Home My Page Projects Code Snippets Project Openings SML/NJ
Summary Activity Forums Tracker Lists Tasks Docs Surveys News SCM Files

SCM Repository

[smlnj] Annotation of /sml/trunk/HISTORY
ViewVC logotype

Annotation of /sml/trunk/HISTORY

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1079 - (view) (download)

1 : dbm 570 This is the HISTORY file for the Yale SML/NJ CVS repository.
2 :    
3 :     An entry should be made for _every_ commit to the repository.
4 :     The entries in this file will be used when creating the README
5 :     for new versions, so keep that in mind when writing the
6 :     description.
7 :    
8 :     The form of an entry should be:
9 :    
10 :     Name:
11 : george 1003 Date: yyyy/mm/dd
12 : dbm 570 Tag: <post-commit CVS tag>
13 :     Description:
14 : leunga 1025
15 : george 1003 ----------------------------------------------------------------------
16 : blume 1066 Name: Matthias Blume
17 : leunga 1079 Date: 2002/02/19 20:20:00 EST
18 :     Tag: leunga-20020219-mlrisc-tools
19 :     Description:
20 :    
21 :     Minor bug fixes to the mlrisc-tools library:
22 :    
23 :     1. Fixed up parsing colon suffixed keywords
24 :     2. Added the ability to shut the error messages up
25 :     3. Reimplemented the pretty printer and fixed up/improved
26 :     the pretty printing of handle and -> types.
27 :     4. Fixed up generation of literal symbols in the nowhere tool.
28 :     5. Added some SML keywords to to sml.sty
29 :    
30 :     ----------------------------------------------------------------------
31 :     Name: Matthias Blume
32 : blume 1078 Date: 2002/02/19 16:20:00 EST
33 :     Tag: blume-20020219-cmffi
34 :     Description:
35 :    
36 :     A wild mix of changes, some minor, some major:
37 :    
38 :     * All C FFI-related libraries are now anchored under $c:
39 :     $/c.cm --> $c/c.cm
40 :     $/c-int.cm --> $c/internals/c-int.cm
41 :     $/memory.cm --> $c/memory/memory.cm
42 :    
43 :     * "make" tool (in CM) now treats its argument pathname slightly
44 :     differently:
45 :     1. If the native expansion is an absolute name, then before invoking
46 :     the "make" command on it, CM will apply OS.Path.mkRelative
47 :     (with relativeTo = OS.FileSys.getDir()) to it.
48 :     2. The argument will be passed through to subsequent phases of CM
49 :     processing without "going native". In particular, if the argument
50 :     was an anchored path, then "make" will not lose track of that anchor.
51 :    
52 :     * Compiler backends now "know" their respective C calling conventions
53 :     instead of having to be told about it by ml-nlffigen. This relieves
54 :     ml-nlffigen from one of its burdens.
55 :    
56 :     * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
57 :    
58 :     * Export C_DEBUG and C_Debug from $c/c.cm.
59 :    
60 :     * C type encoding in ml-nlffi-lib has been improved to model the conceptual
61 :     subtyping relationship between incomplete pointers and their complete
62 :     counterparts. For this, ('t, 'c) ptr has been changed to 'o ptr --
63 :     with the convention of instantiating 'o with ('t, 'c) obj whenever
64 :     the pointer target type is complete. In the incomplete case, 'o
65 :     will be instantiated with some "'c iobj" -- a type obtained by
66 :     using one of the functors PointerToIncompleteType or PointerToCompleteType.
67 :    
68 :     Operations that work on both incomplete and complete pointer types are
69 :     typed as taking an 'o ptr while operations that require the target to
70 :     be known are typed as taking some ('t, 'c) obj ptr.
71 :    
72 :     voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
73 :     where void is an eqtype without any values. This makes it possible
74 :     to work on voidptr values using functions meant to operate on light
75 :     incomplete pointers.
76 :    
77 :     * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
78 :     been vastly simplified.
79 :    
80 :     ----------------------------------------------------------------------
81 :     Name: Matthias Blume
82 : blume 1076 Date: 2002/02/19 10:48:00 EST
83 :     Tag: blume-20020219-pqfix
84 :     Description:
85 :    
86 :     Applied Chris Okasaki's bug fix for priority queues.
87 :    
88 :     ----------------------------------------------------------------------
89 :     Name: Matthias Blume
90 : blume 1073 Date: 2002/02/15 17:05:00
91 : blume 1069 Tag: Release_110_39
92 :     Description:
93 :    
94 : blume 1073 Last-minute retagging is becoming a tradition... :-(
95 :    
96 :     This is the working release 110.39.
97 :    
98 :     ----------------------------------------------------------------------
99 :     Name: Matthias Blume
100 :     Date: 2002/02/15 16:00:00 EST
101 :     Tag: Release_110_39-orig
102 :     Description:
103 :    
104 : blume 1069 Working release 110.39. New bootfiles.
105 :    
106 : blume 1073 (Update: There was a small bug in the installer so it wouldn't work
107 :     with all shells. So I retagged. -Matthias)
108 :    
109 : blume 1069 ----------------------------------------------------------------------
110 :     Name: Matthias Blume
111 : blume 1068 Date: 2002/02/15 14:17:00 EST
112 :     Tag: blume-20020215-showbindings
113 :     Description:
114 :    
115 :     Added EnvRef.listBoundSymbols and CM.State.showBindings. Especially
116 :     the latter can be useful for exploring what bindings are available at
117 :     the interactive prompt. (The first function returns only the list
118 :     of symbols that are really bound, the second prints those but also the
119 :     ones that CM's autoloading mechanism knows about.)
120 :    
121 :     ----------------------------------------------------------------------
122 :     Name: Matthias Blume
123 : blume 1067 Date: 2002/02/15 12:08:00 EST
124 :     Tag: blume-20020215-iptrs
125 :     Description:
126 :    
127 :     Two improvements to ml-nlffigen:
128 :    
129 :     1. Write files only if they do not exist or if their current contents
130 :     do not coincide with what's being written. (That is, avoid messing
131 :     with the time stamps unless absolutely necessary.)
132 :    
133 :     2. Implement a "repository" mechanism for generated files related
134 :     to "incomplete pointer types". See the README file for details.
135 :    
136 :     ----------------------------------------------------------------------
137 :     Name: Matthias Blume
138 : blume 1066 Date: 2002/02/14 11:50:00 EST
139 :     Tag: blume-20020214-quote
140 :     Description:
141 :    
142 :     Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm). This is required
143 :     because of the new and improved tag generation scheme. (Thanks to Allen
144 :     Leung for pointing it out.)
145 :    
146 :     ----------------------------------------------------------------------
147 : george 1065 Name: Lal George
148 :     Date: 2002/02/14 09:55:27 EST 2002
149 :     Tag: george-20020214-isabelle-bug
150 :     Description:
151 :    
152 :     Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
153 :     of Isabelle on the x86.
154 :    
155 :     From Allen:
156 :     -----------
157 :     I've found the problem:
158 :    
159 :     in ra-core.sml, I use the counter "blocked" to keep track of the
160 :     true number of elements in the freeze queue. When the counter goes
161 :     to zero, I skip examining the queue. But I've messed up the
162 :     bookkeeping in combine():
163 :    
164 :     else ();
165 :     case !ucol of
166 :     PSEUDO => (if !cntv > 0 then
167 :     (if !cntu > 0 then blocked := !blocked - 1 else ();
168 :     ^^^^^^^^^^^^^^^^^^^^^^^
169 :     moveu := mergeMoveList(!movev, !moveu)
170 :     )
171 :     else ();
172 :    
173 :     combine() is called to coalesce two nodes u and v.
174 :     I think I was thinking that if the move counts of u and v are both
175 :     greater than zero then after they are coalesced then one node is
176 :     removed from the freeze queue. Apparently I was thinking that
177 :     both u and v are of low degree, but that's clearly not necessarily true.
178 :    
179 :    
180 :     02/12/2002:
181 :     Here's the patch. HOL now compiles.
182 :    
183 :     I don't know how this impact on performance (compile
184 :     time or runtime). This bug caused the RA (especially on the x86)
185 :     to go thru the potential spill phase when there are still nodes on the
186 :     freeze queue.
187 :    
188 :    
189 :    
190 :    
191 :     ----------------------------------------------------------------------
192 : blume 1058 Name: Matthias Blume
193 : blume 1064 Date: 2002/02/13 22:40:00 EST
194 :     Tag: blume-20020213-fptr-rtti
195 :     Description:
196 :    
197 :     Fixed a bug in ml-nlffigen that was introduced with one of the previous
198 :     updates.
199 :    
200 :     ----------------------------------------------------------------------
201 :     Name: Matthias Blume
202 : blume 1063 Date: 2002/02/13 16:41:00 EST
203 :     Tag: blume-20020213-cmlpq
204 :     Description:
205 :    
206 :     Added new priority queue export symbols (which have just been added to
207 :     smlnj-lib.cm) to CML's version of smlnj-lib.cm. (Otherwise CML would
208 :     not compile and the installer would choke.)
209 :    
210 :     ----------------------------------------------------------------------
211 :     Name: Matthias Blume
212 : blume 1062 Date: 2002/02/13 16:15:00 EST
213 :     Tag: blume-20020213-various
214 :     Description:
215 :    
216 :     1. More tweaks to ml-nlffigen:
217 :    
218 :     - better internal datastructures (resulting in slight speedup)
219 :     - "-match" option requires exact match
220 :     - "localized" gensym counters (untagged structs/unions nested within
221 :     other structs/unions or within typedefs get a fresh counter; their
222 :     tag will be prefixed by a concatenation of their parents' tags)
223 :     - bug fixes (related to calculation of transitive closure of types
224 :     to be included in the output)
225 :    
226 :     2. Minor Basis updates:
227 :    
228 :     - added implementations for List.collate and Option.app
229 :    
230 :     ----------------------------------------------------------------------
231 :     Name: Matthias Blume
232 : blume 1060 Date: 2002/02/11 15:55:00 EST
233 :     Tag: blume-20020211-gensym
234 :     Description:
235 :    
236 :     Added a "-gensym" option to command line of ml-nlffigen. This can be
237 :     used to specify a "stem" -- a string that is inserted in all "gensym'd"
238 :     names (ML structure names that correspond to unnamed C structs, unions,
239 :     and enums), so that separate runs of ml-nlffigen do not clash.
240 :    
241 :     ----------------------------------------------------------------------
242 :     Name: Matthias Blume
243 : blume 1059 Date: 2002/02/11 12:05:00 EST
244 :     Tag: blume-20020211-gensml
245 :     Description:
246 :    
247 :     A quick fix for a problem with GenSML (in the pgraph-util library):
248 :     Make generation of toplevel "local" optional. (Strictly speaking,
249 :     signature definitions within "local" are not legal SML.)
250 :    
251 :     Other than that: updates to INSTALL and cm/TODO.
252 :    
253 :     ----------------------------------------------------------------------
254 :     Name: Matthias Blume
255 : blume 1058 Date: 2002/02/08 15:00:00 EST
256 :     Tag: blume-20020208-uniquepid
257 :     Description:
258 :    
259 :     0. Version number has been bumped to 110.38.1. NEW BOOTFILES!!!
260 :    
261 :     1. The installer (config/install.sh) has gotten smarter:
262 :    
263 :     - Configuration options are a bit easier to specify now
264 :     (in config/targets).
265 :     - Bug in recognizing .tar.bz2 files fixed.
266 :     - Installer automatically resolves dependencies between
267 :     configuration options (e.g., if you ask for eXene, you will
268 :     also get cml -- regardless whether you asked for it or not).
269 :     - Installer can run in "quieter mode" by setting the environment
270 :     variable INSTALL_QUIETLY to "true". "Quieter" does not mean
271 :     "completely silent", though.
272 :     - Build HashCons library as part of smlnj-lib.
273 :    
274 :     2. A new scheme for assigning persistent identifiers to compilation
275 :     units (and, by extension, to types etc.) has been put into place.
276 :     This fixes a long-standing bug where types and even dynamic values
277 :     can get internally confused, thereby compromising type safety
278 :     (abstraction) and dynamic correctness. See
279 :    
280 :     http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
281 :    
282 :     for an example of how things could go wrong until now.
283 :    
284 :     The downside of the new scheme is that pids are not quite as
285 :     persistent as they used to be: CM will generate a fresh pid
286 :     for every compilation unit that it thinks it sees for the first
287 :     time. That means that if you compile starting from a clean, fresh
288 :     source tree at two different times, you end up with different
289 :     binaries.
290 :    
291 :     Cutoff recompilation, however, has not been compromised because
292 :     CM keeps pid information in special caches between runs.
293 :    
294 :     ----------------------------------------------------------------------
295 : george 1053 Name: Lal George
296 : george 1055 Date: 2002/02/07 15:34:13 EST 2002
297 :     Tag: <none>
298 :     Description:
299 :    
300 :     Compilers that generate assembly code may produce global labels
301 :     whose value is resolved at link time. The various peephole optimization
302 :     modules did not take this in account.
303 :    
304 :     TODO. The Labels.addrOf function should really return an option
305 :     type so that clients are forced to deal with this issue, rather
306 :     than an exception being raised.
307 :    
308 :     ----------------------------------------------------------------------
309 :     Name: Lal George
310 : george 1053 Date: 2002/02/06 13:55:02 EST
311 :     Tag: george-20020206-ra-breakup
312 :     Description:
313 :    
314 :     1. A bug fix from Allen.
315 :    
316 :     A typo causes extra fstp %st(0)'s to be generated at compensation
317 :     edges, which might cause stack underflow traps at runtime. This
318 :     occurs in fft where there are extraneous fstps right before the 'into'
319 :     trap instruction (in this case they are harmless since none of the
320 :     integers overflow.)
321 :    
322 : george 1054 2. Pulled out various utility modules that were embedded in the modules
323 : george 1053 of the register allocator. I need these modules for other purposes, but
324 : george 1054 they are not complete enough to put into a library (just yet).
325 : george 1053 ----------------------------------------------------------------------
326 : blume 1049 Name: Matthias Blume
327 :     Date: 2002/01/31 16:05:00 EST
328 :     Tag: blume-20020131-sparc-ccalls
329 :     Description:
330 :    
331 :     1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
332 :     of extra stack space by mistake. Fixed.
333 :    
334 :     2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
335 :    
336 :     ----------------------------------------------------------------------
337 : leunga 1048 Name: Allen Leung
338 :     Date: 2002/01/30
339 :     Tag: leunga-20020130-nowhere-bug-fix
340 :     Description:
341 :    
342 :     MLRISC bug fixes:
343 :     1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
344 :     2. MachineInt.fromString was negating its value.
345 :    
346 :     ----------------------------------------------------------------------
347 : blume 1039 Name: Matthias Blume
348 : blume 1047 Date: 2002/01/29
349 :     Tag: blume-20020129-INSTALL
350 :     Description:
351 :    
352 :     - Added somewhat detailed installation instructions (file INSTALL).
353 :     - Fixed curl-detection bug in config/install.sh.
354 :     - It is now possible to select the URL getter using the URLGETTER
355 :     environment variable:
356 :    
357 :     not set / "unknown" --> automatic detection (script tries wget,
358 :     curl, and lynx)
359 :     "wget" / "curl" / "lynx" --> use the specified program (script "knows"
360 :     how to properly invoke them)
361 :     other --> use $URLGETTER directly, it must take
362 :     precisely two command-line arguments
363 :     (source URL and destination file name)
364 :    
365 :     ----------------------------------------------------------------------
366 :     Name: Matthias Blume
367 : blume 1044 Date: 2002/01/28
368 : blume 1045 Tag: blume-20020128-sparc-ccalls
369 :     Description:
370 :    
371 :     - Fixed problem with calculation of "used" registers in sparc-c-calls.
372 :     - Make use of the allocParam argument in sparc-c-calls.
373 :    
374 :     ----------------------------------------------------------------------
375 :     Name: Matthias Blume
376 :     Date: 2002/01/28
377 : blume 1044 Tag: blume-20020128-allocParam
378 :     Description:
379 :    
380 :     John Reppy: Changes c-calls API to accept client-callback for
381 :     allocating extra stack space.
382 :     me: Corresponding changes to mlriscGen (using a dummy argument that
383 :     does not change the current behavior).
384 :    
385 :     ----------------------------------------------------------------------
386 :     Name: Matthias Blume
387 : blume 1040 Date: 2002/01/28 12:00:00
388 :     Tag: Release_110_38
389 :     Description:
390 :    
391 :     This time for real!!!
392 :    
393 :     ----------------------------------------------------------------------
394 :     Name: Matthias Blume
395 : blume 1039 Date: 2002/01/28 10:56:00 EST
396 :     Tag: blume-20020128-retraction
397 :     Description:
398 :    
399 :     0. Retracted earlier 110.38. (The Release_110_38 tag has been replaced
400 :     with blume-Release_110_38-retracted.)
401 :    
402 :     1. Fixed a problem with incorrect rounding modes in real64.sml.
403 :     (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
404 :    
405 :     2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
406 :     and enums fixed. The naming of corresponding ML identifiers should
407 :     now be consistent again.
408 :    
409 :     ----------------------------------------------------------------------
410 : leunga 1038 Name: Allen Leung
411 :     Date: 2002/01/27
412 :     Tag: leunga-20020127-nowhere
413 :     Description:
414 :    
415 :     Added a target called nowhere in the configuration scripts.
416 :     Enabling this will build the MLRISC 'nowhere' tool (for translating
417 :     programs with where-clauses into legal SML code) during installation.
418 :    
419 :     ----------------------------------------------------------------------
420 : blume 1036 Name: Matthias Blume
421 : leunga 1038 Date: 2002/01/25 21:27:00 EST
422 : blume 1039 Tag: blume-Release_110_38-retracted
423 : blume 1037 Description:
424 :    
425 :     Call it a (working) release! Version is 110.38. Bootfiles are ready.
426 :    
427 :     README will be added later.
428 :    
429 : blume 1039 !!! NOTE: Re-tagged as blume-Release_110_38-retracted. Original tag
430 :     (Release_110_38) removed. Reason: Last-minute bug fixes.
431 :    
432 : blume 1037 ----------------------------------------------------------------------
433 :     Name: Matthias Blume
434 : blume 1036 Date: 2002/01/25
435 :     Tag: blume-20020125-ffi
436 :     Description:
437 :    
438 :     A large number of tweaks and improvements to ml-nlffi-lib and
439 :     ml-nlffigen:
440 :    
441 :     - ML represenation types have been streamlined
442 :     - getter and setter functions work with concrete values, not abstract
443 :     ones where possible
444 :     - ml-nlffigen command line more flexible (see README file there)
445 :     - some bugs have been fixed (hopefully)
446 :    
447 :     ----------------------------------------------------------------------
448 : george 1034 Name: Lal George
449 :     Date: 2002/01/24
450 :     Tag: george-20020124-risc-ra-interface
451 :     Description:
452 :    
453 :     There is a dramatic simplification in the interface to the
454 :     register allocator for RISC architectures as a result of making
455 :     parallel copy instructions explicit.
456 :    
457 :     ----------------------------------------------------------------------
458 : blume 1028 Name: Matthias Blume
459 : blume 1032 Date: 2002/01/22
460 :     Tag: blume-20020122-x86-ccalls
461 :     Description:
462 :    
463 :     Bug fix for c-calls on x86 (having to do with how char- and
464 :     short-arguments are being handled).
465 :    
466 :     ----------------------------------------------------------------------
467 :     Name: Matthias Blume
468 : blume 1031 Date: 2002/01/21
469 :     Tag: blume-20020121-ff
470 :     Description:
471 :    
472 :     Another day of fiddling with the FFI...
473 :    
474 :     1. Bug fix/workaround: CKIT does not complain about negative array
475 :     dimensions, so ml-nlffigen has to guard itself against this possibility.
476 :     (Otherwise a negative dimension would send it into an infinite loop.)
477 :    
478 :     2. Some of the abstract types (light objects, light pointers, most "base"
479 :     types) in structure C are now eqtypes.
480 :    
481 :     3. Added constructors and test functions for NULL function pointers.
482 :    
483 :     ----------------------------------------------------------------------
484 :     Name: Matthias Blume
485 : blume 1028 Date: 2002/01/18
486 : blume 1030 Tag: blume-20020118-ready-for-new-release
487 :     Description:
488 :    
489 :     Made config/srcarchiveurl point to a new place. (Will provide boot
490 :     files shortly.)
491 :    
492 :     Maybe we christen this to be 110.38?
493 :    
494 :     ----------------------------------------------------------------------
495 :     Name: Matthias Blume
496 :     Date: 2002/01/18
497 : blume 1029 Tag: blume-20020118-more-ffifiddle
498 :     Description:
499 :    
500 :     Today's FFI fiddling:
501 :    
502 :     - Provided a structure CGetSet with "convenient" versions of C.Get.* and
503 :     C.Set.* that use concrete (MLRep.*) arguments and results instead
504 :     of abstract ones.
505 :    
506 :     - Provided word-style bit operations etc. for "int" representation
507 :     types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
508 :     and Long.
509 :    
510 :     ----------------------------------------------------------------------
511 :     Name: Matthias Blume
512 :     Date: 2002/01/18
513 : blume 1028 Tag: blume-20020118-use-x86-fp
514 :     Description:
515 :    
516 :     Now that x86-fast-fp seems to be working, I turned it back on again
517 :     by default. (Seems to work fine now, even with the FFI.)
518 :    
519 :     Other than that, I added some documentation about the FFI to
520 :     src/ml-nlffigen/README and updated the FFI test examples in
521 :     src/ml-nlffi-lib/Tests/*.
522 :    
523 :     ----------------------------------------------------------------------
524 : leunga 1025 Name: Allen Leung
525 :     Date: 2002/01/17
526 :     Tag: leunga-20020117-x86-fast-fp-call
527 :     Description:
528 :    
529 :     1. Fixed a problem with handling return fp values when x86's fast fp
530 :     mode is turned on.
531 :    
532 :     2. Minor pretty printing fix for cellset. Print %st(0) as %st(0) instead
533 :     of %f32.
534 :    
535 :     3. Added a constructor INT32lit to the ast of MLRISC tools.
536 :    
537 :     ----------------------------------------------------------------------
538 : blume 1021 Name: Matthias Blume
539 :     Date: 2002/01/16
540 :     Tag: blume-20020116-ffifiddle
541 :     Description:
542 :    
543 :     More fiddling with the FFI interface:
544 :    
545 :     - Make constness 'c instead of rw wherever possible. This eliminates
546 :     the need for certain explicit coercions. (However, due to ML's
547 :     value polymorphism, there will still be many cases where explicit
548 :     coercions are necessary. Phantom types are not the whole answer
549 :     to modeling a subtyping relationship in ML.)
550 :    
551 :     - ro/rw coersions for pointers added. (Avoids the detour through */&.)
552 :    
553 :     - "printf" test example added to src/ml-nlffi-lib/Tests. (Demonstrates
554 :     clumsy workaround for varargs problem.)
555 :    
556 :     ----------------------------------------------------------------------
557 : george 1016 Name: Lal George
558 :     Date: 2002/01/15
559 :     Tag: <none>
560 :     Description:
561 :    
562 :     1. Since COPY instructions are no longer native to the architecture,
563 :     a generic functor can be used to implement the expandCopies function.
564 :    
565 :     2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
566 :     TEXT segment.
567 :    
568 :     ----------------------------------------------------------------------
569 : blume 1011 Name: Matthias Blume
570 : blume 1015 Date: 2002/01/15
571 :     Tag: blume-20020115-ffiupdates
572 :     Description:
573 :    
574 :     1. Fix for bug resulting in single-precision float values being returned
575 :     incorrectly from FFI calls.
576 :    
577 :     2. Small modifications to C FFI API:
578 :    
579 :     - memory-allocation routines return straight objects (no options)
580 :     and raise an exception in out-of-memory situations
581 :     - unsafe extensions to cast between function pointers and pointers
582 :     from/to ints
583 :     - added structure C_Debug as an alternative to structure C where
584 :     pointer-dereferencing (|*| and |*!) always check for null-pointers
585 :     - added open_lib' to DynLinkage; open_lib' works like open_lib
586 :     but also takes a (possibly empty) list of existing library handles
587 :     that the current library depends on
588 :    
589 :     ----------------------------------------------------------------------
590 :     Name: Matthias Blume
591 : blume 1011 Date: 2002/01/10
592 :     Tag: blume-20020110-newffigen
593 :     Description:
594 :    
595 :     1. Updates to portable graph code.
