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 1650 - (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 : mblume 1615 Name:
11 : mblume 1480 Date: yyyy/mm/dd
12 :     Tag: <post-commit CVS tag>
13 :     Description:
14 :    
15 :     ----------------------------------------------------------------------
16 : mblume 1622 Name: Matthias Blume (blume (at) tti - c (dot) org)
17 : mblume 1650 Date: 2004/10/13 16:34:00 CDT
18 :     Tag: blume-20041013-tdp
19 :     Description:
20 :    
21 :     Some rationalization of names:
22 :    
23 :     structure BTrace -> structure TDPInstrument
24 :     etc.
25 :    
26 :     This is is preparation of using the original back-trace
27 :     instrumentation for other purposes. "TDP" stands for
28 :     Trace/Debug/Profile.
29 :    
30 :     The control flag controlling whether instrumentation is on or off is now
31 :     registered under a different name, so instead of running sml as
32 :    
33 :     sml -Cinstrument.btrace-mode=true
34 :    
35 :     one has to say
36 :    
37 :     sml -Ctdp.instrument=true
38 :    
39 :     ----------------------------------------------------------------------
40 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
41 : mblume 1645 Date: 2004/10/11 16:37:00 CDT
42 :     Tag: blume-20041011-regions
43 :     Description:
44 :    
45 :     Made some minor modifications to elabcore.sml to have source regions
46 :     be propagated more tightly -- resulting in better (i.e., smaller)
47 :     regions being reported in error- and debug messages.
48 :    
49 :     ----------------------------------------------------------------------
50 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
51 : mblume 1644 Date: 2004/10/08 22:50:00 CDT
52 :     Tag: blume-20041008-cmkw
53 :     Description:
54 :    
55 :     Fixed handling of keywords in .cm files: After seeing "is" the lexer
56 :     treats subsequent occurrences of "group", "library", "source", "is",
57 :     "*", and "-" as ordinary identifiers rather than keywords.
58 :    
59 :     Most seriously, this fixes a problem with CM's "shell" tool. The tool
60 :     is supposed to accept a tool argument called "source", but this did
61 :     not work because of the clash with the keyword.
62 :    
63 :     ----------------------------------------------------------------------
64 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
65 : mblume 1642 Date: 2004/10/07 16:00:00 CDT
66 :     Tag: blume-20041007-cleanup
67 :     Description:
68 :    
69 :     Assorted cleanup work:
70 :    
71 :     - got rid of intstrmap in favor of using the library's
72 :     hash table implementation
73 :     - threw out most of the pathnames stuff, as it was not used anyway
74 :     - simplified tokentable implementation
75 :     - fixed some minor spelling errors
76 :    
77 :     ----------------------------------------------------------------------
78 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
79 : mblume 1641 Date: 2004/10/06 15:15:15 CDT
80 :     Tag: blume-20041006-handler
81 :     Description:
82 :    
83 :     Cleaned up the absyn to reflect the invariant that HANDLE always
84 :     carries a FNexp as part of the type definition. This eliminates some
85 :     superfluous sanity checks at runtime down the road.
86 :    
87 :     Some minor cleanup of the btrace code.
88 :    
89 :     ----------------------------------------------------------------------
90 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
91 : mblume 1640 Date: 2004/10/01 10:20:30 CDT
92 :     Tag: blume-20041001-slave
93 :     Description:
94 :    
95 :     Added hack to make slave mode work in the presence of the version
96 :     tool. (Still, since the master does two passes over the code for
97 :     CMB.make, the release number gets bumped twice when slaves are
98 :     attached. I don't know if this is worth fixing...)
99 :    
100 :     ----------------------------------------------------------------------
101 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
102 : mblume 1639 Date: 2004/09/30 10:55:00 CDT
103 :     Tag: blume-20040930-version
104 :     Description:
105 :    
106 :     * Moved the "version" magic into its own little library under
107 :     src/system/smlnj/internal. This avoids expensive reconstruction
108 :     of a stable src/compiler/core.cm.
109 :    
110 :     * At the same time, structure CompilerVersion is now known as structure
111 :     SMLNJVersion.
112 :    
113 :     * Arranged for the version tool to NOT kick in when rebuilding the system
114 :     (makeml -rebuild, fixpt). Otherwise one would never reach a fixpoint.
115 :     Also, loading the versiontool does not work when rebuilding the system
116 :     because CM is not properly initialized at that time.
117 :    
118 :     ----------------------------------------------------------------------
119 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
120 : mblume 1638 Date: 2004/09/29 14:00:00 CDT
121 :     Tag: blume-20040929-autoversion
122 :     Description:
123 :    
124 :     Implemented some CM magic to have
125 :     file src/compiler/TopLevel/main/version.sml
126 :     generated automagically.
127 :     The version is taken from two files: config/version and config/release.
128 :     The first is expected to contain a two-part version number such as 110.49.
129 :     The second should contain a single number, but it may be missing.
130 :    
131 :     If the environment variable VERSIONTOOL_BUMP_RELEASE is defined at the
132 :     time the version tool is loaded (which is the first time you say CMB.make),
133 :     then the tool will increment the value stored in config/release every
134 :     time CMB.make is invoked.
135 :    
136 :     The binfile format is now insensitive to anything beyond the first
137 :     two components of a version number, so bumping the release does not render
138 :     binfiles incompatible. Auto-bumping can be used to keep track of versions
139 :     during development without invalidating existing binfiles.
140 :    
141 :     In any case, every CMB.make updates the date information in version.sml.
142 :     (This is the date that is printed in the banner.)
143 :    
144 :     ----------------------------------------------------------------------
145 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
146 : mblume 1634 Date: 2004/09/28 10:53:00 CDT
147 :     Tag: blume-20040928-controls
148 :     Description:
149 :    
150 :     Some cleanup of the controls code.
151 :    
152 :     ----------------------------------------------------------------------
153 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
154 : mblume 1633 Date: 2004/09/27 22:08:00 CDT
155 :     Tag: blume-20040927-controls
156 :     Description:
157 :    
158 :     Added two pieces of functionality to the Controls interface:
159 :    
160 :     1. val save'restore: 'a control -> unit -> unit
161 :    
162 :     grabs the current value of the control in stage 1 and restores it
163 :     in stage 2.
164 :    
165 :     2. val set' : 'a control * 'a -> unit -> unit
166 :    
167 :     stores the given value into the control in stage 2 (i.e., delayed)
168 :     but does all error checking in stage 1.
169 :     (This is for string controls that need to do parse their argument --
170 :     something that might fail. In some cases, notably in CM, one
171 :     already knows the intended argument but wants to delay the actual
172 :     assignment until a time when error recovery would be more difficult.)
173 :    
174 :     Changed the handling of controls in tool arguments to classes "sml" and
175 :     "lazysml":
176 :     - use Controls.save'restore as a more robust way of restoring the
177 :     old value (in particular: without having to re-parse the string)
178 :     - use controls to handle the "overload" keyword in the init group
179 :     (I believe this change actually fixes a long-standing obscure bug.)
180 :    
181 :     ----------------------------------------------------------------------
182 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
183 : mblume 1632 Date: 2004/09/27 17:00:00 CDT
184 :     Tag: blume-20040927-lazysml
185 :     Description:
186 :    
187 :     Added a new tool class called "lazysml" to CM's tool chest. The only
188 :     difference to "sml" is that compilation is done with Control.lazysml
189 :     set to true. A source of class "lazysml" is automatically recognized
190 :     by a file name suffix of ".lml".
191 :    
192 :     In addition to the above feature, the original class "sml" now also
193 :     supports a tool argument "lazy" which has the same effect. As a
194 :     result, the following three lines are equivalent:
195 :    
196 :     foo.sml : lazysml
197 :     foo.sml : sml (lazy)
198 :     foo.sml (lazy)
199 :    
200 :     The setting goes into effect both during parsing and during
201 :     compilation. The original setting is restored right after parsing and
202 :     after compilation, respectively.
203 :    
204 :     In addition to all the above, there is also a general mechanism to set
205 :     ANY of the "controls" that are available at the command line via
206 :     "-C..." on a per-sml-file basis. The same rules that apply for "lazy"
207 :     apply as well. (In fact, "lazy" is implemented as a special case of
208 :     the general mechanism.)
209 :    
210 :     The .cm file syntax uses a new keyword tool argument called "with".
211 :     There are several ways of indicating the desired settings:
212 :    
213 :     foo.sml (with:parser.quotations=true)
214 :     foo.sml (with:(name:parser.quotations value:true))
215 :     foo.sml (with:(name:name1 value:value1 name:name2 value:value2 ...))
216 :     foo.sml (with:(name1=value1 name2=value2 ...))
217 :     foo.sml (with:(name1=value1 name:name2 value:value2 name3=value3 ...))
218 :    
219 :     etc.
220 :    
221 :     Another possible abbreviation is to leave out the =v or value:v part
222 :     if the name refers to a boolean control (in which case the value is
223 :     taken to be true). Thus, one could get lazy sml also by saying:
224 :    
225 :     foo.sml (with:parser.lazy-keyword=true)
226 :     foo.sml (with:parser.lazy-keyword)
227 :     foo.sml (with:(name:parser.lazy-keyword value:true))
228 :     foo.sml (with:(name:parser.lazy-keyword))
229 :    
230 :     ----------------------------------------------------------------------
231 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
232 : mblume 1631 Date: 2004/09/24 16:22:00 CDT
233 :     Tag: blume-20040924-ppc-long-branch
234 :     Description:
235 :    
236 :     Turned message about "emiting long form of branch" off by default.
237 :     Added a control flag to turn it back on when desired.
