/foo/bar instead of smlnj/foo/bar.  In
1835    particular, when you fire up sml with a command-line argument, say,
1836    e.g.:
1837    
1838       sml '$smlnj/cmb.cm'
1839    
1840    At the ML toplevel prompt:
1841    
1842       CM.autoload "$smlnj/cmb.cm";
1843    
1844    There is also a new controller in CM.Control that can be used to turn
1845    off all remaining support for implicit anchors by saying:
1846    
1847        CM.autoload "$smlnj/
1848        #set CM.Control.implicit_anchors false;
1849    
1850    This causes CM to reject implicitly anchored paths.  This is (for the
1851    time being) less permissive than the "final" version where there will
1852    be no more such implicit anchors and relative paths will be just that:
1853    relative.
1854    
1855    The next step (version after next version?) will be to make the
1856    default for CM.Control.implicit_anchors false.  After the dust has
1857    settled, I can then produce the "final" version of this...
1858    
1859    Note: Since bootstrapping is a bit tricky, I provided new boot files.
1860    
1861    ----------------------------------------------------------------------
1862    Name: Matthias Blume
1863    Date: 2000/05/11 16:30:00 JST
1864    Tag: blume-20000511-sources
1865    Description:
1866    
1867    The main change is that I added function CM.sources as a generalized
1868    version of the earlier CM.makedepend.  This entails the following
1869    additional changes:
1870    
1871      - CM.makedepend has been dropped.
1872    
1873      - CM manual has been updated.
1874    
1875      - TOOLS signature and API have been changed.
1876    
1877    ----------------------------------------------------------------------
1878    Name: Allen Leung
1879    Date: 2000/05/10 21:17 EDT
1880    Tag: leunga-20000510-moby-c--ssa
1881    Description:
1882    
1883      Various bug fixes and new features for C--, Moby and MLRISC optimizations.
1884    None of these affect SML/NJ.
1885    
1886    1. Register Allocation
1887    
1888        a. A new ra spilling module (ra/ra-spill-with-renaming) is implemented.
1889           This module tries to remove local (i.e. basic block level) redundancies
1890           during spilling.
1891    
1892        b. A new framework for performing region based register allocation.
1893           Not yet entirely functional.
1894    
1895    2. X86
1896    
1897       a. DefUse for POP was missing the stack pointer [found by Lal]
1898       b. Reload for CALL was incorrect in X86Spill [found by John]
1899       c. Various fixes in X86Spill so that it can be used correctly for
1900          the new spilling module.
1901    
1902    3. SSA/IR
1903    
1904       a. New module ir/dj-dataflow.sml implements elimination based
1905          data flow analysis.
1906    
1907    4. MLRiscGen
1908    
1909       a. Fix for gc type annotation
1910    
1911    5. MDGen
1912    
1913       Various fixes for machine description -> ml code translation.  For ssa
1914       only.
1915    
1916    ----------------------------------------------------------------------
1917    Name: Allen Leung
1918    Date: 2000/05/08 22:17 EDT
1919    Tag: leunga-20000508-labexp
1920    Description:
1921    
1922      Fermin has found a few assembly problems with constant expressions
1923      generated in LabelExp.  Mostly, the problems involve extra parentheses,
1924      which choke on dumb assemblers.  This is his fix.
1925    
1926    ----------------------------------------------------------------------
1927    Name: Dave MacQueen
1928    Date: 2000/04/09 14:00 EDT
1929    Tag: dbm-20000502-Version_110_28
1930    Description:
1931    
1932    1. Updated src/compiler/TopLevel/main/version.sml to version 110.28
1933    
1934    2. Updated config/version to 110.28
1935    
1936    3. Updated config/srcarchiveurl
1937    
1938    3. New boot files!
1939       ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.28/
1940    
1941    ----------------------------------------------------------------------
1942    Name: Matthias Blume
1943    Date: 2000/05/01 19:05:00 JST
1944    Tag: blume-20000501-noweb
1945    Description:
1946    
1947    A new noweb tool has been added.  The existing system is entirely
1948    unaffected by this, but some CM users have asked for renewed noweb
1949    support.  Everything is documented in the CM manual.