596 :    
597 :     2. Major update to ml-nlffigen and ml-nlffi-lib. Things are much
598 :     more scalable now so that even huge interfaces such as the one
599 :     for GTK compile in finite time and space. :-)
600 :     See src/ml-nlffigen/README for details on what's new.
601 :    
602 :     ----------------------------------------------------------------------
603 : george 1003 Name: Lal George
604 : george 1009 Date: 2001/01/09 14:31:35 EST 2002
605 :     Tag: george-20011206-rm-native-copy
606 :     Description:
607 :    
608 :     Removed the native COPY and FCOPY instructions
609 :     from all the architectures and replaced it with the
610 :     explicit COPY instruction from the previous commit.
611 :    
612 :     It is now possible to simplify many of the optimizations
613 :     modules that manipulate copies. This has not been
614 :     done in this change.
615 :    
616 :     ----------------------------------------------------------------------
617 :     Name: Lal George
618 : george 1003 Date: 2001/12/06 16:50:13 EST 2001
619 :     Tag: george-20011206-mlrisc-instruction
620 :     Description:
621 : george 824
622 : george 1003 Changed the representation of instructions from being fully abstract
623 :     to being partially concrete. That is to say:
624 :    
625 :     from
626 :     type instruction
627 :    
628 :     to
629 :     type instr (* machine instruction *)
630 :    
631 :     datatype instruction =
632 :     LIVE of {regs: C.cellset, spilled: C.cellset}
633 :     | KILL of {regs: C.cellset, spilled: C.cellset}
634 :     | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
635 :     | ANNOTATION of {i: instruction, a: Annotations.annotation}
636 :     | INSTR of instr
637 :    
638 :     This makes the handling of certain special instructions that appear on
639 :     all architectures easier and uniform.
640 :    
641 :     LIVE and KILL say that a list of registers are live or killed at the
642 :     program point where they appear. No spill code is generated when an
643 :     element of the 'regs' field is spilled, but the register is moved to
644 :     the 'spilled' (which is present, more for debugging than anything else).
645 :    
646 :     LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
647 :     We used to generate:
648 :    
649 :     DEFFREG f1
650 :     f1 := f2 + f3
651 :     trapb
652 :    
653 :     but now generate:
654 :    
655 :     f1 := f2 + f3
656 :     trapb
657 :     LIVE {regs=[f1,f2,f3], spilled=[]}
658 :    
659 :     Furthermore, the DEFFREG (hack) required that all floating point instruction
660 :     use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
661 :     defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
662 :     in a cleaner alpha implementation. (Hopefully, intel will not get rid of
663 :     this architecture).
664 :    
665 :     COPYXXX is intended to replace the parallel COPY and FCOPY available on
666 :     all the architectures. This will result in further simplification of the
667 :     register allocator that must be aware of them for coalescing purposes, and
668 :     will also simplify certain aspects of the machine description that provides
669 :     callbacks related to parallel copies.
670 :    
671 :     ANNOTATION should be obvious, and now INSTR represents the honest to God
672 :     machine instruction set!
673 :    
674 :     The <arch>/instructions/<arch>Instr.sml files define certain utility
675 :     functions for making porting easier -- essentially converting upper case
676 :     to lower case. All machine instructions (of type instr) are in upper case,
677 :     and the lower case form generates an MLRISC instruction. For example on
678 :     the alpha we have:
679 :    
680 :     datatype instr =
681 :     LDA of {r:cell, b:cell, d:operand}
682 :     | ...
683 :    
684 :     val lda : {r:cell, b:cell, d:operand} -> instruction
685 :     ...
686 :    
687 :     where lda is just (INSTR o LDA), etc.
688 :    
689 : leunga 815 ----------------------------------------------------------------------
690 : blume 986 Name: Matthias Blume
691 : blume 993 Date: 2001/11/22 21:40:00 EST
692 :     Tag: Release_110_37
693 :     Description:
694 :    
695 :     Release 110.37. This time for real.
696 :    
697 :     ----------------------------------------------------------------------
698 :     Name: Matthias Blume
699 : blume 988 Date: 2001/11/21 16:35:00 EST
700 :     Tag: blume-20011121-foot-in-mouth
701 :     Description:
702 :    
703 :     Removed the "Release_110_37" tag because of a serious bug.
704 :     This will be re-tagged once the bug is fixed.
705 :    
706 :     ----------------------------------------------------------------------
707 :     Name: Matthias Blume
708 : blume 987 Date: 2001/11/21 16:14:00 EST
709 :     Tag: blume-20011121-forgottenfile
710 :     Description:
711 :    
712 :     Forgot to add a file. (Just a .tex-file -- part of
713 :     the CM manual source.)
714 :    
715 :     ----------------------------------------------------------------------
716 :     Name: Matthias Blume
717 : blume 986 Date: 2001/11/21 16:10:00 EST
718 : blume 988 Tag: blume-20011121-invalid_110_37
719 : blume 986 Description:
720 :    
721 : blume 988 Note: I removed the original tag "Release_110_37" from this commit
722 :     because we found a serious bug in all non-x86 backends.
723 :     - Matthias
724 :    
725 : blume 986 1. Modifications to the SML/NJ code generator and to the runtime system
726 :     so that code object name strings are directly inserted into code
727 :     objects at code generation time. The only business the runtime system
728 :     has with this is now to read the name strings on occasions.
729 :     (The encoding of the name string has also changed somewhat.)
730 :    
731 :     2. CM now implements a simple "set calculus" for specifying export lists.
732 :     In particular, it is now possible to refer to the export lists of
733 :     other libraries/groups/sources and form unions as well as differences.
734 :     See the latest CM manual for details.
735 :    
736 :     3. An separate notion of "proxy" libraries has again be eliminated from
737 :     CM's model. (Proxy libraries are now simply a special case of using
738 :     the export list calculus.)
739 :    
740 :     4. Some of the existing libraries now take advantage of the new set
741 :     calculus.
742 :     (Notice that not all libraries have been converted because some
743 :     of the existing .cm-files are supposed to be backward compatible
744 :     with 110.0.x.)
745 :    
746 :     5. Some cleanup in stand-alone programs. (Don't use "exnMessage" -- use
747 :     "General.exnMessage"! The former relies on a certain hook to be
748 :     initialized, and that often does not happen in the stand-alone case.)
749 :    
750 :     ----------------------------------------------------------------------
751 : george 984 Name: Lal George
752 :     Date: 2001/11/21 13:56:18 EST
753 :     Tag: george-2001121-pseudo-ops
754 :     Description:
755 :    
756 :     Implemented a complete redesign of MLRISC pseudo-ops. Now there
757 :     ought to never be any question of incompatabilities with
758 :     pseudo-op syntax expected by host assemblers.
759 :    
760 :     For now, only modules supporting GAS syntax are implemented
761 :     but more should follow, such as MASM, and vendor assembler
762 :     syntax, e.g. IBM as, Sun as, etc.
763 :    
764 :     ----------------------------------------------------------------------
765 : blume 977 Name: Matthias Blume
766 :     Date: 2001/11/14 11:52:00 EST
767 :     Tag: blume-20011114-srcname
768 :     Description:
769 :    
770 :     1. Routed the name of the current source file to mlriscgen where it
771 :     should be directly emitted into the code object. (This last part
772 :     is yet to be done.)
773 :    
774 :     2. Some cleanup of the pgraph code to make it match the proposal that
775 :     I put out the other day. (The proposal notwithstanding, things are
776 :     still in flux here.)
777 :    
778 :     ----------------------------------------------------------------------
779 : george 976 Name: Lal George
780 :     Date: 2001/11/14 09:44:04 EST
781 :     Tag:
782 :     Description:
783 :    
784 :     Fix for a backpatching bug reported by Allen.
785 :    
786 :     Because the boundary between short and long span-dependent
787 :     instructions is +/- 128, there are an astounding number of
788 :     span-dependent instructions whose size is over estimated.
789 :    
790 :     Allen came up with the idea of letting the size of span
791 :     dependent instructions be non-monotonic, for a maxIter
792 :     number of times, after which the size must be monotonically
793 :     increasing.
794 :    
795 :     This table shows the number of span-dependent instructions
796 :     whose size was over-estimated as a function of maxIter, for the
797 :     file Parse/parse/ml.grm.sml:
798 :    
799 :     maxIter # of instructions:
800 :     10 687
801 :     20 438
802 :     30 198
803 :     40 0
804 :    
805 :     In compiling the compiler, there is no significant difference in
806 :     compilation speed between maxIter=10 and maxIter=40. Actually,
807 :     my measurements showed that maxIter=40 was a tad faster than
808 :     maxIter=10! Also 96% of the files in the compiler reach a fix
809 :     point within 13 iterations, so fixing maxIter at 40, while high,
810 :     is okay.
811 :    
812 :     ----------------------------------------------------------------------
813 : blume 975 Name: Matthias Blume
814 :     Date: 2001/10/31 15:25:00 EST
815 :     Tag: blume-20011031-pgraph
816 :     Description:
817 :    
818 :     CKIT:
819 :     * Changed the "Function" constructor of type Ast.ctype to carry optional
820 :     argument identifiers.
821 :     * Changed the return type of TypeUtil.getFunction accordingly.
822 :     * Type equality ignores the argument names.
823 :     * TypeUtil.composite tries to preserve argument names but gives up quickly
824 :     if there is a mismatch.
825 :    
826 :     installation script:
827 :     * attempts to use "curl" if available (unless "wget" is available as well)
828 :    
829 :     CM:
830 :     * has an experimental implementation of "portable graphs" which I will
831 :     soon propose as an implementation-independent library format
832 :     * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
833 :    
834 :     NLFFI-LIB:
835 :     * some cleanup (all cosmetic)
836 :    
837 :     NLFFIGEN:
838 :     * temporarily disabled the mechanism that suppresses ML output for
839 :     C definitions whose identifiers start with an underscore character
840 :     * generate val bindings for enum constants
841 :     * user can request that only one style (light or heavy) is being used;
842 :     default is to use both (command-line arguments: -heavy and -light)
843 :     * fixed bug in handling of function types involving incomplete pointers
844 :     * generate ML entry points that take record arguments (i.e., using
845 :     named arguments) for C functions that have a prototype with named
846 :     arguments
847 :     (see changes to CKIT)
848 :    
849 :     ----------------------------------------------------------------------
850 : leunga 972 Name: Allen Leung
851 :     Date: 2001/10/27 20:34:00 EDT
852 :     Tag: leunga-20011027-x86-fast-fp-call
853 :     Description:
854 :    
855 :     Fixed the bug described in blume-20010920-slowfp.
856 :    
857 :     The fix involves
858 :     1. generating FCOPYs in FSTP in ia32-svid
859 :     2. marking a CALL with the appropriate annotation
860 :    
861 :     ----------------------------------------------------------------------
862 : blume 961 Name: Matthias Blume
863 :     Date: 2001/10/16 11:32:00 EDT
864 :     Tag: blume-20011016-netbsd
865 :     Description:
866 :    
867 :     Underscore patch from Chris Richards (fixing problem with compiling
868 :     runtime system under recent NetBSD).
869 :    
870 :     ----------------------------------------------------------------------
871 : george 959 Name: Allen Leung
872 :     Date: 2001/10/12 17:18:32 EDT 2001
873 :     Tag: leung-20011012-x86-printflowgraph
874 :     Description:
875 :    
876 :     X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
877 :    
878 :     ----------------------------------------------------------------------
879 : george 952 Name: Lal George
880 : george 958 Date: 2001/10/11 23:51:34 EDT
881 :     Tag: george-20011011-too-many-instrs
882 :     Description:
883 :    
884 :     The representation of a program point never expected to see more
885 :     than 65536 instructions in a basic block!
886 :    
887 :     ----------------------------------------------------------------------
888 :     Name: Lal George
889 : george 952 Date: 2001/10/09 09:41:37 EDT
890 :     Tag: george-20011008-mlrisc-labels
891 :     Description:
892 :    
893 :     Changed the machine description files to support printing of
894 :     local and global labels in assembly code, based on host assembler
895 :     conventions.
896 :    
897 :     ----------------------------------------------------------------------
898 : blume 916 Name: Matthias Blume
899 : blume 939 Date: 2001/09/25 15:25:00 EDT
900 :     Tag: blume-20010925-exninfo
901 :     Description:
902 :    
903 :     I provided a non-hook implementation of exnName (at the toplevel) and
904 :     made the "dummy" implementation of exnMessage (at the toplevel) more
905 :     useful: if nothing gets "hooked in", then at least you are going to
906 :     see the exception name and a message indicating why you don't see more.
907 :    
908 :     [For the time being, programs that need exnMessage and want to use
909 :     ml-build should either use General.exnMessage (strongly recommended) or
910 :     refer to structure General at some other point so that CM sees a
911 :     static dependency.]
912 :    
913 :     [Similar remarks go for "print" and "use": If you want to use their
914 :     functionality in stand-alone programs generated by ml-build, then use
915 :     TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
916 :    
917 :     ----------------------------------------------------------------------
918 :     Name: Matthias Blume
919 : blume 938 Date: 2001/09/20 17:28:00 EDT
920 :     Tag: blume-20010920-slowfp
921 :     Description:
922 :    
923 :     Allen says that x86-fast-fp is not safe yet, so I turned it off again...
924 :    
925 :     ----------------------------------------------------------------------
926 :     Name: Matthias Blume
927 : blume 937 Date: 2001/09/20 17:20:00 EDT
928 :     Tag: blume-20010920-canonicalpaths
929 :     Description:
930 :    
931 :     0. Updated the BOOT file (something that I forgot to do earlier).
932 :    
933 :     1. Small internal change to CM so that it avoids "/../" in filenames
934 :     as much as possible (but only where it is safe).
935 :    
936 :     2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
937 :     that arguments that contain delimiters are passed through correctly.
938 :     This change also means that all "special" arguments of the form
939 :     @SMLxxx... must come first.
940 :    
941 :     3. Changed install script to put relative anchor names for tool commands
942 :     into pathconfig.
943 :    
944 :     ----------------------------------------------------------------------
945 :     Name: Matthias Blume
946 : blume 932 Date: 2001/09/18 15:35:00 EDT
947 :     Tag: blume-20010918-readme11036
948 :     Description:
949 :    
950 :     Added README files.
951 :    
952 :     ----------------------------------------------------------------------
953 :     Name: Matthias Blume
954 : blume 929 Date: 2001/09/18 11:45:00 EDT
955 :     Tag: Release_110_36 (retag)
956 :     Description:
957 :    
958 :     Fixed mistake in config/preloads. Retagged as 110.36.
959 :    
960 :     ----------------------------------------------------------------------
961 :     Name: Matthias Blume
962 : blume 928 Date: 2001/09/18 09:40:00 EDT
963 : blume 929 Tag: Release_110_36_orig (tag changed)
964 : blume 928 Description:
965 :    
966 :     New version (110.36). New bootfiles.
967 :    
968 :     ----------------------------------------------------------------------
969 :     Name: Matthias Blume
970 : blume 927 Date: 2001/09/14 16:15:00 EDT
971 :     Tag: blume-20010914-x86fastfp
972 :     Description:
973 :    
974 :     John committed some changes that Allen made, in particular a (hopefully)
975 :     correctly working version of the x86-fp module.
976 :    
977 :     I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
978 :     flag to "true". Everything seems to compile to a fixpoint ok, and
979 :     "mandelbrot" speeds up by about 15%.
980 :    
981 :     ----------------------------------------------------------------------
982 :     Name: Matthias Blume
983 : blume 923 Date: 2001/09/13 11:20:00 EDT
984 :     Tag: blume-20010913-minimal
985 :     Description:
986 :    
987 :     1. Stefan Monnier's patch to fix a miscompilation problem that
988 :     was brought to light by John Reppy's work on Moby.
989 :    
990 :     2. Implemented a minimal "structure Compiler" that contains just
991 :     "version" and "architecture". The minimal version will be
992 :     available when the full version is not. This is for backward-
993 :     compatibility with code that wants to test Compiler.version.
994 :    
995 :     ----------------------------------------------------------------------
996 :     Name: Matthias Blume
997 : blume 920 Date: 2001/08/28 14:03:00 EDT
998 :     Tag: blume-20010828-ml-lex
999 :     Description:
1000 :    
1001 :     Fix for bug 1581, received from Neophytos Michael.
1002 :    
1003 :     ----------------------------------------------------------------------
1004 :     Name: Matthias Blume
1005 : blume 918 Date: 2001/08/27 11:20:00 EDT
1006 :     Tag: blume-20010827-readme11035
1007 :     Description:
1008 :    
1009 :     Fleshed out the README file for 110.35.
1010 :    
1011 :     ----------------------------------------------------------------------
1012 :     Name: Matthias Blume
1013 : blume 916 Date: 2001/08/24 17:10:00 EDT
1014 :     Tag: Release_110_35
1015 :     Description:
1016 :    
1017 :     New version number (110.35). New bootfiles.
1018 :    
1019 :     ----------------------------------------------------------------------
1020 : george 915 Name: Lal George
1021 : george 909 Date: 2001/08/24 13:47:18 EDT 2001
1022 :     Tag: george-20010824-MLRISC-graphs
1023 :     Description:
1024 :    
1025 :     removed clusters from MLRISC completely and replaced with graphs.
1026 :    
1027 :     ----------------------------------------------------------------------
1028 :     Name: Matthias Blume
1029 : blume 905 Date: 2001/08/23 17:50:00 EDT
1030 :     Tag: blume-20010823-toplevel
1031 :     Description:
1032 :    
1033 :     - some reorganization of the code that implements various kinds of
1034 :     environments in the compiler (static, dynamic, symbolic, combined)
1035 :     - re-implemented the EnvRef module so that evalStream works properly
1036 :     (if the stream contains references to "use", "CM.make", etc.)
1037 :     - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
1038 :    
1039 :     ----------------------------------------------------------------------
1040 :     Name: Matthias Blume
1041 : blume 904 Date: 2001/08/20 15:50 EDT
1042 :     Tag: blume20010820-slipup
1043 :     Description:
1044 :    
1045 :     I forgot to commit a few files. Here they are...
1046 :    
1047 :     ----------------------------------------------------------------------
1048 :     Name: Matthias Blume
1049 : blume 903 Date: 2001/08/20 15:35:00 EDT
1050 :     Tag: blume-20010820-debugprof
1051 :     Description:
1052 :    
1053 :     !!!! NEW BOOTFILES !!!!
1054 :    
1055 :     This is another round of reorganizing the compiler sources. This
1056 :     time the main goal was to factor out all the "instrumentation"
1057 :     passes (for profiling and backtracing) into their own library.
1058 :     The difficulty was to do it in such a way that it does not depend
1059 :     on elaborate.cm but only on elabdata.cm.
1060 :    
1061 :     Therefore there have been further changes to both elaborate.cm and
1062 :     elabdata.cm -- more "generic" things have been moved from the former
1063 :     to the latter. As a result, I was forced to split the assignment
1064 :     of numbers indicating "primtyc"s into two portions: SML-generic and
1065 :     SML/NJ-specific. Since it would have been awkward to maintain,
1066 :     I bit the bullet and actually _changed_ the mapping between these
1067 :     numbers and primtycs. The bottom line of this is that you need
1068 :     a new set of bin- and bootfiles.
1069 :    
1070 :     I have built new bootfiles for all architectures, so doing a fresh
1071 :     checkout and config/install.sh should be all you need.
1072 :    
1073 :     The newly created library's name is
1074 :    
1075 :     $smlnj/viscomp/debugprof.cm
1076 :    
1077 :     and its sources live under
1078 :    
1079 :     src/compiler/DebugProf
1080 :    
1081 :     ----------------------------------------------------------------------
1082 :     Name: Matthias Blume
1083 : blume 902 Date: 2001/08/15 17:15:00 EDT
1084 :     Tag: blume-20010815-compreorg
1085 :     Description:
1086 :    
1087 :     This is a first cut at reorganizing the CM libraries that make up the
1088 :     core of the compiler. The idea is to separate out pieces that could
1089 :     be used independently by tools, e.g., the parser, the typechecker, etc.
1090 :    
1091 :     The current status is a step in this direction, but it is not quite
1092 :     satisfactory yet. Expect more changes in the future.
1093 :    
1094 :     Here is the current (new) organization...
1095 :    
1096 :     What used to be $smlnj/viscomp/core.cm is now divided into
1097 :     six CM libraries:
1098 :    
1099 :     $smlnj/viscomp/basics.cm
1100 :     /parser.cm
1101 :     /elabdata.cm
1102 :     /elaborate.cm
1103 :     /execute.cm
1104 :     /core.cm
1105 :    
1106 :     The CM files for these libraries live under src/system/smlnj/viscomp.
1107 :     All these libraries are proxy libraries that contain precisely
1108 :     one CM library component. Here are the locations of the components
1109 :     (all within the src/compiler tree):
1110 :    
1111 :     Basics/basics.cm
1112 :     Parse/parser.cm
1113 :     ElabData/elabdata.cm
1114 :     Elaborator/elaborate.cm
1115 :     Execution/execute.cm
1116 :     core.cm
1117 :    
1118 :     [This organization is the same that has been used already
1119 :     for a while for the architecture-specific parts of the visible
1120 :     compiler and for the old version of core.cm.]
1121 :    
1122 :     As you will notice, many source files have been moved from their
1123 :     respective original locations to a new home in one of the above
1124 :     subtrees.
1125 :    
1126 :     The division of labor between the new libraries is the following:
1127 :    
1128 :     basics.cm:
1129 :     - Simple, basic definitions that pertain to many (or all) of
1130 :     the other libraries.
1131 :     parser.cm:
1132 :     - The SML parser, producing output of type Ast.dec.
1133 :     - The type family for Ast is also defined and exported here.
1134 :     elabdata.cm:
1135 :     - The datatypes that describe input and output of the elaborator.
1136 :     This includes types, absyn, and static environments.
1137 :     elaborator.cm:
1138 :     - The SML/NJ type checker and elaborator.
1139 :     This maps an Ast.dec (with a given static environment) to
1140 :     an Absyn.dec (with a new static environment).
1141 :     - This libraries implements certain modules that used to be
1142 :     structures as functors (to remove dependencies on FLINT).
1143 :     execute.cm:
1144 :     - Everything having to do with executing binary code objects.
1145 :     - Dynamic environments.
1146 :     core.cm:
1147 :     - SML/NJ-specific instantiations of the elaborator and MLRISC.
1148 :     - Top-level modules.
1149 :     - FLINT (this should eventually become its own library)
1150 :    
1151 :     Notes:
1152 :    
1153 :     I am not 100% happy with the way I separated the elaborator (and its
1154 :     data structures) from FLINT. Two instances of the same problem:
1155 :    
1156 :     1. Data structures contain certain fields that carry FLINT-specific
1157 :     information. I hacked around this using exn and the property list
1158 :     module from smlnj-lib. But the fact that there are middle-end
1159 :     specific fields around at all is a bit annoying.
1160 :    
1161 :     2. The elaborator calculates certain FLINT-related information. I tried
1162 :     to make this as abstract as I could using functorization, but, again,
1163 :     the fact that the elaborator has to perform calculations on behalf
1164 :     of the middle-end at all is not nice.
1165 :    
1166 :     3. Having to used exn and property lists is unfortunate because it
1167 :     weakens type checking. The other alternative (parameterizing
1168 :     nearly *everything*) is not appealing, though.
1169 :    
1170 :     I removed the "rebinding =" warning hack because due to the new organization
1171 :     it was awkward to maintain it. As a result, the compiler now issues some of
1172 :     these warnings when compiling init.cmi during bootstrap compilation. On
1173 :     the plus side, you also get a warning when you do, for example:
1174 :     val op = = Int32.+
1175 :     which was not the case up to now.
1176 :    
1177 :     I placed "assign" and "deref" into the _Core structure so that the
1178 :     code that deals with the "lazy" keyword can find them there. This
1179 :     removes the need for having access to the primitive environment
1180 :     during elaboration.