238 :    
239 :     ----------------------------------------------------------------------
240 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
241 : mblume 1630 Date: 2004/09/24 16:05:00 CDT
242 :     Tag: blume-20040924-rounding
243 :     Description:
244 :    
245 :     Applied patch for setting rounding modes under Mac OS X. Thanks to
246 :     Melissa O'Neill for providing the code!
247 :    
248 :     ----------------------------------------------------------------------
249 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
250 : mblume 1629 Date: 2004/09/23 17:30:00 CDT
251 :     Tag: blume-20040923-envvars
252 :     Description:
253 :    
254 :     1. Changed definition of type ControlRegistry.registry_tree to
255 :     include control_info (i.e., the name of the controlling environment
256 :     variable).
257 :    
258 :     2. Added command-line flags -e and -E to print the names of environment
259 :     variables that can be used to control internal settings.
260 :     (This uses the new API mentioned in 1.)
261 :    
262 :     ----------------------------------------------------------------------
263 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
264 : mblume 1622 Date: 2004/09/13 16:50:00 CDT
265 :     Tag: Release_110_49
266 :     Description:
267 :    
268 :     New working version (110.49). NEW BOOTFILES!
269 :    
270 :     ----------------------------------------------------------------------
271 : mblume 1621 Name: Matthias Blume
272 :     Date: 2004/09/13 16:20:00 CDT
273 :     Tag: blume-20040913-config-mlrisc
274 :     Description:
275 :    
276 :     Put target "mlrisc" back into the default list.
277 :     (There is no harm in having it, and some users have expressed their
278 :     wish to have "mlrisc" included by default.)
279 :    
280 :     ----------------------------------------------------------------------
281 : jhr 1620 Name: John Reppy
282 :     Date: 2004/09/13
283 :     Tag: jhr-20040913-signals
284 :     Description:
285 :    
286 :     Fixed the signal masking code to properly nest mask/unmask operations
287 :     on a per-signal basis.
288 :    
289 :     ----------------------------------------------------------------------
290 : mblume 1619 Name: Matthias Blume (blume (at) tti - c (dot) org)
291 :     Date: 2004/09/08 13:20:00 CDT
292 :     Tag: blume-20040908-heap-magic
293 :     Description:
294 :    
295 :     Bumped the heap macig number to 0x09082004 to account for the changed
296 :     layout of the ML frame under MacOS X.
297 :    
298 :     ----------------------------------------------------------------------
299 : allenleung 1618 Name: Allen Leung (leunga (at) reservoir (dot) com)
300 :     Date: 2004/09/03 11:26:00 EST
301 :     Tag: leunga-20040903-cygwin-install
302 :     Description:
303 :    
304 :     Added a patch to _arch-n-opsys to enable the Cygwin runtime.
305 :     The Cygwin runtime is turned on by setting the environment
306 :     variable SMLNJ_CYGWIN_RUNTIME to 1.
307 :    
308 :     ----------------------------------------------------------------------
309 : mblume 1615 Name: Matthias Blume (blume (at) tti - c (dot) org)
310 : mblume 1617 Date: 2004/08/31 17:15:00 CDT
311 :     Tag: blume-20040831-core
312 :     Description:
313 :    
314 :     Added some exports to src/compiler/core.cm upon request by J. Joemann.
315 :    
316 :     ----------------------------------------------------------------------
317 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
318 : mblume 1616 Date: 2004/08/30 17:55:00 CDT
319 :     Tag: blume-20040830-installer
320 :     Description:
321 :    
322 :     Upon request by Johannes Joemann:
323 :    
324 :     - improved ML code of installer to fall back to coping when renaming
325 :     fails (i.e., when source and target are on different file systems);
326 :     the code compiles but has yet to be tested in anger
327 :     - removed mlrisc from list of default targets (config/targets)
328 :    
329 :     ----------------------------------------------------------------------
330 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
331 : mblume 1615 Date: 2004/08/27 17:20:00 CDT
332 :     Tag: blume-20040827-ptreql
333 :     Description:
334 :    
335 :     Added ptreql primop to structure InlineT (upon request from Larry
336 :     Paulson).
337 :    
338 :     ----------------------------------------------------------------------
339 : allenleung 1613 Name: Allen Leung (leunga (at) reservoir (dot) com)
340 : allenleung 1614 Date: 2004/08/15 21:21:00 EST
341 :     Tag: leunga-110_48-udgraph
342 :     Description:
343 :    
344 :     Another bug fix from Carl Hauser:
345 :    
346 :     diff /net/niflab/smlnj48/src/MLRISC/graphs/udgraph.sml udgraph.sml
347 :     > 48c48
348 :     > < | rmv((e as (k,_))::es,L) = rmv(es,if k = i then es else
349 :     > e::L)
350 :     > ---
351 :     > > | rmv((e as (k,_))::es,L) = rmv(es,if k = i then L else e::L)
352 :     > Without this, any deletion of an edge in an undirected graph does severe
353 :     > violence to the graph.
354 :    
355 :     ----------------------------------------------------------------------
356 :     Name: Allen Leung (leunga (at) reservoir (dot) com)
357 : allenleung 1613 Date: 2004/08/10 23:35:00 EST
358 :     Tag: leunga-110_48-ppc
359 :     Description:
360 :    
361 :     The IBM/MacOS syntax switch on PPC was incorrectly swapped.
362 :    
363 :     ----------------------------------------------------------------------
364 : mblume 1602 Name: Matthias Blume (blume (at) tti - c (dot) org)
365 : mblume 1615 Date: 2004/08/10 12:00:00 CDT
366 : mblume 1602 Tag: Release_110_48
367 :     Description:
368 :    
369 :     New working version (110.48). NEW BOOTFILES!
370 :    
371 :     ----------------------------------------------------------------------
372 : allenleung 1599 Name: Allen Leung (leunga (at) reservoir (dot) com)
373 :     Date: 2004/08/09 12:21:00 EST
374 :     Tag: leunga-110_47-dijsktra
375 :     Description:
376 :    
377 :     Bug fix from Carl Hauser:
378 :    
379 :     single_source_shortest_paths in dijkstra.sml was observed to get wrong
380 :     answers (by comparing to single_source_shortest_paths in bellman-ford.sml).
381 :    
382 :     The problem is that following the expression A.update(dist,s,Num.zero)
383 :     it is necessary to update the priority queue using Q.decreaseWeight(Q,s).
384 :    
385 :     ----------------------------------------------------------------------
386 : mblume 1595 Name: Matthias Blume (blume (at) tti - c (dot) org)
387 :     Date: 2004/08/06 18:10:00 CDT
388 :     Tag: blume-20040806-cmdline
389 :     Description:
390 :    
391 :     Fiddled with handling of command-line options:
392 :    
393 :     * sml now quits after processing the command line
394 :     if -H, -S, -h<n>, or -s<n> appears as the last
395 :     command-line argument
396 :     * a new option -q terminates the session when encountered on
397 :     the command line; subsequent arguments will be ignored
398 :     * bug fixes: short (erroneous) arguments are no longer ignored
399 :     completely
400 :    
401 :     ----------------------------------------------------------------------
402 : allenleung 1593 Name: Allen Leung (leunga (at) reservoir (dot) com)
403 :     Date: 2004/08/04 18:17:00 EST
404 :     Tag: leunga-110_47-ppc-ibm-asm
405 :     Description:
406 :    
407 :     - Added minimal IBM assembly syntax support for PowerPC.
408 :    
409 :     - Cygwin: manually changed the file cygwin.def. Some exported symbols have
410 :     been altered in the runtime. We need an automatic way to keep the file
411 :     in sync.
412 :    
413 :     ----------------------------------------------------------------------
414 : mblume 1548 Name: Matthias Blume (blume (at) tti - c (dot) org)
415 : mblume 1581 Date: 2004/08/04 14:00:00 CDT
416 :     Tag: Release_110_47
417 :     Description:
418 :    
419 :     New working version (110.47). NEW BOOTFILES!
420 :    
421 :     ----------------------------------------------------------------------
422 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
423 : mblume 1580 Date: 2004/08/03 14:25:00 CDT
424 :     Tag: blume-20040803-callingconv
425 :     Description:
426 :    
427 :     Added low-level support for choosing C calling conventions by
428 :     twiddling the type of rawccall. (See
429 :     src/compiler/Semant/types/cproto.sml for details.)
430 :    
431 :     ----------------------------------------------------------------------
432 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
433 : mblume 1579 Date: 2004/08/02 15:55:00 CDT
434 :     Tag: blume-20040802-backout
435 :     Description:
436 :    
437 :     Backed out of change to win32-filesys.c. The earlier patch to
438 :     get_file_time caused CM to produce files with the wrong time stamp.
439 :    
440 :     ----------------------------------------------------------------------
441 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
442 : mblume 1573 Date: 2004/08/02 14:45:00 CDT
443 :     Tag: blume-20040802-nlffi-win32
444 :     Description:
445 :    
446 :     Added NLFFI support for Win32, adapted from a patch provided by David
447 :     Hansel. This is currently completely untested. Also, the issue
448 :     concerning stdcall vs. ccall is still unresolved.
449 :    
450 :     ----------------------------------------------------------------------
451 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
452 : mblume 1572 Date: 2004/07/30 17:55:00 CDT
453 :     Tag: blume-20040730-various
454 :     Description:
455 :    
456 :     Gearing up towards 110.47...
457 :    
458 :     - various minor bugfixes to ml-nlffigen
459 :     - a beginning of a manual for nlffi
460 :    
461 :     - eliminated 'export name=value' in config/install.sh as this does
462 :     not work with certain versions of /bin/sh
463 :     (Thanks to David King at Motorola for catching this.)