1950    
1951    New (plugin) libraries:
1952    
1953       noweb-tool.cm
1954       nw-ext.cm
1955    
1956    ----------------------------------------------------------------------
1957    Name: Dave MacQueen
1958    Date: 2000/04/30 12:40PM EDT
1959    Tag: dbm-20000430-bug_fixes
1960    Description:
1961    
1962    1. Fix for bug 1498
1963       smlnj/src/system/Basis/Implementation/Unsafe/object.sig
1964       smlnj/src/system/Basis/Implementation/Unsafe/object.sml
1965         added toRealArray function
1966       smlnj/src/compiler/MiscUtil/print/ppobj.sml
1967         added check for tag Obj.RealArray to array printing case in ppObj
1968    
1969    2. Fix for bug 1510
1970       smlnj/src/compiler/Semant/types/typesutil.sml
1971         fixed definition of dummyargs (used by equalTycon) so that
1972         dummy args are distinct types
1973    
1974    ----------------------------------------------------------------------
1975    Name: Matthias Blume
1976    Date: 2000/04/30 01:00:00 JST
1977    Tag: blume-20000430-versions
1978    Description:
1979    
1980    1. CM version numbering added.  This is an implementation of Lal's
1981       proposal for adding version numbers and version checking to .cm
1982       files.  Lal said that his proposal was just that -- a proposal.
1983       For the time being I went ahead and implemented it so that people
1984       can comment on it.  Everything is completely backward-compatible
1985       (except for the stable library format, i.e., new bootfiles!).
1986    
1987       As usual, see the CM manual for details.
1988    
1989    2. An alternative syntax for anchored paths has been implemented.
1990       Dave has recently voiced the same concerns that I had when I did
1991       this, so there should be some support.  My take is that eventually
1992       I will let support for the current syntax (where anchors are
1993       "implicit") fade out in favor of the new, explicit syntax.
1994       In order to be backward-compatible, both old and new syntax are
1995       currently supported.
1996    
1997       Again, see the CM manual for details.
1998    
1999    3. Parallel make is trying to be slightly smarter:  When the master
2000       process finds a "bottleneck", i.e., when there is only one
2001       compilation unit that can be compiled and everybody else is
2002       waiting on it, then it will simply compile it directly instead
2003       of clumsily telling one of the slaves to do it.
2004    
2005    4. Support for "unsharing" added.  This is necessary in order to be
2006       able to have two different versions of the same library running
2007       at the same time (e.g., for trying out a new MLRISC while still
2008       having the old MLRISC linked into the current compiler, etc.)
2009       See the CM manual.
2010    
2011    5. Simple "makedepend" functionality added for generating Makefile
2012       dependency information.  (This is rather crude at the moment.
2013       Expect some changes here in the future.)
2014    
2015    6. ".fun" added as a recognized suffix for ML files. Also documented
2016       explicitly in the manual that the fallback behavior (unknown suffix
2017       -> ML file) is not an official feature!
2018    
2019    7. Small changes to the pickler for stable libraries.
2020    
2021    8. Several internal changes to CM (for cleanup/improvement).
2022    
2023    
2024    !!!! NEW BINFILES !!!!
2025    
2026    ----------------------------------------------------------------------
2027    Name: Matthias Blume
2028    Date: 2000/04/28 17:30:00 JST
2029    Tag: blume-20000428-pathconfig
2030    Description:
2031    
2032    1. I changed config/install.sh to remove duplicate entries from the
2033       lib/pathconfig file at the end.  Moreover, the final version of
2034       lib/pathconfig is sorted alphabetically.  The same (sorting) is done
2035       in src/system/installml.
2036    
2037    2. The config/install.sh script now consistently uses relative
2038       pathnames in lib/pathconfig whenever the anchor is in the lib
2039       directory.  (So far this was true for the libraries that come
2040       pre-compiled and bundled as part of the bootfiles but not for
2041       libraries that are compiled by the script itself.)
2042    
2043    ----------------------------------------------------------------------
2044    Name: Matthias Blume
2045    Date: 2000/04/26 13:10:00 JST
2046    Tag: blume-20000426-fun_suffix
2047    Description:
2048    
2049    Added ".fun" as a recognized file name suffix (for ML code).
2050    
2051    ----------------------------------------------------------------------
2052    Name: Allen Leung
2053    Date: 2000/04/25 17:00:00 EST
2054    Tag: leunga-20000425-alpha-ra
2055    Description:
2056    
2057    1. Alpha
2058    
2059        PSEUDOARITH was missing in AlphaRewrite.  This causes an endless loop
2060    in C--.
2061    
2062    2. RA
2063    
2064       Added a flag "ra-dump-size" to print out the size of the flowgraph
2065       and the interference graph.