1181 :    
1182 :     ----------------------------------------------------------------------
1183 :     Name: Matthias Blume
1184 : blume 898 Date: 2001/08/13
1185 :     Tag: blume-20010813-closures
1186 :     Description:
1187 :    
1188 :     This fix was sent to us by Zhong Shao. It is supposed to improve the
1189 :     performance of certain loops by avoiding needless closure allocation.
1190 :    
1191 :     ----------------------------------------------------------------------
1192 : george 890 Name: Lal George
1193 : george 896 Date: 2001/07/31 10:03:23 EDT 2001
1194 :     Tag: george-20010731-x86-fmalloc
1195 :     Description: Fixed bug in x86 calls
1196 :    
1197 :     There was a bug where call instructions would mysteriously
1198 :     vanish. The call instruction had to be one that returned
1199 :     a floating point value.
1200 :    
1201 :     ----------------------------------------------------------------------
1202 :     Name: Lal George
1203 : george 890 Date: 2001/07/19 16:36:29 EDT 2001
1204 :     Tag: george-20010719-simple-cells
1205 :     Description:
1206 :    
1207 :     I have dramatically simplified the interface for CELLS in MLRISC.
1208 :    
1209 :     In summary, the cells interface is broken up into three parts:
1210 :    
1211 :     1. CellsBasis : CELLS_BASIS
1212 :    
1213 :     CellsBasis is a top level structure and common for all
1214 :     architectures. it contains the definitions of basic datatypes
1215 :     and utility functions over these types.
1216 :    
1217 :     2. functor Cells() : CELLS
1218 :    
1219 :     Cells generates an interface for CELLS that incorporates the
1220 :     specific resources on the target architecture, such as the
1221 :     presence of special register classes, their number and size,
1222 :     and various useful substructures.
1223 :    
1224 :     3. <ARCH>CELLS
1225 :    
1226 :     e.g. SparcCells: SPARCCELLS
1227 :    
1228 :     <ARCH>CELLS usually contains additional bindings for special
1229 :     registers on the architecture, such as:
1230 :    
1231 :     val r0 : cell (* register zero *)
1232 :     val y : cell (* Y register *)
1233 :     val psr : cell (* processor status register *)
1234 :     ...
1235 :    
1236 :     The structure returned by applying the Cells functor is opened
1237 :     in this interface.
1238 :    
1239 :     The main implication of all this is that the datatypes for cells is
1240 :     split between CellsBasis and CELLS -- a fairly simple change for user
1241 :     code.
1242 :    
1243 :     In the old scheme the CELLS interface had a definitional binding of
1244 :     the form:
1245 :    
1246 :     signature CELLS = sig
1247 :    
1248 :     structure CellsBasis = CellsBasis
1249 :    
1250 :     ...
1251 :    
1252 :     end
1253 :    
1254 :     With all the sharing constraints that goes on in MLRISC, this old
1255 :     design quickly leads to errors such as:
1256 :    
1257 :     "structure definition spec inside of sharing ... "
1258 :    
1259 :    
1260 :     and appears to require an unacceptable amount of sharing and where
1261 :     constraint hackery.
1262 :    
1263 :     I think this error message (the interaction of definitional specs and
1264 :     sharing) requires more explanation on our web page.
1265 :    
1266 :     ----------------------------------------------------------------------
1267 : blume 826 Name: Matthias Blume
1268 : blume 879 Date: 2001/07/19 15:00:00 EDT
1269 :     Tag: blume-20010719-libreorg
1270 :     Description:
1271 :    
1272 :     This update puts together a fairly extensive but straightforward change
1273 :     to the way the libraries that implement the interactive system are
1274 :     organized:
1275 :    
1276 :     The biggest change is the elimination of structure Compiler. As a
1277 :     replacement for this structure, there is now a CM library
1278 :     (known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
1279 :     that exports all the substructures of the original structure Compiler
1280 :     directly. So instead of saying Compiler.Foo.bar one now simply
1281 :     says Foo.bar. (The CM libraries actually export a collection of
1282 :     structures that is richer than the collection of substructures of
1283 :     structure Compiler.)
1284 :    
1285 :     To make the transition smooth, there is a separate library called
1286 :     $smlnj/compiler/compiler.cm which puts together and exports the
1287 :     original structure Compiler (or at least something very close to it).
1288 :    
1289 :     There are five members of the original structure Compiler
1290 :     that are not exported directly but which instead became members
1291 :     of a new structure Backend (described by signature BACKEND). These are:
1292 :     structure Profile (: PROFILE), structure Compile (: COMPILE), structure
1293 :     Interact (: INTERACT), structure Machine (: MACHINE), and val
1294 :     architecture (: string).
1295 :    
1296 :     Structure Compiler.Version has become structure CompilerVersion.
1297 :    
1298 :     Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
1299 :     by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
1300 :     or x86, respectively.
1301 :     Each of these exports the same frontend structures that
1302 :     $smlnj/compiler.cm exports. But they do not have a structure Backend
1303 :     and instead export some structure <Arch>Backend where <Arch> is Alpha32,
1304 :     Hppa, PPC, Sparc, or X86, respectively.
1305 :    
1306 :     Library $smlnj/compiler/all.cm exports the union of the exports of
1307 :     $smlnj/compiler/<arch>.cm
1308 :    
1309 :     There are no structures <Arch>Compiler anymore, use
1310 :     $smlnj/compiler/<arch>.cm instead.
1311 :    
1312 :     Library host-compiler-0.cm is gone. Instead, the internal library
1313 :     that instantiates CM is now called cm0.cm. Selection of the host
1314 :     compiler (backend) is no longer done here but. (Responsibility for it
1315 :     now lies with $smlnj/compiler/current.cm. This seems to be more
1316 :     logical.)
1317 :    
1318 :     Many individual files have been moved or renamed. Some files have
1319 :     been split into multiple files, and some "dead" files have been deleted.
1320 :    
1321 :     Aside from these changes to library organization, there are also changes
1322 :     to the way the code itself is organized:
1323 :    
1324 :     Structure Binfile has been re-implemented in such a way that it no
1325 :     longer needs any knowledge of the compiler. It exclusively deals
1326 :     with the details of binfile layout. It no longer invokes the
1327 :     compiler (for the purpose of creating new prospective binfile
1328 :     content), and it no longer has any knowledge of how to interpret
1329 :     pickles.
1330 :    
1331 :     Structure Compile (: COMPILE) has been stripped down to the bare
1332 :     essentials of compilation. It no longer deals with linking/execution.
1333 :     The interface has been cleaned up considerably.
1334 :    
1335 :     Utility routines for dealing with linking and execution have been
1336 :     moved into their own substructures.
1337 :    
1338 :     (The ultimate goal of these changes is to provide a light-weight
1339 :     binfile loader/linker (at least for, e.g., stable libraries) that
1340 :     does not require CM or the compiler to be present.)
1341 :    
1342 :     CM documentation has been updated to reflect the changes to library
1343 :     organization.
1344 :    
1345 :     ----------------------------------------------------------------------
1346 :     Name: Matthias Blume
1347 : blume 875 Date: 2001/07/10 17:30:00 EDT
1348 :     Tag: Release_110_34
1349 :     Description:
1350 :    
1351 : blume 876 Minor tweak to 110.34 (re-tagged):
1352 :    
1353 :     - README.html file added to CVS repository
1354 :     - runtime compiles properly under FreeBSD 3.X and 4.X
1355 :    
1356 :     ----------------------------------------------------------------------
1357 :     Name: Matthias Blume
1358 :     Date: 2001/07/10 17:30:00 EDT
1359 :     Tag: Release_110_34
1360 :     Description:
1361 :    
1362 : blume 875 New version number (110.34). New bootfiles.
1363 :    
1364 :     ----------------------------------------------------------------------
1365 :     Name: Matthias Blume
1366 : blume 874 Date: 2001/07/09 16:00:00 EDT
1367 :     Tag: blume-20010709-more-varargs
1368 :     Description:
1369 :    
1370 :     I changed the handling of varargs in ml-nlffigen again:
1371 :     The ellipsis ... will now simply be ignored (with an accompanying warning).
1372 :    
1373 :     The immediate effect is that you can actually call a varargs function
1374 :     from ML -- but you can't actually supply any arguments beyond the ones
1375 :     specified explicitly. (For example, you can call printf with its format
1376 :     string, but you cannot pass additional arguments.)
1377 :    
1378 :     This behavior is only marginally more useful than the one before, but
1379 :     it has the advantage that a function or, more importantly, a function
1380 :     type never gets dropped on the floor, thus avoiding follow-up problems with
1381 :     other types that refer to the offending one.
1382 :    
1383 :     ----------------------------------------------------------------------
1384 :     Name: Matthias Blume
1385 : blume 873 Date: 2001/07/09 11:25:00 EDT
1386 :     Tag: blume-20010709-varargs
1387 :     Description:
1388 :    
1389 :     1. ckit-lib.cm now exports structure Error
1390 :     2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
1391 :     with a warning accompanied by a source location. Moreover, it
1392 :     merely skips the offending function or type and proceeds with the
1393 :     rest of its work.u As a result, one can safely feed C code containing
1394 :     "..." to ml-nlffigen.
1395 :     3. There are some internal improvements to CM, providing slightly
1396 :     more general string substitutions in the tools subsystem.
1397 :    
1398 :     ----------------------------------------------------------------------
1399 :     Name: Matthias Blume
1400 : blume 854 Date: 2001/06/27 15:10:00 EDT
1401 :     Tag: blume-20010627-concur
1402 :     Description:
1403 :    
1404 :     Fixed a small bug in CM's handling of parallel compilation.
1405 :     (You could observe the bug by Control-C-interrupting an ordinary
1406 :     CMB.make or CM.stabilize and then attaching some compile servers.
1407 :     The result was that all of a sudden the previously interrupted
1408 :     compilation would continue on its own. This was because of
1409 :     an over-optimization: CM did not bother to clean out certain queues
1410 :     when no servers were attached "anyway", resulting in the contents
1411 :     of these queues to grab control when new servers did get attached.)
1412 :    
1413 :     There is also another minor update to the CM manual.
1414 :    
1415 :     ----------------------------------------------------------------------
1416 :     Name: Matthias Blume
1417 : blume 853 Date: 2001/06/26 16:15:00 EDT
1418 :     Tag: blume-20010626-cmdoc
1419 :     Description:
1420 :    
1421 :     Minor typo fixed in CM manual (syntax diagram for libraries).
1422 :    
1423 :     ----------------------------------------------------------------------
1424 :     Name: Matthias Blume
1425 : blume 852 Date: 2001/06/25 22:55:00 EDT
1426 :     Tag: blume-20010625-x86pc
1427 :     Description:
1428 :    
1429 :     Fixed a nasty bug in the X86 assembly code that caused signal
1430 :     handlers to fail (crash) randomly.
1431 :    
1432 :     ----------------------------------------------------------------------
1433 :     Name: Matthias Blume
1434 : blume 846 Date: 2001/06/25 12:05:00 EDT
1435 :     Tag: blume-20010625-nlffigen
1436 :     Description:
1437 :    
1438 :     This update fixes a number of minor bugs in ml-nlffigen as reported by
1439 :     Nick Carter <nbc@andrew.cmu.edu>.
1440 :    
1441 :     1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
1442 :     2. Default names for generated files are now derived from the name of
1443 :     the C file *without its directory*. In particular, this causes generated
1444 :     files to be placed locally even if the C file is in some system directory.
1445 :     3. Default names for generated signatures and structures are also derived
1446 :     from the C file name without its directory. This avoids silly things
1447 :     like "structure GL/GL".
1448 :     (Other silly names are still possible because ml-nlffigen does not do
1449 : blume 847 a thorough check of whether generated names are legal ML identifiers.
1450 :     When in doubt, use command line arguments to force particular names.)
1451 : blume 846
1452 :     ----------------------------------------------------------------------
1453 :     Name: Matthias Blume
1454 : blume 845 Date: 2001/06/21 12:25:00 EDT
1455 :     Tag: blume-20010621-eXene
1456 :     Description:
1457 :    
1458 :     eXene now compiles and (sort of) works again.
1459 :    
1460 :     The library name (for version > 110.33) is $/eXene.cm.
1461 :    
1462 :     I also added an new example in src/eXene/examples/nbody. See the
1463 :     README file there for details.
1464 :    
1465 :     ----------------------------------------------------------------------
1466 :     Name: Matthias Blume
1467 : blume 844 Date: 2001/06/20 16:40:00 EDT
1468 :     Tag: blume-20010620-cml
1469 :     Description:
1470 :    
1471 :     CML now compiles and works again.
1472 :    
1473 :     Libraries (for version > 110.33):
1474 :    
1475 :     $cml/cml.cm Main CML library.
1476 :     $cml/basis.cm CML's version of $/basis.cm.
1477 :     $cml/cml-internal.cm Internal helper library.
1478 :     $cml/core-cml.cm Internal helper library.
1479 :     $cml-lib/trace-cml.cm Tracing facility.
1480 :     $cml-lib/smlnj-lib.cm CML's version of $/smlnj-lib.cm
1481 :    
1482 :     The installer (config/install.sh) has been taught how to properly
1483 :     install this stuff.
1484 :    
1485 :     ----------------------------------------------------------------------
1486 :     Name: Matthias Blume
1487 : blume 843 Date: 2001/06/19 17:55:00 EDT
1488 :     Tag: blume-20010619-instantiate
1489 :     Description:
1490 :    
1491 :     This un-breaks the fix for bug 1432.
1492 :     (The bug was originally fixed in 110.9 but I broke it again some
1493 :     time after that.)
1494 :    
1495 :     ----------------------------------------------------------------------
1496 :     Name: Matthias Blume
1497 : blume 842 Date: 2001/06/19 17:25:00 EDT
1498 :     Tag: blume-20010619-signals
1499 :     Description:
1500 :    
1501 :     This should (hopefully) fix the long-standing signal handling bug.
1502 :     (The runtime system was constructing a continuation record with an
1503 :     incorrect descriptor which would cause the GC to drop data on the floor...)
1504 :    
1505 :     ----------------------------------------------------------------------
1506 :     Name: Matthias Blume
1507 : blume 840 Date: 2001/06/15 15:05:00 EDT
1508 : blume 841 Tag: blume-20010615-moresparc
1509 :     Description:
1510 :    
1511 :     Here is a short late-hour update related to Sparc c-calls:
1512 :    
1513 :     -- made handling of double-word arguments a bit smarter
1514 :    
1515 :     -- instruction selection phase tries to collapse certain clumsily
1516 :     constructed ML-Trees; typical example:
1517 :    
1518 :     ADD(ty,ADD(_,e,LI d1),LI d2) -> ADD(ty,e,LI(d1+d2))
1519 :    
1520 :     This currently has no further impact on SML/NJ since mlriscGen does
1521 :     not seem to generate such patterns in the first place, and c-calls
1522 :     (which did generate them in the beginning) has meanwhile been fixed
1523 :     so as to avoid them as well.
1524 :    
1525 :     ----------------------------------------------------------------------
1526 :     Name: Matthias Blume
1527 :     Date: 2001/06/15 15:05:00 EDT
1528 : blume 840 Tag: blume-20010615-sparc
1529 :     Description:
1530 :    
1531 :     The purpose of this update is to provide an implementation of NLFFI
1532 :     on Sparc machines.
1533 :    
1534 :     Here are the changes in detail:
1535 :    
1536 :     * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
1537 :     the Sparc implementation of the c-calls API.
1538 :     * The Sparc backend of SML/NJ has been modified to uniformely use %fp
1539 :     for accessing the ML frame. Thus, we have a real frame pointer and
1540 :     can freely modify %sp without need for an omit-frame-ptr phase.
1541 :     The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
1542 :     this case.
1543 :     * ml-nlffigen has been taught to produce code for different architectures
1544 :     and calling conventions.
1545 :     * In a way similar to what was done in the x86 case, the Sparc
1546 :     backend uses its own specific extension to mltree. (For example,
1547 :     it needs to be able to generate UNIMP instructions which are part
1548 :     of the calling convention.)
1549 :     * ml-nlffi-lib was reorganized to make it more modular (in particular,
1550 :     to make it easier to plug in new machine- and os-dependent parts).
1551 :    
1552 :     There are some other fairly unrelated bug fixes and cleanups as well:
1553 :    
1554 :     * I further hacked the .cm files for MLRISC tools (like MDLGen) so
1555 :     that they properly share their libraries with existing SML/NJ libraries.
1556 :     * I fixed a minor cosmetic bug in CM, supressing certain spurious
1557 :     follow-up error messages.
1558 :     * Updates to CM/CMB documentation.
1559 :    
1560 :     TODO items:
1561 :    
1562 :     * MLRISC should use a different register as its asmTemp on the Sparc.
1563 :     (The current %o2 is a really bad choice because it is part of the
1564 :     calling conventions, so things might interfere in unexpected ways.)
1565 :    
1566 :     ----------------------------------------------------------------------
1567 :     Name: Matthias Blume
1568 : blume 839 Date: 2001/06/07
1569 :     Tag: blume-20010607-calls
1570 :     Description:
1571 :    
1572 :     A number of internal changes related to C calls and calling conventions:
1573 :    
1574 :     1. ML-Tree CALL statements now carry a "pops" field. It indicates the
1575 :     number of bytes popped implicitly (by the callee). In most cases
1576 :     this field is 0 but on x86/win32 it is some non-zero value. This
1577 :     is information provided for the benefit of the "omit-frameptr" pass.
1578 :     2. The CALL instruction on the x86 carries a similar "pops" field.
1579 :     The instruction selection phase copies its value from the ML-Tree
1580 :     CALL statement.
1581 :     3. On all other architectures, the instruction selection phase checks
1582 :     whether "pops=0" and complains if not.
1583 :     4. The c-calls implementation for x86 now accepts two calling conventions:
1584 :     "ccall" and "stdcall". When "ccall" is selected, the caller cleans
1585 :     up after the call and pops is set to 0. For "stdcall", the caller
1586 :     does nothing, leaving the cleanup to the callee; pops is set to
1587 :     the number of bytes that were pushed onto the stack.
1588 :     5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
1589 :     distinguish between "ccall" and "stdcall".
1590 :     6. The UNIMP instruction has been added to the supported Sparc instruction
1591 :     set. (This is needed for implementing the official C calling convention
1592 :     on this architecture.)
1593 :     7. I fixed some of the .cm files under src/MLRISC/Tools to make them
1594 :     work with the latest CM.
1595 :    
1596 :     ----------------------------------------------------------------------
1597 :     Name: Matthias Blume
1598 : blume 838 Date: 2001/06/05 15:10:00 EDT
1599 :     Tag: blume-20010605-cm-index
1600 :     Description:
1601 :    
1602 :     0. The "lambdasplit" parameter for class "sml" in CM has been documented.
1603 :    
1604 :     1. CM can now generate "index files". These are human-readable files
1605 :     that list on a per-.cm-file basis each toplevel symbol defined or
1606 :     imported. The location of the index file for
1607 :     <p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
1608 :     To enable index-file generation, set CM.Control.generate_index to true
1609 :     or export an environment-symbol: export CM_GENERATE_INDEX=true.
1610 :    
1611 :     The CM manual has been updated accordingly.
1612 :    
1613 :     2. I made some slight modifications to the c-calls API in MLRISC.
1614 :    
1615 :     a) There is now a callback to support saving/restoring of
1616 :     dedicated but caller-save registers around the actual call
1617 :     instruction.
1618 :     b) One can optionally specify a comment-annotation for the
1619 :     call instruction.
1620 :    
1621 :     3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
1622 :     (For example, the comment annotation shows the C prototype of
1623 :     the function being called.)
1624 :    
1625 :     ----------------------------------------------------------------------
1626 :     Name: Matthias Blume
1627 : blume 837 Date: 2001/06/01 13:30:00 EDT
1628 :     Tag: blume-20010601-nlffi-cleanup
1629 :     Description:
1630 :    
1631 :     This is mostly a cleanup of MLFFI stuff:
1632 :    
1633 :     - some signature files have been put into a more exposed place
1634 :     - the ugly 'f type parameter is gone (simplifies types tremendously!)
1635 :     - ml-nlffigen changed accordingly
1636 :     - tutorial updated
1637 :    
1638 :     Other changes:
1639 :    
1640 :     - author's affiliation in CM manual(s) updated
1641 :     - some more recognized keywords added to Allen's sml.sty
1642 :    
1643 :     ----------------------------------------------------------------------
1644 :     Name: Matthias Blume
1645 : blume 836 Date: 2001/05/25 15:30:00 EDT
1646 :     Tag: blume-20010525-iptr
1647 :     Description:
1648 :    
1649 :     - put the official 110.33-README (as it appears on the ftp server) under
1650 :     CVS
1651 :     - fixed a small bug related to incomplete pointer types in
1652 :     ml-nlffigen
1653 :     - small cosmetic change to the ml-nlffi-lib's "arr" type constructor
1654 :     (it does not need the 'f type parameter)
1655 :    
1656 :     ----------------------------------------------------------------------
1657 :     Name: Matthias Blume
1658 : blume 834 Date: 2001/05/23 14:30:00 EDT
1659 :     Tag: Release_110_33
1660 :     Description:
1661 :    
1662 :     New version number (110.33). New bootfiles.
1663 :    
1664 :     ----------------------------------------------------------------------
1665 :     Name: Matthias Blume
1666 : blume 833 Date: 2001/05/22 18:06:00 EDT
1667 :     Tag: blume-20010522-targets
1668 :     Description:
1669 :    
1670 :     Made install.sh use file config/targets.customized if it exists, falling
1671 :     back to config/targets if it doesn't. This way one can have a customized
1672 :     version of the targets file without touching the "real thing", thus
1673 :     eliminating the constant fear of accidentally checking something bogus
1674 :     back into the CVS repository... (File config/targets.customized must
1675 :     not be added to the repository!)
1676 :    
1677 :     ----------------------------------------------------------------------
1678 :     Name: Matthias Blume
1679 : blume 832 Date: 2001/05/22 16:30:00 EDT
1680 :     Tag: blume-20010522-minitut
1681 :     Description:
1682 :    
1683 :     1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
1684 :     struct returns.
1685 :     2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt. This is some very
1686 :     incomplete, preliminary documentation for NLFFI.
1687 :    
1688 :     ----------------------------------------------------------------------
1689 :     Name: Matthias Blume
1690 : blume 829 Date: 2001/05/14 11:30:00 EDT
1691 :     Tag: blume-20010514-script
1692 :     Description:
1693 :    
1694 :     Some bugs in install script fixed.
1695 :    
1696 :     In addition to that I also made a slight change to the NLFFI API:
1697 :     Functors generated by ml-nlffigen now take the dynamic library as a
1698 :     straight functor argument, not as a suspended one. (The original
1699 :     functor code used to force the suspension right away anyway, so there
1700 :     was nothing gained by this complication of the interface.)
1701 :    
1702 :     ----------------------------------------------------------------------
1703 :     Name: Matthias Blume
1704 : blume 828 Date: 2001/05/11 14:35:00 EDT
1705 :     Tag: blume-20010511-ml-nlffi
1706 :     Description:
1707 :    
1708 :     I finally took the plunge and added my new FFI code to the main
1709 :     repository. For x86-linux it is now ready for prime-time.
1710 :    
1711 :     There are two new subdirectories of "src":
1712 :    
1713 :     - ml-nlffi-lib:
1714 :     The utility library for programs using the FFI interface.
1715 :     Here is the implementation of $/c.cm and its associated low-level
1716 :     partners $/c-int.cm and $/memory.cm.
1717 :     - ml-nlffigen:
1718 :     A stand-alone program for generating ML glue code from C source
1719 :     code.
1720 :    
1721 :     Building ml-nlffigen requires $/ckit-lib.cm.
1722 :    
1723 :     The config/install.sh script has been updates to do the Right Thing
1724 :     (hopefully).
1725 :    
1726 :     Notice that the source tree for the C-Kit will not be put under "src"
1727 :     but directly under the installation root directory. (This is the
1728 :     structure that currently exists on the CVS server when you check out
1729 :     module "sml".) Fortunately, config/install.sh knows about this oddity.