464 :    
465 :     - several bugfixes provided or suggested by David Hansel at Reactive Systems:
466 :     - added a test for tm==NULL to gmtime.c and localtime.c
467 :     - applied patch for incorrect GetFileTime under win32
468 :     - toSeconds -> toMilliseconds in Win32/win32-process.sml
469 :    
470 :     ----------------------------------------------------------------------
471 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
472 : mblume 1557 Date: 2004/07/21 18:20:00 CDT
473 :     Tag: blume-20040721-nlffigen
474 :     Description:
475 :    
476 :     - Fixed minor issue in ml-nlffigen: Now generate structure T_foo
477 :     for a typedef to an incomplete type, but leave out the "typ" member.
478 :     (This is just for consistency.)
479 :     - Started to produce what is supposed to become better (i.e., comprehensive)
480 :     documentation of what ml-nlffigen does and produces.
481 :    
482 :     ----------------------------------------------------------------------
483 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
484 : mblume 1548 Date: 2004/07/14 16:25:00 CDT
485 :     Tag: blume-20040714-union
486 :     Description:
487 :    
488 :     Added C_UNION to c-calls/c-types.sml and updated the machinery
489 :     (ml-nlffigen, cproto.sml) that conveys C function interface
490 :     information to the code generator.
491 :    
492 :     However, the actual architecture-specific implementation of function
493 :     arguments and results that are C unions is still not implemented.
494 :    
495 :     ----------------------------------------------------------------------
496 : allenleung 1539 Name: Allen Leung (leunga (at) reservoir (dot) com)
497 : allenleung 1546 Date: 2004/07/14 14:38:00 EST
498 :     Tag: leunga-110_46_1-ppc-lwzu
499 :     Description:
500 :    
501 :     Added these instructions to the PowerPC architecture:
502 :     LBZU(X), LHZU(X), LWZU(X),
503 :     STWU(X), STFDU, STFSU
504 :    
505 :     etc...
506 :    
507 :     Note: I haven't added their instruction encoding into the description.
508 :    
509 :     ----------------------------------------------------------------------
510 :     Name: Allen Leung (leunga (at) reservoir (dot) com)
511 : allenleung 1539 Date: 2004/07/13 15:04:00 EST
512 :     Tag: leunga-110_46_1-ppc-lwarx
513 :     Description:
514 :    
515 :     Added the two instructions LWARX and STWCX to the PowerPC
516 :     instruction set.
517 :    
518 :     A (untested) rewrite of loop-structure.sml. The old version
519 :     is completely broken.
520 :    
521 :     ----------------------------------------------------------------------
522 : mblume 1479 Name: Matthias Blume (blume (at) tti - c (dot) org)
523 : mblume 1537 Date: 2004/07/13 13:50:00 CDT
524 :     Tag: blume-20040713-nlffi
525 :     Description:
526 :    
527 :     - use paramAlloc to report c-calls with too many arguments
528 :     (for PPC version where parameter area is pre-allocated)
529 :     - added ccall_maxargspace to machspec (to implement the above)
530 :     - made "make" commend in CM's "make" tool configurable
531 :     - added option (default: on) for passing the name of the SML/NJ's "bin"
532 :     directory to "make"; the call looks like this:
533 :    
534 :     make <options> SMLNJ_BINDIR=<dir> <target>
535 :    
536 :     This can be used by the Makefile to, e.g., pick the "right" version
537 :     of ml-nlffigen.
538 :     - minor code tweaks
539 :    
540 :     ----------------------------------------------------------------------
541 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
542 : mblume 1532 Date: 2004/07/12 22:50:00 CDT
543 :     Tag: blume-110_46_1-macosx-nlffi
544 :     Description:
545 :    
546 :     NLFFI under Mac OS X now working (sort of). This is largely untested,
547 :     though.
548 :    
549 :     Note: 1. You have to make a new, clean build of the runtime system.
550 :     2. There are new BOOTFILES, you have to use them!
551 :     (Doing the bootstrap process yourself would be *very* painful!
552 :     If you absolutely have to do it, build the system under
553 :     a different architecture and then cross-compile.)
554 :    
555 :     Version bumped to 110.46.1 to account for runtime data format changes.
556 :    
557 :     ----------------------------------------------------------------------
558 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
559 : mblume 1515 Date: 2004/06/18 14:30:00 CDT
560 :     Tag: blume-20040618-unix
561 :     Description:
562 :    
563 :     Changed the implementation of structure Unix so that the same stream
564 :     is returned every time one of the {text,bin}{In,Out}streamOf functions
565 :     is invoked on the same proc. This is not what the spec currently
566 :     says -- although IMO it arguably should. (See discussion below.)
567 :    
568 :     ----------------------------------------------------------------------
569 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
570 : mblume 1507 Date: 2004/06/17 18:15:00 CDT
571 :     Tag: Release_110_46
572 :     Description:
573 :    
574 :     New working version (110.46). NEW BOOTFILES!
575 :    
576 :     ----------------------------------------------------------------------
577 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
578 : mblume 1506 Date: 2004/06/17 17:20:00 CDT
579 :     Tag: blume-20040617-timer-unix
580 :     Description:
581 :    
582 :     Changed the interface of structures Timer and Unix to match the most
583 :     recent Basis spec.
584 :    
585 :     In the case of Unix there still seems to be an open/weird issue:
586 :    
587 :     The {text,bin}{In,Out}streamOf functions are supposed to create
588 :     fresh streams whenever they are called -- as opposed to have them
589 :     return the same stream every time. This design is supposed to
590 :     prevent space leaks caused by proc values hanging on to streams.
591 :    
592 :     The reap function, on the other hand, is supposed to close the
593 :     streams. This cannot be done without having a handle on the
594 :     stream in proc after all...
595 :    
596 :     I took the liberty to implement the following stopgap solution:
597 :    
598 :     The proc value hangs on to the most recently created stream(s).
599 :     Reap closes those. If either or both of the two streams hadn't
600 :     been created at all yet, then reap will close the corresponding
601 :     file descriptors directly.
602 :    
603 :     PS: I don't understand the original space leak argument anymore. If
604 :     a proc hangs on to the imperative stream, then I/O operations on those
605 :     will advance the state of the cached stream and avoid the space leak.
606 :    
607 :     ----------------------------------------------------------------------
608 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
609 : mblume 1500 Date: 2004/05/28 16:45:00 CDT
610 :     Tag: blume-20040528-basis
611 :     Description:
612 :    
613 :     Added signature PACK_REAL and exported functor PrimIO.
614 :    
615 :     ----------------------------------------------------------------------
616 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
617 : mblume 1498 Date: 2004/05/25 16:00:00 CDT
618 :     Tag: blume-20040525-group-owner
619 :     Description:
620 :    
621 :     CM now ignores (but still accepts) the "owner" information in group
622 :     descriptions. The owner of a group is its next enclosing
623 :     library. Each group must have a unique owner. (There is a virtual
624 : mblume 1505 "toplevel" library that owns groups which are not nested within a real
625 : mblume 1498 library.) Previously, each group had to explicitly declare its owner,
626 :     and CM would check that such a declaration is correct. The new scheme
627 :     is to have CM check that for each group there is precisely one owning
628 :     library.
629 :    
630 :     The advantage of the new scheme is that the programmer no longer needs
631 :     to maintain the somewhat annoying owner information. The downside is
632 :     that CM cannot enforce the ownership rule across multiple runs of
633 :     CM.make. Fortunately, enclosing the same group in two different
634 :     libraries A and B which are not part of the same program does not
635 :     cause real problems.
636 :    
637 :     ----------------------------------------------------------------------
638 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
639 : mblume 1496 Date: 2004/05/20 16:00:00 CDT
640 :     Tag: blume-20040520-win32
641 :     Description:
642 :    
643 :     Made the win32 version work again. (Strangely, a misplaced comma had
644 :     slipped into win32-process.c which prevented the runtime from being
645 :     compiled correctly.)
646 :    
647 :     Also, included a minor addition to ml-build.bat analogous to what was
648 :     done in blume-20040519-ml-build.
649 :    
650 :     ----------------------------------------------------------------------
651 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
652 : mblume 1494 Date: 2004/05/19 22:10:00 CDT
653 :     Tag: blume-20040519-ml-build
654 :     Description:
655 :    
656 :     Arranged for ml-build to clean up after itself a little bit better.
657 :     The script generates a temporary SML source file and compiles it using
658 :     CM, so CM generates metadata (GUID, SKEL, objectfile) for it. It now
659 :     gets rid of those at the end, so they don't accumulate under .cm.
660 :    
661 :     This required a minor change to install.sh because the name of the
662 :     metadata directory (default: .cm) is actually configurable at
663 :     installation time.
664 :    
665 :     ----------------------------------------------------------------------
666 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
667 : mblume 1491 Date: 2004/05/18 15:50:00 CDT
668 :     Tag: blume-20040518-mkreader
669 :     Description:
670 :    
671 :     Added Posix.IO.mk{Bin,Text}{Reader,Writer} by lifting their respective
672 :     implementations from internal modules PosixBinPrimIO and PosixTextPrimIO.
673 :    
674 :     ----------------------------------------------------------------------
675 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
676 : mblume 1489 Date: 2004/05/11 14:35:00 CDT
677 :     Tag: blume-20040511-win32sock
678 :     Description:
679 :    
680 :     Added previously missing support for many socket-related functions
681 :     under win32. Thanks to David Hansel <hansel@reactive-systems.com>
682 :     for the voluminous patch!