2066    
2067    ----------------------------------------------------------------------
2068    Name: Dave MacQueen
2069    Date: 2000/04/25/
2070    Tag: dbm-20000425-mlyacc_doc_examples
2071    Description:
2072      Updated mlyacc.tex sections 5 and 7 for SML '97 and CM.
2073      Updated all three examples in src/ml-yacc/examples to run
2074      under 110.* using CM.make.
2075    
2076    ----------------------------------------------------------------------
2077    Name: Allen Leung
2078    Date: 2000/04/20 23:04:00 EST
2079    Tag: leunga-20000420-ssa-c---stuff
2080    Description:
2081    
2082      This update synchronizes my repository with Yale's.  Most of these
2083    changes, however, do not affect SML/NJ at all (the RA is an exception).
2084    
2085    1. Register Allocator
2086    
2087       a. An improvement in the interference graph construction:
2088          Given a copy
2089    
2090                s <- t
2091    
2092          no interference edge between s and t is added for this definition of s.
2093    
2094       b. I've added two new spill heuristic modules that Fermin and I developed
2095          (in the new library RA.cm). These are unused in SML/NJ but maybe
2096          useful for others (Moby?)
2097    
2098    2. X86
2099    
2100       a. Various fixes in the backend provided by Fermin [C--] and Lal.
2101    
2102    3. Alpha
2103    
2104       a. Added the BSR instruction and code generation that goes with it [C--]
2105       b. Other fixes too numerous to recount provided by Fermin [C--]
2106    
2107    4. Regmaps
2108    
2109       a. The regmaps are not initialized with the identity physical bindings
2110          at creation time.  This is unneeded.
2111    
2112    5. MLRISC Optimizations
2113    
2114       a. The DJ-Graph module can now compute the iterated dominance frontiers
2115          intersects with liveness incrementally in linear time! Woohoo!
2116          This is now used in my new SSA construction algorithm.
2117    
2118       b. THe branch reorganization module is now smarter about linear chains of
2119          basic blocks.
2120    
2121    
2122    ----------------------------------------------------------------------
2123    Name: Matthias Blume
2124    Date: 2000/04/12 13:52:00 JST
2125    Tag: blume_main_v110p27_1
2126    Description:
2127    
2128    Changed install.sh script to handle archive files without version number
2129    and to use "boot.<arch>-<os>" instead of "sml.boot.<arch>-<os>" for the
2130    name of the boot file archive.
2131    
2132    ----------------------------------------------------------------------
2133    Name: Dave MacQueen
2134    Date: 2000/04/09 14:00 EDT
2135    Tag: dbm-20000410-Version_110_27
2136    Description:
2137    
2138    1. Updated src/compiler/TopLevel/main/version.sml to version 110.27
2139    
2140    2. Updated src/config/version to 110.27
2141    
2142    3. New boot files!
2143    
2144    ----------------------------------------------------------------------
2145    Name: Allen Leung
2146    Date: 2000/04/09 19:09:00 EST
2147    Tag: leunga-20000409-misc
2148    Description:
2149    
2150    1.  Yet another fix for x86 assembly for idivl, imull, mull and friends.
2151    
2152    2.  Miscellaneous improvements to MLRISC (unused in sml/nj)
2153    
2154    ----------------------------------------------------------------------
2155    Name: Stefan
2156    Date: 2000/04/07 10:00:00 EDT
2157    Tag: monnier-20000406-branch-handling
2158    Description:
2159    
2160    Improved handling of branches (mostly those generated from
2161    polymorphic equality), removed switchoff and changed the
2162    default optimization settings (more cpsopt and less flintopt).
2163    
2164    ----------------------------------------------------------------------
2165    Name: Allen Leung
2166    Date: 2000/04/06 01:30:00 EST
2167    Tag: leunga-20000406-peephole-x86-SSA-2
2168    Description:
2169    
2170       Forgot a few files.
2171    
2172  ----------------------------------------------------------------------  ----------------------------------------------------------------------
2173  Name: Allen Leung  Name: Allen Leung
2174  Date: 2000/04/06 00:36:00 EST  Date: 2000/04/06 00:36:00 EST
# Line 79  Line 2237 
2237    
2238  3.  Assembly  3.  Assembly
2239    
2240      When generating assemby, resolve the value of client defined constants,      When generating assembly, resolve the value of client defined constants,
2241      instead of generating symbolic values.  This is controlled by the      instead of generating symbolic values.  This is controlled by the
2242      new flag "asm-resolve-constants", which is default to true.      new flag "asm-resolve-constants", which is default to true.
