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 657, Fri Jun 9 05:20:54 2000 UTC revision 840, Fri Jun 15 19:05:19 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-sparc
19    Description:
20    
21    The purpose of this update is to provide an implementation of NLFFI
22    on Sparc machines.
23    
24    Here are the changes in detail:
25    
26       * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
27       the Sparc implementation of the c-calls API.
28       * The Sparc backend of SML/NJ has been modified to uniformely use %fp
29       for accessing the ML frame.  Thus, we have a real frame pointer and
30       can freely modify %sp without need for an omit-frame-ptr phase.
31       The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
32       this case.
33       * ml-nlffigen has been taught to produce code for different architectures
34       and calling conventions.
35       * In a way similar to what was done in the x86 case, the Sparc
36       backend uses its own specific extension to mltree.  (For example,
37       it needs to be able to generate UNIMP instructions which are part
38       of the calling convention.)
39       * ml-nlffi-lib was reorganized to make it more modular (in particular,
40       to make it easier to plug in new machine- and os-dependent parts).
41    
42    There are some other fairly unrelated bug fixes and cleanups as well:
43    
44       * I further hacked the .cm files for MLRISC tools (like MDLGen) so
45       that they properly share their libraries with existing SML/NJ libraries.
46       * I fixed a minor cosmetic bug in CM, supressing certain spurious
47       follow-up error messages.
48       * Updates to CM/CMB documentation.
49    
50    TODO items:
51    
52       * MLRISC should use a different register as its asmTemp on the Sparc.
53         (The current %o2 is a really bad choice because it is part of the
54         calling conventions, so things might interfere in unexpected ways.)
55    
56    ----------------------------------------------------------------------
57    Name: Matthias Blume
58    Date: 2001/06/07
59    Tag: blume-20010607-calls
60    Description:
61    
62    A number of internal changes related to C calls and calling conventions:
63    
64    1. ML-Tree CALL statements now carry a "pops" field.  It indicates the
65       number of bytes popped implicitly (by the callee).  In most cases
66       this field is 0 but on x86/win32 it is some non-zero value.  This
67       is information provided for the benefit of the "omit-frameptr" pass.
68    2. The CALL instruction on the x86 carries a similar "pops" field.
69       The instruction selection phase copies its value from the ML-Tree
70       CALL statement.
71    3. On all other architectures, the instruction selection phase checks
72       whether "pops=0" and complains if not.
73    4. The c-calls implementation for x86 now accepts two calling conventions:
74       "ccall" and "stdcall".  When "ccall" is selected, the caller cleans
75       up after the call and pops is set to 0.  For "stdcall", the caller
76       does nothing, leaving the cleanup to the callee; pops is set to
77       the number of bytes that were pushed onto the stack.
78    5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
79       distinguish between "ccall" and "stdcall".
80    6. The UNIMP instruction has been added to the supported Sparc instruction
81       set. (This is needed for implementing the official C calling convention
82       on this architecture.)
83    7. I fixed some of the .cm files under src/MLRISC/Tools to make them
84       work with the latest CM.
85    
86    ----------------------------------------------------------------------
87    Name: Matthias Blume
88    Date: 2001/06/05 15:10:00 EDT
89    Tag: blume-20010605-cm-index
90    Description:
91    
92    0. The "lambdasplit" parameter for class "sml" in CM has been documented.
93    
94    1. CM can now generate "index files".  These are human-readable files
95       that list on a per-.cm-file basis each toplevel symbol defined or
96       imported.  The location of the index file for
97       <p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
98       To enable index-file generation, set CM.Control.generate_index to true
99       or export an environment-symbol: export CM_GENERATE_INDEX=true.
100    
101       The CM manual has been updated accordingly.
102    
103    2. I made some slight modifications to the c-calls API in MLRISC.
104    
105         a) There is now a callback to support saving/restoring of
106            dedicated but caller-save registers around the actual call
107            instruction.
108         b) One can optionally specify a comment-annotation for the
109            call instruction.
110    
111    3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
112       (For example, the comment annotation shows the C prototype of
113        the function being called.)
114    
115    ----------------------------------------------------------------------
116    Name: Matthias Blume
117    Date: 2001/06/01 13:30:00 EDT
118    Tag: blume-20010601-nlffi-cleanup
119    Description:
120    
121    This is mostly a cleanup of MLFFI stuff:
122    
123       - some signature files have been put into a more exposed place
124       - the ugly 'f type parameter is gone (simplifies types tremendously!)
125       - ml-nlffigen changed accordingly
126       - tutorial updated
127    
128    Other changes:
129    
130       - author's affiliation in CM manual(s) updated
131       - some more recognized keywords added to Allen's sml.sty
132    
133    ----------------------------------------------------------------------
134    Name: Matthias Blume
135    Date: 2001/05/25 15:30:00 EDT
136    Tag: blume-20010525-iptr
137    Description:
138    
139      - put the official 110.33-README (as it appears on the ftp server) under
140        CVS
141      - fixed a small bug related to incomplete pointer types in
142        ml-nlffigen
143      - small cosmetic change to the ml-nlffi-lib's "arr" type constructor
144        (it does not need the 'f type parameter)
145    
146    ----------------------------------------------------------------------
147    Name: Matthias Blume
148    Date: 2001/05/23 14:30:00 EDT
149    Tag: Release_110_33
150    Description:
151    
152    New version number (110.33).  New bootfiles.
153    
154    ----------------------------------------------------------------------
155    Name: Matthias Blume
156    Date: 2001/05/22 18:06:00 EDT
157    Tag: blume-20010522-targets
158    Description:
159    
160    Made install.sh use file config/targets.customized if it exists, falling
161    back to config/targets if it doesn't.  This way one can have a customized
162    version of the targets file without touching the "real thing", thus
163    eliminating the constant fear of accidentally checking something bogus
164    back into the CVS repository...  (File config/targets.customized must
165    not be added to the repository!)
166    
167    ----------------------------------------------------------------------
168    Name: Matthias Blume
169    Date: 2001/05/22 16:30:00 EDT
170    Tag: blume-20010522-minitut
171    Description:
172    
173    1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
174       struct returns.
175    2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt.  This is some very
176       incomplete, preliminary documentation for NLFFI.
177    
178    ----------------------------------------------------------------------
179    Name: Matthias Blume
180    Date: 2001/05/14 11:30:00 EDT
181    Tag: blume-20010514-script
182    Description:
183    
184    Some bugs in install script fixed.
185    
186    In addition to that I also made a slight change to the NLFFI API:
187    Functors generated by ml-nlffigen now take the dynamic library as a
188    straight functor argument, not as a suspended one.  (The original
189    functor code used to force the suspension right away anyway, so there
190    was nothing gained by this complication of the interface.)
191    
192    ----------------------------------------------------------------------
193    Name: Matthias Blume
194    Date: 2001/05/11 14:35:00 EDT
195    Tag: blume-20010511-ml-nlffi
196    Description:
197    
198    I finally took the plunge and added my new FFI code to the main
199    repository.  For x86-linux it is now ready for prime-time.
200    
201    There are two new subdirectories of "src":
202    
203      - ml-nlffi-lib:
204           The utility library for programs using the FFI interface.
205           Here is the implementation of $/c.cm and its associated low-level
206           partners $/c-int.cm and $/memory.cm.
207      - ml-nlffigen:
208           A stand-alone program for generating ML glue code from C source
209           code.
210    
211    Building ml-nlffigen requires $/ckit-lib.cm.
212    
213    The config/install.sh script has been updates to do the Right Thing
214    (hopefully).
215    
216    Notice that the source tree for the C-Kit will not be put under "src"
217    but directly under the installation root directory.  (This is the
218    structure that currently exists on the CVS server when you check out
219    module "sml".)  Fortunately, config/install.sh knows about this oddity.
220    
221    Bugs: No documentation yet.
222    
223    ----------------------------------------------------------------------
224    Name: Matthias Blume
225    Date: 2001/05/09 16:35:00 EDT
226    Tag: blume-20010509-cpscontract
227    Description:
228    
229    Fixed a bug in the accounting code in cpsopt/contract.sml.  (The
230    wrapper/unwrapper elimination did not decrement usage counts and some
231    dead variables got overlooked by the dead-up logic.)
232    
233    ----------------------------------------------------------------------
234    Name: Lal George
235    Date: 2001/05/08  17:26:09 EDT
236    Tag: george-20010508-omit-frameptr
237    Description:
238    
239    Changes to implement the omit-frame-pointer optimization to support
240    raw C calls. For now, there is only support on the Intel x86, but
241    other architectures will follow as more experience is gained with this.