1730 :    
1731 :     Bugs: No documentation yet.
1732 :    
1733 :     ----------------------------------------------------------------------
1734 :     Name: Matthias Blume
1735 : blume 826 Date: 2001/05/09 16:35:00 EDT
1736 :     Tag: blume-20010509-cpscontract
1737 :     Description:
1738 :    
1739 :     Fixed a bug in the accounting code in cpsopt/contract.sml. (The
1740 :     wrapper/unwrapper elimination did not decrement usage counts and some
1741 :     dead variables got overlooked by the dead-up logic.)
1742 :    
1743 :     ----------------------------------------------------------------------
1744 : george 824 Name: Lal George
1745 :     Date: 2001/05/08 17:26:09 EDT
1746 :     Tag: george-20010508-omit-frameptr
1747 :     Description:
1748 :    
1749 :     Changes to implement the omit-frame-pointer optimization to support
1750 :     raw C calls. For now, there is only support on the Intel x86, but
1751 :     other architectures will follow as more experience is gained with this.
1752 :    
1753 :    
1754 :     ----------------------------------------------------------------------
1755 : blume 818 Name: Matthias Blume
1756 : blume 819 Date: 2001/05/07 14:40:00 EDT
1757 :     Tag: blume-20010507-proxies
1758 :     Description:
1759 :    
1760 :     I made into "proxy libraries" all libraries that qualify for such a
1761 :     change. (A qualifying library is a library that has another library or
1762 :     groups as its sole member and repeats that member's export list
1763 :     verbatim. A proxy library avoids this repetition by omitting its export
1764 :     list, effectively inheriting the list that its (only) member exports.
1765 :     See the CM manual for more explanation.)
1766 :     The main effect is that explicit export lists for these libraries
1767 :     do not have to be kepts in sync, making maintenance a bit easier.
1768 :    
1769 :     I also added copyright notices to many .cm-files.
1770 :    
1771 :     Last but not least, I made a new set of bootfiles.
1772 :    
1773 :     ----------------------------------------------------------------------
1774 :     Name: Matthias Blume
1775 : blume 818 Date: 2001/05/04 17:00:00 EDT
1776 :     Tag: blume-20010504-cm-lsplit
1777 :     Description:
1778 :    
1779 :     0. John merged pending changes to $/smlnj-lib.cm
1780 :    
1781 :     1. Allen's previous change accidentally backed out of one of Lal's
1782 :     earlier changes. I undid this mistake (re-introducing Lal's change).
1783 :    
1784 :     2. I used the new topOrder' function from graph-scc.sml (from $/smlnj-lib.cm)
1785 :     within the compiler where applicable. There is some code simplification
1786 :     because of that.
1787 :    
1788 :     3. The "split" phase (in FLINT) is now part of the default list of phases.
1789 :     Compiler.Control.LambdaSplitting.* can be used to globally control the
1790 :     lambda-splitting (cross-module-inlining) engine. In addition to that,
1791 :     it can now also be controlled on a per-source basis: CM has been taught
1792 :     a new tool parameter applicable to ML source files.
1793 :    
1794 :     - To turn lambda-splitting off completely:
1795 :     local open Compiler.Control.LambdaSplitting in
1796 :     val _ = set Off
1797 :     end
1798 :     - To make "no lambda-splitting" the global default (but allow per-source
1799 :     overriding); this is the initial setting:
1800 :     local open Compiler.Control.LambdaSplitting in
1801 :     val _ = set (Default NONE)
1802 :     end
1803 :     - To make "lambda-splitting with aggressiveness a" the global default
1804 :     (and allow per-source overriding):
1805 :     local open Compiler.Control.LambdaSplitting in
1806 :     val _ = set (Default (SOME a))
1807 :     end
1808 :    
1809 :     - To turn lambda-splitting off for a given ML souce file (say: a.sml)
1810 :     write (in the respective .cm-file):
1811 :     a.sml (lambdasplitting:off)
1812 :     - To turn lambda-splitting for a.sml on with minimal aggressiveness:
1813 :     a.sml (lambdasplitting:on)
1814 :     - To turn lambda-splitting for a.sml on with aggressiveness <a> (where
1815 :     <a> is a decimal non-negative integer):
1816 :     a.sml (lambdasplitting:<a>)
1817 :     - To turn lambda-splitting for a.sml on with maximal aggressiveness:
1818 :     a.sml (lambdasplitting:infinity)
1819 :     - To use the global default for a.sml:
1820 :     a.sml (lambdasplitting:default)
1821 :     or simply
1822 :     a.sml
1823 :    
1824 :     ----------------------------------------------------------------------
1825 : leunga 815 Name: Allen Leung
1826 :     Date: 2001/05/04 01:57:00 EDT
1827 :     Tag: leunga-20010504-sync
1828 :     Description:
1829 : leunga 788
1830 : leunga 815 MLRISC features.
1831 :    
1832 :     1. Fix to CMPXCHG instructions.
1833 :     2. Changed RA interface to allow annotations in callbacks.
1834 :     3. Added a new method to the stream interface to allow annotations updates.
1835 :    
1836 : george 749 ----------------------------------------------------------------------
1837 : blume 812 Name: Matthias Blume
1838 : blume 814 Date: 2001/05/01 11:45:00 EDT
1839 :     Tag: blume-20010501-pcedittmp
1840 :     Description:
1841 :    
1842 :     Changed install.sh to use the current working directory instead of
1843 :     /usr/tmp for a temporary file (pcedittmp). The previous choice
1844 :     of /usr/tmp caused trouble with MacOS X because of file premission
1845 :     problems.
1846 :    
1847 :     ----------------------------------------------------------------------
1848 :     Name: Matthias Blume
1849 : blume 812 Date: 2001/04/20 11:10:00 EDT
1850 :     Tag: blume-20010420-inMLflag
1851 :     Description:
1852 :    
1853 :     - added vp_limitPtrMask to vproc-state.h
1854 :     (for use by the raw-C-calls mechanism to implement proper interrupt
1855 :     handling)
1856 :     - made the ML compiler aware of various data-structure offsets so it
1857 :     can generate code for accessing the vp_inML flag and vp_limitPtrMask
1858 :     - tweaked mlriscGen.sml to have it emit interrupt-handling code for
1859 :     raw C-calls
1860 :    
1861 :     ----------------------------------------------------------------------
1862 : george 810 Name: Lal George
1863 :     Date: 2001/04/20 09:15:28 EDT
1864 :     Tag: george-20010420-macosX
1865 :     Description:
1866 :    
1867 :     - Changes to port to Mac OS X; Darwin.
1868 :    
1869 :     - In the process I found that sqrt was broken on the PPC, because the
1870 :     fsqrt instruction is not implemented.
1871 :    
1872 :     ----------------------------------------------------------------------
1873 : blume 799 Name: Matthias Blume
1874 : blume 808 Date: 2001/04/18 12:45:00 EDT
1875 :     Tag: blume-20010418-ccalls
1876 :     Description:
1877 :    
1878 :     - fixed two off-by-4 errors in the x86-specific c-calls implementation
1879 :     (this bug prevented structure arguments containing pointers from being
1880 :     passed correctly)
1881 :     - changed the raw-C-call code in mlriscGen.sml in such a way that
1882 :     structure arguments are represented as a pointer to the beginning
1883 :     of the structure (instead of having a series of synthesized arguments,
1884 :     one for each structure member)
1885 :    
1886 :     - made makeml script's verbosity level configurable via environment
1887 :     variable (MAKEML_VERBOSITY)
1888 :    
1889 : blume 813 - eliminated placeholder implementations for f32l, w16s, i16s, and f32s
1890 : blume 808 in rawmem-x86.sml; we are now using the real thing
1891 :    
1892 :     ----------------------------------------------------------------------
1893 :     Name: Matthias Blume
1894 : blume 806 Date: 2001/03/22 16:25:00 EST
1895 :     Tag: blume-20010322-bootfiles
1896 :     Description:
1897 :    
1898 :     Created a new set of bootfiles (for your automatic installation convenience).
1899 :    
1900 :     ----------------------------------------------------------------------
1901 :     Name: Matthias Blume
1902 : blume 805 Date: 2001/03/22 15:10:00 EST
1903 :     Tag: blume-20010322-rawmem-parcm
1904 :     Description:
1905 :    
1906 :     1. All "raw memory access" primitives for the new FFI are implemented now
1907 :     (at least on the x86).
1908 :     2. Some further cleanup of CM's parallel make mechanism.
1909 :    
1910 :     ----------------------------------------------------------------------
1911 :     Name: Matthias Blume
1912 : blume 801 Date: 2001/03/19 17:53:00 EST
1913 :     Tag: blume-20010319-parallel
1914 :     Description:
1915 :    
1916 :     Parallel make (using compile servers) now works again.
1917 :    
1918 :     To this end, CM.stabilize and CMB.make have been modified to work in
1919 :     two passes when compile servers are attached:
1920 :     1. Compile everything, do not perform stabilization; this pass
1921 :     uses compile servers
1922 :     2. Stabilize everything; this pass does not use compile servers
1923 :     If there are no compile servers, the two passes are combined into one
1924 :     (as before). Splitting the passes increases the inherent parallelism
1925 :     in the dependency graph because the entire graph including all
1926 :     libraries is available at the same time. This, in turn, improves
1927 :     server utilization. The downside is that the master process will
1928 :     have to do some extra work after compilation is done (because for
1929 :     technical reasons it must re-read all the binfiles during stabilization).
1930 :    
1931 :     ----------------------------------------------------------------------
1932 :     Name: Matthias Blume
1933 : blume 800 Date: 2001/03/16 12:22:00 EST
1934 :     Tag: blume-20010316-bootfiles
1935 :     Description:
1936 :    
1937 :     Created a new set of bootfiles (for your automatic installation convenience).
1938 :    
1939 :     ----------------------------------------------------------------------
1940 :     Name: Matthias Blume
1941 : blume 799 Date: 2001/03/16 11:00:00 EST
1942 :     Tag: blume-20010316-MLTREE-fixup
1943 :     Description:
1944 :    
1945 :     This is a minor fixup for an (untagged) earlier commit by Allen.
1946 :     (A file was missing).
1947 :    
1948 :     ----------------------------------------------------------------------
1949 : leunga 788 Name: Allen Leung
1950 : leunga 796 Date: Mon Mar 5 18:54:57 EST 2001
1951 :     Tag: leunga-20010305-cut-support
1952 :    
1953 :     1. New support for alternative control-flow in MLTREE.
1954 :     Currently we support
1955 :    
1956 :     FLOW_TO(CALL ...., [k1,...,kn])
1957 :    
1958 :     This is needed for 'cuts to' in C-- and try/handle-like constructs
1959 :     in Moby
1960 :    
1961 :     New assembler flag "asm-show-cutsto" to turn on control-flow debugging.
1962 :    
1963 :     2. Register Allocator
1964 :    
1965 :     Changes in interface [from Fermin, John]
1966 :    
1967 :     3. Alpha 8-bit SLL support [Fermin]
1968 :    
1969 :     4. All architectures
1970 :    
1971 :     A new module (ClusterExpandCopies) for expanding parallel copies.
1972 :    
1973 :     ----------------------------------------------------------------------
1974 :     Name: Allen Leung
1975 : leunga 788 Date: 2001/02/27 23:07:00 EST
1976 :     Tag: leunga-20010227-minor-stuff
1977 :    
1978 :     1. Alpha bug fix for CMOVNE
1979 :     2. Handle mltree COND(..,FCMP ...,...)
1980 :     3. Bug fix in simplifier
1981 :    
1982 :     ----------------------------------------------------------------------
1983 : blume 777 Name: Matthias Blume
1984 : blume 781 Date: 2001/01/30 17:50:00 EST
1985 :     Tag: blume-20010130-sync
1986 :     Description:
1987 :    
1988 :     This is just a minor update to sync my devel branch with the main brach.
1989 :     The only visible change is the addition of some README files.
1990 :    
1991 :     ----------------------------------------------------------------------
1992 :     Name: Matthias Blume
1993 : blume 778 Date: 2001/01/12 23:30:00 JST
1994 :     Tag: blume-20010112-bootfiles
1995 :     Description:
1996 :    
1997 :     Made a new set of bootfiles that goes with the current state of the
1998 :     repository.
1999 :    
2000 :     ----------------------------------------------------------------------
2001 :     Name: Matthias Blume
2002 : blume 777 Date: 2001/01/12 21:20:00 JST
2003 :     Tag: blume-20010112-sync
2004 :     Description:
2005 :    
2006 :     I am just flushing out some minor changes that had accumulated in
2007 :     my private branch in order to sync with the main tree. (This is
2008 :     mainly because I had CVS trouble when trying to merge _into_ my
2009 :     private branch.)
2010 :    
2011 :     Most people should be completely unaffected by this.
2012 :    
2013 :     ----------------------------------------------------------------------
2014 : leunga 775 Name: Allen Leung
2015 :     Date: Thu Jan 11 21:03:00 EST 2001
2016 :     Tag: leunga-20010111-labexp=mltree
2017 :     Description:
2018 :    
2019 :     1. Removed the type LabelExp and replace it by MLTree.
2020 :     2. Rewritten mltree-simplify with the pattern matcher tool.
2021 :     3. There were some bugs in alpha code generator which would break
2022 :     64-bit code generation.
2023 :     4. Redo the tools to generate code with the
2024 :     5. The CM files in MLRISC (and in src/system/smlnj/MLRISC)
2025 :     are now generated by perl scripts.
2026 :    
2027 :     ----------------------------------------------------------------------
2028 : blume 771 Name: Matthias Blume
2029 : blume 774 Date: 2001/01/10 21:55:00 JST
2030 :     Tag: blume-20010110-rcc
2031 :     Description:
2032 :    
2033 :     The RCC stuff now seems to work (but only on the x86).
2034 :     This required hacking of the c-calls interface (and -implementation) in
2035 :     MLRISC.
2036 :    
2037 :     Normal compiler users should be unaffected.
2038 :    
2039 :     ----------------------------------------------------------------------
2040 :     Name: Matthias Blume
2041 : blume 773 Date: 2001/01/09 01:20:00 JST
2042 :     Tag: blume-20010109-rcc
2043 :     Description:
2044 :    
2045 :     This is a fairly big patch, flushing out a large number of pending
2046 :     changes that I made to my development copy over the last couple of days.
2047 :    
2048 :     Of practical relevance at this moment is a workaround for a pickling
2049 :     bug that Allen ran into the other day. The cause of the bug itself is
2050 :     still unknown and it might be hard to fix it properly, but the
2051 :     workaround has some merits of its own (namely somewhat reducing pickling
2052 :     overhead for certain libraries). Therefore, I think this solution should
2053 :     be satisfactory at this time.
2054 :    
2055 :     The rest of the changes (i.e., the vast majority) has to do with my
2056 :     ongoing efforts of providing direct support for C function calls from
2057 :     ML. At the moment there is a new primop "RAW_CCALL", typing magic
2058 :     in types/cproto.sml (invoked from FLINT/trans/translate.sml), a new
2059 :     case in the FLINT CPS datatype (RCC), changes to cps/convert.sml to
2060 :     translate uses of RAW_CCALL into RCC, and changes to mlriscGen.sml to
2061 :     handle RCC.
2062 :    
2063 :     The last part (the changes to mlriscGen.sml) are still known to be
2064 :     wrong on the x86 and not implemented on all other architectures. But
2065 :     the infrastructure is in place. I had to change a few functor
2066 :     signatures in the backend to be able to route the CCalls interface
2067 :     from MLRISC there, and I had to specialize the mltree type (on the
2068 :     x86) to include the necessary extensions. (The extensions themselves
2069 :     were already there and redy to go in MLRISC/x86).
2070 :    
2071 :     Everything should be very happy as soon as someone helps me with
2072 :     mlriscGen.sml...
2073 :    
2074 :     In any case, nothing of this should matter to anyone as long as the
2075 :     new primop is not being used (which is going to be the case unless you
2076 :     find it where I hid it :). The rest of the compiler is completely
2077 :     unaffected.
2078 :    
2079 :     ----------------------------------------------------------------------
2080 :     Name: Matthias Blume
2081 : blume 772 Date: 2001/01/05 00:30:00 JST
2082 :     Tag: blume-20010105-primops
2083 :     Description:
2084 :    
2085 :     Added some experimental support for work that I am doing right now.
2086 :     These changes mostly concern added primops, but there is also a new
2087 :     experimental C library in the runtime system (but currently not enabled
2088 :     anywhere except on Linux/X86).
2089 :    
2090 :     In the course of adding primops (and playing with them), I discovered that
2091 :     Zhong's INL_PRIM hack (no type info for certain primops) was, in fact, badly
2092 :     broken. (Zhong was very right he labeled this stuff as "major gross hack".)
2093 :     To recover, I made type information in INL_PRIM mandatory and changed
2094 :     prim.sml as well as built-in.sml accordingly. The InLine structure now
2095 :     has complete, correct type information (i.e., no bottom types).
2096 :    
2097 :     Since all these changes mean that we need new binfiles, I also bumped the
2098 :     version number to 110.32.1.
2099 :    
2100 :     ----------------------------------------------------------------------
2101 :     Name: Matthias Blume
2102 : blume 771 Date: 2000/12/30 22:10:00 JST
2103 :     Tag: blume-20001230-various
2104 :     Description:
2105 :    
2106 :     Added proxy libraries for MLRISC and let MLRISC libraries refer
2107 :     to each other using path anchors. (See CM manual for explanation.)
2108 :    
2109 :     Updated CM documentation.
2110 :    
2111 :     Fixed some bugs in CM.
2112 :    
2113 :     Implemented "proxy" libraries (= syntactic sugar for CM).
2114 :    
2115 :     Added "-quiet" option to makeml and changed runtime system accordingly.
2116 :    
2117 :     Added cleanup handler for exportML to reset timers and compiler stats.
2118 :    
2119 :     ----------------------------------------------------------------------
2120 : george 761 Name: Lal George
2121 :     Date: 2000/12/22 22:22:58 EST 2000
2122 :     Tag: Release_110_32
2123 :     Description:
2124 :    
2125 :     Infinite precision used throughout MLRISC.
2126 :     see MLRISC/mltree/machine-int.sig
2127 :    
2128 :     ----------------------------------------------------------------------
2129 : blume 756 Name: Matthias Blume
2130 : blume 760 Date: 2000/12/22 23:16:00 JST
2131 :     Tag: blume-20001222-warn
2132 :     Description:
2133 : blume 759
2134 : blume 760 Corrected wording and formatting of some CM warning message which I
2135 :     broke in my previous patch.
2136 : blume 759
2137 :     ----------------------------------------------------------------------
2138 :     Name: Matthias Blume
2139 : blume 763 Date: 2000/12/22 21:20:00 JST
2140 :     Tag: blume-20001222-anchorenv
2141 :     Description:
2142 :    
2143 :     Fixed CM's handling of anchor environments in connection with CMB.make.
2144 :    
2145 :     ----------------------------------------------------------------------
2146 :     Name: Matthias Blume
2147 : blume 758 Date: 2000/12/22 13:15:00 JST
2148 :     Tag: blume-20001222-cleanup
2149 :     Description:
2150 :    
2151 :     Removed src/cm/ffi which does not (and did not) belong here.
2152 :    
2153 :     ----------------------------------------------------------------------
2154 :     Name: Matthias Blume
2155 : blume 757 Date: 2000/12/21 23:55:00 JST
2156 :     Tag: blume-20001221-exn
2157 :     Description:
2158 :    
2159 :     Probably most important: CM no longer silently swallows all exceptions
2160 :     in the compiler.
2161 :     Plus: some other minor CM changes. For example, CM now reports some
2162 :     sizes for generated binfiles (code, data, envpickle, lambdapickle).
2163 :    
2164 :     ----------------------------------------------------------------------
2165 :     Name: Matthias Blume
2166 : blume 756 Date: 2000/12/15 00:01:05 JST
2167 :     Tag: blume-20001215-dirtool
2168 :     Description:
2169 :    
2170 :     - "dir" tool added.
2171 :     - improvements and cleanup to Tools structure
2172 :     - documentation updates
2173 :    
2174 :     ----------------------------------------------------------------------
2175 : leunga 752 Name: Allen Leung
2176 : leunga 755 Date: Thu Dec 14 03:45:24 EST 2000
2177 :     Description:
2178 :     Tag: leunga-20001214-int-inf
2179 :     Description:
2180 :    
2181 :     In IntInf, added these standard functions, which are missing from our
2182 :     implementation:
2183 :    
2184 :     andb : int * int -> int
2185 :     xorb : int * int -> int
2186 :     orb : int * int -> int
2187 :     notb : int -> int
2188 :     << : int * word -> int
2189 :     ~>> : int * word -> int
2190 :    
2191 :     Not tested, I hope they are correct.
2192 :    
2193 :     ----------------------------------------------------------------------
2194 :     Name: Allen Leung
2195 : leunga 752 Date: Fri Dec 8 19:23:26 EST 2000
2196 :     Description:
2197 :     Tag: leunga-20001208-nowhere
2198 :     Description:
2199 :    
2200 :     Slight improvements to the 'nowhere' tool to handle OR-patterns,
2201 :     to generate better error messages etc. Plus a brief manual.
2202 :    
2203 :     ----------------------------------------------------------------------
2204 : george 749 Name: Lal George
2205 :     Date: 2000/12/08 09:54:02 EST 2000
2206 :     Tag: Release_110_31
2207 :     Description:
2208 : george 705
2209 : george 749 - Version 110.31
2210 : leunga 591 ----------------------------------------------------------------------
2211 : leunga 744 Name: Allen Leung
2212 :     Date: Thu Dec 7 22:01:04 EST 2000
2213 :     Tag: leunga-20001207-cell-monster-hack
2214 :     Description:
2215 :    
2216 :     Major MLRISC internal changes. Affect all clients.
2217 :     Summary:
2218 :    
2219 :     1. Type CELLS.cell = int is now replaced by a datatype.
2220 :     As a result, the old regmap is now gone. Almost all interfaces
2221 :     in MLRISC change as a consequence.
2222 :    
2223 :     2. A new brand version of machine description tool (v3.0) that generates
2224 :     modules expecting the new interface. The old version is removed.
2225 :    
2226 :     3. The RA interface has been further abstracted into two new functors.
2227 :     RISC_RA and X86RA. These functors have much simpler interfaces.
2228 :     [See also directory MLRISC/demo.]
2229 :    
2230 :     4. Some other new source->source code generation tools are available:
2231 :    
2232 :     a. MLRISC/Tools/RewriteGen -- generate rewriters from rules.
2233 :     b. MLRISC/Tools/WhereGen -- expands conditional pattern matching rules.
2234 :     I use this tool to generate the peephole optimizers---with the new
2235 :     cell type changes, peephole rules are becoming difficult to write
2236 :     without conditional pattern matching.
2237 :    
2238 :     5. More Intmap -> IntHashTable change. Previous changes by Matthias didn't
2239 :     cover the entire MLRISC source tree so many things broke.
2240 :    
2241 :     6. CM files have been moved to the subdirectory MLRISC/cm.
2242 :     They are moved because there are a lot of them and they clutter up the
2243 :     root dir.
2244 :    
2245 :     7. More detailed documentation to come...
2246 :    
2247 :     NOTE: To rebuild from 110.30 (ftp distribution), you'll have to do
2248 :     a makeml -rebuild first. This is because of other other
2249 :     changes that Matthias has made (see below).
2250 :    
2251 :    
2252 :     ----------------------------------------------------------------------
2253 : blume 733 Name: Matthias Blume
2254 : blume 742 Date: 2000/11/30 23:12:00 JST
2255 :     Tag: blume-20001130-filereorg
2256 :     Description:
2257 :    
2258 :     Some manual updates and some file reorganizations in CM.
2259 :    
2260 :     ----------------------------------------------------------------------
2261 :     Name: Matthias Blume
2262 : blume 737 Date: 2000/11/24 17:45:00 JST
2263 :     Tag: blume-20001124-link
2264 :     Description:
2265 :    
2266 :     Drastically improved link traversal code for the case that the dynamic
2267 :     value was already loaded at bootstrap time. As a result, CM and CMB
2268 :     now both load blazingly fast -- even on a very slow machine. Also,
2269 :     memory consumption has been further reduced by this.