683 :    
684 :     (I have not tested this patch under win32 yet.)
685 :    
686 : mblume 1490 Here is David's e-mail:
687 :    
688 :     Hi,
689 :    
690 :     Attached to this email you find a diff against sml/nj 110.45
691 :     that will enable socket support under Windows.
692 :    
693 :     To apply the patch (using unix or cygwin)
694 :     1) gunzip runtime.diff.gz
695 :     2) "cd" into "src/runtime" in the source tree of a fresh
696 :     110.45 installation.
697 :     3) patch -p 1 < [your/path/to]runtime.diff
698 :    
699 :     The code compiles fine but has NOT yet been extensively tested.
700 :     I only ran a few tests for basic socket client functionality
701 :     (which worked fine). Especially the functions that use ioctl
702 :     are not tested at all and might not work (see below).
703 :    
704 :     I implemented this since we want to move to a newer version of sml/nj
705 :     but need socket support in order to use it. This is the first time I
706 :     even had a look at the sml/nj source, so please review my changes
707 :     before making this part of the distribution! Here are a few issues
708 :     that I think might be better for someone to solve who is more
709 :     familiar with the sml/nj source (and socket programming):
710 :    
711 :     - getnetbyaddr.c and getnetbyname.c will raise a "not implemented"
712 :     exception since I could not figure out what the windows equivalent
713 :     of these functions is
714 :    
715 :     - In sockets-osdep.h there are a some #include statements that are
716 :     only used in a few files that include sockets-osdep.h
717 :    
718 :     - In smlnj-sock-lib.c, function init_fn() calls WSAStartup() but
719 :     does not process its return value since I don't know how to
720 :     report an error upwards.
721 :    
722 :     - It would probably be good to have a call to WSACleanup() when
723 :     the library is unloaded (if there is such a possibility).
724 :     Otherwise I think Windows will take care of this automatically
725 :     when the process finishes.
726 :    
727 :     - I used ioctlsocket() as a replacement for ioctl() but I have
728 :     no idea if that is actually the proper replacement on Windows.
729 :    
730 :     - All these issues are marked in the code by "FIXME" comments.
731 :    
732 :     We use sml/nj extensively in our products and are quite happy
733 :     with it. I hope this contribution will help you.
734 :    
735 :     Keep up the good work!
736 :    
737 :     David
738 :    
739 : mblume 1489 ----------------------------------------------------------------------
740 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
741 : mblume 1488 Date: 2004/05/11 14:20:00 CDT
742 :     Tag: blume-20040511-installml
743 :     Description:
744 :    
745 :     Fixed two bugs in installml script. (Thanks to Vesa A. Norrman for
746 :     the patch.)
747 :    
748 :     ----------------------------------------------------------------------
749 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
750 : mblume 1487 Date: 2004/05/11 14:05:00 CDT
751 :     Tag: blume-20040511-nlffi-netbsd
752 :     Description:
753 :    
754 :     Added support for nlffi under netbsd. (Thanks to Vesa A. Norrman for
755 :     the patch.)
756 :    
757 :     ----------------------------------------------------------------------
758 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
759 : mblume 1486 Date: 2004/05/11 12:05:00 CDT
760 :     Tag: blume-20040511-exports
761 :     Description:
762 :    
763 :     As per request by Adam Chlipala <adam@hcoop.net>, extended various
764 :     export lists in compiler-related .cm-files.
765 :    
766 :     ----------------------------------------------------------------------
767 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
768 : mblume 1485 Date: 2004/05/11 11:35:00 CDT
769 :     Tag: blume-20040511-allsource
770 :     Description:
771 :    
772 :     The installer now honors the "src-smlnj" target again, although its meaning
773 :     has changed from "all sources required for the compiler" to "all sources
774 :     the installer knows about". In other words, if you enable "src-smlnj"
775 :     in the "targets" file, then the installer will pull in sources for
776 :     everything. (Notice that this refers to source code only. Compiled
777 :     code is still only installed for modules that were requested explicitly
778 :     or which are required for other modules that were requested explicitly.)
779 :    
780 :     ----------------------------------------------------------------------
781 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
782 : mblume 1482 Date: 2004/04/23 17:40:00 CDT
783 :     Tag: blume-20040423-ieee-scan
784 :     Description:
785 :    
786 :     Fixed IEEEReal.scan (and .fromString) so that if there is an overflow
787 :     in the exponent calculation we get INF or ZERO (depending on the mantissa
788 :     and the sign of the exponent).
789 :    
790 :     ----------------------------------------------------------------------
791 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
792 : mblume 1480 Date: 2004/04/23 10:40:00 CDT
793 :     Tag: blume-20040423-ml-build
794 :     Description:
795 :    
796 :     The ml-build script now terminates with a non-0 status when something
797 :     goes wrong.
798 :    
799 :     ----------------------------------------------------------------------
800 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
801 : mblume 1479 Date: 2004/04/22 16:35:00 CDT
802 :     Tag: blume-20040422-Option
803 :     Description:
804 :    
805 :     Made exception Option to be the same as exception Option.Option
806 :     (as it should be).
807 :    
808 :     ----------------------------------------------------------------------
809 : allenleung 1468 Name: Allen Leung (leunga (at) reservoir (dot) com)
810 :     Date: 2004/03/19 14:40:00 EST
811 :     Tag: leunga-20040319-cygwin-nlffi
812 :     Description:
813 :    
814 :     Fixed the runtime so that ml-nlffi-lib runs on the cygwin version
815 :     of SML/NJ. The problem is that
816 :    
817 :     lib = dlopen(NULL, ...)
818 :     f = dlsym(lib, "malloc");
819 :    
820 :     does not work on Windows unless we explicitly export symbols
821 :     such as 'malloc' during linking. We fixed this by explicitly
822 :     exporting the required symbols with the magic gcc incantation:
823 :    
824 :     -Wl,--export-all cygwin.def
825 :    
826 :     where cygwin.def is a file containing all the symbols that
827 :     we wish to export.
828 :    
829 :     I suspect this is a Windows problem and we'll have to
830 :     do the same (somehow with windows compilers) when
831 :     we build the native win32 version with the system
832 :     calls LoadLibrary/GetProcAddress.
833 :    
834 :     ----------------------------------------------------------------------
835 : mblume 1384 Name: Matthias Blume (blume (at) tti - c (dot) org)
836 : mblume 1465 Date: 2004/03/04 16:35:00 CST
837 :     Tag: blume-20040304-intinf-fmt
838 :     Description:
839 :    
840 :     Fixed problem with IntInf.fmt (sign would show up on the right instead
841 :     of on the left for BIN, OCT, and HEX).
842 :    
843 :     ----------------------------------------------------------------------
844 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
845 : mblume 1464 Date: 2004/03/04 11:25:00 CST
846 :     Tag: blume-20040304-symlinks
847 :     Description:
848 :    
849 :     Fixed problem with installer script (unix only) where bin/ml-yacc and
850 :     friends pointed (via symlinks) to absolute locations instead of just
851 :     .run-sml. This was reported by Vesa A Norrman.
852 :    
853 :     ----------------------------------------------------------------------
854 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
855 : mblume 1455 Date: 2004/02/13 14:50:00 CST
856 :     Tag: Release_110_45
857 :     Description:
858 :    
859 :     New working version (110.45). New bootfiles.
860 :    
861 :     ----------------------------------------------------------------------
862 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
863 : mblume 1448 Date: 2004/01/26 15:15:15 CST
864 :     Tag: blume-20040126-toplevel
865 :     Description:
866 :    
867 :     Improved handling of exceptions at the interactive toplevel.
868 :    
869 :     ----------------------------------------------------------------------
870 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
871 :     Date: 2004/01/26 11:25:00
872 : mblume 1446 Tag: blume-20040126-app
873 :     Description:
874 :    
875 :     Type of top-level "app" corrected.
876 : mblume 1447 Added code for setting vp_limitPtrMask to Win32-specific runtime.
877 : mblume 1446
878 :     ----------------------------------------------------------------------
879 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
880 : mblume 1427 Date: 2003/11/18 17:10 CST
881 :     Tag: blume-20031118-basis-fiddle
882 :     Description:
883 :    
884 :     - changed Timer interface to what might become the spec
885 :     - POSIX_FLAGS -> BIT_FLAGS according to spec
886 :     - some other minor discrepancies wrt. spec eliminated
887 :    
888 :     ----------------------------------------------------------------------
889 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
890 : mblume 1414 Date: 2003/11/06 12:00:00 CST
891 :     Tag: Release_110_44
892 :     Description:
893 :    
894 :     New working version (110.44). New bootfiles.
895 :    
896 :     ----------------------------------------------------------------------
897 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
898 : mblume 1411 Date: 2003/11/04 11:50:00 CST
899 :     Tag: blume-20031104-move-libraries
900 :     Description:
901 :    
902 :     Eliminated the "dont_move_libraries" directive in config/targets.
903 :     (The mechanism was broken and could not be fixed easily. Moreover,
904 :     there does not seem to be any reason not to move all libraries into
905 :     lib during installation. I originally implemented this directive as a
906 :     backward-compatibility feature when I first introduced the new CM.
907 :     Now that things have been stable for a long time and going back to the
908 :     old CM is not an option, there is no reason to keep it around.)
909 :    
910 :     ----------------------------------------------------------------------
911 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
912 : mblume 1409 Date: 2003/11/03 16:00:00 CST
913 :     Tag: blume-20031103-installdir
914 :     Description:
915 :    
916 :     Made installer honor INSTALLDIR variable again. (Thanks to Chris
917 :     Richards for pointing out the problem and providing the solution.)