242    
243    
244    ----------------------------------------------------------------------
245    Name: Matthias Blume
246    Date: 2001/05/07 14:40:00 EDT
247    Tag: blume-20010507-proxies
248    Description:
249    
250    I made into "proxy libraries" all libraries that qualify for such a
251    change.  (A qualifying library is a library that has another library or
252    groups as its sole member and repeats that member's export list
253    verbatim.  A proxy library avoids this repetition by omitting its export
254    list, effectively inheriting the list that its (only) member exports.
255    See the CM manual for more explanation.)
256    The main effect is that explicit export lists for these libraries
257    do not have to be kepts in sync, making maintenance a bit easier.
258    
259    I also added copyright notices to many .cm-files.
260    
261    Last but not least, I made a new set of bootfiles.
262    
263    ----------------------------------------------------------------------
264    Name: Matthias Blume
265    Date: 2001/05/04 17:00:00 EDT
266    Tag: blume-20010504-cm-lsplit
267    Description:
268    
269    0. John merged pending changes to $/smlnj-lib.cm
270    
271    1. Allen's previous change accidentally backed out of one of Lal's
272       earlier changes.  I undid this mistake (re-introducing Lal's change).
273    
274    2. I used the new topOrder' function from graph-scc.sml (from $/smlnj-lib.cm)
275       within the compiler where applicable.  There is some code simplification
276       because of that.
277    
278    3. The "split" phase (in FLINT) is now part of the default list of phases.
279       Compiler.Control.LambdaSplitting.* can be used to globally control the
280       lambda-splitting (cross-module-inlining) engine.  In addition to that,
281       it can now also be controlled on a per-source basis: CM has been taught
282       a new tool parameter applicable to ML source files.
283    
284       - To turn lambda-splitting off completely:
285            local open Compiler.Control.LambdaSplitting in
286                val _ = set Off
287            end
288       - To make "no lambda-splitting" the global default (but allow per-source
289         overriding); this is the initial setting:
290            local open Compiler.Control.LambdaSplitting in
291                val _ = set (Default NONE)
292            end
293       - To make "lambda-splitting with aggressiveness a" the global default
294         (and allow per-source overriding):
295            local open Compiler.Control.LambdaSplitting in
296                val _ = set (Default (SOME a))
297            end
298    
299       - To turn lambda-splitting off for a given ML souce file (say: a.sml)
300         write (in the respective .cm-file):
301            a.sml (lambdasplitting:off)
302       - To turn lambda-splitting for a.sml on with minimal aggressiveness:
303            a.sml (lambdasplitting:on)
304       - To turn lambda-splitting for a.sml on with aggressiveness <a> (where
305         <a> is a decimal non-negative integer):
306            a.sml (lambdasplitting:<a>)
307       - To turn lambda-splitting for a.sml on with maximal aggressiveness:
308            a.sml (lambdasplitting:infinity)
309       - To use the global default for a.sml:
310            a.sml (lambdasplitting:default)
311         or simply
312            a.sml
313    
314    ----------------------------------------------------------------------
315    Name: Allen Leung
316    Date: 2001/05/04 01:57:00 EDT
317    Tag: leunga-20010504-sync
318    Description:
319    
320      MLRISC features.
321    
322      1. Fix to CMPXCHG instructions.
323      2. Changed RA interface to allow annotations in callbacks.
324      3. Added a new method to the stream interface to allow annotations updates.
325    
326    ----------------------------------------------------------------------
327    Name: Matthias Blume
328    Date: 2001/05/01 11:45:00 EDT
329    Tag: blume-20010501-pcedittmp
330    Description:
331    
332    Changed install.sh to use the current working directory instead of
333    /usr/tmp for a temporary file (pcedittmp).  The previous choice
334    of /usr/tmp caused trouble with MacOS X because of file premission
335    problems.
336    
337    ----------------------------------------------------------------------
338    Name: Matthias Blume
339    Date: 2001/04/20 11:10:00 EDT
340    Tag: blume-20010420-inMLflag
341    Description:
342    
343     - added vp_limitPtrMask to vproc-state.h
344       (for use by the raw-C-calls mechanism to implement proper interrupt
345        handling)
346     - made the ML compiler aware of various data-structure offsets so it
347       can generate code for accessing the vp_inML flag and vp_limitPtrMask
348     - tweaked mlriscGen.sml to have it emit interrupt-handling code for
349       raw C-calls
350    
351    ----------------------------------------------------------------------
352    Name: Lal George
353    Date: 2001/04/20 09:15:28 EDT
354    Tag: george-20010420-macosX
355    Description:
356    
357     - Changes to port to Mac OS X; Darwin.
358    
359     - In the process I found that sqrt was broken on the PPC, because the
360       fsqrt instruction is not implemented.
361    
362    ----------------------------------------------------------------------
363    Name: Matthias Blume
364    Date: 2001/04/18 12:45:00 EDT
365    Tag: blume-20010418-ccalls
366    Description:
367    
368     - fixed two off-by-4 errors in the x86-specific c-calls implementation
369       (this bug prevented structure arguments containing pointers from being
370        passed correctly)
371     - changed the raw-C-call code in mlriscGen.sml in such a way that
372       structure arguments are represented as a pointer to the beginning
373       of the structure (instead of having a series of synthesized arguments,
374       one for each structure member)
375    
376     - made makeml script's verbosity level configurable via environment
377       variable (MAKEML_VERBOSITY)
378    
379     - eliminated placeholder implementations for f32l, w16s, i16s, and f32s
380       in rawmem-x86.sml; we are now using the real thing
381    
382    ----------------------------------------------------------------------
383    Name: Matthias Blume
384    Date: 2001/03/22 16:25:00 EST
385    Tag: blume-20010322-bootfiles
386    Description:
387    
388    Created a new set of bootfiles (for your automatic installation convenience).
389    
390    ----------------------------------------------------------------------
391    Name: Matthias Blume
392    Date: 2001/03/22 15:10:00 EST
393    Tag: blume-20010322-rawmem-parcm
394    Description:
395    
396    1. All "raw memory access" primitives for the new FFI are implemented now
397       (at least on the x86).
398    2. Some further cleanup of CM's parallel make mechanism.
399    
400    ----------------------------------------------------------------------
401    Name: Matthias Blume
402    Date: 2001/03/19 17:53:00 EST
403    Tag: blume-20010319-parallel
404    Description:
405    
406    Parallel make (using compile servers) now works again.
407    
408    To this end, CM.stabilize and CMB.make have been modified to work in
409    two passes when compile servers are attached:
410       1. Compile everything, do not perform stabilization; this pass
411          uses compile servers
412       2. Stabilize everything; this pass does not use compile servers
413    If there are no compile servers, the two passes are combined into one
414    (as before).  Splitting the passes increases the inherent parallelism
415    in the dependency graph because the entire graph including all
416    libraries is available at the same time.  This, in turn, improves
417    server utilization.  The downside is that the master process will
418    have to do some extra work after compilation is done (because for
419    technical reasons it must re-read all the binfiles during stabilization).
420    
421    ----------------------------------------------------------------------
422    Name: Matthias Blume
423    Date: 2001/03/16 12:22:00 EST
424    Tag: blume-20010316-bootfiles
425    Description:
426    
427    Created a new set of bootfiles (for your automatic installation convenience).
428    
429    ----------------------------------------------------------------------
430    Name: Matthias Blume
431    Date: 2001/03/16 11:00:00 EST
432    Tag: blume-20010316-MLTREE-fixup
433    Description:
434    
435    This is a minor fixup for an (untagged) earlier commit by Allen.
436    (A file was missing).
437    
438    ----------------------------------------------------------------------
439    Name: Allen Leung
440    Date: Mon Mar  5 18:54:57 EST 2001
441    Tag: leunga-20010305-cut-support
442    
443    1. New support for alternative control-flow in MLTREE.
444       Currently we support
445    
446          FLOW_TO(CALL ...., [k1,...,kn])
447    
448       This is needed for 'cuts to' in C-- and try/handle-like constructs
449       in Moby
450    
451       New assembler flag "asm-show-cutsto" to turn on control-flow debugging.
452    
453    2. Register Allocator
454    
455       Changes in interface [from Fermin, John]
456    
457    3. Alpha 8-bit SLL support [Fermin]
458    
459    4. All architectures
460    
461       A new module (ClusterExpandCopies) for expanding parallel copies.
462    
463    ----------------------------------------------------------------------
464    Name: Allen Leung
465    Date: 2001/02/27 23:07:00 EST
466    Tag: leunga-20010227-minor-stuff
467    
468    1. Alpha bug fix for CMOVNE
469    2. Handle mltree COND(..,FCMP ...,...)