2270 :    
2271 :     Warning: The format of the PIDMAP file has changed. THerefore, to
2272 :     bootstrap you have to do this:
2273 :    
2274 :     1. Run CMB.make
2275 :     2. Make a symbolic link for the boot directory:
2276 :     ln -s sml.boot.ARCH-OS xxx
2277 :     3. "Rebuild" the boot directory:
2278 :     ./makeml -boot xxx -rebuild sml ; rm xxx
2279 :     4. Boot normally:
2280 :     ./makeml
2281 :    
2282 :     ----------------------------------------------------------------------
2283 :     Name: Matthias Blume
2284 : blume 735 Date: 2000/11/21 21:20:00 JST
2285 :     Tag: blume-20001121-tools
2286 :     Description:
2287 :    
2288 :     Continued hacking on autoloading problem -- with success this time.
2289 :     Also changed tool-plugin mechanism. See new CM manual.
2290 :    
2291 :     ----------------------------------------------------------------------
2292 :     Name: Matthias Blume
2293 : blume 734 Date: 2000/11/19 14:30:00 JST
2294 :     Tag: blume-20001119-autoload
2295 :     Description:
2296 :    
2297 :     Some hacking to make autoloading faster. Success for CMB, no success
2298 :     so far for CM. There is a reduced structure CM' that autoloads faster.
2299 :     (This is a temporary, non-documented hack to be eliminated again when
2300 :     the general problem is solved.)
2301 :    
2302 :     ----------------------------------------------------------------------
2303 :     Name: Matthias Blume
2304 : blume 733 Date: 2000/11/17 14:10:00 JST
2305 :     Tag: blume-20001117-pickle-lib
2306 :     Description:
2307 :    
2308 :     1. Eliminated comp-lib.cm
2309 :     2. Made pickle-lib.cm
2310 :     3. Eliminated all uses of intset.sml (from comp-lib.cm)
2311 :     4. Replaced all uses of intmap.{sig,sml} (from comp-lib.cm) with
2312 :     equivalent constructs from smlnj-lib.cm (INtHashTable).
2313 :     5. Point 4. also goes for those uses of intmap.* in MLRISC.
2314 :     Duplicated intmap modules thrown out.
2315 :     6. Hunted down all duplicated SCC code and replaced it with
2316 :     equivalent stuff (GraphSCCFn from smlnj-lib.cm).
2317 :     7. Rewrote Feedback module.
2318 :     8. Moved sortedlist.sml into viscomp-lib.cm. Eventually it
2319 :     should be thrown out and equivalent modules from smlnj-lib.cm
2320 :     should be used (IntRedBlackSet, IntListSet, ...).
2321 :    
2322 :     Confirmed that compiler compiles to fixpoint.
2323 :    
2324 :     ----------------------------------------------------------------------
2325 : leunga 731 Name: Allen Leung
2326 :     Date: 2000/11/10 18:00:00
2327 :     Tag: leunga-20001110-new-x86-fp
2328 :    
2329 :     A new x86 floating point code generator has been added.
2330 :     By default this is turned off. To turn this on, do:
2331 :    
2332 :     CM.autoload "$smlnj/compiler.cm";
2333 :     Compiler.Control.MLRISC.getFlag "x86-fast-fp" := true;
2334 :    
2335 :     Changes:
2336 :    
2337 :     1. Changed FTAN to FPTAN so that the assembly output is correct.
2338 :     2. Changed the extension callback for FTANGENT to generate:
2339 :    
2340 :     fptan
2341 :     fstp %st(0)
2342 :     instead of
2343 :     fptan
2344 :     fstpl ftempmem
2345 :    
2346 :     3. Numerous assembly fixes for x86.
2347 :    
2348 :     5. Cleaned up the machine code output module x86/x86MC.sml and added
2349 :     support for a whole bunch of instructions and addressing modes:
2350 :    
2351 :     fadd/fsub/fsubr/fmul/fdiv/fdivr %st, %st(n)
2352 :     faddp/fsubp/fsubrp/fmulp/fdivp/fdivrp %st, %st(n)
2353 :     fadd/fsub/fsubr/fmul/fdiv/fdivr %st(n), %st
2354 :     fiadd/fisub/fisubr/fimul/fidiv/fidivr mem
2355 :     fxch %st(n)
2356 :     fld %st(n)
2357 :     fst %st(n)
2358 :     fst mem
2359 :     fstp %st(n)
2360 :     fucom %st(n)
2361 :     fucomp %st(n)
2362 :    
2363 :     All these are now generated when the fast fp mode is turned on.
2364 :    
2365 :     6. Removed the dedicated registers %st(0), ..., %st(7) from X86CpsRegs
2366 :    
2367 :     ----------------------------------------------------------------------
2368 : blume 729 Name: Matthias Blume
2369 :     Date: 2000/11/09 11:20:00 JST
2370 :     Tag: blume-20001109-scc
2371 :     Description:
2372 :    
2373 :     Eliminated some code duplication:
2374 :    
2375 :     1. Added "where" clause to GraphSCCFn in SML/NJ Library.
2376 :     (Otherwise the functor is useless.)
2377 :     2. Used GraphSCCFn where SCCUtilFun was used previously.
2378 :     3. Got rid of SCCUtilFun (in comp-lib.cm).
2379 :    
2380 :     ----------------------------------------------------------------------
2381 : george 721 Name: Lal George
2382 :     Date: 2000/11/06 09:02:21 EST 2000
2383 :     Tag: Release_110_30
2384 :     Description:
2385 :    
2386 :     - Version 110.30
2387 :     ----------------------------------------------------------------------
2388 : blume 715 Name: Matthias Blume
2389 : blume 716 Date: 2000/11/04 14:45:00
2390 :     Tag: blume-20001104-mlbuild
2391 :     Description:
2392 :    
2393 :     - Made ml-build faster on startup.
2394 :     - Documentation fixes.
2395 :    
2396 :     ----------------------------------------------------------------------
2397 :     Name: Matthias Blume
2398 : blume 715 Date: 2000/11/02 17:00:00 JST
2399 :     Tag: blume-20001102-condcomp
2400 :     Description:
2401 :    
2402 :     - Small tweaks to pickler -- new BOOTFILES!
2403 :     - Version bumped to 110.29.2.
2404 :     - Added conditional compilation facility to init.cmi (see comment there).
2405 :     ----------------------------------------------------------------------
2406 : leunga 713 Name: Allen Leung
2407 :     Date: 2000/10/23 19:31:00
2408 :     Tag: leunga-20001023-demo-ra
2409 :    
2410 :     1. Minor RA changes that improves spilling on x86 (affects Moby and C-- only)
2411 :     2. Test programs for the graph library updated
2412 :     3. Some new MLRISC demo programs added
2413 :    
2414 :     ----------------------------------------------------------------------
2415 : blume 710 Name: Matthias Blume
2416 : blume 711 Date: 2000/08/31 22:15:00 JST
2417 :     Tag: blume-20001017-errmsg
2418 :     Description:
2419 :    
2420 :     More error message grief: Where there used to be no messages, there
2421 :     now were some that had bogus error regions. Fixed.
2422 :    
2423 :     ----------------------------------------------------------------------
2424 :     Name: Matthias Blume
2425 : blume 710 Date: 2000/08/31 17:30:00 JST
2426 :     Tag: blume-20001017-v110p29p1
2427 :     Description:
2428 :    
2429 :     I made a version 110.29.1 with new bootfiles.
2430 :    
2431 :     Changes: Modified pickler/unpickler for faster and leaner unpickling.
2432 :     CM documentation changes and a small bugfix in CM's error reporting.
2433 :    
2434 :     ----------------------------------------------------------------------
2435 : george 705 Name: Lal George
2436 :     Date: 2000/09/27 14:42:35 EDT
2437 :     Tag: george-20000927-nodestatus
2438 :     Description:
2439 :    
2440 :     Changed the type of the nodestatus, so that:
2441 :    
2442 :     SPILLED(~1) is now SPILLED
2443 :     SPILLED(m) where m>=0 is now MEMREG(m)
2444 :     SPILLED(s) where s<~1 is now SPILL_LOC(~s)
2445 :    
2446 :     ----------------------------------------------------------------------
2447 : blume 697 Name: Matthias Blume
2448 : blume 703 Date: 2000/09/07 14:45:00 JST
2449 :     Tag: blume-20000907-cmerrmsg
2450 :     Description:
2451 :    
2452 :     Small tweak to CM to avoid getting ML syntax error messages twice.
2453 :    
2454 :     ----------------------------------------------------------------------
2455 :     Name: Matthias Blume
2456 : blume 701 Date: 2000/08/31 18:00:00 JST
2457 :     Tag: blume-20000831-cvsbootfiles
2458 :     Description:
2459 :    
2460 :     New URL for boot files (because the 110.29 files on the BL server do
2461 :     now work correctly with my updated install scripts for yacc and lex).
2462 :    
2463 :     ----------------------------------------------------------------------
2464 :     Name: Matthias Blume
2465 : blume 697 Date: 2000/08/08 12:33:00 JST
2466 :     Tag: blume-20000808-manual
2467 :     Description:
2468 :    
2469 :     Tiny update to CM manual.
2470 :    
2471 :     ----------------------------------------------------------------------
2472 : leunga 695 Name: Allen Leung
2473 :     Date: 2000/08/7 19:31:00
2474 :     Tag: leunga-20000807-a-whole-bunch-of-stuff
2475 :    
2476 :     Moby, C--, SSA, x86, machine descriptions etc. Should only affect C--
2477 :     and Mobdy.
2478 :    
2479 :     1. x86
2480 :    
2481 :     a. Fixes to peephole module by John and Dan.
2482 :     b. Assembly fix to SETcc by Allen.
2483 :     c. Fix to c-call by John.
2484 :     d. Fix to spilling by John. (This one deals with the missing FSTPT case)
2485 :     e. Instruction selection optimization to SETcc as suggested by John.
2486 :    
2487 :     For example,
2488 :    
2489 :     MV(32, x, COND(32, CMP(32, LT, a, b), LI 1, LI 0))
2490 :    
2491 :     should generate:
2492 :    
2493 :     MOVL a, x
2494 :     SUBL b, x
2495 :     SHRL 31, x
2496 :    
2497 :     2. IR stuff
2498 :    
2499 :     A bunch of new DJ-graph related algorithms added. These
2500 :     speed up SSA construction.
2501 :    
2502 :     3. SSA + Scheduling
2503 :    
2504 :     Added code for SSA and scheduling to the repository
2505 :    
2506 :     ----------------------------------------------------------------------
2507 : george 694 Name: Lal George
2508 :     Date: 2000/07/27 11:53:14 EDT
2509 :    
2510 :     Tag: lal-20000727-linux-ppc
2511 :     Description:
2512 :    
2513 :     Made changes to support Linux PPC.
2514 :     p.s. I have confirmation that the 110.29 boot files work fine.
2515 :    
2516 :     ----------------------------------------------------------------------
2517 : blume 684 Name: Matthias Blume
2518 : blume 692 Date: 2000/07/27 17:40:00 JST
2519 :     Tag: blume-20000727-scripts
2520 :     Description:
2521 :    
2522 :     !!!! WARNING !!!!
2523 :     You must recompile the runtime system!
2524 :     !!!! WARNING !!!!
2525 :    
2526 :     This is basically another round of script-enhancements:
2527 :    
2528 :     1. sml, ml-build, and ml-makedepend accept options -D and -U to define
2529 :     and undefine CM preprocessor symbols.
2530 :    
2531 :     2. ml-build avoids generating a new heap image if it finds that the
2532 :     existing one is still ok. (The condition is that no ML file had to
2533 :     be recompiled and all ML files are found to be older that the heap
2534 :     file.)
2535 :    
2536 :     To make this work smoothly, I also hacked the runtime system as
2537 :     well as SMLofNJ.SysInfo to get access to the heap image suffix
2538 :     (.sparc-solaris, ...) that is currently being used.
2539 :    
2540 :     Moreover, the signature of CM.mk_standalone has changed. See the
2541 :     CM manual.
2542 :    
2543 :     3. ml-makedepend accepts additional options -n, -a, and -o. (See the
2544 :     CM manual for details.)
2545 :    
2546 :     4. More CM manual updates:
2547 :     - all of the above has been documented.
2548 :     - there is now a section describing the (CM-related) command line
2549 :     arguments that are accepted by the "sml" command
2550 :    
2551 :     ----------------------------------------------------------------------
2552 :     Name: Matthias Blume
2553 : blume 691 Date: 2000/07/25 16:20:00 JST
2554 :     Tag: blume-20000725-makedepend
2555 :     Description:
2556 :    
2557 :     Added a script called ml-makedepend. This can be used in makefiles
2558 :     for Unix' make in a way very similar to the "makedepend" command for
2559 :     C.
2560 :    
2561 :     The script internally uses function CM.sources.
2562 :    
2563 :     Synopsis:
2564 :    
2565 :     ml-makedepend [-f makefile] cmfile targetname
2566 :    
2567 :     The default for the makefile is "makefile" (or "Makefile" should
2568 :     "makefile" not exist).
2569 :    
2570 :     ml-makedepend adds a cmfile/targetname-specific section to this
2571 :     makefile (after removing the previous version of this section). The
2572 :     section contains a single dependency specification with targetname on
2573 :     the LHS (targetname is an arbitrary name), and a list of files derived
2574 :     from the cmfile on the RHS. Some of the files on the RHS are
2575 :     ARCH/OPSYS-specific. Therefore, ml-makedepend inserts references to
2576 :     "make" variables $(ARCH) and $(OPSYS) in place of the corresponding
2577 :     path names. The makefile writer is responsible for making sure that
2578 :     these variables have correct at the time "make" is invoked.
2579 :    
2580 :     ----------------------------------------------------------------------
2581 :     Name: Matthias Blume
2582 : blume 690 Date: 2000/07/22 23:30:00 JST
2583 :     Tag: blume-20000722-urlupdate
2584 :     Description:
2585 :    
2586 :     Changed BOOT and config/srcarchiveurl to point to BL server:
2587 :    
2588 :     ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.29/
2589 :    
2590 :     ----------------------------------------------------------------------
2591 :     Name: Matthias Blume
2592 : blume 686 Date: 2000/07/18 18:00:00 JST
2593 :     Tag: blume-20000718-Version_110_29
2594 :     Description:
2595 :    
2596 :     1. Updated src/compiler/TopLevel/main/version.sml to version 110.29
2597 :    
2598 :     2. Updated config/version to 110.29
2599 :    
2600 :     3. Updated config/srcarchiveurl
2601 :    
2602 :     3. New boot files!
2603 :     ftp://ftp.cs.princeton.edu/pub/people/blume/sml/110.29-autofetch
2604 :    
2605 :     ----------------------------------------------------------------------
2606 :     Name: Matthias Blume
2607 : blume 684 Date: 2000/07/11 13:58:00 JST
2608 :     Tag: blume-20000711-doctypo
2609 :     Description:
2610 :    
2611 :     Fixed a few typos in CM manual.
2612 :    
2613 :     ----------------------------------------------------------------------
2614 : leunga 683 Name: Allen Leung
2615 :     Date: 2000/06/15 00:38:00
2616 :     Tag: leunga-20000704-sparc-x86
2617 :    
2618 :     1. x86 peephole improvement sp += k; sp -= k => nop [from John]
2619 :     2. fix to x86 RET bug [found by Dan Grossman]
2620 :     3. sparc assembly bug fix for ticc instructions [found by Fermin]
2621 :    
2622 :     Affects c-- and moby only
2623 :    
2624 :     ----------------------------------------------------------------------
2625 : blume 680 Name: Matthias Blume
2626 : blume 682 Date: 2000/07/04 15:26:00
2627 :     Tag: blume-20000704-trigger
2628 :     Description:
2629 :    
2630 :     1. Improvements to CM manual.
2631 :     2. SMLofNJ.Internals.BTrace.trigger reinstated as an alternative way
2632 :     of getting a back-trace. The function, when called, raises an
2633 :     internal exception which explicitly carries the full back-trace history,
2634 :     so it is unaffected by any intervening handle-raise pairs ("trivial"
2635 :     or not). The interactive loop will print that history once it arrives
2636 :     at top level.
2637 :     Short of having all exceptions implicitly carry the full history, the
2638 :     recommended way of using this facility is:
2639 :     - compile your program with instrumentation "on"
2640 :     - run it, when it raises an exception, look at the history
2641 :     - if the history is "cut off" because of some handler, go and modify
2642 :     your program so that it explicitly calls BTrace.trigger
2643 :     - recompile (still instrumented), and rerun; look at the full history
2644 :    
2645 :     ----------------------------------------------------------------------
2646 :     Name: Matthias Blume
2647 :     Date: 2000/07/03 15:36:00 JST
2648 : blume 680 Tag: blume-20000702-manual
2649 :     Description:
2650 :    
2651 :     Small corrections and updates to CM manual.
2652 :    
2653 :     ----------------------------------------------------------------------
2654 : blume 682 Name: Matthias Blume
2655 : blume 679 Date: 2000/06/29 16:04:00 JST
2656 :     Tag: blume-20000629-yacctool
2657 :     Description:
2658 :    
2659 :     Changes:
2660 :    
2661 :     1. Class "mlyacc" now takes separate arguments to pass options to
2662 :     generated .sml- and .sig-files independently.
2663 :     2. Corresponding CM manual updates.
2664 :     3. BTrace module now also reports call sites. (However, for loop clusters
2665 :     it only shows from where the cluster was entered.) There are associated
2666 :     modifications to core.sml, internals.{sig,sml}, btrace.sml, and btimp.sml.
2667 :    
2668 :     ----------------------------------------------------------------------
2669 : blume 675 Name: Matthias Blume
2670 : blume 678 Date: 2000/06/27 16:51:00 JST
2671 :     Tag: blume-20000627-noweb
2672 :     Description:
2673 :    
2674 :     Changes:
2675 :    
2676 :     1. Implemented "subdir" and "witness" options for noweb tool.
2677 :     This caused some slight internal changes in CM's tool implementation.
2678 :     2. Fixed bug in "tool plugin" mechanism. This is essentially cleaning
2679 :     some remaining issues from earlier path anchor changes.
2680 :     3. Updated CM manual accordingly.
2681 :    
2682 :     4. Changed implementation of back-tracing so that I now consider it
2683 :     ready for prime-time.
2684 :    
2685 :     In particular, you don't have to explicitly trigger the back-trace
2686 :     anymore. Instead, if you are running BTrace-instrumented code and
2687 :     there is an uncaught exception (regardless of whether or not it was
2688 :     raised in instrumented code), the top-level evalloop will print
2689 :     the back-trace.
2690 :    
2691 :     Features:
2692 :    
2693 :     - Instrumented and uninstrumented code work together seemlessly.
2694 :     (Of course, uninstrumented code is never mentioned in actual
2695 :     back-traces.)
2696 :    
2697 :     - Asymptotic time- and space-complexity of instrumented code is
2698 :     equal to that of uninstrumented code. (This means that
2699 :     tail-recursion is preserved by the instrumentation phase.)
2700 :    
2701 :     - Modules whose code has been instrumented in different sessions
2702 :     work together without problem.
2703 :    
2704 :     - There is no penalty whatsoever on uninstrumented code.
2705 :    
2706 :     - There is no penalty on "raise" expressions, even in
2707 :     instrumented code.
2708 :    
2709 :     A potential bug (or perhaps it is a feature, too):
2710 :    
2711 :     A back-trace reaches no further than the outermost instrumented
2712 :     non-trivial "raise". Here, a "trivial" raise is one that is the
2713 :     sole RHS of a "handle" rule. Thus, back-traces reach trough
2714 :    
2715 :     <exp> handle e => raise e
2716 :    
2717 :     and even
2718 :    
2719 :     <exp> handle Foo => raise Bar
2720 :    
2721 :     and, of course, through
2722 :    
2723 :     <exp> handle Foo => ...
2724 :    
2725 :     if the exception was not Foo.
2726 :    
2727 :     Back-traces always reach right through any un-instrumented code
2728 :     including any of its "handle" expressions, trivial or not.
2729 :    
2730 :     To try this out, do the following:
2731 :    
2732 :     - Erase all existing binfiles for your program.
2733 :     (You may keep binfiles for those modules where you think you
2734 :     definitely don't need back-tracing.)
2735 :     - Turn on back-trace instrumentation:
2736 :     SMLofNJ.Internals.BTrace.mode (SOME true);
2737 :     - Recompile your program. (I.e., run "CM.make" or "use".)
2738 :     - You may now turn instrumentation off again (if you want):
2739 :     SMLofNJ.Internals.BTrace.mode (SOME false);
2740 :     - Run your program as usual. If it raises an exception that
2741 :     reaches the interactive toplevel, then a back-trace will
2742 :     automatically be printed. After that, the toplevel loop
2743 :     will print the exception history as usual.
2744 :    
2745 :     ----------------------------------------------------------------------
2746 :     Name: Matthias Blume
2747 : blume 677 Date: 2000/06/26 09:56:46 JST
2748 :     Tag: blume-20000626-setup
2749 :     Description:
2750 :    
2751 :     CM: - setup-parameter to "sml" added; this can be used to run arbitrary
2752 :     ML code before and after compiling a file (e.g., to set compiler
2753 :     flags)
2754 :    
2755 :     Compiler: - improved btrace API (in core.sml, internals.{sig,sml})
2756 :     - associated changes to btrace.sml (BTrace instrumentation pass)
2757 :     - cleaner implementation of btimp.sml (BTrace tracing and report
2758 :     module)
2759 :    
2760 :     CM manual: * new path encoding documented
2761 :     * description of setup-parameter to "sml" added
2762 :    
2763 :     The biggest user-visible change to back-tracing is that it is no
2764 :     longer necessary to compile all traced modules within the same
2765 :     session. (This was a real limitation.)
2766 :    
2767 :     ----------------------------------------------------------------------
2768 :     Name: Matthias Blume
2769 : blume 676 Date: 2000/06/24 12:40:00 JST
2770 :     Tag: blume-20000624-startup
2771 :     Description:
2772 :    
2773 :     Fixes startup slowdown problem. (I was calling SrcPath.sync a _tad_
2774 :     bit too often -- to put it mildly. :)
2775 :    
2776 :     ----------------------------------------------------------------------
2777 :     Name: Matthias Blume
2778 : blume 675 Date: 2000/06/23 18:20:00 JST
2779 :     Tag: blume-20000623-btrace
2780 :     Description:
2781 :    
2782 :     This updates adds a backtrace facility to aid programmers in debugging
2783 :     their programs. This involves the following changes:
2784 :    
2785 :     1. Module system/smlnj/init/core.sml (structure _Core) now has hooks for
2786 :     keeping track of the current call stack. When programs are compiled
2787 :     in a special mode, the compiler will insert calls to these hooks
2788 :     into the user program.
2789 :     "Hook" means that it is possible for different implementations of
2790 :     back-tracing to register themselves (at different times).
2791 :    
2792 :     2. compiler/MiscUtil/profile/btrace.sml implements the annotation phase
2793 :     as an Absyn.dec->Absyn.dec rewrite. Normally this phase is turned off.
2794 :     It can be turned on using this call:
2795 :     SMLofNJ.Internals.BTrace.mode (SOME true);
2796 :     Turning it off again:
2797 :     SMLofNJ.Internals.BTrace.mode (SOME false);
2798 :     Querying the current status:
2799 :     SMLofNJ.Internals.BTrace.mode NONE;
2800 :     Annotated programs are about twice as big as normal ones, and they
2801 :     run a factor of 2 to 4 slower with a dummy back-trace plugin (one
2802 :     where all hooks do nothing). The slowdown with a plugin that is
2803 :     actually useful (such as the one supplied by default) is even greater,
2804 :     but in the case of the default plugin it is still only an constant
2805 :     factor (amortized).
2806 :    
2807 :     3. system/Basis/Implementation/NJ/internals.{sig,sml} have been augmented
2808 :     with a sub-structure BTrace for controlling back-tracing. In particular,
2809 :     the above-mentioned function "mode" controls whether the annotation
2810 :     phase is invoked by the compiler. Another important function is
2811 :     "trigger": when called it aborts the current execution and causes
2812 :     the top-level loop to print a full back-trace.