918 :    
919 :     ----------------------------------------------------------------------
920 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
921 : mblume 1400 Date: 2003/10/01 17:05:00 CDT
922 :     Tag: blume-20031001-lal-mlrisc
923 :     Description:
924 :    
925 :     MLRISC bug fix from Lal.
926 :    
927 :     ----------------------------------------------------------------------
928 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
929 : mblume 1399 Date: 2003/09/30 16:10:00 CDT
930 :     Tag: blume-20030930-primio-bat
931 : mblume 1398 Description:
932 :    
933 :     1. Added openVector, nullRd, and nullWr to PRIM_IO.
934 :     2. Improved .bat files (for Win32 port) to make things work under Win95.
935 :     (thanks to Aaron S. Hawley for this one)
936 :    
937 :     ----------------------------------------------------------------------
938 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
939 : mblume 1395 Date: 2003/09/26 16:05:00 CDT
940 :     Tag: blume-20030926-wrappriv
941 :     Description:
942 :    
943 :     Added missing wrapper for privilege "primitive" in $smlnj/viscomp/core.cm.
944 :    
945 :     ----------------------------------------------------------------------
946 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
947 : mblume 1394 Date: 2003/09/26 15:00:00 CDT
948 :     Tag: blume-20030926-110_43_3
949 :     Description:
950 :    
951 :     - additional cleanup
952 :     - version number bump, NEW BOOTFILES
953 :    
954 :     ----------------------------------------------------------------------
955 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
956 : mblume 1393 Date: 2003/09/26 12:00:00 CDT
957 :     Tag: blume-20030926-ppautoload
958 :     Description:
959 :    
960 :     I modified the read-eval-print loop so that the autoloader gets
961 :     invoked whenever the prettyprinter tries to look up a symbol that
962 :     is not currently defined in the toplevel environment but which
963 :     appears in CM's autoload registry. As a result, we see far fewer of
964 :     those ?.Foo.Bar.xxx names in the prettyprinter's output.
965 :    
966 :     In addition to this I tried to clean up some pieces of the Basis
967 :     implementation (e.g., Socket, Word8Array) in order to prevent other
968 :     instances of these ?.Foo.Bar.xxx names from being printed.
969 :    
970 :     The mechanism that picks names for types still needs some work, though.
971 :     (Right now it seems that if there is a type A.t which is defined to
972 : mblume 1394 be B.u, but B is unavailable at toplevel, then A.t gets printed as
973 :     "?.B.u" although the perhaps more sensible solution would be to use
974 : mblume 1393 "A.t" in this case. In other words, the prettyprinter should follow
975 :     a chain of DEFtycs not farther than there are corresponding toplevel
976 :     names in the current environment.)
977 :    
978 :     ----------------------------------------------------------------------
979 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
980 : mblume 1391 Date: 2003/09/24 16:31:00 CDT
981 :     Tag: blume-20030924-installer
982 :     Description:
983 :    
984 :     Another installer tweak: All the ML code for the installer is now
985 :     compiled during CMB.make and put into a little library called
986 :     $smlnj/installer.cm. The installation then simply invokes
987 :    
988 :     sml -m $smlnj/installer.cm
989 :    
990 :     and everything happens automagically.
991 :    
992 :     Win32: ML code senses value of environment variable SMLNJ_HOME.
993 :     Unix: ML code senses values of environment variables ROOT, CONFIGDIR,
994 :     and BINDIR.
995 :    
996 :     The new scheme guarantees that the ML code responsible for the installation
997 :     is in sync with the APIs of the main system. Also, the installer is
998 :     somewhat faster because the installer script is precompiled.
999 :    
1000 :     ----------------------------------------------------------------------
1001 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
1002 : mblume 1390 Date: 2003/09/24 15:35:00 CDT
1003 :     Tag: blume-20030924-synsock
1004 :     Description:
1005 :    
1006 :     Added a signature SYNCHRONOUS_SOCKET to basis.cm. This is like SOCKET
1007 :     but excludes all non-blocking operations. Defined SOCKET (in Basis)
1008 :     and CML_SOCKET in terms of SYNCHRONOUS_SOCKET. Removed superfluous
1009 :     implementations of non-blocking operations from CML's Socket
1010 :     structure.
1011 :    
1012 :     ----------------------------------------------------------------------
1013 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
1014 : mblume 1389 Date: 2003/09/24 15:10:05 CDT
1015 :     Tag: blume-20030924-sockets
1016 :     Description:
1017 :    
1018 :     1. Fixed SOCKET API and implementation to match Basis spec.
1019 :     This required changing the internal representation of sockets to one
1020 :     that remembers (for each socket file descriptor) whether it is currently
1021 :     blocking or non-blocking. This state is maintained lazily (i.e., a system
1022 :     call is made only if the state actually needs to change).
1023 :    
1024 :     2. OS-specific details of sockets were moved into separate files, thus
1025 :     making it possible to unify the bulk of the socket implementations
1026 :     between Unix and Win32.
1027 :    
1028 :     3. CML's socket API changed accordingly.
1029 :     (Note that we need to remove non-blocking functions from this API
1030 :     since they are redundant in the case of CML!)
1031 :    
1032 :     4. CML's socket implementation now makes use of non-blocking functions
1033 :     provided by Basis, thus removing all OS-dependent code from this part
1034 :     of CML.
1035 :    
1036 :     5. Changed Real64.precision from 52 to 53. Minor cleanup in Real64 code.
1037 :    
1038 :     ----------------------------------------------------------------------
1039 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
1040 : mblume 1387 Date: 2003/09/22 12:10:00 CDT
1041 : mblume 1388 Tag: blume-20030922-110_43_2
1042 : mblume 1387 Description:
1043 :    
1044 :     Made a new interim version and bootfiles for developer's bootstrapping
1045 :     convenience.
1046 :    
1047 :     110.43.2 -- NEW BOOTFILES
1048 :    
1049 :     ----------------------------------------------------------------------
1050 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
1051 : mblume 1385 Date: 2003/09/19 15:55:00 CDT
1052 :     Tag: blume-20030919-cmdir
1053 :     Description:
1054 :    
1055 :     1. new-install.sh -> install.sh
1056 :     2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
1057 :     3. tweaked installer so that another name instead of .cm can be chosen
1058 :     at install time (by setting the CM_DIR_ARC environment variable
1059 :     during installation); once installation is complete, the name is
1060 :     fixed
1061 :    
1062 :     ----------------------------------------------------------------------
1063 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
1064 : mblume 1384 Date: 2003/09/18 16:00:00 CDT
1065 :     Tag: blume-20030918-110_43_1
1066 :     Description:
1067 :    
1068 :     Made a new interim version and bootfiles for developer's bootstrapping
1069 :     convenience.
1070 :    
1071 :     110.43.1 -- NEW BOOTFILES
1072 :    
1073 :     ----------------------------------------------------------------------
1074 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
1075 : mblume 1383 Date: 2003/09/18 15:20:00 CDT
1076 :     Tag: blume-20030918-misc
1077 :     Description:
1078 :    
1079 :     1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
1080 :     this seems to be controversial at the moment) and used those in
1081 :     Posix.ProcEnv.times.
1082 :    
1083 :     2. Added Time.{from,to}Nanoseconds to Time.
1084 :    
1085 :     3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
1086 :     For example, fromLargeInt never needs to look at more than 3 "big
1087 :     digits" to get its 53 bits of precision.
1088 :    
1089 :     ----------------------------------------------------------------------
1090 : mblume 1384 Name: Matthias Blume (blume (at) tti - c (dot) org)
1091 : mblume 1381 Date: 2003/09/17 16:30:00 CDT
1092 :     Tag: blume-20030917-real32-slices
1093 :     Description:
1094 :    
1095 :     Added an entry to the primitive environment
1096 :     (compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
1097 :     added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
1098 :    
1099 :     Removed some of the "magic" constants in real64.sml and replaced them
1100 :     with code that generates these values from their corresponding
1101 :     integer counterparts.
1102 :    
1103 :     Made all(?) the slice-related changes to the Basis and made everything
1104 :     compile again...
1105 :    
1106 :     ----------------------------------------------------------------------
1107 : mblume 1368 Name: Matthias Blume (blume (at) tti - c (dot) org)
1108 : mblume 1375 Date: 2003/09/15 17:45:00 CDT
1109 :     Tag: blume-20030915-rbase
1110 :     Description:
1111 :    
1112 :     Fixed bug in Real.fromLargeInt.
1113 :    
1114 :     ----------------------------------------------------------------------
1115 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
1116 : mblume 1369 Date: 2003/09/13 18:11:00 CDT
1117 :     Tag: blume-20030913-libinstall
1118 :     Description:
1119 :    
1120 :     Minor bugfix in config/libinstall (set anchor with path to
1121 :     standalone tool after installing it, otherwise libraries that
1122 :     need ml-lex or ml-yacc won't compile the first time the installer
1123 :     runs).
1124 :    
1125 :     ----------------------------------------------------------------------
1126 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
1127 : mblume 1368 Date: 2003/09/12 11:45:00 CDT
1128 :     Tag: blume-20030912-various
1129 :     Description:
1130 :    
1131 :     - fixed bug in Real.toLargeInt
1132 :     - fixed bug in Posix.ProcEnv.times
1133 :     - changed inputLine functions to return an option
1134 :     - minor installer improvements / bugfixes
1135 :     - changed default @SMLalloc parameter for x86/celeron to 64k
1136 :    
1137 :     ----------------------------------------------------------------------
1138 :     Name: Matthias Blume (blume (at) tti - c (dot) org)
1139 : mblume 1356 Date: 2003/09/09 22:00:00 CDT
1140 :     Tag: Release_110_43
1141 :     Description:
1142 :    
1143 :     New working release 110.43. New bootfiles.