470    3. Bug fix in simplifier
471    
472    ----------------------------------------------------------------------
473    Name: Matthias Blume
474    Date: 2001/01/30 17:50:00 EST
475    Tag: blume-20010130-sync
476    Description:
477    
478    This is just a minor update to sync my devel branch with the main brach.
479    The only visible change is the addition of some README files.
480    
481    ----------------------------------------------------------------------
482    Name: Matthias Blume
483    Date: 2001/01/12 23:30:00 JST
484    Tag: blume-20010112-bootfiles
485    Description:
486    
487    Made a new set of bootfiles that goes with the current state of the
488    repository.
489    
490    ----------------------------------------------------------------------
491    Name: Matthias Blume
492    Date: 2001/01/12 21:20:00 JST
493    Tag: blume-20010112-sync
494    Description:
495    
496    I am just flushing out some minor changes that had accumulated in
497    my private branch in order to sync with the main tree.  (This is
498    mainly because I had CVS trouble when trying to merge _into_ my
499    private branch.)
500    
501    Most people should be completely unaffected by this.
502    
503    ----------------------------------------------------------------------
504    Name: Allen Leung
505    Date: Thu Jan 11 21:03:00 EST 2001
506    Tag: leunga-20010111-labexp=mltree
507    Description:
508    
509    1.  Removed the type LabelExp and replace it by MLTree.
510    2.  Rewritten mltree-simplify with the pattern matcher tool.
511    3.  There were some bugs in alpha code generator which would break
512        64-bit code generation.
513    4.  Redo the tools to generate code with the
514    5.  The CM files in MLRISC (and in src/system/smlnj/MLRISC)
515        are now generated by perl scripts.
516    
517    ----------------------------------------------------------------------
518    Name: Matthias Blume
519    Date: 2001/01/10 21:55:00 JST
520    Tag: blume-20010110-rcc
521    Description:
522    
523    The RCC stuff now seems to work (but only on the x86).
524    This required hacking of the c-calls interface (and -implementation) in
525    MLRISC.
526    
527    Normal compiler users should be unaffected.
528    
529    ----------------------------------------------------------------------
530    Name: Matthias Blume
531    Date: 2001/01/09 01:20:00 JST
532    Tag: blume-20010109-rcc
533    Description:
534    
535    This is a fairly big patch, flushing out a large number of pending
536    changes that I made to my development copy over the last couple of days.
537    
538    Of practical relevance at this moment is a workaround for a pickling
539    bug that Allen ran into the other day.  The cause of the bug itself is
540    still unknown and it might be hard to fix it properly, but the
541    workaround has some merits of its own (namely somewhat reducing pickling
542    overhead for certain libraries).  Therefore, I think this solution should
543    be satisfactory at this time.
544    
545    The rest of the changes (i.e., the vast majority) has to do with my
546    ongoing efforts of providing direct support for C function calls from
547    ML.  At the moment there is a new primop "RAW_CCALL", typing magic
548    in types/cproto.sml (invoked from FLINT/trans/translate.sml), a new
549    case in the FLINT CPS datatype (RCC), changes to cps/convert.sml to
550    translate uses of RAW_CCALL into RCC, and changes to mlriscGen.sml to
551    handle RCC.
552    
553    The last part (the changes to mlriscGen.sml) are still known to be
554    wrong on the x86 and not implemented on all other architectures.  But
555    the infrastructure is in place. I had to change a few functor
556    signatures in the backend to be able to route the CCalls interface
557    from MLRISC there, and I had to specialize the mltree type (on the
558    x86) to include the necessary extensions. (The extensions themselves
559    were already there and redy to go in MLRISC/x86).
560    
561    Everything should be very happy as soon as someone helps me with
562    mlriscGen.sml...
563    
564    In any case, nothing of this should matter to anyone as long as the
565    new primop is not being used (which is going to be the case unless you
566    find it where I hid it :). The rest of the compiler is completely
567    unaffected.
568    
569    ----------------------------------------------------------------------
570    Name: Matthias Blume
571    Date: 2001/01/05 00:30:00 JST
572    Tag: blume-20010105-primops
573    Description:
574    
575    Added some experimental support for work that I am doing right now.
576    These changes mostly concern added primops, but there is also a new
577    experimental C library in the runtime system (but currently not enabled
578    anywhere except on Linux/X86).
579    
580    In the course of adding primops (and playing with them), I discovered that
581    Zhong's INL_PRIM hack (no type info for certain primops) was, in fact, badly
582    broken.  (Zhong was very right he labeled this stuff as "major gross hack".)
583    To recover, I made type information in INL_PRIM mandatory and changed
584    prim.sml as well as built-in.sml accordingly.  The InLine structure now
585    has complete, correct type information (i.e., no bottom types).
586    
587    Since all these changes mean that we need new binfiles, I also bumped the
588    version number to 110.32.1.
589    
590    ----------------------------------------------------------------------
591    Name: Matthias Blume
592    Date: 2000/12/30 22:10:00 JST
593    Tag: blume-20001230-various
594    Description:
595    
596    Added proxy libraries for MLRISC and let MLRISC libraries refer
597    to each other using path anchors.  (See CM manual for explanation.)
598    
599    Updated CM documentation.
600    
601    Fixed some bugs in CM.
602    
603    Implemented "proxy" libraries (= syntactic sugar for CM).
604    
605    Added "-quiet" option to makeml and changed runtime system accordingly.
606    
607    Added cleanup handler for exportML to reset timers and compiler stats.
608    
609    ----------------------------------------------------------------------
610    Name: Lal George
611    Date: 2000/12/22 22:22:58 EST 2000
612    Tag: Release_110_32
613    Description:
614    
615            Infinite precision used throughout MLRISC.
616            see MLRISC/mltree/machine-int.sig
617    
618    ----------------------------------------------------------------------
619    Name: Matthias Blume
620    Date: 2000/12/22 23:16:00 JST
621    Tag: blume-20001222-warn
622    Description:
623    
624    Corrected wording and formatting of some CM warning message which I
625    broke in my previous patch.
626    
627    ----------------------------------------------------------------------
628    Name: Matthias Blume
629    Date: 2000/12/22 21:20:00 JST
630    Tag: blume-20001222-anchorenv
631    Description:
632    
633    Fixed CM's handling of anchor environments in connection with CMB.make.
634    
635    ----------------------------------------------------------------------
636    Name: Matthias Blume
637    Date: 2000/12/22 13:15:00 JST
638    Tag: blume-20001222-cleanup
639    Description:
640    
641    Removed src/cm/ffi which does not (and did not) belong here.
642    
643    ----------------------------------------------------------------------
644    Name: Matthias Blume
645    Date: 2000/12/21 23:55:00 JST
646    Tag: blume-20001221-exn
647    Description:
648    
649    Probably most important: CM no longer silently swallows all exceptions
650    in the compiler.
651    Plus: some other minor CM changes.  For example, CM now reports some
652    sizes for generated binfiles (code, data, envpickle, lambdapickle).
653    
654    ----------------------------------------------------------------------
655    Name: Matthias Blume
656    Date: 2000/12/15 00:01:05 JST
657    Tag: blume-20001215-dirtool
658    Description:
659    
660    - "dir" tool added.
661    - improvements and cleanup to Tools structure
662    - documentation updates
663    
664    ----------------------------------------------------------------------
665    Name: Allen Leung
666    Date: Thu Dec 14 03:45:24 EST 2000
667    Description:
668    Tag:  leunga-20001214-int-inf
669    Description:
670    
671       In IntInf, added these standard functions, which are missing from our
672    implementation:
673    
674        andb : int * int -> int
675        xorb : int * int -> int
676        orb  : int * int -> int
677        notb : int -> int
678         <<   : int * word -> int
679        ~>>  : int * word -> int
680    
681       Not tested, I hope they are correct.
682    
683    ----------------------------------------------------------------------
684    Name: Allen Leung
685    Date: Fri Dec  8 19:23:26 EST 2000
686    Description:
687    Tag:  leunga-20001208-nowhere
688    Description:
689    
690      Slight improvements to the 'nowhere' tool to handle OR-patterns,
691    to generate better error messages etc.  Plus a brief manual.
692    
693    ----------------------------------------------------------------------
694    Name: Lal George
695    Date: 2000/12/08 09:54:02 EST 2000
696    Tag: Release_110_31
697    Description:
698    
699    - Version 110.31
700    ----------------------------------------------------------------------
701    Name: Allen Leung
702    Date: Thu Dec  7 22:01:04 EST 2000
703    Tag:  leunga-20001207-cell-monster-hack
704    Description:
705    
706    Major MLRISC internal changes.  Affect all clients.
707    Summary:
708    
709    1.  Type CELLS.cell = int is now replaced by a datatype.