2813 :    
2814 :     4. compiler/MiscUtil/profile/btimp.sml is the current default plugin
2815 :     for back-tracing. It keeps track of the dynamic call stack and in
2816 :     addition to that it keeps a partial history at each "level" of that
2817 :     stack. For example, if a tail-calls b, b tail-calls c, and c tail-calls
2818 :     d and b (at separate times, dynamically), then the report will show:
2819 :    
2820 :     GOTO d
2821 :     /c
2822 :     GOTO \b
2823 :     CALL a
2824 :    
2825 :     This shows that there was an initial non-tail call of a, then a
2826 :     tail-call to b or c, looping behavior in a cluster of functions that
2827 :     consist of b and c, and then a goto from that cluster (i.e., either from
2828 :     b or from c) to d.
2829 :    
2830 :     Note that (depending on the user program) the amount of information
2831 :     that the back-trace module has to keep track of at each level is bounded
2832 :     by a constant. Thus, the whole implementation has the same asymptotical
2833 :     complexity as the original program (both in space and in time).
2834 :    
2835 :     5. compiler/TopLevel/interact/evalloop.sml has been modified to
2836 :     handle the special exception SMLofNJ.Internals.BTrace.BTrace
2837 :     which is raised by the "trigger" function mentioned above.
2838 :    
2839 :     Notes on usage:
2840 :    
2841 :     - Annotated code works well together with unannotated code:
2842 :     Unannotated calls simply do not show up at all in the backtrace.
2843 :    
2844 :     - It is not a good idea to let modules that were annotated during
2845 :     different sessions run at the same time. This is because the compiler
2846 :     chooses small integers to identify individual functions, and there
2847 :     will be clashes if different modules were compiled in separate sessions.
2848 :     (Nothing will crash, and you will even be told about the clashes, but
2849 :     back-trace information will in general not be useful.)
2850 :    
2851 :     - Back-tracing can be confused by callcc and capture.
2852 :    
2853 :     - The only way of getting a back-trace right now is to explicitly
2854 :     invoke the "trigger" function from your user program. Eventually, we
2855 :     should make every exception carry back-trace information (if
2856 :     available). But since this creates more overhead at "raise"-time
2857 :     (similar to the current exnHistory overhead), I have not yet
2858 :     implemented this. (The implementation will be rather easy.) With
2859 :     exceptions carrying back-trace information, this facility will be even
2860 :     more useful because users don't need to modify their programs...
2861 :    
2862 :     - While it is possible to compile the compiler with back-trace
2863 :     annotations turned on (I did it to get some confidence in
2864 :     correctness), you must make absolutely sure that core.sml and
2865 :     btimp.sml are compiled WITHOUT annotation! (core.sml cannot actually
2866 :     be compiled with annotation because there is no core access yet, but
2867 :     if you compile btimp.sml with annotation, then the system will go into
2868 :     an infinite recursion and crash.)
2869 :     Since CM currently does not know about BTrace, the only way to turn
2870 :     annotations on and off for different modules of the compiler is to
2871 :     interrupt CMB.make, change the settings, and re-invoke it. Of course,
2872 :     this is awkward and clumsy.
2873 :    
2874 :     Sample sessions:
2875 :    
2876 :     Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
2877 :     - SMLofNJ.Internals.BTrace.mode (SOME true);
2878 :     [autoloading]
2879 :     [autoloading done]
2880 :     val it = false : bool
2881 :     - structure X = struct
2882 :     - fun main n = let
2883 :     - fun a (x, 0) = d x
2884 :     - | a (x, n) = b (x, n - 1)
2885 :     - and b (x, n) = c (x, n)
2886 :     - and c (x, n) = a (x, n)
2887 :     - and d x = e (x, 3)
2888 :     - and e (x, 0) = f x
2889 :     - | e (x, n) = e (x, n - 1)
2890 :     - and f 0 = SMLofNJ.Internals.BTrace.trigger ()
2891 :     - | f n = n * g (n - 1)
2892 :     - and g n = a (n, 3)
2893 :     - in
2894 :     - f n
2895 :     - end
2896 :     - end;
2897 :     structure X : sig val main : int -> int end
2898 :     - X.main 3;
2899 :     *** BACK-TRACE ***
2900 :     GOTO stdIn:4.2-13.20: X.main[2].f
2901 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
2902 :     GOTO stdIn:4.2-13.20: X.main[2].d
2903 :     / stdIn:4.2-13.20: X.main[2].a
2904 :     | stdIn:4.2-13.20: X.main[2].b
2905 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
2906 :     CALL stdIn:4.2-13.20: X.main[2].g
2907 :     GOTO stdIn:4.2-13.20: X.main[2].f
2908 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
2909 :     GOTO stdIn:4.2-13.20: X.main[2].d
2910 :     / stdIn:4.2-13.20: X.main[2].a
2911 :     | stdIn:4.2-13.20: X.main[2].b
2912 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
2913 :     CALL stdIn:4.2-13.20: X.main[2].g
2914 :     GOTO stdIn:4.2-13.20: X.main[2].f
2915 :     GOTO-( stdIn:4.2-13.20: X.main[2].e
2916 :     GOTO stdIn:4.2-13.20: X.main[2].d
2917 :     / stdIn:4.2-13.20: X.main[2].a
2918 :     | stdIn:4.2-13.20: X.main[2].b
2919 :     GOTO-\ stdIn:4.2-13.20: X.main[2].c
2920 :     CALL stdIn:4.2-13.20: X.main[2].g
2921 :     GOTO stdIn:4.2-13.20: X.main[2].f
2922 :     CALL stdIn:2.15-17.4: X.main[2]
2923 :     -
2924 :    
2925 :     (Note that because of a FLINt bug the above code currently does not
2926 :     compile without BTrace turned on.)
2927 :    
2928 :     Here is another example, using my modified Tiger compiler:
2929 :    
2930 :     Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
2931 :     - SMLofNJ.Internals.BTrace.mode (SOME true);
2932 :     [autoloading]
2933 :     [autoloading done]
2934 :     val it = false : bool
2935 :     - CM.make "sources.cm";
2936 :     [autoloading]
2937 :     ...
2938 :     [autoloading done]
2939 :     [scanning sources.cm]
2940 :     [parsing (sources.cm):parse.sml]
2941 :     [creating directory CM/SKEL ...]
2942 :     [parsing (sources.cm):tiger.lex.sml]
2943 :     ...
2944 :     [wrote CM/sparc-unix/semant.sml]
2945 :     [compiling (sources.cm):main.sml]
2946 :     [wrote CM/sparc-unix/main.sml]
2947 :     [New bindings added.]
2948 :     val it = true : bool
2949 :     - Main.compile ("../testcases/merge.tig", "foo.out");
2950 :     *** BACK-TRACE ***
2951 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trvar
2952 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
2953 :     CALL lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
2954 :     GOTO lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
2955 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
2956 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
2957 :     CALL lib/semant.sml:488.3-505.6: SemantFun[2].transDec.trdec[2].transBody[2]
2958 :     / lib/semant.sml:411.65-543.8: SemantFun[2].transDec
2959 :     CALL-\ lib/semant.sml:413.2-540.9: SemantFun[2].transDec.trdec[2]
2960 :     CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
2961 :     CALL lib/semant.sml:8.52-558.4: SemantFun[2].transProg[2]
2962 :     CALL main.sml:1.18-118.4: Main.compile[2]
2963 :     -
2964 :    
2965 :     ----------------------------------------------------------------------
2966 : blume 673 Name: Matthias Blumen
2967 :     Date: 2000/06/21 18:00:00 JST
2968 :     Tag: blume-20000621-manual
2969 :     Description:
2970 :    
2971 :     CM manual update: Path environments documented.
2972 :    
2973 :     ----------------------------------------------------------------------
2974 : blume 666 Name: Matthias Blume
2975 : blume 672 Date: 2000/06/19 13:40:00
2976 :     Tag: blume-20000619-manual
2977 :     Description:
2978 :    
2979 :     CM manual and system/README update. This only covers the fact that
2980 :     there are no more implicit anchors. (Path environments and the "bind"
2981 :     option to "cm" have yet to be documented.)
2982 :    
2983 :     ----------------------------------------------------------------------
2984 :     Name: Matthias Blume
2985 : blume 671 Date: 2000/06/19 11:05:00 JST
2986 :     Tag: blume-20000619-chdir-bugfix
2987 :     Description:
2988 :    
2989 :     Fixed a bug in new SrcPath module that sometimes led to a bad chDir call.
2990 :    
2991 :     ----------------------------------------------------------------------
2992 :     Name: Matthias Blume
2993 : blume 670 Date: 2000/06/18 22:00:10 JST
2994 :     Tag: blume-20000618-implicit-anchors-really-gone
2995 :     Description:
2996 :    
2997 :     I updates the previous HISTORY entry where I forgot to mention that
2998 :     implicit anchors are no longer with us.
2999 :    
3000 :     The current update also gets rid of the (now useless) controller
3001 :     CM.Control.implicit_anchors.
3002 :    
3003 :     ----------------------------------------------------------------------
3004 :     Name: Matthias Blume
3005 : blume 666 Date: 2000/06/16 17:30:00 JST
3006 :     Tag: blume-20000616-anchorenv
3007 :     Description:
3008 :    
3009 :     This patch implements the long anticipated (just kidding :) "anchor
3010 :     environment" mechanism. In the course of doing this, I also
3011 :     re-implemented CM's internal "SrcPath" module from scratch. The new
3012 :     one should be more robust in certain boundary cases. In any case, it
3013 :     is a lot cleaner than its predecessor (IMHO).
3014 :    
3015 :     This time, although there is yet another boot file format change, I
3016 :     kept the unpickler backward-compatible. As a result, no new bootfiles
3017 :     are necessary and bootstrapping is straightforward. (You cannot read
3018 :     new bootfiles into an old system, but the other way around is no
3019 :     problem.)
3020 :    
3021 :     Visible changes:
3022 :    
3023 : blume 670 ** 0. Implicit path anchors (without the leading $-symbol) are no
3024 :     longer recognized at all. This means that such path names are not
3025 :     illegal either. For example, the name basis.cm simply refers to a
3026 :     local file called "basis.cm" (i.e, the name is an ordinary path
3027 :     relative to .cm-files directory). Or, to put it differently, only
3028 :     names that start with $ are anchored paths.
3029 : blume 666
3030 : blume 670 ** 1. The $<singlearc> abbreviation for $/<singlearc> has finally
3031 :     vanished.
3032 :    
3033 :     John (Reppy) had critizised this as soon as I originally proposed and
3034 : blume 666 implemented it, but at that time I did not really deeply believe
3035 :     him. :) Now I came full-circle because I need the $<singlearc> syntax
3036 :     in another place where it cannot be seen as an abbreviation for
3037 :     $/<singlearc>. To avoid the confusion, $<singlearc> now means what it
3038 :     seems to mean (i.e., it "expands" into the corresponding anchor
3039 :     value).
3040 :    
3041 :     However, when paths are used as members in CM description files, it
3042 :     continues to be true that there must be at least another arc after the
3043 :     anchor. This is now enforced separately during semantic analysis
3044 : blume 670 (i.e., from a lexical/syntactical point of view, the notation is ok.)
3045 : blume 666
3046 :     ** 2. The "cm" class now accepts an option "bind". The option's value
3047 :     is a sub-option list of precisely two items -- one labeled "anchor"
3048 :     and the other one labeled "value". As you might expect, "anchor" is
3049 :     used to specify an anchor name to be bound, and "value" specifies what
3050 :     the anchor is being bound to.
3051 :    
3052 :     The value must be a directory name and can be given in either standard
3053 :     syntax (including the possibility that it is itself an anchored path)
3054 :     or native syntax.
3055 :    
3056 :     Examples:
3057 :    
3058 :     foo.cm (bind:(anchor:bar value:$mystuff/bar))
3059 :     lib.cm (bind:(anchor:a value:"H:\\x\\y\\z")) (* only works under windows *)
3060 :    
3061 :     and so on.
3062 :    
3063 :     The meaning of this is that the .cm-file will be processed with an
3064 :     augmented anchor environment where the given anchor(s) is/are bound to
3065 :     the given values(s).
3066 :    
3067 :     The rationale for having this feature is this: Suppose you are trying
3068 :     to use two different (already stable) libraries a.cm and b.cm (that
3069 :     you perhaps didn't write yourself). Further, suppose each of these
3070 :     two libraries internally uses its own auxiliary library $aux/lib.cm.
3071 :     Normally you would now have a problem because the anchor "lib" can not
3072 :     be bound to more than one value globally. Therefore, the project that
3073 :     uses both a.cm and b.cm must locally redirect the anchor to some other
3074 :     place:
3075 :    
3076 :     a.cm (bind:(anchor:lib value:/usr/lib/smlnj/a-stuff))
3077 :     b.cm (bind:(anchor:lib value:/usr/lib/smlnj/b-stuff))
3078 :    
3079 :     This hard-wires $lib/aux.cm to /usr/lib/smlnj/a-stuff/aux.cm or
3080 :     /usr/lib/smlnj/b-stuff/aux.cm, respectively.
3081 :    
3082 :     Hard-wiring path names is a bit inflexible (and CM will verbosely warn
3083 :     you when you do so at the time of CM.stabilize). Therefore, you can
3084 :     also use an anchored path as the value:
3085 :    
3086 :     a.cm (bind:(anchor:lib value:$a-lib))
3087 :     b.cm (bind:(anchor:lib value:$b-lib))
3088 :    
3089 :     Now you can globally configure (using the usual CM.Anchor.anchor or
3090 :     pathconfig machinery) bindings for "a-lib" and "b-lib". Since "lib"
3091 :     itself is always locally bound, setting it globally is no longer
3092 :     meaningful or necessary (but it does not hurt either). In fact, "lib"
3093 :     can still be used as a global anchor for separate purposes. As a
3094 :     matter of fact, one can locally define "lib" in terms of a global
3095 :     "lib":
3096 :    
3097 :     a.cm (bind:(anchor:lib value:$lib/a))
3098 :     b.cm (bind:(anchor:lib value:$lib/b))
3099 :    
3100 :     ** 3: The encoding of path names has changed. This affects the way
3101 :     path names are shown in CM's progress report and also the internal
3102 :     protocol encoding used for parallel make.
3103 :    
3104 :     The encoding now uses one or more ':'-separated segments. Each
3105 :     segments corresponds to a file that has been specified relative to the
3106 :     file given by its preceding segment. The first segment is either
3107 :     relative to the CWD, absolute, or anchored. Each segment itself is
3108 :     basically a Unix pathname; all segments but the first are relative.
3109 :    
3110 :     Example:
3111 :    
3112 :     $foo/bar/baz.cm:a/b/c.sml
3113 :    
3114 :     This path denotes the file bar/a/b/c.sml relative to the directory
3115 :     denoted by anchor "foo". Notice that the encoding also includes
3116 :     baz.cm which is the .cm-file that listed a/b/c.sml. As usual, such
3117 :     paths are resolved relative to the .cm-files directory, so baz.cm must
3118 :     be ignored to get the "real" pathname.
3119 :    
3120 :     To make this fact more obvious, CM puts the names of such "virtual
3121 :     arcs" into parentheses when they appear in progress reports. (No
3122 :     parentheses will appear in the internal protocol encoding.) Thus,
3123 :     what you really see is:
3124 :    
3125 :     $foo/bar/(baz.cm):a/b/c.sml
3126 :    
3127 :     I find this notation to be much more informative than before.
3128 :    
3129 :     Another new feature of the encoding is that special characters
3130 :     including parentheses, colons, (back)slashes, and white space are
3131 :     written as \ddd (where ddd is the decimal encoding of the character).
3132 :    
3133 : blume 670 *** The CM manual still needs to be updated.
3134 :    
3135 : blume 666 ----------------------------------------------------------------------
3136 : leunga 660 Name: Allen Leung
3137 :     Date: 2000/06/15 00:38:00
3138 :     Tag: leunga-20000615-x86-peephole
3139 :    
3140 :     x86 Peephole fix by Fermin. Affects c-- and moby only.
3141 :    
3142 :     ----------------------------------------------------------------------
3143 : blume 658 Name: Matthias Blume
3144 :     Date: 2000/06/12 11:40:00
3145 :     Tag: blume-20000612-parmakefix
3146 :     Description:
3147 :    
3148 :     More cleanup after changing the file naming scheme: This time I
3149 :     repaired the parallel make mechanism for CMB.make which I broke earlier.
3150 :    
3151 :     ----------------------------------------------------------------------
3152 : leunga 657 Name: Allen Leung
3153 :     Date: 2000/06/09 01:25:00
3154 :     Tag: leunga-20000609-various
3155 :    
3156 :     None of these things should affect normal SML/NJ operations
3157 :    
3158 :     1. Peephole improvements provided by Fermin (c--)
3159 :     2. New annotation DEFUSE for adding extra dependence (moby)
3160 :     3. New X86 LOCK instructions (moby)
3161 :     4. New machine description language for reservation tables (scheduling)
3162 :     5. Fixes to various optimization/analysis modules (branch chaining, dominator
3163 :     trees etc.)
3164 :     6. I've changed the CM files so that they can work with versions
3165 :     110.0.6, 110.25 and 110.28
3166 :    
3167 :     ----------------------------------------------------------------------
3168 : blume 652 Name: Matthias Blume
3169 : blume 656 Date: 2000/06/09 12:40:00
3170 :     Tag: blume-20000609-log
3171 :     Description:
3172 :    
3173 :     - Removed all(?) remaining RCS Log entries from sources.
3174 :    
3175 :     - Fixed bug in ml-yacc and ml-lex sources (use explicit anchors for
3176 :     anchored paths).
3177 :    
3178 :     ----------------------------------------------------------------------
3179 :     Name: Matthias Blume
3180 : blume 653 Date: 2000/06/07 17:00:00 JST
3181 :     Tag: blume-20000607-no-implicit-anchors
3182 :     Description:
3183 :    
3184 :     1. This update changes the default setting for
3185 :     CM.Control.implicit_anchors from true to false. This means that
3186 :     implicit anchors are no longer permitted by default. I also tried to
3187 :     make sure that nothing else still relies on implicit anchors.
3188 :     (This is the next step on the schedule towards a CM that does not even
3189 :     have the notion of implicit anchors anymore.)
3190 :    
3191 :     2. More CM manual updates.
3192 :    
3193 :     3. I managed to track down and fix the pickling bug I mentioned last
3194 :     time. Because of the previously existing workaround, this entails no
3195 :     immediate practical changes.
3196 :    
3197 :     ----------------------------------------------------------------------
3198 :     Name: Matthias Blume
3199 : blume 652 Date: 2000/06/06 11:15:00 JST
3200 :     Tag: blume-20000606-lazierpickle
3201 :     Description:
3202 :    
3203 :     !!!! NEW BOOT FILES !!!!
3204 :    
3205 :     * The main purpose of this update is to make library pickles lazier in
3206 :     order to reduce the initial space penalty for autoloading a library.
3207 :     As a result, it is now possible to have $smlnj/compiler.cm
3208 :     pre-registered. This should take care of the many complaints or
3209 :     inquiries about missing structure Compiler. This required changes to
3210 :     CM's internal data structures and small tweaks to some algorithms.
3211 :    
3212 :     As a neat additional effect, it is no longer necessary (for the sake
3213 :     of lean heap image files) to distinguish between a "minimal" CM and a
3214 :     "full" CM. Now, there is only one CM (i.e., the "full" version:
3215 :     $smlnj/cm.cm aka $smlnj/cm/full.cm), and it is always available at the
3216 :     interactive top level. ($smlnj/cm/minimal.cm is gone.)
3217 :    
3218 :     To make the life of compiler-hackers easier, "makeml" now also
3219 :     pre-registers $smlnj/cmb.cm (aka $smlnj/cmb/current.cm). In other
3220 :     words, after you bootstrap a new sml for the first time, you will not
3221 :     have to autoload $smlnj/cmb.cm again afterwards. (The first time
3222 :     around you will still have to do it, though.)
3223 :    
3224 :     * A second change consists of major updates to the CM manual. There
3225 :     are now several appendices with summary information and also a full
3226 :     specification of the CM description file syntax.
3227 :    
3228 :     * In directory src/system I added the script "allcross". This script
3229 :     invokes sml and cross-compiles the compiler for all supported
3230 :     architectures. (Useful when providing a new set of boot files.)
3231 :    
3232 :     * There seems to be a latent bug in my "lazy pickles" mechanism. I
3233 :     added a small tweak to pickle-util.sml to work around this problem,
3234 :     but it is not a proper fix yet. I will investigate further. (The
3235 :     effect of the bug was an inflation of library pickle size.)
3236 :    
3237 :     * Version number increased to 110.28.1 (to avoid compatibility problems).
3238 :    
3239 :     ----------------------------------------------------------------------
3240 : leunga 646 Name: Allen Leung
3241 : leunga 648 Date: 2000/05/25 17:28 EDT
3242 :     Tag: leunga-20000525-ra
3243 :     Description:
3244 :    
3245 :     Fixed a bug in freezing phase of the register allocator.
3246 :    
3247 :     ----------------------------------------------------------------------
3248 :     Name: Allen Leung
3249 : leunga 647 Date: 2000/05/15 22:53 EDT
3250 : leunga 646 Tag: leunga-20000515-alpha-x86-ra
3251 :     Description:
3252 :    
3253 :     1. Alpha
3254 :    
3255 :     Slight cleanup. Removed the instruction SGNXL
3256 :    
3257 :     2. X86
3258 :    
3259 :     Added the following instructions to the instruction set:
3260 :    
3261 :     ROLx, RORx,
3262 :     BTx, BTSx, BTLx, BTRx,
3263 :     XCHGx, and variants with the LOCK prefix
3264 :    
3265 :     3. Register Allocation
3266 :    
3267 :     The module ra-rewrite-with-renaming has been improved.
3268 :    
3269 : leunga 647 These have no effect on SML/NJ.
3270 :    
3271 : leunga 646 ----------------------------------------------------------------------
3272 : blume 642 Name: Matthias Blume
3273 : blume 645 Date: 2000/05/15 16:20:00 JST
3274 :     Tag: blume-20000515-lightrebuild
3275 :     Description:
3276 :    
3277 :     1. I added an alternative to "-rebuild" to "makeml". The difference is
3278 :     that prior to calling CMB.make' the CM-variable "LIGHT" will be
3279 :     defined. In effect, the command will not build any cross-compiler
3280 :     backends and therefore finish more quickly.
3281 :    
3282 :     The "fixpt" script also takes a "-light" switch to be able to use
3283 :     this new facility while compiling for a fixpoint.
3284 :    
3285 :     2. I replaced all mentions of anchored paths in group owner specifications
3286 :     with simple relative paths (usually starting with "..").
3287 :     The rationale is that a library's internal workings should not be
3288 :     compromised by the lack of some anchor. (An anchor is necessary
3289 :     for someone who wants to refer to the library by an anchored path,
3290 :     but it should not be necessary to build the same library in the first
3291 :     place.)
3292 :    
3293 :     3. I changed the way CM's tool mechanism determines the shell command
3294 :     string used for things like ml-yacc etc. so that it does not break
3295 :     when CM.Control.implicit_anchors is turned off.
3296 :    
3297 :     ----------------------------------------------------------------------
3298 :     Name: Matthias Blume
3299 : blume 644 Date: 2000/05/12 18:20:00 JST
3300 :     Tag: blume-20000512-ml-build
3301 :     Description:
3302 :    
3303 :     Fixed a bug in config/_ml-build that prevented ml-yacc and ml-lex from
3304 :     getting installed properly (by config/install.sh).
3305 :    
3306 :     ----------------------------------------------------------------------
3307 :     Name: Matthias Blume
3308 : blume 643 Date: 2000/05/12 17:30:00 JST
3309 :     Tag: blume-20000512-anchors
3310 :     Description:
3311 :    
3312 :     !!! NEW BOOT FILES !!!
3313 :    
3314 :     This change is in preparation of fading out support for "implicitly
3315 :     anchored path names". I went through all sources and used the
3316 :     explicit (and relatively new) $-notation. See system/README and the
3317 :     CM manual for more info on this.