1144 :    
1145 :     ----------------------------------------------------------------------
1146 : mblume 1368 Name: Matthias Blume (blume (at) tti - c (dot) org)
1147 : mblume 1354 Date: 2003/09/09 19:20:00 CDT
1148 :     Tag: blume-20030909-installer
1149 :     Description:
1150 :    
1151 :     Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
1152 :     Modified config/install.bat to take advantage of it. Also modified
1153 :     config/install.sh (and called it config/new-install.sh) to take advantage
1154 :     of it on Unix systems. (The SML code is (supposed to be) platform-
1155 :     independent.)
1156 :    
1157 :     The installer can now install everything under Win32
1158 :     as well as under *nix as long as it compiles.
1159 :    
1160 :     Other changes:
1161 :    
1162 :     - made CML compile again under Win32
1163 :     - made eXene compile under Win32 (by providing a fake structure UnixSock
1164 :     and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
1165 :     - fixed a bug in nowhere: it assumed that type OS.Process.status is the
1166 :     same as type int; under Win32 it isn't
1167 :     - fixed some slice-related problems in the win32-specific parts of CML
1168 :     - added a functor argument "sameVol" to os-path-fn.sml in the Basis
1169 :     (under Win32, the volume name is case-insensitive, and the
1170 :     OS.Path code compares volume names for equality)
1171 :    
1172 :     ----------------------------------------------------------------------
1173 : mblume 1368 Name: Matthias Blume (blume (at) tti - c (dot) org)
1174 : mblume 1353 Date: 2003/09/08 11:55:00 CDT
1175 :     Tag: blume-20030908-fullpath
1176 :     Description:
1177 :    
1178 :     Made Win32 version of OS.FileSys.fullPath return current directory
1179 :     when given an empty string. This is what the spec says, and incidentally,
1180 :     CM depends on it. (CM otherwise goes into an infinite loop in certain
1181 :     cases when presented with the name of a non-existing .cm file.)
1182 :    
1183 :     ----------------------------------------------------------------------
1184 : mblume 1368 Name: Matthias Blume (blume (at) tti - c (dot) org)
1185 : mblume 1350 Date: 2003/09/04 16:30:00 CDT
1186 : mblume 1352 Tag: blume-20030905-slices-etc
1187 : mblume 1350 Description:
1188 :    
1189 :     1. Changed interface to vectors and arrays in Basis to match
1190 :     (draft) Basis spec.
1191 :     2. Added signatures and implementations of slices according to
1192 :     Basis spec.
1193 :     3. Edited source code throughout the system to make it compile again
1194 :     under 1. and 2. (In some cases code had to be added to have it
1195 :     match the new signatures.)
1196 :     4. MLRISC should be backward-compatible: the copies of the originals
1197 :     of files that needed to change under 3. were retained, the .cm files
1198 :     check the compiler version number and use old versions when
1199 :     appropriate.
1200 :     5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
1201 :     dirstream -> string option (in accordance with Basis spec).
1202 :     6. When generating code that counts lines, ml-lex used function
1203 :     CharVector.foldli, taking advantage of its old interface.
1204 :     This has been replaced with the corresponding code from
1205 :     CharVectorSlice. (html-lex must be re-lexed!)
1206 :     7. BitArray in smlnj-lib/Util has been extended/modified to match the
1207 :     new MONO_ARRAY signature. (Do we need BitArraySlice?)
1208 :     8. Removed temporary additions (fromInternal, toInternal) from the
1209 :     (now obsolete) IntInf in smlnj-lib/Util.
1210 :     9. Cleaned up structure Byte.
1211 :     10. Added localOffset, scan, and fromString to Date (according to spec).
1212 :     Cleaned/corrected implementation of Date.
1213 :     (Still need to check for correctness; implement better canonicalizeDate.)
1214 :     11. Added "scan" to signature IEEE_REAL.
1215 :     12. Some improvements to IntInf [in particular: efficiency-hack for
1216 :     mod and rem when second operand is 2 (for parity checks).]
1217 :     13. Changed representation of type Time.time, using a single IntInf.int
1218 :     value counting microseconds. This considerably simplified the
1219 :     implementation of structure Time. We now support negative time
1220 :     values; scan and fromString handle signs.
1221 :     14. Functor PrimIO now takes two additional arguments (VectorSlice and
1222 :     ArraySlice).
1223 :    
1224 :     ----------------------------------------------------------------------
1225 : mblume 1368 Name: Matthias Blume (blume (at) tti - c (dot) org)
1226 : mblume 1347 Date: 2003/08/28 17:00:00 CDT
1227 :     Tag: blume-20030828-intinf
1228 :     Description:
1229 :    
1230 :     This is a major update which comes with a version number bump
1231 :     (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
1232 :     and an implementation of IntInf in the Basis.
1233 :    
1234 :     There are a fairly large number of related changes and updates throughout
1235 :     the system:
1236 :    
1237 :     Basis:
1238 :     - Implemented IntInf.
1239 :     - Made LargeInt a projection of IntInf (by filtering through INTEGER).
1240 : mblume 1348 - Added some missing Real64 operations, most notably Real.toLargeInt.
1241 : mblume 1347 - Added FixedInt as a synonym for Int32.
1242 :    
1243 :     compiler:
1244 :     * Added support for a built-in intinf type.
1245 :     - literals
1246 :     - pattern matching
1247 :     - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
1248 :     - overloading on literals and operations
1249 :    
1250 :     This required adding a primitive type intinf, some additional
1251 :     primops, and implementations for several non-trivial intinf
1252 :     operations in Core. (The intinf type is completely abstract
1253 :     to the compiler; all operations get delegated back to the Core.)
1254 :    
1255 :     * Intinf equality is handled by polyequal. However, the compiler
1256 :     does not print its usual warning in this case (since polyequal
1257 :     is the right thing to do there).
1258 :    
1259 :     * Improved the organization of structure InlineT.
1260 :    
1261 :     * A word about conversion primops:
1262 :     If conversions involving intinf do not cancel out during
1263 : mblume 1348 CPS contract, then the compiler must insert calls to Core functions.
1264 : mblume 1347 Since all core access must be resolved already during the FLINT
1265 :     translate phase, it would be too late a the time of CPS contract
1266 : mblume 1348 to add new Core calls. For this reason, conversion primops
1267 : mblume 1347 for intinf carry two arguments: 1. the numeric argument that
1268 :     they are supposed to convert, and 2. the Core function that
1269 :     can help with this conversion if necessary. If CPS contract
1270 : mblume 1348 eliminates a primop, then the associated Core function becomes
1271 :     dead and goes away. Intinf conversion primops that do not get
1272 :     eliminated by CPS contract get rewritten into calls of their
1273 :     core functions by a separate, new phase.
1274 : mblume 1347
1275 :     interactive system:
1276 :     - Control.Print.intinfDepth controls max length of intinf constants
1277 :     being printed. (Analogous to Control.Print.stringDepth.)
1278 :     - Cleanup in printutil and pputil: got rid of unused stuff and
1279 :     duplicates; replaced some of the code with code that makes better
1280 :     use of library functionality.
1281 :    
1282 :     CM:
1283 :     Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
1284 :     are no longer silent.
1285 :    
1286 :     CKIT:
1287 :     Fixed mismatched uses of Int32 and LargeInt. I always decided
1288 :     in favor of LargeInt -- which is now the same as IntInf.
1289 :     CKIT-knowledgable people should check whether this is what's
1290 :     intended and otherwise change things back to using Int32 or
1291 :     FixedInt.
1292 :    
1293 :     Throughout the code:
1294 : mblume 1348 Started using IntInf.int literals and built-in operations
1295 : mblume 1347 (e.g., comparison with 0) where this seems appropriate.
1296 :    
1297 :    
1298 :     ----------------------------------------------------------------------
1299 : macqueen 1343 Name: Dave MacQueen (dbm@cs.uchicago.edu)
1300 :     Date: 2003/08/13 11:36:00 CDT
1301 :     Tag: dbm-20030813-mcz-merge1
1302 :     Description:
1303 :    
1304 :     Merging changes from the mcz-branch development branch into trunk.
1305 :     These changes involve replacement of the emulated old prettyprinter
1306 :     interface with direct use of the SML/NJ Lib PP library, and fixing
1307 :     of a couple of bugs (895, 1186) relating to error messages. A new
1308 :     prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
1309 :     has been added.
1310 :    
1311 :     ----------------------------------------------------------------------
1312 : mblume 1368 Name: Matthias Blume (blume (at) tti - c (dot) org)
1313 : mblume 1342 Date: 2003/08/11 15:45:00 CDT
1314 :     Tag: blume-20030811-windows
1315 :     Description:
1316 :    
1317 :     Version number bumped to 110.42.9. NEW BOOTFILES!!!
1318 :    
1319 :     http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
1320 :    
1321 :     This patch restores SML/NJ's ability to run under win32. There are a
1322 :     number of changes, including fixes for several bugs that had gone
1323 :     unnoticed until now:
1324 :    
1325 :     - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
1326 :     trying to run the win32 version from within cygwin.)
1327 :    
1328 :     - There are a number of simple .bat scripts that substitute for their
1329 :     corresponding Unix shell-scripts. (See below.)
1330 :    
1331 :     - The internals of ml-build have been modified slightly. The main
1332 :     difference is that instead of calling ".link-sml" (or link-sml.bat)
1333 :     using OS.Process.system, the ML process delegates this task back
1334 :     to the script. Otherwise problems arise in mixed environments such
1335 :     as Cygwin where scripts look and work like Unix scripts, but
1336 :     where OS.Process.system cannot run them.