710        As a result, the old regmap is now gone.  Almost all interfaces
711        in MLRISC change as a consequence.
712    
713    2.  A new brand version of machine description tool (v3.0) that generates
714        modules expecting the new interface.  The old version is removed.
715    
716    3.  The RA interface has been further abstracted into two new functors.
717        RISC_RA and X86RA.  These functors have much simpler interfaces.
718        [See also directory MLRISC/demo.]
719    
720    4.  Some other new source->source code generation tools are available:
721    
722        a. MLRISC/Tools/RewriteGen -- generate rewriters from rules.
723        b. MLRISC/Tools/WhereGen -- expands conditional pattern matching rules.
724           I use this tool to generate the peephole optimizers---with the new
725           cell type changes, peephole rules are becoming difficult to write
726           without conditional pattern matching.
727    
728    5.  More Intmap -> IntHashTable change.  Previous changes by Matthias didn't
729        cover the entire MLRISC source tree so many things broke.
730    
731    6.  CM files have been moved to the subdirectory MLRISC/cm.
732        They are moved because there are a lot of them and they clutter up the
733        root dir.
734    
735    7.  More detailed documentation to come...
736    
737        NOTE: To rebuild from 110.30 (ftp distribution), you'll have to do
738        a makeml -rebuild first.  This is because of other other
739        changes that Matthias has made (see below).
740    
741    
742    ----------------------------------------------------------------------
743    Name: Matthias Blume
744    Date: 2000/11/30 23:12:00 JST
745    Tag: blume-20001130-filereorg
746    Description:
747    
748    Some manual updates and some file reorganizations in CM.
749    
750    ----------------------------------------------------------------------
751    Name: Matthias Blume
752    Date: 2000/11/24 17:45:00 JST
753    Tag: blume-20001124-link
754    Description:
755    
756    Drastically improved link traversal code for the case that the dynamic
757    value was already loaded at bootstrap time.  As a result, CM and CMB
758    now both load blazingly fast -- even on a very slow machine.  Also,
759    memory consumption has been further reduced by this.
760    
761    Warning: The format of the PIDMAP file has changed.  THerefore, to
762    bootstrap you have to do this:
763    
764    1. Run CMB.make
765    2. Make a symbolic link for the boot directory:
766         ln -s sml.boot.ARCH-OS xxx
767    3. "Rebuild" the boot directory:
768         ./makeml -boot xxx -rebuild sml ; rm xxx
769    4. Boot normally:
770          ./makeml
771    
772    ----------------------------------------------------------------------
773    Name: Matthias Blume
774    Date: 2000/11/21 21:20:00 JST
775    Tag: blume-20001121-tools
776    Description:
777    
778    Continued hacking on autoloading problem -- with success this time.
779    Also changed tool-plugin mechanism.  See new CM manual.
780    
781    ----------------------------------------------------------------------
782    Name: Matthias Blume
783    Date: 2000/11/19 14:30:00 JST
784    Tag:  blume-20001119-autoload
785    Description:
786    
787    Some hacking to make autoloading faster.  Success for CMB, no success
788    so far for CM.  There is a reduced structure CM' that autoloads faster.
789    (This is a temporary, non-documented hack to be eliminated again when
790    the general problem is solved.)
791    
792    ----------------------------------------------------------------------
793    Name: Matthias Blume
794    Date: 2000/11/17 14:10:00 JST
795    Tag: blume-20001117-pickle-lib
796    Description:
797    
798    1. Eliminated comp-lib.cm
799    2. Made pickle-lib.cm
800    3. Eliminated all uses of intset.sml (from comp-lib.cm)
801    4. Replaced all uses of intmap.{sig,sml} (from comp-lib.cm) with
802       equivalent constructs from smlnj-lib.cm (INtHashTable).
803    5. Point 4. also goes for those uses of intmap.* in MLRISC.
804       Duplicated intmap modules thrown out.
805    6. Hunted down all duplicated SCC code and replaced it with
806       equivalent stuff (GraphSCCFn from smlnj-lib.cm).
807    7. Rewrote Feedback module.
808    8. Moved sortedlist.sml into viscomp-lib.cm.  Eventually it
809       should be thrown out and equivalent modules from smlnj-lib.cm
810       should be used (IntRedBlackSet, IntListSet, ...).
811    
812    Confirmed that compiler compiles to fixpoint.
813    
814    ----------------------------------------------------------------------
815    Name: Allen Leung
816    Date: 2000/11/10 18:00:00
817    Tag: leunga-20001110-new-x86-fp
818    
819    A new x86 floating point code generator has been added.
820    By default this is turned off.  To turn this on, do:
821    
822        CM.autoload "$smlnj/compiler.cm";
823        Compiler.Control.MLRISC.getFlag "x86-fast-fp" := true;
824    
825    Changes:
826    
827    1.  Changed FTAN to FPTAN so that the assembly output is correct.
828    2.  Changed the extension callback for FTANGENT to generate:
829    
830              fptan
831              fstp  %st(0)
832        instead of
833              fptan
834              fstpl ftempmem
835    
836    3.  Numerous assembly fixes for x86.
837    
838    5.  Cleaned up the machine code output module x86/x86MC.sml and added
839        support for a whole bunch of instructions and addressing modes:
840    
841          fadd/fsub/fsubr/fmul/fdiv/fdivr  %st, %st(n)
842          faddp/fsubp/fsubrp/fmulp/fdivp/fdivrp  %st, %st(n)
843          fadd/fsub/fsubr/fmul/fdiv/fdivr  %st(n), %st
844          fiadd/fisub/fisubr/fimul/fidiv/fidivr mem
845          fxch %st(n)
846          fld %st(n)
847          fst %st(n)
848          fst mem
849          fstp %st(n)
850          fucom %st(n)
851          fucomp %st(n)
852    
853        All these are now generated when the fast fp mode is turned on.
854    
855    6.  Removed the dedicated registers %st(0), ..., %st(7) from X86CpsRegs
856    
857    ----------------------------------------------------------------------
858    Name: Matthias Blume
859    Date: 2000/11/09 11:20:00 JST
860    Tag: blume-20001109-scc
861    Description:
862    
863    Eliminated some code duplication:
864    
865    1. Added "where" clause to GraphSCCFn in SML/NJ Library.
866       (Otherwise the functor is useless.)
867    2. Used GraphSCCFn where SCCUtilFun was used previously.
868    3. Got rid of SCCUtilFun (in comp-lib.cm).
869    
870    ----------------------------------------------------------------------
871    Name: Lal George
872    Date: 2000/11/06 09:02:21 EST 2000
873    Tag: Release_110_30
874    Description:
875    
876    - Version 110.30
877    ----------------------------------------------------------------------
878    Name: Matthias Blume
879    Date: 2000/11/04 14:45:00
880    Tag: blume-20001104-mlbuild
881    Description:
882    
883    - Made ml-build faster on startup.
884    - Documentation fixes.
885    
886    ----------------------------------------------------------------------
887    Name: Matthias Blume
888    Date: 2000/11/02 17:00:00 JST
889    Tag: blume-20001102-condcomp
890    Description:
891    
892    - Small tweaks to pickler -- new BOOTFILES!
893    - Version bumped to 110.29.2.
894    - Added conditional compilation facility to init.cmi (see comment there).
895    ----------------------------------------------------------------------
896    Name: Allen Leung
897    Date: 2000/10/23 19:31:00
898    Tag: leunga-20001023-demo-ra
899    
900    1. Minor RA changes that improves spilling on x86 (affects Moby and C-- only)
901    2. Test programs for the graph library updated
902    3. Some new MLRISC demo programs added
903    
904    ----------------------------------------------------------------------
905    Name: Matthias Blume
906    Date: 2000/08/31 22:15:00 JST
907    Tag: blume-20001017-errmsg
908    Description:
909    
910    More error message grief: Where there used to be no messages, there
911    now were some that had bogus error regions.  Fixed.
912    
913    ----------------------------------------------------------------------
914    Name: Matthias Blume
915    Date: 2000/08/31 17:30:00 JST
916    Tag: blume-20001017-v110p29p1
917    Description:
918    
919    I made a version 110.29.1 with new bootfiles.
920    
921    Changes:  Modified pickler/unpickler for faster and leaner unpickling.
922              CM documentation changes and a small bugfix in CM's error reporting.
923    
924    ----------------------------------------------------------------------
925    Name: Lal George
926    Date: 2000/09/27 14:42:35 EDT
927    Tag: george-20000927-nodestatus
928    Description:
929    
930    Changed the type of the nodestatus, so that:
931    
932            SPILLED(~1)             is now SPILLED
933            SPILLED(m) where m>=0   is now MEMREG(m)
934            SPILLED(s) where s<~1   is now SPILL_LOC(~s)
935    
936    ----------------------------------------------------------------------
937    Name: Matthias Blume
938    Date: 2000/09/07 14:45:00 JST
939    Tag: blume-20000907-cmerrmsg
940    Description:
941    
942    Small tweak to CM to avoid getting ML syntax error messages twice.
943    
944    ----------------------------------------------------------------------
945    Name: Matthias Blume
946    Date: 2000/08/31 18:00:00 JST
947    Tag: blume-20000831-cvsbootfiles
948    Description:
949    
950    New URL for boot files (because the 110.29 files on the BL server do
951    now work correctly with my updated install scripts for yacc and lex).