3318 :    
3319 :     I also modified the anchoring scheme for some things such as "smlnj",
3320 :     "MLRISC", "cm", etc. to take advantage of the fact that explicit
3321 :     anchors are more expressive: anchor name and first arc do not have to
3322 :     coincide. This entails the following user-visible change:
3323 :    
3324 :     You have to write $smlnj/foo/bar instead of smlnj/foo/bar. In
3325 :     particular, when you fire up sml with a command-line argument, say,
3326 :     e.g.:
3327 :    
3328 :     sml '$smlnj/cmb.cm'
3329 :    
3330 :     At the ML toplevel prompt:
3331 :    
3332 :     CM.autoload "$smlnj/cmb.cm";
3333 :    
3334 :     There is also a new controller in CM.Control that can be used to turn
3335 :     off all remaining support for implicit anchors by saying:
3336 :    
3337 :     CM.autoload "$smlnj/
3338 :     #set CM.Control.implicit_anchors false;
3339 :    
3340 :     This causes CM to reject implicitly anchored paths. This is (for the
3341 :     time being) less permissive than the "final" version where there will
3342 :     be no more such implicit anchors and relative paths will be just that:
3343 :     relative.
3344 :    
3345 :     The next step (version after next version?) will be to make the
3346 :     default for CM.Control.implicit_anchors false. After the dust has
3347 :     settled, I can then produce the "final" version of this...
3348 :    
3349 :     Note: Since bootstrapping is a bit tricky, I provided new boot files.
3350 :    
3351 :     ----------------------------------------------------------------------
3352 :     Name: Matthias Blume
3353 : blume 642 Date: 2000/05/11 16:30:00 JST
3354 :     Tag: blume-20000511-sources
3355 :     Description:
3356 :    
3357 :     The main change is that I added function CM.sources as a generalized
3358 :     version of the earlier CM.makedepend. This entails the following
3359 :     additional changes:
3360 :    
3361 :     - CM.makedepend has been dropped.
3362 :    
3363 :     - CM manual has been updated.
3364 :    
3365 :     - TOOLS signature and API have been changed.
3366 :    
3367 :     ----------------------------------------------------------------------
3368 : leunga 640 Name: Allen Leung
3369 : leunga 641 Date: 2000/05/10 21:17 EDT
3370 :     Tag: leunga-20000510-moby-c--ssa
3371 :     Description:
3372 :    
3373 :     Various bug fixes and new features for C--, Moby and MLRISC optimizations.
3374 :     None of these affect SML/NJ.
3375 :    
3376 :     1. Register Allocation
3377 :    
3378 :     a. A new ra spilling module (ra/ra-spill-with-renaming) is implemented.
3379 :     This module tries to remove local (i.e. basic block level) redundancies
3380 :     during spilling.
3381 :    
3382 :     b. A new framework for performing region based register allocation.
3383 :     Not yet entirely functional.
3384 :    
3385 :     2. X86
3386 :    
3387 :     a. DefUse for POP was missing the stack pointer [found by Lal]
3388 :     b. Reload for CALL was incorrect in X86Spill [found by John]
3389 :     c. Various fixes in X86Spill so that it can be used correctly for
3390 :     the new spilling module.
3391 :    
3392 :     3. SSA/IR
3393 :    
3394 :     a. New module ir/dj-dataflow.sml implements elimination based
3395 :     data flow analysis.
3396 :    
3397 :     4. MLRiscGen
3398 :    
3399 :     a. Fix for gc type annotation
3400 :    
3401 :     5. MDGen
3402 :    
3403 :     Various fixes for machine description -> ml code translation. For ssa
3404 :     only.
3405 :    
3406 :     ----------------------------------------------------------------------
3407 :     Name: Allen Leung
3408 : leunga 640 Date: 2000/05/08 22:17 EDT
3409 :     Tag: leunga-20000508-labexp
3410 :     Description:
3411 :    
3412 :     Fermin has found a few assembly problems with constant expressions
3413 :     generated in LabelExp. Mostly, the problems involve extra parentheses,
3414 :     which choke on dumb assemblers. This is his fix.
3415 :    
3416 :     ----------------------------------------------------------------------
3417 : dbm 635 Name: Dave MacQueen
3418 :     Date: 2000/04/09 14:00 EDT
3419 :     Tag: dbm-20000502-Version_110_28
3420 :     Description:
3421 :    
3422 :     1. Updated src/compiler/TopLevel/main/version.sml to version 110.28
3423 :    
3424 :     2. Updated config/version to 110.28
3425 :    
3426 :     3. Updated config/srcarchiveurl
3427 :    
3428 :     3. New boot files!
3429 :     ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.28/
3430 :    
3431 :     ----------------------------------------------------------------------
3432 : blume 634 Name: Matthias Blume
3433 :     Date: 2000/05/01 19:05:00 JST
3434 :     Tag: blume-20000501-noweb
3435 :     Description:
3436 :    
3437 :     A new noweb tool has been added. The existing system is entirely
3438 :     unaffected by this, but some CM users have asked for renewed noweb
3439 :     support. Everything is documented in the CM manual.
3440 :    
3441 :     New (plugin) libraries:
3442 :    
3443 :     noweb-tool.cm
3444 :     nw-ext.cm
3445 :    
3446 :     ----------------------------------------------------------------------
3447 : dbm 633 Name: Dave MacQueen
3448 :     Date: 2000/04/30 12:40PM EDT
3449 :     Tag: dbm-20000430-bug_fixes
3450 :     Description:
3451 :    
3452 :     1. Fix for bug 1498
3453 :     smlnj/src/system/Basis/Implementation/Unsafe/object.sig
3454 :     smlnj/src/system/Basis/Implementation/Unsafe/object.sml
3455 :     added toRealArray function
3456 :     smlnj/src/compiler/MiscUtil/print/ppobj.sml
3457 :     added check for tag Obj.RealArray to array printing case in ppObj
3458 :    
3459 :     2. Fix for bug 1510
3460 :     smlnj/src/compiler/Semant/types/typesutil.sml
3461 :     fixed definition of dummyargs (used by equalTycon) so that
3462 :     dummy args are distinct types
3463 :    
3464 :     ----------------------------------------------------------------------
3465 : blume 629 Name: Matthias Blume
3466 : blume 632 Date: 2000/04/30 01:00:00 JST
3467 :     Tag: blume-20000430-versions
3468 :     Description:
3469 :    
3470 :     1. CM version numbering added. This is an implementation of Lal's
3471 :     proposal for adding version numbers and version checking to .cm
3472 :     files. Lal said that his proposal was just that -- a proposal.
3473 :     For the time being I went ahead and implemented it so that people
3474 :     can comment on it. Everything is completely backward-compatible
3475 :     (except for the stable library format, i.e., new bootfiles!).
3476 :    
3477 :     As usual, see the CM manual for details.
3478 :    
3479 :     2. An alternative syntax for anchored paths has been implemented.
3480 :     Dave has recently voiced the same concerns that I had when I did
3481 :     this, so there should be some support. My take is that eventually
3482 :     I will let support for the current syntax (where anchors are
3483 :     "implicit") fade out in favor of the new, explicit syntax.
3484 :     In order to be backward-compatible, both old and new syntax are
3485 :     currently supported.
3486 :    
3487 :     Again, see the CM manual for details.
3488 :    
3489 :     3. Parallel make is trying to be slightly smarter: When the master
3490 :     process finds a "bottleneck", i.e., when there is only one
3491 :     compilation unit that can be compiled and everybody else is
3492 :     waiting on it, then it will simply compile it directly instead
3493 :     of clumsily telling one of the slaves to do it.
3494 :    
3495 :     4. Support for "unsharing" added. This is necessary in order to be
3496 :     able to have two different versions of the same library running
3497 :     at the same time (e.g., for trying out a new MLRISC while still
3498 :     having the old MLRISC linked into the current compiler, etc.)
3499 :     See the CM manual.
3500 :    
3501 :     5. Simple "makedepend" functionality added for generating Makefile
3502 :     dependency information. (This is rather crude at the moment.
3503 :     Expect some changes here in the future.)
3504 :    
3505 :     6. ".fun" added as a recognized suffix for ML files. Also documented
3506 :     explicitly in the manual that the fallback behavior (unknown suffix
3507 :     -> ML file) is not an official feature!
3508 :    
3509 :     7. Small changes to the pickler for stable libraries.
3510 :    
3511 :     8. Several internal changes to CM (for cleanup/improvement).
3512 :    
3513 :    
3514 :     !!!! NEW BINFILES !!!!
3515 :    
3516 :     ----------------------------------------------------------------------
3517 :     Name: Matthias Blume
3518 : blume 631 Date: 2000/04/28 17:30:00 JST
3519 :     Tag: blume-20000428-pathconfig
3520 :     Description:
3521 :    
3522 :     1. I changed config/install.sh to remove duplicate entries from the
3523 :     lib/pathconfig file at the end. Moreover, the final version of
3524 :     lib/pathconfig is sorted alphabetically. The same (sorting) is done
3525 :     in src/system/installml.
3526 :    
3527 :     2. The config/install.sh script now consistently uses relative
3528 :     pathnames in lib/pathconfig whenever the anchor is in the lib
3529 :     directory. (So far this was true for the libraries that come
3530 :     pre-compiled and bundled as part of the bootfiles but not for
3531 :     libraries that are compiled by the script itself.)
3532 :    
3533 :     ----------------------------------------------------------------------
3534 :     Name: Matthias Blume
3535 : blume 629 Date: 2000/04/26 13:10:00 JST
3536 :     Tag: blume-20000426-fun_suffix
3537 :     Description:
3538 :    
3539 :     Added ".fun" as a recognized file name suffix (for ML code).
3540 :    
3541 :     ----------------------------------------------------------------------
3542 : leunga 628 Name: Allen Leung
3543 :     Date: 2000/04/25 17:00:00 EST
3544 :     Tag: leunga-20000425-alpha-ra
3545 :     Description:
3546 :    
3547 :     1. Alpha
3548 :    
3549 :     PSEUDOARITH was missing in AlphaRewrite. This causes an endless loop
3550 :     in C--.
3551 :    
3552 :     2. RA
3553 :    
3554 :     Added a flag "ra-dump-size" to print out the size of the flowgraph
3555 :     and the interference graph.
3556 :    
3557 :     ----------------------------------------------------------------------
3558 : dbm 627 Name: Dave MacQueen
3559 :     Date: 2000/04/25/
3560 :     Tag: dbm-20000425-mlyacc_doc_examples
3561 :     Description:
3562 :     Updated mlyacc.tex sections 5 and 7 for SML '97 and CM.
3563 :     Updated all three examples in src/ml-yacc/examples to run
3564 :     under 110.* using CM.make.
3565 :    
3566 :     ----------------------------------------------------------------------
3567 : leunga 624 Name: Allen Leung
3568 :     Date: 2000/04/20 23:04:00 EST
3569 :     Tag: leunga-20000420-ssa-c---stuff
3570 :     Description:
3571 :    
3572 :     This update synchronizes my repository with Yale's. Most of these
3573 :     changes, however, do not affect SML/NJ at all (the RA is an exception).
3574 :    
3575 :     1. Register Allocator
3576 :    
3577 :     a. An improvement in the interference graph construction:
3578 :     Given a copy
3579 :    
3580 :     s <- t
3581 :    
3582 : leunga 628 no interference edge between s and t is added for this definition of s.
3583 : leunga 624
3584 :     b. I've added two new spill heuristic modules that Fermin and I developed
3585 :     (in the new library RA.cm). These are unused in SML/NJ but maybe
3586 :     useful for others (Moby?)
3587 :    
3588 :     2. X86
3589 :    
3590 :     a. Various fixes in the backend provided by Fermin [C--] and Lal.
3591 :    
3592 :     3. Alpha
3593 :    
3594 :     a. Added the BSR instruction and code generation that goes with it [C--]
3595 :     b. Other fixes too numerous to recount provided by Fermin [C--]
3596 :    
3597 :     4. Regmaps
3598 :    
3599 :     a. The regmaps are not initialized with the identity physical bindings
3600 :     at creation time. This is unneeded.
3601 :    
3602 :     5. MLRISC Optimizations
3603 :    
3604 :     a. The DJ-Graph module can now compute the iterated dominance frontiers
3605 :     intersects with liveness incrementally in linear time! Woohoo!
3606 :     This is now used in my new SSA construction algorithm.
3607 :    
3608 :     b. THe branch reorganization module is now smarter about linear chains of
3609 :     basic blocks.
3610 :    
3611 :    
3612 :     ----------------------------------------------------------------------
3613 : blume 615 Name: Matthias Blume
3614 :     Date: 2000/04/12 13:52:00 JST
3615 :     Tag: blume_main_v110p27_1
3616 :     Description:
3617 :    
3618 :     Changed install.sh script to handle archive files without version number
3619 :     and to use "boot.<arch>-<os>" instead of "sml.boot.<arch>-<os>" for the
3620 :     name of the boot file archive.
3621 :    
3622 :     ----------------------------------------------------------------------
3623 : dbm 607 Name: Dave MacQueen
3624 :     Date: 2000/04/09 14:00 EDT
3625 :     Tag: dbm-20000410-Version_110_27
3626 :     Description:
3627 :    
3628 :     1. Updated src/compiler/TopLevel/main/version.sml to version 110.27
3629 :    
3630 :     2. Updated src/config/version to 110.27
3631 :    
3632 :     3. New boot files!
3633 :    
3634 :     ----------------------------------------------------------------------
3635 : leunga 606 Name: Allen Leung
3636 :     Date: 2000/04/09 19:09:00 EST
3637 :     Tag: leunga-20000409-misc
3638 :     Description:
3639 :    
3640 :     1. Yet another fix for x86 assembly for idivl, imull, mull and friends.
3641 :    
3642 :     2. Miscellaneous improvements to MLRISC (unused in sml/nj)
3643 :    
3644 :     ----------------------------------------------------------------------
3645 : monnier 605 Name: Stefan
3646 :     Date: 2000/04/07 10:00:00 EDT
3647 :     Tag: monnier-20000406-branch-handling
3648 :     Description:
3649 :    
3650 :     Improved handling of branches (mostly those generated from
3651 :     polymorphic equality), removed switchoff and changed the
3652 :     default optimization settings (more cpsopt and less flintopt).
3653 :    
3654 :     ----------------------------------------------------------------------
3655 : leunga 601 Name: Allen Leung
3656 : leunga 602 Date: 2000/04/06 01:30:00 EST
3657 :     Tag: leunga-20000406-peephole-x86-SSA-2
3658 :     Description:
3659 :    
3660 :     Forgot a few files.
3661 :    
3662 :     ----------------------------------------------------------------------
3663 :     Name: Allen Leung
3664 : leunga 601 Date: 2000/04/06 00:36:00 EST
3665 :     Tag: leunga-20000406-peephole-x86-SSA
3666 :     Description:
3667 :    
3668 :     1. New Peephole code
3669 :    
3670 :     2. Minor improvement to X86 instruction selection
3671 :    
3672 :     3. Various fixes to SSA and machine description -> code translator
3673 :    
3674 :     ----------------------------------------------------------------------
3675 : blume 595 Name: Matthias Blume
3676 :     Date: 2000/04/05 12:30:00 JST
3677 :     Tag: blume_main_v110p26p2_3
3678 :     Description:
3679 :    
3680 :     This update just merges three minor cosmetic updates to CM's sources
3681 :     to get ready for the 110.27 code freeze on Friday. No functionality
3682 :     has changed.
3683 :    
3684 :     ----------------------------------------------------------------------
3685 : leunga 593 Name: Allen Leung
3686 : leunga 594 Date: 2000/04/04 19:39:00 EST
3687 :     Tag: leunga-20000404-x86-asm
3688 :     Description:
3689 :    
3690 :     1. Fixed a problem in X86 assembly.
3691 :    
3692 :     Things like
3693 :    
3694 :     jmp %eax
3695 :     jmp (%eax)
3696 :    
3697 :     should be output as
3698 :    
3699 :     jmp *%eax
3700 :     jmp *(%eax)
3701 :    
3702 :     2. Assembly output
3703 :    
3704 :     Added a new flag
3705 :    
3706 :     "asm-indent-copies" (default to false)
3707 :    
3708 :     When this flag is on, parallel copies will be indented an extra level.
3709 :    
3710 :     ----------------------------------------------------------------------
3711 :     Name: Allen Leung
3712 : leunga 593 Date: 2000/04/04 03:18:00 EST
3713 :     Tag: leunga-20000404-C--Moby
3714 :     Description:
3715 :    
3716 :     All of these fixes are related to C--, Moby, and my own optimization
3717 :     stuff; so they shouldn't affect SML/NJ.
3718 :    
3719 :     1. X86
3720 :    
3721 :     Various fixes related floating point, and extensions.
3722 :    
3723 :     2. Alpha
3724 :    
3725 :     Some extra patterns related to loads with signed/zero extension
3726 :     provided by Fermin.
3727 :    
3728 :     3. Assembly
3729 :    
3730 : monnier 659 When generating assembly, resolve the value of client defined constants,
3731 : leunga 593 instead of generating symbolic values. This is controlled by the
3732 :     new flag "asm-resolve-constants", which is default to true.
3733 :    
3734 :     4. Machine Descriptions
3735 :    
3736 :     a. The precedence parser was slightly broken when parsing infixr symbols.
3737 :     b. The type generalizing code had the bound variables reversed, resulting
3738 :     in a problem during arity raising.
3739 :     c. Various fixes in machine descriptions.
3740 :    
3741 :     ----------------------------------------------------------------------
3742 : blume 592 Name: Matthias Blume
3743 :     Date: 2000/04/03 16:05:00 JST
3744 :     Tag: blume_main_v110p26p2_2
3745 :     Description:
3746 :    
3747 :     I eliminated coreEnv from compInfo. Access to the "Core" structure is
3748 :     now done via the ordinary static environment that is context to each
3749 :     compilation unit.
3750 :    
3751 :     To this end, I arranged that instead of "structure Core" as "structure
3752 :     _Core" is bound in the pervasive environment. Core access is done via
3753 : monnier 659 _Core (which can never be accidentally rebound because _Core is not a
3754 : blume 592 legal surface-syntax symbol).
3755 :    
3756 :     The current solution is much cleaner because the core environment is
3757 :     now simply part of the pervasive environment which is part of every
3758 :     compilation unit's context anyway. In particular, this eliminates all
3759 :     special-case handling that was necessary until now in order to deal
3760 :     with dynamic and symbolic parts of the core environment.
3761 :    
3762 :     Remaining hackery (to bind the "magic" symbol _Core) is localized in the
3763 : monnier 659 compilation manager's bootstrap compiler (actually: in the "init group"
3764 : blume 592 handling). See the comments in src/system/smlnj/init/init.cmi for
3765 :     more details.
3766 :    
3767 :     I also tried to track down all mentions of "Core" (as string argument
3768 :     to Symbol.strSymbol) in the compiler and replaced them with a
3769 :     reference to the new CoreSym.coreSym. Seems cleaner since the actual
3770 :     name appears in one place only.
3771 :    
3772 :     Binfile and bootfile format have not changed, but the switchover from
3773 :     the old "init.cmi" to the new one is a bit tricky, so I supplied new
3774 :     bootfiles anyway.
3775 :    
3776 :     ----------------------------------------------------------------------
3777 : leunga 591 Name: Allen Leung
3778 :     Date: 2000/04/02 21:17:00 EST
3779 :     Tag: leunga-20000402-mltree
3780 :     Description:
3781 : leunga 585
3782 : leunga 591 1. Renamed the constructor CALL in MLTREE by popular demand.
3783 :     2. Added a bunch of files from my repository. These are currently
3784 :     used by other non-SMLNJ backends.
3785 :    
3786 : leunga 576 ----------------------------------------------------------------------
3787 : leunga 590 Name: Allen Leung
3788 :     Date: 2000/03/31 21:15:00 EST
3789 :     Tag: leunga-20000331-aliasing
3790 :     Description:
3791 :    
3792 :     This update contains a rewritten (and hopefully more correct) module
3793 :     for extracting aliasing information from CPS.
3794 :    
3795 :     To turn on this feature:
3796 :    
3797 :     Compiler.Control.CG.memDisambiguate := true
3798 :    
3799 :     To pretty print the region information with assembly
3800 :    
3801 :     Compiler.Control.MLRISC.getFlag "asm-show-region" := true;
3802 :    
3803 :     To control how many levels of aliasing information are printed, use:
3804 :    
3805 :     Compiler.Control.MLRISC.getInt "points-to-show-level" := n
3806 :    
3807 :     The default of n is 3.
3808 :    
3809 :     ----------------------------------------------------------------------
3810 : dbm 589 Name: David MacQueen
3811 :     Date: 2000/03/31 11:15:00 EST
3812 :     Tag: dbm-20000331-runtime_fix
3813 :     Description:
3814 :    
3815 :     This update contains:
3816 :    
3817 :     1. runtime/c-lib/c-libraries.c
3818 :     includes added in revision 1.2 caused compilation errors on hppa-hpux
3819 :    
3820 :     2. fix for bug 1556
3821 :     system/Basis/Implementation/NJ/internal-signals.sml
3822 :    
3823 :     ----------------------------------------------------------------------
3824 : blume 587 Name: Matthias Blume
3825 : blume 588 Date: 2000/03/31 18:00:00 JST
3826 :     Tag: blume_main_v110p26p2_1
3827 :     Description:
3828 :    
3829 :     This update contains:
3830 :    
3831 :     1. A small change to CM's handling of stable libraries:
3832 :     CM now maintains one "global" modmap that is used for all stable
3833 :     libraries. The use of such a global modmap maximizes sharing and
3834 :     minimizes the need for re-traversing parts of environments during
3835 :     modmap construction. (However, this has minor impact since modmap
3836 :     construction seems to account for just one percent or less of total
3837 :     compile time.)
3838 :    
3839 :     2. I added a "genmap" phase to the statistics. This is where I got the
3840 :     "one percent" number (see above).
3841 :    
3842 :     3. CM's new tool parameter mechanism just became _even_ better. :)
3843 :     - The parser understands named parameters and recursive options.
3844 :     - The "make" and "shell" tools use these new features.
3845 :     (This makes it a lot easier to cascade these tools.)
3846 :     - There is a small syntax change: named parameters use a
3847 :    
3848 :     <name> : ( <option> ... ) or
3849 :     <name> : <string>
3850 :    
3851 :     syntax. Previously, named parameters were implemented in an
3852 :     ad-hoc fashion by each tool individually (by parsing strings)
3853 :     and had the form
3854 :    
3855 :     <name>=<string>
3856 :    
3857 :     See the CM manual for a full description of these issues.
3858 :    
3859 :     ----------------------------------------------------------------------
3860 :     Name: Matthias Blume
3861 : blume 587 Date: 2000/03/30 18:00:00 JST
3862 :     Tag: blume_main_v110p26p2_0
3863 :     Description:
3864 :    
3865 :     !!!!! WARNING !!!!!!
3866 :     !! New binfiles !!
3867 :     !!!!!!!!!!!!!!!!!!!!
3868 :    
3869 :     This update contains:
3870 :    
3871 :     1. Moderate changes to CM:
3872 :    
3873 :     - Changes to CM's tools mechanism. In particular, it is now possible
3874 :     to have tools that accept additional "command line" parameters
3875 :     (specified in the .cm file at each instance where the tool's class is
3876 :     used).
3877 :    
3878 : monnier 659 This was done to accommodate the new "make" and "shell" tools which
3879 :     facilitate fairly seamless hookup to portions of code managed using
3880 : blume 587 Makefiles or Shell scripts.
3881 :    
3882 :     There are no classes "shared" or "private" anymore. Instead, the
3883 :     sharing annotation is now a parameter to the "sml" class.
3884 :    
3885 :     There is a bit of generic machinery for implementing one's own
3886 :     tools that accept command-line parameters. However, I am not yet fully
3887 :     satisfied with that part, so expect changes here in the future.
3888 :    
3889 :     All existing tools are described in the CM manual.
3890 :    
3891 : monnier 659 - Slightly better error handling. (CM now suppresses many followup
3892 : blume 587 error messages that tended to be more annoying than helpful.)