1337 :    
1338 :     - In CM, the srcpath pickler used native pathname syntax -- which
1339 :     is incorrect in the case of cross-compilation. The new pickle format
1340 :     is independent of platform-specific naming conventions.
1341 :    
1342 :     - Path configuration files (such as lib/pathconfig) can now choose
1343 :     between native and standard syntax. Placing a line of the form
1344 :    
1345 :     standard!
1346 :    
1347 :     into the file causes all subsequent paths to be interpreted using
1348 :     CM standard pathname syntax (= Unix conventions); a line
1349 :    
1350 :     native!
1351 :    
1352 :     switches back to native style. This was needed so that
1353 :     path config files can be written portably, see src/system/pathconfig.
1354 :    
1355 :     - Runtime system:
1356 :    
1357 :     - win32-filesys.c: get_file_time and set_file_time now
1358 :     access modification time, not creation time.
1359 :    
1360 :     - I/O code made aware of new array representation.
1361 :    
1362 :     - Bug fixes in X86.prim.masm.
1363 :    
1364 :     - src/system/makeml made aware of win32. (For use under cygwin
1365 :     and other Unix-environments for windows.)
1366 :    
1367 :     - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
1368 :     which caused BinIO.inputAll to fail consistently.
1369 :    
1370 :     .bat scripts:
1371 :    
1372 :     Windows .bat scripts assume that SMLNJ_HOME is defined.
1373 :    
1374 :     - sml.bat, ml-yacc.bat, ml-lex.bat: Driver scripts for standalone
1375 :     applications (sml, ml-yacc, ml-lex).
1376 :     - ml-build.bat: analogous to ml-build.
1377 :     - config\install.bat: Analogous to config/install.sh. This requires
1378 :     that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
1379 :     (nmake etc. must be on the path, and vcvars32 must have been run.)
1380 :     Moreover, sources for ml-lex and ml-yacc need to exist under src,
1381 :     and the bootfile hierarchy must have been unpacked under
1382 :     sml.boot.x86-win32.
1383 :     The script is very primitive and does a poor job at error checking.
1384 :     It only installs the base system, ml-lex, and ml-yacc. No other
1385 :     libraries are being installed (i.e., you get only those that
1386 :     are part of the compiler.)
1387 :     - link-sml.bat: analogous to .link-sml, but not currently used
1388 :    
1389 :     Unrelated bug fixes:
1390 :    
1391 :     - ml-nlffigen now exports structures ST_* corresponding to incomplete
1392 :     types.
1393 :     - Added getDevice to PP/src/pp-debug-fn.sml. (Would not compile
1394 :     otherwise.)
1395 :    
1396 :     ----------------------------------------------------------------------
1397 : macqueen 1337 Name: Dave MacQueen (macqueen@cs.uchicago.edu)
1398 :     Date: 2003/06/17
1399 :     Tag: macqueen-20030617-bug895
1400 :     Description:
1401 :    
1402 :     Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
1403 :     Tag will be used for new development branch (mcz-branch) for
1404 :     use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
1405 :    
1406 :     ----------------------------------------------------------------------
1407 : mblume 1368 Name: Matthias Blume (blume (at) tti - c (dot) org)
1408 : mblume 1335 Date: 2003/05/27 16:55:00 CDT
1409 :     Tag: blume-20030527-polyeq
1410 :     Description:
1411 :    
1412 :     Tried to eliminated most cases of polymorphic equality.
1413 :    
1414 :     ----------------------------------------------------------------------
1415 : mblume 1368 Name: Matthias Blume (blume (at) tti - c (dot) org)
1416 : mblume 1334 Date: 2003/05/21 17:45:00 CDT
1417 :     Tag: blume-20030517-complete
1418 :     Description:
1419 :    
1420 :     Two changes:
1421 :    
1422 :     1. Added a flag for controlling whether non-exhaustive bindings will
1423 :     be treated as errors (default is false).
1424 :     2. Cleaned up the *entire* source tree so that CMB.make goes through
1425 :     without a single non-exhaustive match- or bind warning.
1426 :    
1427 :     ----------------------------------------------------------------------
1428 : mblume 1368 Name: Matthias Blume (blume (at) tti - c (dot) org)
1429 : mblume 1332 Date: 2003/05/17 10:20:00 CDT
1430 :     Tag: blume-20030517-absyn
1431 :     Description:
1432 :    
1433 :     1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
1434 :    
1435 :     This mainly affects the quality of error messages. However, some
1436 :     of the code is now more straightforward than before. (Treatment of
1437 :     the above four constructs in translate.sml is much simpler than
1438 :     the "macro-expansion" that was going on before. Plus, the mach-
1439 :     compiler no longer gets invoked just to be able to compile an
1440 :     if-expression.)
1441 :    
1442 :     2. The ErrorMsg.Error exception is now caught and absorbed by the
1443 :     interactive loop.
1444 :    
1445 :     ----------------------------------------------------------------------
1446 : allenleung 1327 Name: Allen Leung
1447 :     Date: 2003/05/16 13:05:00 CDT
1448 :     Tag: leunga-20030516-cygwin-runtime
1449 :     Description:
1450 :    
1451 :     Ported the runtime system to cygwin, which uses the unix
1452 :     x86-unix bin files. Missing/buggy features:
1453 :    
1454 :     o getnetbyname, getnetbyaddr: these functions seem to be missing in
1455 :     the Cygwin library.
1456 :     o Ctrl-C handling may be flaky.
1457 :     o Windows system calls and Windows I/O are not supported.
1458 :    
1459 :     A new set of binfiles is located at:
1460 :    
1461 :     http://www.dorsai.org/~leunga/boot.x86-unix.tgz
1462 :    
1463 :     This is only needed for bootstrapping the cygwin version of smlnj.
1464 :     Other x86 versions can use the existing binfiles.
1465 :    
1466 :     ----------------------------------------------------------------------
1467 : mblume 1317 Name: Matthias Blume
1468 :     Date: 2003/04/08 15:42:00 CDT
1469 :     Tag: blume-20030408-listpair
1470 :     Description:
1471 :    
1472 :     1. Added a target 'mlrisc' to installer.
1473 :    
1474 :     2. Added missing elements to structure ListPair.
1475 :    
1476 :     ----------------------------------------------------------------------
1477 : leunga 1296 Name: Allen Leung
1478 :     Date: 2003/01/07 10:40:00 EST
1479 :     Tag: leunga-20030107-int-rem
1480 :     Description:
1481 :    
1482 :     Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
1483 :     The arguments to the SUBL instruction were swapped.
1484 :    
1485 :     ----------------------------------------------------------------------
1486 : blume 1229 Name: Matthias Blume
1487 : blume 1294 Date: 2002/12/12 16:25:00 EST
1488 :     Tag: blume-20021212-risc-ra
1489 :     Description:
1490 :    
1491 :     Fixed a serious bug in the rewrite code for FP spilling/reloading that
1492 :     sent the RA into an infinite loop when floating point registers get
1493 :     spilled. (Because of this bug, e.g., nucleic stopped compiling between
1494 :     110.37 and 110.38.)
1495 :     There was another set of potential problems related to the handling of
1496 :     MLRISC annotations (but those did not yet cause real problems, apparently).
1497 :    
1498 :     ----------------------------------------------------------------------
1499 :     Name: Matthias Blume
1500 : blume 1291 Date: 2002/12/06 22:40:00 EST
1501 :     Tag: blume-20021206-cm-fileid
1502 :     Description:
1503 :    
1504 :     Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
1505 :     This fixes the problem of CM getting confused by files that suddenly
1506 :     change their identity (e.g., by getting unlinked and recreated by some
1507 :     text editor such as vi). There might be a better/cheaper/cleaner way
1508 :     of doing this, but for now this will have to do.
1509 :    
1510 :     ----------------------------------------------------------------------
1511 :     Name: Matthias Blume
1512 : blume 1288 Date: 2002/10/28 09:50:00 EST
1513 :     Tag: blume-20021028-typecheck
1514 :     Description:
1515 :    
1516 :     Exported structure Typecheck from $smlnj/viscomp/core.cm.
1517 :    
1518 :     ----------------------------------------------------------------------
1519 :     Name: Matthias Blume
1520 : blume 1282 Date: 2002/10/17 09:10:00 EDT
1521 : blume 1279 Tag: Release_110_42
1522 :     Description:
1523 :    
1524 : blume 1282 In good old tradition, there has been a slight hiccup so that we have
1525 :     to patch 110.42 after the fact. The old release tag has been replaced
1526 :     (see below).
1527 :    
1528 :     The change solves a problem with two competing approaches the
1529 :     configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
1530 :     each other's way.
1531 :    
1532 :     This change only affects the runtime system code and the installer script.
1533 :     (No new bootfiles.)
1534 :    
1535 :     ----------------------------------------------------------------------
1536 :     Name: Matthias Blume
1537 :     Date: 2002/10/16 12:00:00 EDT
1538 :     Tag: Release_110_42_removed
1539 :     Description:
1540 :    
1541 : blume 1279 New working release. New bootfiles.
1542 :    
1543 :     ----------------------------------------------------------------------
1544 :     Name: Matthias Blume
1545 : blume 1273 Date: 2002/10/10 13:10:00 EDT
1546 :     Tag: blume-20021010-ppc-divs
1547 :     Description:
1548 :    
1549 :     The mltree operator DIVS must be implemented with an overflow check on
1550 :     the PPC because the hardware indicates divide-by-zero using "overflow" as
1551 :     well.