952    
953    ----------------------------------------------------------------------
954    Name: Matthias Blume
955    Date: 2000/08/08 12:33:00 JST
956    Tag: blume-20000808-manual
957    Description:
958    
959    Tiny update to CM manual.
960    
961    ----------------------------------------------------------------------
962    Name: Allen Leung
963    Date: 2000/08/7 19:31:00
964    Tag: leunga-20000807-a-whole-bunch-of-stuff
965    
966      Moby, C--, SSA, x86, machine descriptions etc.  Should only affect C--
967    and Mobdy.
968    
969    1.  x86
970    
971       a.  Fixes to peephole module by John and Dan.
972       b.  Assembly fix to SETcc by Allen.
973       c.  Fix to c-call by John.
974       d.  Fix to spilling by John.  (This one deals with the missing FSTPT case)
975       e.  Instruction selection optimization to SETcc as suggested by John.
976    
977           For example,
978    
979            MV(32, x, COND(32, CMP(32, LT, a, b), LI 1, LI 0))
980    
981           should generate:
982    
983            MOVL a, x
984            SUBL b, x
985            SHRL 31, x
986    
987    2.  IR stuff
988    
989         A bunch of new DJ-graph related algorithms added.  These
990         speed up SSA construction.
991    
992    3.  SSA + Scheduling
993    
994         Added code for SSA and scheduling to the repository
995    
996    ----------------------------------------------------------------------
997    Name: Lal George
998    Date: 2000/07/27 11:53:14 EDT
999    
1000    Tag: lal-20000727-linux-ppc
1001    Description:
1002    
1003     Made changes to support Linux PPC.
1004     p.s. I have confirmation that the 110.29 boot files work fine.
1005    
1006    ----------------------------------------------------------------------
1007    Name: Matthias Blume
1008    Date: 2000/07/27 17:40:00 JST
1009    Tag: blume-20000727-scripts
1010    Description:
1011    
1012    !!!! WARNING !!!!
1013    You must recompile the runtime system!
1014    !!!! WARNING !!!!
1015    
1016    This is basically another round of script-enhancements:
1017    
1018    1. sml, ml-build, and ml-makedepend accept options -D and -U to define
1019       and undefine CM preprocessor symbols.
1020    
1021    2. ml-build avoids generating a new heap image if it finds that the
1022       existing one is still ok.  (The condition is that no ML file had to
1023       be recompiled and all ML files are found to be older that the heap
1024       file.)
1025    
1026       To make this work smoothly, I also hacked the runtime system as
1027       well as SMLofNJ.SysInfo to get access to the heap image suffix
1028       (.sparc-solaris, ...) that is currently being used.
1029    
1030       Moreover, the signature of CM.mk_standalone has changed.  See the
1031       CM manual.
1032    
1033    3. ml-makedepend accepts additional options -n, -a, and -o.  (See the
1034       CM manual for details.)
1035    
1036    4. More CM manual updates:
1037        - all of the above has been documented.
1038        - there is now a section describing the (CM-related) command line
1039          arguments that are accepted by the "sml" command
1040    
1041    ----------------------------------------------------------------------
1042    Name: Matthias Blume
1043    Date: 2000/07/25 16:20:00 JST
1044    Tag: blume-20000725-makedepend
1045    Description:
1046    
1047    Added a script called ml-makedepend.  This can be used in makefiles
1048    for Unix' make in a way very similar to the "makedepend" command for
1049    C.
1050    
1051    The script internally uses function CM.sources.
1052    
1053    Synopsis:
1054    
1055        ml-makedepend [-f makefile] cmfile targetname
1056    
1057    The default for the makefile is "makefile" (or "Makefile" should
1058    "makefile" not exist).
1059    
1060    ml-makedepend adds a cmfile/targetname-specific section to this
1061    makefile (after removing the previous version of this section).  The
1062    section contains a single dependency specification with targetname on
1063    the LHS (targetname is an arbitrary name), and a list of files derived
1064    from the cmfile on the RHS.  Some of the files on the RHS are
1065    ARCH/OPSYS-specific.  Therefore, ml-makedepend inserts references to
1066    "make" variables $(ARCH) and $(OPSYS) in place of the corresponding
1067    path names.  The makefile writer is responsible for making sure that
1068    these variables have correct at the time "make" is invoked.
1069    
1070    ----------------------------------------------------------------------
1071    Name: Matthias Blume
1072    Date: 2000/07/22 23:30:00 JST
1073    Tag: blume-20000722-urlupdate
1074    Description:
1075    
1076    Changed BOOT and config/srcarchiveurl to point to BL server:
1077    
1078        ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.29/
1079    
1080    ----------------------------------------------------------------------
1081    Name: Matthias Blume
1082    Date: 2000/07/18 18:00:00 JST
1083    Tag: blume-20000718-Version_110_29
1084    Description:
1085    
1086    1. Updated src/compiler/TopLevel/main/version.sml to version 110.29
1087    
1088    2. Updated config/version to 110.29
1089    
1090    3. Updated config/srcarchiveurl
1091    
1092    3. New boot files!
1093       ftp://ftp.cs.princeton.edu/pub/people/blume/sml/110.29-autofetch
1094    
1095    ----------------------------------------------------------------------
1096    Name: Matthias Blume
1097    Date: 2000/07/11 13:58:00 JST
1098    Tag: blume-20000711-doctypo
1099    Description:
1100    
1101    Fixed a few typos in CM manual.
1102    
1103    ----------------------------------------------------------------------
1104    Name: Allen Leung
1105    Date: 2000/06/15 00:38:00
1106    Tag: leunga-20000704-sparc-x86
1107    
1108    1. x86 peephole improvement sp += k; sp -= k => nop  [from John]
1109    2. fix to x86 RET bug [found by Dan Grossman]
1110    3. sparc assembly bug fix for ticc instructions [found by Fermin]
1111    
1112       Affects c-- and moby only
1113    
1114    ----------------------------------------------------------------------
1115    Name: Matthias Blume
1116    Date: 2000/07/04 15:26:00
1117    Tag: blume-20000704-trigger
1118    Description:
1119    
1120    1. Improvements to CM manual.
1121    2. SMLofNJ.Internals.BTrace.trigger reinstated as an alternative way
1122       of getting a back-trace.  The function, when called, raises an
1123       internal exception which explicitly carries the full back-trace history,
1124       so it is unaffected by any intervening handle-raise pairs ("trivial"
1125       or not).  The interactive loop will print that history once it arrives
1126       at top level.
1127       Short of having all exceptions implicitly carry the full history, the
1128       recommended way of using this facility is:
1129         - compile your program with instrumentation "on"
1130         - run it, when it raises an exception, look at the history
1131         - if the history is "cut off" because of some handler, go and modify
1132           your program so that it explicitly calls BTrace.trigger
1133         - recompile (still instrumented), and rerun; look at the full history
1134    
1135    ----------------------------------------------------------------------
1136    Name: Matthias Blume
1137    Date: 2000/07/03 15:36:00 JST
1138    Tag: blume-20000702-manual
1139    Description:
1140    
1141    Small corrections and updates to CM manual.
1142    
1143    ----------------------------------------------------------------------
1144    Name: Matthias Blume
1145    Date: 2000/06/29 16:04:00 JST
1146    Tag: blume-20000629-yacctool
1147    Description:
1148    
1149    Changes:
1150    
1151    1. Class "mlyacc" now takes separate arguments to pass options to
1152       generated .sml- and .sig-files independently.
1153    2. Corresponding CM manual updates.
1154    3. BTrace module now also reports call sites.  (However, for loop clusters
1155       it only shows from where the cluster was entered.)  There are associated
1156       modifications to core.sml, internals.{sig,sml}, btrace.sml, and btimp.sml.
1157    
1158    ----------------------------------------------------------------------
1159    Name: Matthias Blume
1160    Date: 2000/06/27 16:51:00 JST
1161    Tag: blume-20000627-noweb
1162    Description:
1163    
1164    Changes:
1165    
1166     1. Implemented "subdir" and "witness" options for noweb tool.