3893 :    
3894 :     2. Major changes to the compiler's static environment data structures.
3895 :    
3896 :     - no CMStaticEnv anymore.
3897 :     - no CMEnv, no "BareEnvironment" (actually, _only_ BareEnvironment,
3898 :     but it is called Environment), no conversions between different
3899 :     kinds of static environments
3900 :    
3901 :     - There is still a notion of a "modmap", but such modmaps are generated
3902 :     on demand at the time when they are needed. This sounds slow, but I
3903 :     sped up the code that generates modmaps enough for this not to lead to
3904 :     a slowdown of the compiler (at least I didn't detect any).
3905 :    
3906 :     - To facilitate rapid modmap generation, static environments now
3907 :     contain an (optional) "modtree" structure. Modtree annotations are
3908 :     constructed by the unpickler during unpickling. (This means that
3909 :     the elaborator does not have to worry about modtrees at all.)
3910 :     Modtrees have the advantage that they are compositional in the same
3911 :     way as the environment data structure itself is compositional.
3912 :     As a result, modtrees never hang on to parts of an environment that
3913 :     has already been rendered "stale" by filtering or rebinding.
3914 :    
3915 :     - I went through many, many trials and errors before arriving at the
3916 :     current solution. (The initial idea of "linkpaths" did not work.)
3917 :     But the result of all this is that I have touched a lot of files that
3918 :     depend on the "modules" and "types" data structures (most of the
3919 :     elaborator). There were a lot of changes during my "linkpath" trials
3920 :     that could have been reverted to their original state but weren't.
3921 :     Please, don't be too harsh on me for messing with this code a bit more
3922 : monnier 921 than what was strictly necessary... (I _did_ resist the temptation
3923 : blume 587 of doing any "global reformatting" to avoid an untimely death at
3924 :     Dave's hands. :)
3925 :    
3926 :     - One positive aspect of the previous point: At least I made sure that
3927 :     all files that I touched now compile without warnings (other than
3928 :     "polyEqual").
3929 :    
3930 :     - compiler now tends to run "leaner" (i.e., ties up less memory in
3931 :     redundant modmaps)
3932 :    
3933 :     ----------------------------------------------------------------------
3934 : leunga 580 Name: Allen Leung
3935 : leunga 585 Date: 2000/03/29 18:00:00
3936 :     Tag: leunga-20000327-mlriscGen_hppa_alpha_x86
3937 : leunga 586 Boot files (optional): ftp://react-ilp.cs.nyu.edu/leunga/110.26.1-sml.boot.x86-unix-20000330.tar.gz
3938 : leunga 585 Description:
3939 :    
3940 :     This update contains *MAJOR* changes to the way code is generated from CPS
3941 :     in the module mlriscGen, and in various backend modules.
3942 :    
3943 :     CHANGES
3944 :     =======
3945 :    
3946 :     1. MLRiscGen: forward propagation fix.
3947 :    
3948 :     There was a bug in forward propagation introduced at about the same time
3949 :     as the MLRISC x86 backend, which prohibits coalescing to be
3950 :     performed effectively in loops.
3951 :    
3952 :     Effect: speed up of loops in RISC architectures.
3953 :     By itself, this actually slowed down certain benchmarks on the x86.
3954 :    
3955 :     2. MLRiscGen: forward propagating addresses from consing.
3956 :    
3957 :     I've changed the way consing code is generated. Basically I separated
3958 :     out the initialization part:
3959 :    
3960 :     store tag, offset(allocptr)
3961 :     store elem1, offset+4(allocptr)
3962 :     store elem2, offset+8(allocptr)
3963 :     ...
3964 :     store elemn, offset+4n(allocptr)
3965 :    
3966 :     and the address computation part:
3967 :    
3968 :     celladdr <- offset+4+alloctpr
3969 :    
3970 :     and move the address computation part
3971 :    
3972 :     Effect: register pressure is generally lower as a result. This
3973 :     makes compilation of certain expressions much faster, such as
3974 :     long lists with non-trivial elements.
3975 :    
3976 :     [(0,0), (0,0), .... (0,0)]
3977 :    
3978 :     3. MLRiscGen: base pointer elimination.
3979 :    
3980 :     As part of the linkage mechanism, we generate the sequence:
3981 :    
3982 :     L: ... <- start of the code fragment
3983 :    
3984 :     L1:
3985 :     base pointer <- linkreg - L1 + L
3986 :    
3987 :     The base pointer was then used for computing relocatable addresses
3988 :     in the code fragment. Frequently (such as in lots of continuations)
3989 :     this is not needed. We now eliminate this sequence whenever possible.
3990 :    
3991 :     For compile time efficiency, I'm using a very stupid local heuristic.
3992 :     But in general, this should be done as a control flow analysis.
3993 :    
3994 :     Effect: Smaller code size. Speed up of most programs.
3995 :    
3996 :     4. Hppa back end
3997 :    
3998 :     Long jumps in span dependence resolution used to depend on the existence
3999 :     of the base pointer.
4000 :    
4001 :     A jump to a long label L was expanded into the following sequence:
4002 :    
4003 :     LDIL %hi(L-8192), %r29
4004 :     LDO %lo(L-8192)(%r29), %r29
4005 :     ADD %r29, baseptr, %r29
4006 :     BV,n %r0(%r29)
4007 :    
4008 :     In the presence of change (3) above, this will not work. I've changed
4009 :     it so that the following sequence of instructions are generated, which
4010 :     doesn't mention the base pointer at all:
4011 :    
4012 :     BL,n L', %r29 /* branch and link, L' + 4 -> %r29 */
4013 :     L': ADDIL L-(L'+4), %r29 /* Compute address of L */
4014 :     BV,n %r0(%r29) /* Jump */
4015 :    
4016 :     5. Alpha back end
4017 :    
4018 :     New alpha instructions LDB/LDW have been added, as per Fermin's
4019 :     suggestions. This is unrelated to all other changes.
4020 :    
4021 :     6. X86 back end
4022 :    
4023 :     I've changed andl to testl in the floating point test sequence
4024 :     whenever appropriate. The Intel optimization guide states that
4025 : monnier 659 testl is preferable to andl.
4026 : leunga 585
4027 :     7. RA (x86 only)
4028 :    
4029 :     I've improved the spill propagation algorithm, using an approximation
4030 :     of maximal weighted independent sets. This seems to be necessary to
4031 :     alleviate the negative effect in light of the slow down in (1).
4032 :    
4033 :     I'll write down the algorithm one of these days.
4034 :    
4035 :     8. MLRiscGen: frequencies
4036 :    
4037 :     I've added an annotation that states that all call gc blocks have zero
4038 :     execution frequencies. This improves register allocation on the x86.
4039 :    
4040 :     BENCHMARKS
4041 :     ==========
4042 :    
4043 :     I've only perform the comparison on 110.25.
4044 :    
4045 :     The platforms are:
4046 :    
4047 :     HPPA A four processor HP machine (E9000) with 5G of memory.
4048 :     X86 A 300Hhz Pentium II with 128M of memory, and
4049 :     SPARC An Ultra sparc 2 with 512M of memory.
4050 :    
4051 :     I used the following parameters for the SML benchmarks:
4052 :    
4053 :     @SMLalloc
4054 :     HPPA 256k
4055 :     SPARC 512k
4056 :     X86 256k
4057 :    
4058 :     COMPILATION TIME
4059 :     ----------------
4060 :     Here are the numbers comparing the compilation times of the compilers.
4061 :     I've only compared 110.25 compiling the new sources versus
4062 :     a fixpoint version of the new compiler compiling the same.
4063 :    
4064 :     110.25 New
4065 :     Total Time in RA Spill+Reload Total Time In RA Spill+Reload
4066 :     HPPA 627s 116s 2684+3584 599s 95s 1003+1879
4067 :     SPARC 892s 173s 2891+3870 708s 116s 1004+1880
4068 :     X86 999s 315s 94006+130691 987s 296s 108877+141957
4069 :    
4070 :     110.25 New
4071 :     Code Size Code Size
4072 :     HPPA 8596736 8561421
4073 :     SPARC 8974299 8785143
4074 :     X86 9029180 8716783
4075 :    
4076 :     So in summary, things are at least as good as before. Dramatic
4077 :     reduction in compilation is obtained on the Sparc; I can't explain it,
4078 :     but it is reproducible. Perhaps someone should try to reproduce this
4079 :     on their own machines.
4080 :    
4081 :     SML BENCHMARKS
4082 :     --------------
4083 :    
4084 :     On the average, all benchmarks perform at least as well as before.
4085 :    
4086 :     HPPA Compilation Time Spill+Reload Run Time
4087 :     110.25 New 110.25 New 110.25 New
4088 :    
4089 :     barnesHut 3.158 3.015 4.75% 1+1 0+0 2.980 2.922 2.00%
4090 :     boyer 6.152 5.708 7.77% 0+0 0+0 0.218 0.213 2.34%
4091 :     count-graphs 1.168 1.120 4.32% 0+0 0+0 22.705 23.073 -1.60%
4092 :     fft 0.877 0.792 10.74% 1+3 1+3 0.602 0.587 2.56%
4093 :     knuthBendix 3.180 2.857 11.32% 0+0 0+0 0.675 0.662 2.02%
4094 :     lexgen 6.190 5.290 17.01% 0+0 0+0 0.913 0.788 15.86%
4095 :     life 0.803 0.703 14.22% 25+25 0+0 0.153 0.140 9.52%
4096 :     logic 2.048 2.007 2.08% 6+6 1+1 4.133 4.008 3.12%
4097 :     mandelbrot 0.077 0.080 -4.17% 0+0 0+0 0.765 0.712 7.49%
4098 :     mlyacc 22.932 20.937 9.53% 154+181 32+57 0.468 0.430 8.91%
4099 :     nucleic 5.183 5.060 2.44% 2+2 0+0 0.125 0.120 4.17%
4100 :     ratio-regions 3.357 3.142 6.84% 0+0 0+0 116.225 113.173 2.70%
4101 :     ray 1.283 1.290 -0.52% 0+0 0+0 2.887 2.855 1.11%
4102 :     simple 6.307 6.032 4.56% 28+30 5+7 3.705 3.658 1.28%
4103 :     tsp 0.888 0.862 3.09% 0+0 0+0 7.040 6.893 2.13%
4104 :     vliw 24.378 23.455 3.94% 106+127 25+45 2.758 2.707 1.91%
4105 :     --------------------------------------------------------------------------
4106 :     Average 6.12% 4.09%
4107 :    
4108 :     SPARC Compilation Time Spill+Reload Run Time
4109 :     110.25 New 110.25 New 110.25 New
4110 :    
4111 :     barnesHut 3.778 3.592 5.20% 2+2 0+0 3.648 3.453 5.65%
4112 :     boyer 6.632 6.110 8.54% 0+0 0+0 0.258 0.242 6.90%
4113 :     count-graphs 1.435 1.325 8.30% 0+0 0+0 33.672 34.737 -3.07%
4114 :     fft 0.980 0.940 4.26% 3+9 2+6 0.838 0.827 1.41%
4115 :     knuthBendix 3.590 3.138 14.39% 0+0 0+0 0.962 0.967 -0.52%
4116 :     lexgen 6.593 6.072 8.59% 1+1 0+0 1.077 1.078 -0.15%
4117 :     life 0.972 0.868 11.90% 26+26 0+0 0.143 0.140 2.38%
4118 :     logic 2.525 2.387 5.80% 7+7 1+1 5.625 5.158 9.05%
4119 :     mandelbrot 0.090 0.093 -3.57% 0+0 0+0 0.855 0.728 17.39%
4120 :     mlyacc 26.732 23.827 12.19% 162+189 32+57 0.550 0.560 -1.79%
4121 :     nucleic 6.233 6.197 0.59% 3+3 0+0 0.163 0.173 -5.77%
4122 :     ratio-regions 3.780 3.507 7.79% 0+0 0+0 133.993 131.035 2.26%
4123 :     ray 1.595 1.550 2.90% 1+1 0+0 3.440 3.418 0.63%
4124 :     simple 6.972 6.487 7.48% 29+32 5+7 3.523 3.525 -0.05%
4125 :     tsp 1.115 1.063 4.86% 0+0 0+0 7.393 7.265 1.77%
4126 :     vliw 27.765 24.818 11.87% 110+135 25+45 2.265 2.135 6.09%
4127 :     ----------------------------------------------------------------------------
4128 :     Average 6.94% 2.64%
4129 :    
4130 :     X86 Compilation Time Spill+Reload Run Time
4131 :     110.25 New 110.25 New 110.25 New
4132 :    
4133 :     barnesHut 5.530 5.420 2.03% 593+893 597+915 3.532 3.440 2.66%
4134 :     boyer 8.768 7.747 13.19% 493+199 301+289 0.327 0.297 10.11%
4135 :     count-graphs 2.040 2.010 1.49% 298+394 315+457 26.578 28.660 -7.26%
4136 :     fft 1.327 1.302 1.92% 112+209 115+210 1.055 0.962 9.71%
4137 :     knuthBendix 5.218 5.475 -4.69% 451+598 510+650 0.928 0.932 -0.36%
4138 :     lexgen 9.970 9.623 3.60% 1014+841 1157+885 0.947 0.928 1.97%
4139 :     life 1.183 1.183 0.00% 162+182 145+148 0.127 0.103 22.58%
4140 :     logic 3.285 3.512 -6.45% 514+684 591+836 5.682 5.577 1.88%
4141 :     mandelbrot 0.147 0.143 2.33% 38+41 33+54 0.703 0.690 1.93%
4142 :     mlyacc 35.457 32.763 8.22% 3496+4564 3611+4860 0.552 0.550 0.30%
4143 :     nucleic 7.100 6.888 3.07% 239+168 201+158 0.175 0.173 0.96%
4144 :     ratio-regions 6.388 6.843 -6.65% 1182+257 981+300 120.142 120.345 -0.17%
4145 :     ray 2.332 2.338 -0.29% 346+398 402+494 3.593 3.540 1.51%
4146 :     simple 9.912 9.903 0.08% 1475+941 1579+1168 3.057 3.178 -3.83%
4147 :     tsp 1.623 1.532 5.98% 266+200 250+211 8.045 7.878 2.12%
4148 :     vliw 33.947 35.470 -4.29% 2629+2774 2877+3171 2.072 1.890 9.61%
4149 :     ----------------------------------------------------------------------------
4150 :     Average 1.22% 3.36%
4151 :    
4152 :     ----------------------------------------------------------------------
4153 :     Name: Allen Leung
4154 : leunga 583 Date: 2000/03/23 16:25:00
4155 :     Tag: leunga-20000323-fix_x86_alpha
4156 :     Description:
4157 :    
4158 :     1. X86 fixes/changes
4159 :    
4160 :     a. The old code generated for SETcc was completely wrong.
4161 :     The Intel optimization guide is VERY misleading.
4162 :    
4163 :     2. ALPHA fixes/changes
4164 :    
4165 :     a. Added the instructions LDBU, LDWU, STB, STW as per Fermin's suggestion.
4166 :     b. Added a new mode byteWordLoadStores to the functor parameter to Alpha()
4167 :     c. Added reassociation code for address computation.
4168 :    
4169 :     ----------------------------------------------------------------------
4170 :     Name: Allen Leung
4171 : leunga 580 Date: 2000/03/22 01:23:00
4172 :     Tag: leunga-20000322-fix_x86_hppa_ra
4173 :     Description:
4174 :    
4175 :     1. X86 fixes/changes
4176 :    
4177 :     a. x86Rewrite bug with MUL3 (found by Lal)
4178 :     b. Added the instructions FSTS, FSTL
4179 :    
4180 :     2. PA-RISC fixes/changes
4181 :    
4182 :     a. B label should not be a delay slot candidate! Why did this work?
4183 :     b. ADDT(32, REG(32, r), LI n) now generates one instruction instead of two,
4184 :     as it should be.
4185 :     c. The assembly syntax for fstds and fstdd was wrong.
4186 :     d. Added the composite instruction COMICLR/LDO, which is the immediate
4187 :     operand variant of COMCLR/LDO.
4188 :    
4189 :     3. Generic MLRISC
4190 :    
4191 :     a. shuffle.sml rewritten to be slightly more efficient
4192 :     b. DIV bug in mltree-simplify fixed (found by Fermin)
4193 :    
4194 :     4. Register Allocator
4195 :    
4196 :     a. I now release the interference graph earlier during spilling.
4197 :     May improve memory usage.
4198 :    
4199 :     ----------------------------------------------------------------------
4200 : blume 577 Name: Matthias Blume
4201 : blume 578 Date: 2000/03/14 14:15:32
4202 :     Tag: blume_main_v110p26p1_2
4203 :     Description:
4204 :    
4205 :     1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an
4206 :     additional argument called "template" which is an optional string that
4207 : monnier 659 specifies the layout of the tool command line. See the CM manual for
4208 : blume 578 explanation.
4209 :    
4210 : monnier 659 2. A special-purpose tool can be "registered" by simply dropping the
4211 : blume 578 corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same
4212 :     directory where the .cm file lives that uses this tool. (The
4213 :     behavior/misfeature until now was to look for the tool description
4214 :     files in the current working directory.) As before, tool description
4215 :     files could also be anchored -- in which case they can live anywhere
4216 :     they like. Following the recent e-mail discussion, this change should
4217 :     make it easier to have special-purpose tools that are shipped together
4218 :     with the sources of the program that uses them.
4219 :    
4220 :     ----------------------------------------------------------------------
4221 :     Name: Matthias Blume
4222 : blume 577 Date: 2000/03/10 07:48:34
4223 :     Tag: blume_main_v110p26p1_1
4224 :     Description:
4225 :    
4226 :     I added a re-written version of Dave's fixpt script to src/system.
4227 :     Changes relative to the original version:
4228 :     - sh-ified (not everybody has ksh)
4229 :     - automatically figures out which architecture it runs on
4230 :     - uses ./makeml a bit more cleverly
4231 :     - never invokes ./installml (and, thus, does not clobber your
4232 :     good and working installation of sml in case something goes wrong)
4233 :     - accepts max iteration count using option "-iter <n>"
4234 :     - accepts a "base" name using option "-base <base>"
4235 :    
4236 :     It does not build any extraneous heap images but directly rebuilds
4237 :     bin- and boot-hierarchies using makeml's "-rebuild" switch. Finally,
4238 :     it can incorporate existing bin- and boot- hierarchies. For example,
4239 :     suppose the base is set to "sml" (which is the default). Then it
4240 :     successively builds
4241 :    
4242 :     sml.bin.<arch>-unix and sml.boot.<arch>-unix
4243 :     then sml1.bin.<arch>-unix and sml1.boot.<arch>-unix
4244 :     then sml2.bin.<arch>-unix and sml2.boot.<arch>-unix
4245 :     ...
4246 :     then sml<n>.bin.<arch>-unix and sml<n>.boot.<arch>-unix
4247 :    
4248 :     and so on. If any of these already exist, it will just use what's
4249 :     there. In particular, many people will have the initial set of bin
4250 :     and boot files around, so this saves time for at least one full
4251 :     rebuild. Having sets of the form <base><k>.{bin,boot}.<arch>-unix for
4252 :     <k>=1,2,... is normally not a good idea when invoking fixpt. However,
4253 :     they might be the result of an earlier partial run of fixpt (which
4254 : monnier 659 perhaps got accidentally killed). In this case, fixpt will quickly
4255 : blume 577 move through what exists before continuing where it left off earlier,
4256 :     and, thus, saves a lot of time.
4257 :    
4258 :     ----------------------------------------------------------------------
4259 : leunga 576 Name: Allen Leung
4260 :     Date: 00/03/10 02:20:00
4261 :     Tag: leunga-20000310-fix_x86_asm_ra
4262 :     Description:
4263 : dbm 570
4264 : leunga 576 More assembly output problems involving the indexed addressing mode
4265 :     on the x86 have been found and corrected. Thanks to Fermin Reig for the
4266 :     fix.
4267 :    
4268 :     The interface and implementation of the register allocator have been changed
4269 :     slightly to accommodate the possibility to skip the register allocation
4270 :     phases completely and go directly to memory allocation. This is needed
4271 :     for C-- use.
4272 :    
4273 : dbm 570 ----------------------------------------------------------------------
4274 : blume 572 Name: Matthias Blume
4275 : blume 575 Date: 00/03/09 10:23:53
4276 :     Tag: blume_main_v110p26p1_0
4277 :     Description:
4278 :    
4279 :     * Complete re-organization of library names. Many libraries have been
4280 :     consolidated so that they share the same path anchor. For example,
4281 :     all MLRISC-related libraries are anchored at MLRISC, most libraries that
4282 :     are SML/NJ-specific are under "smlnj". Notice that names like
4283 :     host-cmb.cm or host-compiler.cm no longer exist. See system/README
4284 :     for a complete description of the new naming scheme. Quick reference:
4285 :    
4286 :     host-cmb.cm -> smlnj/cmb.cm
4287 :     host-compiler.cm -> smlnj/compiler.cm
4288 :     full-cm.cm -> smlnj/cm.cm
4289 :     <arch>-<os>.cm -> smlnj/cmb/<arch>-<os>.cm
4290 :     <arch>-compiler.cm -> smlnj/compiler/<arch>.cm
4291 :    
4292 :     * Bug fixes in CM.
4293 :     - exceptions in user code are being passed through (i.e., reach top level)
4294 :     - more bugs in paranoia mode fixed
4295 :     - bug related to checking group owners fixed
4296 :    
4297 :     * New install.sh script that automagically fetches archive files:
4298 :     The new file config/srcarchiveurl must contain the URL of the
4299 :     (remote) directory that contains bin files (or other source archives).
4300 :     If install.sh does not find the archive locally, it tries to get
4301 :     it from that remote directory.
4302 :     This should simplify installation further: For machines that have
4303 :     access to the internet, just fetch <version>-config.tgz, unpack it,
4304 : monnier 659 edit config/targets, and go (run config/install.sh). The script will
4305 : blume 575 fetch everything else that it might need all by itself.
4306 :    
4307 :     For CVS users, this mechanism is not relevant for source archives, but
4308 :     it is convenient for getting new sets of binfiles.
4309 :    
4310 :     Archives should be tar files compressed with either gzip, compress, or
4311 :     bzip2. The script recognizes .tgz, .tar, tar.gz, tz, .tar.Z, and .tar.bz2.
4312 :    
4313 :     ----------------------------------------------------------------------
4314 :     Name: Matthias Blume
4315 : blume 572 Date: 2000/03/07 04:01:04
4316 :     Tag: blume_main_v110_26_2
4317 : dbm 570 Description:
4318 : blume 572 - size info in BOOTLIST
4319 :     * no fixed upper limits for number of bootfiles or length of
4320 :     bootfile names in runtime
4321 :     * falling back to old behavior if no BOOTLIST size info found
4322 :     - allocation size heuristics in .run-sml
4323 :     * tries to read cache size from /proc/cpuinfo (this is important for
4324 :     small-cache Celeron systems!)
4325 :     - install.sh robustified
4326 :     - CM manual updates
4327 :     - paranoid mode
4328 :     * no more CMB.deliver() (i.e., all done by CMB.make())
4329 :     * can re-use existing sml.boot.* files
4330 :     * init.cmi now treated as library
4331 :     * library stamps for consistency checks
4332 :     - sml.boot.<arch>-<os>/PIDMAP file
4333 :     * This file is read by the CM startup code. This is used to minimize
4334 :     the amount of dynamic state that needs to be stowed away for the
4335 :     purpose of sharing between interactive system and user code.
4336 :     - CM.Anchor.anchor instead of CM.Anchor.{set,cancel}
4337 :     * Upon request by Elsa. Anchors now controlled by get-set-pair
4338 :     like most other CM state variables.
4339 :     - Compiler.CMSA eliminated
4340 :     * No longer supported by CM anyway.
4341 :     - fixed bugs in pickler that kept biting Stefan
4342 :     * past refs to past refs (was caused by the possibility that
4343 :     ad-hoc sharing is more discriminating than hash-cons sharing)
4344 :     * integer overflow on LargeInt.minInt
4345 :     - ml-{lex,yacc} build scripts now use new mechanism
4346 :     for building standalone programs