2243    
# Line 102  Line 2260 
2260    
2261  To this end, I arranged that instead of "structure Core" as "structure  To this end, I arranged that instead of "structure Core" as "structure
2262  _Core" is bound in the pervasive environment.  Core access is done via  _Core" is bound in the pervasive environment.  Core access is done via
2263  _Core (which can never be accidentially rebound because _Core is not a  _Core (which can never be accidentally rebound because _Core is not a
2264  legal surface-syntax symbol).  legal surface-syntax symbol).
2265    
2266  The current solution is much cleaner because the core environment is  The current solution is much cleaner because the core environment is
# Line 112  Line 2270 
2270  with dynamic and symbolic parts of the core environment.  with dynamic and symbolic parts of the core environment.
2271    
2272  Remaining hackery (to bind the "magic" symbol _Core) is localized in the  Remaining hackery (to bind the "magic" symbol _Core) is localized in the
2273  compilation mananger's bootstrap compiler (actually: in the "init group"  compilation manager's bootstrap compiler (actually: in the "init group"
2274  handling).  See the comments in src/system/smlnj/init/init.cmi for  handling).  See the comments in src/system/smlnj/init/init.cmi for
2275  more details.  more details.
2276    
# Line 227  Line 2385 
2385     (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
2386     used).     used).
2387    
2388     This was done to accomodate the new "make" and "shell" tools which     This was done to accommodate the new "make" and "shell" tools which
2389     facilitate fairly seemless hookup to portions of code managed using     facilitate fairly seamless hookup to portions of code managed using
2390     Makefiles or Shell scripts.     Makefiles or Shell scripts.
2391    
2392     There are no classes "shared" or "private" anymore.  Instead, the     There are no classes "shared" or "private" anymore.  Instead, the
# Line 240  Line 2398 
2398    
2399     All existing tools are described in the CM manual.     All existing tools are described in the CM manual.
2400    
2401     - Slightly better error handling.  (CM now surpresses many followup     - Slightly better error handling.  (CM now suppresses many followup
2402     error messages that tended to be more annoying than helpful.)     error messages that tended to be more annoying than helpful.)
2403    
2404  2. Major changes to the compiler's static environment data structures.  2. Major changes to the compiler's static environment data structures.
# Line 374  Line 2532 
2532    
2533       I've changed andl to testl in the floating point test sequence       I've changed andl to testl in the floating point test sequence
2534       whenever appropriate.  The Intel optimization guide states that       whenever appropriate.  The Intel optimization guide states that
2535       testl is perferable to andl.       testl is preferable to andl.
2536    
2537  7. RA (x86 only)  7. RA (x86 only)
2538    
# Line 556  Line 2714 
2714    
2715  1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an  1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an
2716  additional argument called "template" which is an optional string that  additional argument called "template" which is an optional string that
2717  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
2718  explanation.  explanation.
2719    
2720  2. A special-purpose tool can be "regisitered" by simply dropping the  2. A special-purpose tool can be "registered" by simply dropping the
2721  corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same  corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same
2722  directory where the .cm file lives that uses this tool.  (The  directory where the .cm file lives that uses this tool.  (The
2723  behavior/misfeature until now was to look for the tool description  behavior/misfeature until now was to look for the tool description
# Line 603  Line 2761 
2761  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
2762  <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,
2763  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
2764  perhaps got accidentially killed).  In this case, fixpt will quickly  perhaps got accidentally killed).  In this case, fixpt will quickly
2765  move through what exists before continuing where it left off earlier,  move through what exists before continuing where it left off earlier,
2766  and, thus, saves a lot of time.  and, thus, saves a lot of time.
2767    
# Line 653  Line 2811 
2811    it from that remote directory.    it from that remote directory.
2812    This should simplify installation further:  For machines that have    This should simplify installation further:  For machines that have
2813    access to the internet, just fetch <version>-config.tgz, unpack it,    access to the internet, just fetch <version>-config.tgz, unpack it,
2814    edit config/targets, and go (run config/install.sh).  The scipt will    edit config/targets, and go (run config/install.sh).  The script will
2815    fetch everything else that it might need all by itself.    fetch everything else that it might need all by itself.
2816    
2817    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.601  
changed lines
  Added in v.841

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