1167        This caused some slight internal changes in CM's tool implementation.
1168     2. Fixed bug in "tool plugin" mechanism.  This is essentially cleaning
1169        some remaining issues from earlier path anchor changes.
1170     3. Updated CM manual accordingly.
1171    
1172     4. Changed implementation of back-tracing so that I now consider it
1173        ready for prime-time.
1174    
1175        In particular, you don't have to explicitly trigger the back-trace
1176        anymore.  Instead, if you are running BTrace-instrumented code and
1177        there is an uncaught exception (regardless of whether or not it was
1178        raised in instrumented code), the top-level evalloop will print
1179        the back-trace.
1180    
1181        Features:
1182    
1183          - Instrumented and uninstrumented code work together seemlessly.
1184            (Of course, uninstrumented code is never mentioned in actual
1185             back-traces.)
1186    
1187          - Asymptotic time- and space-complexity of instrumented code is
1188            equal to that of uninstrumented code.  (This means that
1189            tail-recursion is preserved by the instrumentation phase.)
1190    
1191          - Modules whose code has been instrumented in different sessions
1192            work together without problem.
1193    
1194          - There is no penalty whatsoever on uninstrumented code.
1195    
1196          - There is no penalty on "raise" expressions, even in
1197            instrumented code.
1198    
1199        A potential bug (or perhaps it is a feature, too):
1200    
1201          A back-trace reaches no further than the outermost instrumented
1202          non-trivial "raise".  Here, a "trivial" raise is one that is the
1203          sole RHS of a "handle" rule.  Thus, back-traces reach trough
1204    
1205               <exp> handle e => raise e
1206    
1207          and even
1208    
1209               <exp> handle Foo => raise Bar
1210    
1211          and, of course, through
1212    
1213               <exp> handle Foo => ...
1214    
1215         if the exception was not Foo.
1216    
1217         Back-traces always reach right through any un-instrumented code
1218         including any of its "handle" expressions, trivial or not.
1219    
1220       To try this out, do the following:
1221    
1222         - Erase all existing binfiles for your program.
1223           (You may keep binfiles for those modules where you think you
1224            definitely don't need back-tracing.)
1225         - Turn on back-trace instrumentation:
1226              SMLofNJ.Internals.BTrace.mode (SOME true);
1227         - Recompile your program.  (I.e., run "CM.make" or "use".)
1228         - You may now turn instrumentation off again (if you want):
1229              SMLofNJ.Internals.BTrace.mode (SOME false);
1230         - Run your program as usual.  If it raises an exception that
1231           reaches the interactive toplevel, then a back-trace will
1232           automatically be printed.  After that, the toplevel loop
1233           will print the exception history as usual.
1234    
1235    ----------------------------------------------------------------------
1236    Name: Matthias Blume
1237    Date: 2000/06/26 09:56:46 JST
1238    Tag: blume-20000626-setup
1239    Description:
1240    
1241    CM: - setup-parameter to "sml" added; this can be used to run arbitrary
1242          ML code before and after compiling a file (e.g., to set compiler
1243          flags)
1244    
1245    Compiler: - improved btrace API (in core.sml, internals.{sig,sml})
1246              - associated changes to btrace.sml (BTrace instrumentation pass)
1247              - cleaner implementation of btimp.sml (BTrace tracing and report
1248                module)
1249    
1250    CM manual: * new path encoding documented
1251               * description of setup-parameter to "sml" added
1252    
1253    The biggest user-visible change to back-tracing is that it is no
1254    longer necessary to compile all traced modules within the same
1255    session.  (This was a real limitation.)
1256    
1257    ----------------------------------------------------------------------
1258    Name: Matthias Blume
1259    Date: 2000/06/24 12:40:00 JST
1260    Tag: blume-20000624-startup
1261    Description:
1262    
1263    Fixes startup slowdown problem.  (I was calling SrcPath.sync a _tad_
1264    bit too often -- to put it mildly. :)
1265    
1266    ----------------------------------------------------------------------
1267    Name: Matthias Blume
1268    Date: 2000/06/23 18:20:00 JST
1269    Tag: blume-20000623-btrace
1270    Description:
1271    
1272    This updates adds a backtrace facility to aid programmers in debugging
1273    their programs.  This involves the following changes:
1274    
1275    1. Module system/smlnj/init/core.sml (structure _Core) now has hooks for
1276       keeping track of the current call stack.  When programs are compiled
1277       in a special mode, the compiler will insert calls to these hooks
1278       into the user program.
1279       "Hook" means that it is possible for different implementations of
1280       back-tracing to register themselves (at different times).
1281    
1282    2. compiler/MiscUtil/profile/btrace.sml implements the annotation phase
1283       as an Absyn.dec->Absyn.dec rewrite.  Normally this phase is turned off.
1284       It can be turned on using this call:
1285         SMLofNJ.Internals.BTrace.mode (SOME true);
1286       Turning it off again:
1287         SMLofNJ.Internals.BTrace.mode (SOME false);
1288       Querying the current status:
1289         SMLofNJ.Internals.BTrace.mode NONE;
1290       Annotated programs are about twice as big as normal ones, and they
1291       run a factor of 2 to 4 slower with a dummy back-trace plugin (one
1292       where all hooks do nothing).  The slowdown with a plugin that is
1293       actually useful (such as the one supplied by default) is even greater,
1294       but in the case of the default plugin it is still only an constant
1295       factor (amortized).
1296    
1297    3. system/Basis/Implementation/NJ/internals.{sig,sml} have been augmented
1298       with a sub-structure BTrace for controlling back-tracing.  In particular,
1299       the above-mentioned function "mode" controls whether the annotation
1300       phase is invoked by the compiler.  Another important function is
1301       "trigger": when called it aborts the current execution and causes
1302       the top-level loop to print a full back-trace.
1303    
1304    4. compiler/MiscUtil/profile/btimp.sml is the current default plugin
1305       for back-tracing.  It keeps track of the dynamic call stack and in
1306       addition to that it keeps a partial history at each "level" of that
1307       stack.  For example, if a tail-calls b, b tail-calls c, and c tail-calls
1308       d and b (at separate times, dynamically), then the report will show:
1309    
1310       GOTO   d
1311             /c
1312       GOTO  \b
1313       CALL   a
1314    
1315       This shows that there was an initial non-tail call of a, then a
1316       tail-call to b or c, looping behavior in a cluster of functions that
1317       consist of b and c, and then a goto from that cluster (i.e., either from
1318       b or from c) to d.
1319    
1320       Note that (depending on the user program) the amount of information
1321       that the back-trace module has to keep track of at each level is bounded
1322       by a constant.  Thus, the whole implementation has the same asymptotical
1323       complexity as the original program (both in space and in time).
1324    
1325    5. compiler/TopLevel/interact/evalloop.sml has been modified to
1326       handle the special exception SMLofNJ.Internals.BTrace.BTrace
1327       which is raised by the "trigger" function mentioned above.
1328    
1329    Notes on usage:
1330    
1331    - Annotated code works well together with unannotated code:
1332    Unannotated calls simply do not show up at all in the backtrace.
1333    
1334    - It is not a good idea to let modules that were annotated during
1335    different sessions run at the same time.  This is because the compiler
1336    chooses small integers to identify individual functions, and there
1337    will be clashes if different modules were compiled in separate sessions.
1338    (Nothing will crash, and you will even be told about the clashes, but
1339    back-trace information will in general not be useful.)
1340    
1341    - Back-tracing can be confused by callcc and capture.
1342    
1343    - The only way of getting a back-trace right now is to explicitly
1344    invoke the "trigger" function from your user program.  Eventually, we
1345    should make every exception carry back-trace information (if
1346    available).  But since this creates more overhead at "raise"-time
1347    (similar to the current exnHistory overhead), I have not yet
1348    implemented this.  (The implementation will be rather easy.)  With
1349    exceptions carrying back-trace information, this facility will be even
1350    more useful because users don't need to modify their programs...
1351    
1352    - While it is possible to compile the compiler with back-trace
1353    annotations turned on (I did it to get some confidence in
1354    correctness), you must make absolutely sure that core.sml and
1355    btimp.sml are compiled WITHOUT annotation!  (core.sml cannot actually
1356    be compiled with annotation because there is no core access yet, but
1357    if you compile btimp.sml with annotation, then the system will go into
1358    an infinite recursion and crash.)
1359    Since CM currently does not know about BTrace, the only way to turn
1360    annotations on and off for different modules of the compiler is to
1361    interrupt CMB.make, change the settings, and re-invoke it.  Of course,
1362    this is awkward and clumsy.
1363    
1364    Sample sessions:
1365    
1366    Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
1367    - SMLofNJ.Internals.BTrace.mode (SOME true);
1368    [autoloading]
1369    [autoloading done]
1370    val it = false : bool
1371    - structure X = struct
1372    -     fun main n = let
1373    -         fun a (x, 0) = d x
1374    -           | a (x, n) = b (x, n - 1)
1375    -         and b (x, n) = c (x, n)
1376    -         and c (x, n) = a (x, n)
1377    -         and d x = e (x, 3)
1378    -         and e (x, 0) = f x
1379    -           | e (x, n) = e (x, n - 1)
1380    -         and f 0 = SMLofNJ.Internals.BTrace.trigger ()
1381    -           | f n = n * g (n - 1)
1382    -         and g n = a (n, 3)
1383    -     in
1384    -         f n
1385    -     end
1386    - end;
1387    structure X : sig val main : int -> int end
1388    - X.main 3;
1389    *** BACK-TRACE ***
1390    GOTO   stdIn:4.2-13.20: X.main[2].f
1391    GOTO-( stdIn:4.2-13.20: X.main[2].e
1392    GOTO   stdIn:4.2-13.20: X.main[2].d
1393         / stdIn:4.2-13.20: X.main[2].a
1394         | stdIn:4.2-13.20: X.main[2].b
1395    GOTO-\ stdIn:4.2-13.20: X.main[2].c
1396    CALL   stdIn:4.2-13.20: X.main[2].g
1397    GOTO   stdIn:4.2-13.20: X.main[2].f
1398    GOTO-( stdIn:4.2-13.20: X.main[2].e
1399    GOTO   stdIn:4.2-13.20: X.main[2].d
1400         / stdIn:4.2-13.20: X.main[2].a
1401         | stdIn:4.2-13.20: X.main[2].b
1402    GOTO-\ stdIn:4.2-13.20: X.main[2].c
1403    CALL   stdIn:4.2-13.20: X.main[2].g
1404    GOTO   stdIn:4.2-13.20: X.main[2].f
1405    GOTO-( stdIn:4.2-13.20: X.main[2].e
1406    GOTO   stdIn:4.2-13.20: X.main[2].d
1407         / stdIn:4.2-13.20: X.main[2].a
1408         | stdIn:4.2-13.20: X.main[2].b
1409    GOTO-\ stdIn:4.2-13.20: X.main[2].c
1410    CALL   stdIn:4.2-13.20: X.main[2].g
1411    GOTO   stdIn:4.2-13.20: X.main[2].f
1412    CALL   stdIn:2.15-17.4: X.main[2]
1413    -
1414    
1415    (Note that because of a FLINt bug the above code currently does not
1416    compile without BTrace turned on.)
1417    
1418    Here is another example, using my modified Tiger compiler:
1419    
1420    Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
1421    - SMLofNJ.Internals.BTrace.mode (SOME true);
1422    [autoloading]
1423    [autoloading done]
1424    val it = false : bool
1425    - CM.make "sources.cm";
1426    [autoloading]
1427    ...
1428    [autoloading done]
1429    [scanning sources.cm]
1430    [parsing (sources.cm):parse.sml]
1431    [creating directory CM/SKEL ...]
1432    [parsing (sources.cm):tiger.lex.sml]
1433    ...
1434    [wrote CM/sparc-unix/semant.sml]
1435    [compiling (sources.cm):main.sml]
1436    [wrote CM/sparc-unix/main.sml]
1437    [New bindings added.]
1438    val it = true : bool
1439    - Main.compile ("../testcases/merge.tig", "foo.out");
1440    *** BACK-TRACE ***
1441    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trvar
1442    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1443    CALL   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
1444    GOTO   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
1445    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1446    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1447    CALL   lib/semant.sml:488.3-505.6: SemantFun[2].transDec.trdec[2].transBody[2]
1448         / lib/semant.sml:411.65-543.8: SemantFun[2].transDec
1449    CALL-\ lib/semant.sml:413.2-540.9: SemantFun[2].transDec.trdec[2]
1450    CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
1451    CALL   lib/semant.sml:8.52-558.4: SemantFun[2].transProg[2]
1452    CALL   main.sml:1.18-118.4: Main.compile[2]
1453    -
1454    
1455    ----------------------------------------------------------------------
1456    Name: Matthias Blumen
1457    Date: 2000/06/21 18:00:00 JST
1458    Tag: blume-20000621-manual
1459    Description:
1460    
1461    CM manual update: Path environments documented.
1462    
1463    ----------------------------------------------------------------------
1464    Name: Matthias Blume
1465    Date: 2000/06/19 13:40:00
1466    Tag: blume-20000619-manual
1467    Description:
1468    
1469    CM manual and system/README update.  This only covers the fact that
1470    there are no more implicit anchors.  (Path environments and the "bind"
1471    option to "cm" have yet to be documented.)
1472    
1473    ----------------------------------------------------------------------
1474    Name: Matthias Blume
1475    Date: 2000/06/19 11:05:00 JST
1476    Tag: blume-20000619-chdir-bugfix
1477    Description:
1478    
1479    Fixed a bug in new SrcPath module that sometimes led to a bad chDir call.
1480    
1481    ----------------------------------------------------------------------
1482    Name: Matthias Blume
1483    Date: 2000/06/18 22:00:10 JST
1484    Tag: blume-20000618-implicit-anchors-really-gone
1485    Description:
1486    
1487    I updates the previous HISTORY entry where I forgot to mention that
1488    implicit anchors are no longer with us.
1489    
1490    The current update also gets rid of the (now useless) controller
1491    CM.Control.implicit_anchors.
1492    
1493    ----------------------------------------------------------------------
1494    Name: Matthias Blume
1495    Date: 2000/06/16 17:30:00 JST
1496    Tag: blume-20000616-anchorenv
1497    Description:
1498    
1499    This patch implements the long anticipated (just kidding :) "anchor
1500    environment" mechanism.  In the course of doing this, I also
1501    re-implemented CM's internal "SrcPath" module from scratch.  The new
1502    one should be more robust in certain boundary cases.  In any case, it
1503    is a lot cleaner than its predecessor (IMHO).
1504    
1505    This time, although there is yet another boot file format change, I
1506    kept the unpickler backward-compatible.  As a result, no new bootfiles
1507    are necessary and bootstrapping is straightforward.  (You cannot read
1508    new bootfiles into an old system, but the other way around is no
1509    problem.)
1510    
1511    Visible changes:
1512    
1513    ** 0. Implicit path anchors (without the leading $-symbol) are no
1514    longer recognized at all. This means that such path names are not
1515    illegal either.  For example, the name basis.cm simply refers to a
1516    local file called "basis.cm" (i.e, the name is an ordinary path
1517    relative to .cm-files directory).  Or, to put it differently, only
1518    names that start with $ are anchored paths.
1519    
1520    ** 1. The $<singlearc> abbreviation for $/<singlearc> has finally
1521    vanished.
1522    
1523    John (Reppy) had critizised this as soon as I originally proposed and
1524    implemented it, but at that time I did not really deeply believe
1525    him. :) Now I came full-circle because I need the $<singlearc> syntax
1526    in another place where it cannot be seen as an abbreviation for
1527    $/<singlearc>.  To avoid the confusion, $<singlearc> now means what it
1528    seems to mean (i.e., it "expands" into the corresponding anchor
1529    value).
1530    
1531    However, when paths are used as members in CM description files, it
1532    continues to be true that there must be at least another arc after the
1533    anchor.  This is now enforced separately during semantic analysis
1534    (i.e., from a lexical/syntactical point of view, the notation is ok.)
1535    
1536    ** 2. The "cm" class now accepts an option "bind".  The option's value
1537    is a sub-option list of precisely two items -- one labeled "anchor"
1538    and the other one labeled "value".  As you might expect, "anchor" is
1539    used to specify an anchor name to be bound, and "value" specifies what
1540    the anchor is being bound to.
1541    
1542    The value must be a directory name and can be given in either standard
1543    syntax (including the possibility that it is itself an anchored path)
1544    or native syntax.
1545    
1546    Examples:
1547    
1548       foo.cm (bind:(anchor:bar value:$mystuff/bar))
1549       lib.cm (bind:(anchor:a value:"H:\\x\\y\\z"))  (* only works under windows *)
1550    
1551    and so on.
1552    
1553    The meaning of this is that the .cm-file will be processed with an
1554    augmented anchor environment where the given anchor(s) is/are bound to
1555    the given values(s).
1556    
1557    The rationale for having this feature is this: Suppose you are trying
1558    to use two different (already stable) libraries a.cm and b.cm (that
1559    you perhaps didn't write yourself).  Further, suppose each of these
1560    two libraries internally uses its own auxiliary library $aux/lib.cm.
1561    Normally you would now have a problem because the anchor "lib" can not
1562    be bound to more than one value globally.  Therefore, the project that
1563    uses both a.cm and b.cm must locally redirect the anchor to some other
1564    place:
1565    
1566       a.cm (bind:(anchor:lib value:/usr/lib/smlnj/a-stuff))
1567       b.cm (bind:(anchor:lib value:/usr/lib/smlnj/b-stuff))
1568    
1569    This hard-wires $lib/aux.cm to /usr/lib/smlnj/a-stuff/aux.cm or
1570    /usr/lib/smlnj/b-stuff/aux.cm, respectively.
1571    
1572    Hard-wiring path names is a bit inflexible (and CM will verbosely warn
1573    you when you do so at the time of CM.stabilize).  Therefore, you can
1574    also use an anchored path as the value:
1575    
1576      a.cm (bind:(anchor:lib value:$a-lib))
1577      b.cm (bind:(anchor:lib value:$b-lib))
1578    
1579    Now you can globally configure (using the usual CM.Anchor.anchor or
1580    pathconfig machinery) bindings for "a-lib" and "b-lib".  Since "lib"
1581    itself is always locally bound, setting it globally is no longer
1582    meaningful or necessary (but it does not hurt either).  In fact, "lib"
1583    can still be used as a global anchor for separate purposes.  As a
1584    matter of fact, one can locally define "lib" in terms of a global
1585    "lib":
1586    
1587      a.cm (bind:(anchor:lib value:$lib/a))
1588      b.cm (bind:(anchor:lib value:$lib/b))
1589    
1590    ** 3: The encoding of path names has changed.  This affects the way
1591    path names are shown in CM's progress report and also the internal
1592    protocol encoding used for parallel make.
1593    
1594    The encoding now uses one or more ':'-separated segments.  Each
1595    segments corresponds to a file that has been specified relative to the
1596    file given by its preceding segment.  The first segment is either
1597    relative to the CWD, absolute, or anchored.  Each segment itself is
1598    basically a Unix pathname; all segments but the first are relative.
1599    
1600    Example:
1601    
1602       $foo/bar/baz.cm:a/b/c.sml
1603    
1604    This path denotes the file bar/a/b/c.sml relative to the directory
1605    denoted by anchor "foo".  Notice that the encoding also includes
1606    baz.cm which is the .cm-file that listed a/b/c.sml.  As usual, such
1607    paths are resolved relative to the .cm-files directory, so baz.cm must
1608    be ignored to get the "real" pathname.
1609    
1610    To make this fact more obvious, CM puts the names of such "virtual
1611    arcs" into parentheses when they appear in progress reports. (No
1612    parentheses will appear in the internal protocol encoding.)  Thus,
1613    what you really see is:
1614    
1615      $foo/bar/(baz.cm):a/b/c.sml
1616    
1617    I find this notation to be much more informative than before.
1618    
1619    Another new feature of the encoding is that special characters
1620    including parentheses, colons, (back)slashes, and white space are
1621    written as \ddd (where ddd is the decimal encoding of the character).
1622    
1623    *** The CM manual still needs to be updated.
1624    
1625    ----------------------------------------------------------------------
1626    Name: Allen Leung
1627    Date: 2000/06/15 00:38:00
1628    Tag: leunga-20000615-x86-peephole
1629    
1630    x86 Peephole fix by Fermin.  Affects c-- and moby only.
1631    
1632    ----------------------------------------------------------------------
1633    Name: Matthias Blume
1634    Date: 2000/06/12 11:40:00
1635    Tag: blume-20000612-parmakefix
1636    Description:
1637    
1638    More cleanup after changing the file naming scheme: This time I
1639    repaired the parallel make mechanism for CMB.make which I broke earlier.
1640    
1641  ----------------------------------------------------------------------  ----------------------------------------------------------------------
1642  Name: Allen Leung  Name: Allen Leung
1643  Date: 2000/06/09 01:25:00  Date: 2000/06/09 01:25:00
# Line 590  Line 2217 
2217    
2218  3.  Assembly  3.  Assembly
2219    
2220      When generating assemby, resolve the value of client defined constants,      When generating assembly, resolve the value of client defined constants,
2221      instead of generating symbolic values.  This is controlled by the      instead of generating symbolic values.  This is controlled by the
2222      new flag "asm-resolve-constants", which is default to true.      new flag "asm-resolve-constants", which is default to true.
2223    
# Line 613  Line 2240 
2240    
2241  To this end, I arranged that instead of "structure Core" as "structure  To this end, I arranged that instead of "structure Core" as "structure
2242  _Core" is bound in the pervasive environment.  Core access is done via  _Core" is bound in the pervasive environment.  Core access is done via
2243  _Core (which can never be accidentially rebound because _Core is not a  _Core (which can never be accidentally rebound because _Core is not a
2244  legal surface-syntax symbol).  legal surface-syntax symbol).
2245    
2246  The current solution is much cleaner because the core environment is  The current solution is much cleaner because the core environment is
# Line 623  Line 2250 
2250  with dynamic and symbolic parts of the core environment.  with dynamic and symbolic parts of the core environment.
2251    
2252  Remaining hackery (to bind the "magic" symbol _Core) is localized in the  Remaining hackery (to bind the "magic" symbol _Core) is localized in the
2253  compilation mananger's bootstrap compiler (actually: in the "init group"  compilation manager's bootstrap compiler (actually: in the "init group"
2254  handling).  See the comments in src/system/smlnj/init/init.cmi for  handling).  See the comments in src/system/smlnj/init/init.cmi for
2255  more details.  more details.
2256    
# Line 738  Line 2365 
2365     (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
2366     used).     used).
2367    
2368     This was done to accomodate the new "make" and "shell" tools which     This was done to accommodate the new "make" and "shell" tools which
2369     facilitate fairly seemless hookup to portions of code managed using     facilitate fairly seamless hookup to portions of code managed using
2370     Makefiles or Shell scripts.     Makefiles or Shell scripts.
2371    
2372     There are no classes "shared" or "private" anymore.  Instead, the     There are no classes "shared" or "private" anymore.  Instead, the
# Line 751  Line 2378 
2378    
2379     All existing tools are described in the CM manual.     All existing tools are described in the CM manual.
2380    
2381     - Slightly better error handling.  (CM now surpresses many followup     - Slightly better error handling.  (CM now suppresses many followup
2382     error messages that tended to be more annoying than helpful.)     error messages that tended to be more annoying than helpful.)
2383    
2384  2. Major changes to the compiler's static environment data structures.  2. Major changes to the compiler's static environment data structures.
# Line 885  Line 2512 
2512    
2513       I've changed andl to testl in the floating point test sequence       I've changed andl to testl in the floating point test sequence
2514       whenever appropriate.  The Intel optimization guide states that       whenever appropriate.  The Intel optimization guide states that
2515       testl is perferable to andl.       testl is preferable to andl.
2516    
2517  7. RA (x86 only)  7. RA (x86 only)
2518    
# Line 1067  Line 2694 
2694    
2695  1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an  1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an
2696  additional argument called "template" which is an optional string that  additional argument called "template" which is an optional string that
2697  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
2698  explanation.  explanation.
2699    
2700  2. A special-purpose tool can be "regisitered" by simply dropping the  2. A special-purpose tool can be "registered" by simply dropping the
2701  corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same  corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same
2702  directory where the .cm file lives that uses this tool.  (The  directory where the .cm file lives that uses this tool.  (The
2703  behavior/misfeature until now was to look for the tool description  behavior/misfeature until now was to look for the tool description
# Line 1114  Line 2741 
2741  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
2742  <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,
2743  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
2744  perhaps got accidentially killed).  In this case, fixpt will quickly  perhaps got accidentally killed).  In this case, fixpt will quickly
2745  move through what exists before continuing where it left off earlier,  move through what exists before continuing where it left off earlier,
2746  and, thus, saves a lot of time.  and, thus, saves a lot of time.
2747    
# Line 1164  Line 2791 
2791    it from that remote directory.    it from that remote directory.
2792    This should simplify installation further:  For machines that have    This should simplify installation further:  For machines that have
2793    access to the internet, just fetch <version>-config.tgz, unpack it,    access to the internet, just fetch <version>-config.tgz, unpack it,
2794    edit config/targets, and go (run config/install.sh).  The scipt will    edit config/targets, and go (run config/install.sh).  The script will
2795    fetch everything else that it might need all by itself.    fetch everything else that it might need all by itself.
2796    
2797    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.657  
changed lines
  Added in v.840

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