1552 :    
1553 :     ----------------------------------------------------------------------
1554 :     Name: Matthias Blume
1555 : blume 1261 Date: 2002/07/23 11:45:00
1556 :     Tag: blume-20020723-smlnj-home
1557 :     Description:
1558 :    
1559 :     Sml now senses the SMLNJ_HOME environment variable. If this is set,
1560 :     then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
1561 :     CM_PATHCONFIG is also set), the path configuration file is assumed
1562 :     to be in $SMLNJ_HOME/lib/pathconfig. This way one can easily move
1563 :     the entire tree to some other place and everything will "just work".
1564 :    
1565 :     (Companion commands such as ml-build and ml-makedepend also sense this
1566 :     variable.)
1567 :    
1568 :     ----------------------------------------------------------------------
1569 :     Name: Matthias Blume
1570 : blume 1259 Date: 2002/07/12 21:19:00 EDT
1571 :     Tag: blume-20020712-liveness
1572 :     Description:
1573 :    
1574 :     Exported two useful "step" functions from liveness module (MLRISC).
1575 :    
1576 :     ----------------------------------------------------------------------
1577 :     Name: Matthias Blume
1578 : blume 1253 Date: 2002/07/05 16:00 EDT
1579 :     Tag: Release_110_41
1580 :     Description:
1581 :    
1582 :     New working release. New bootfiles.
1583 :    
1584 :     ----------------------------------------------------------------------
1585 :     Name: Matthias Blume
1586 : blume 1250 Date: 2002/07/05 10:25:00 EDT
1587 :     Tag: blume-20020705-btimp
1588 :     Description:
1589 :    
1590 :     Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
1591 :     other clients can set up backtracing support.
1592 :    
1593 :     ----------------------------------------------------------------------
1594 :     Name: Matthias Blume
1595 : blume 1249 Date: 2002/06/25 17:23:00 EDT
1596 :     Tag: blume-20020625-fpmax
1597 :     Description:
1598 :    
1599 :     Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
1600 :     case. (The sense of the isNaN test was reversed -- which made min and
1601 :     max always return their first argument.)
1602 :    
1603 :     ----------------------------------------------------------------------
1604 :     Name: Matthias Blume
1605 : blume 1240 Date: 2002/06/11
1606 :     Tag: blume-20020611-unixpath
1607 :     Description:
1608 :    
1609 :     Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
1610 :    
1611 :     ----------------------------------------------------------------------
1612 :     Name: Matthias Blume
1613 : blume 1238 Date: 2002/06/10 16:35:00 EDT
1614 :     Tag: blume-20020610-ieeereal
1615 :     Description:
1616 :    
1617 :     I back-ported my implementation of IEEEReal.fromString from the
1618 :     idlbasis-devel branch so that we can test it.
1619 :    
1620 :     Another small change is that ppDec tries to give more information
1621 :     than just "<sig>" in the case of functors. However, this code is
1622 :     broken in some mysterious way if the functor's body's signature
1623 :     has not been declared by ascription but gets inferred from the
1624 :     implementation. This needs fixing...
1625 :    
1626 :     ----------------------------------------------------------------------
1627 :     Name: Matthias Blume
1628 : blume 1229 Date: 2002/05/31
1629 :     Tag: blume-20020531-btrace-mode
1630 :     Description:
1631 :    
1632 :     Resurrected SMLofNJ.Internals.BTrace.mode. (It accidentally fell by
1633 :     the wayside when I switched over to using Controls everywhere.)
1634 :    
1635 :     ----------------------------------------------------------------------
1636 : george 1222 Name: Lal George
1637 :     Date: 2002/05/23 12:21:40 EDT
1638 :     Tag: george-20020523-visual-labels
1639 :     Description:
1640 :    
1641 :     Labels are now displayed in the graphical output to make
1642 :     the fall-through and target blocks obvious.
1643 :    
1644 :     ----------------------------------------------------------------------
1645 : blume 1204 Name: Matthias Blume
1646 : blume 1220 Date: 2002/05/22 11:03:00 EDT
1647 :     Tag: blume-20020522-shrink
1648 :     Description:
1649 :    
1650 :     John tweaked yesterday's fix for 1131 to handle an out-of-memory
1651 :     situation that comes up when allocating huge arrays.
1652 :    
1653 :     ----------------------------------------------------------------------
1654 :     Name: Matthias Blume
1655 : blume 1215 Date: 2002/05/21 16:00:00 EDT
1656 :     Tag: Release_110_40
1657 :     Description:
1658 :    
1659 :     New working release (110.40). New bootfiles.
1660 :    
1661 :     [Also: John Reppy fixed GC bug 1131.]
1662 :    
1663 :     ----------------------------------------------------------------------
1664 :     Name: Matthias Blume
1665 : blume 1212 Date: 2002/05/21 12:35:00 EDT
1666 :     Tag: blume-20020521-cmdoc
1667 :     Description:
1668 :    
1669 :     CM documentation update.
1670 :    
1671 :     ----------------------------------------------------------------------
1672 :     Name: Matthias Blume
1673 : blume 1211 Date: 2002/05/21 10:55:00 EDT
1674 :     Tag: blume-20020521-misc
1675 :     Description:
1676 :    
1677 :     - John tweaked runtime to be silent on heap export (except when
1678 :     GC messages are on).
1679 :     - I added a few more things (cross-compiling versions of CMB) to
1680 :     config/preloads (as suggestions).
1681 :    
1682 :     ----------------------------------------------------------------------
1683 :     Name: Matthias Blume
1684 : blume 1208 Date: 2002/05/20 22:25:00 EDT
1685 :     Tag: blume-20020520-controls
1686 :     Description:
1687 :    
1688 :     - Added ControlUtil structure to control-lib.cm.
1689 :     - Use it throughout.
1690 :     - Used Controls facility to define MLRISC controls (as opposed to
1691 :     registering MLRISC control ref cells with Controls after the fact)
1692 :     - Fixed messed-up controls priorities.
1693 :    
1694 :     * Removed again all the stuff from config/preloads that one wouldn't
1695 :     be able to preload at the time the initial heap image is built.
1696 :     (Many libraries, e.g., CML, do not exist yet at this time. The
1697 :     only libraries that can be preloaded via config/preloads are those
1698 :     that come bundled with the bootfiles.)
1699 :    
1700 :     ----------------------------------------------------------------------
1701 :     Name: Matthias Blume
1702 : blume 1204 Date: 2002/05/20 10:59:00 EDT
1703 :     Tag: blume-20020520-preloads
1704 :     Description:
1705 :    
1706 :     Added a lot of commented-out suggestions for things to be included
1707 :     in config/preloads.
1708 :    
1709 :     ----------------------------------------------------------------------
1710 : leunga 1203 Name: Allen Leung
1711 :     Date: 2002/05/18 14:20:00 EDT
1712 :     Tag: leunga-20020518-mdl
1713 :     Description:
1714 :    
1715 :     o Made the mdl tool stuff compile and run again.
1716 :     o I've disabled all the stuff that depends on RTL specifications; they
1717 :     are all badly broken anyway.
1718 :    
1719 :     ----------------------------------------------------------------------
1720 : blume 1201 Name: Matthias Blume
1721 :     Date: 2002/05/17 16:49:00 EDT
1722 :     Tag: blume-20020517-controls
1723 :     Description:
1724 :    
1725 :     0. John Reppy made several modifications to the SML/NJ library.
1726 :     In particular, there is a shiny new controls-lib.cm.
1727 :    
1728 :     1. Pushed new controls interface through compiler so that everything
1729 :     compiles again.
1730 :    
1731 :     2. Added FormatComb and FORMAT_COMB to the CML version of the
1732 :     SML/NJ library (so that CML compiles again).
1733 :    
1734 :     3. Modified init scripts because XXX_DEFAULT environment variables
1735 :     are no longer with us. (Boot-time initialization is now done
1736 :     using the same environment variables that are also used for
1737 :     startup-time initialization of controls.)
1738 :    
1739 :     ----------------------------------------------------------------------
1740 : george 1192 Name: Lal George
1741 :     Date: 2002/05/15 09:20:10 EDT
1742 :     Tag: george-20020515-pseudo-op-decls
1743 :     Description:
1744 :    
1745 :     All pseudo-ops emitted before the first segment declaration
1746 :     such as TEXT, DATA, and BSS directives are assumed to be global
1747 :     declarations and are emitted first in the assembly file. This is
1748 :     useful in a number of situations where one has pseudo-ops that are not
1749 :     specific to any segment, and also works around the constraint that one
1750 :     cannot have client pseudo-ops in the TEXT segment.
1751 :    
1752 :     Because no segment is associated with these declarations it is
1753 :     an error to allocate any space or objects before the first segment
1754 :     directive and an exception will be raised. However, we cannot make
1755 :     this check for client pseudo-ops.
1756 :    
1757 :     These top level declarations are a field in the CFG graph_info.
1758 :     In theory you can continue to add to this field after the CFG has been
1759 :     built -- provided you know what you are doing;-)
1760 :    
1761 :     ----------------------------------------------------------------------
1762 : blume 1175 Name: Matthias Blume
1763 : blume 1190 Date: 2002/05/13 16:40:00 EDT
1764 :     Tag: blume-20020513-pp-etc
1765 :     Description:
1766 :    
1767 :     A few minor bugfixes:
1768 :    
1769 :     - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
1770 :     (Bogus printouts for redefined bindings still occur. Compiler
1771 :     bug should no longer occur now. We need to redo the prettyprinter
1772 :     from scratch.)
1773 :    
1774 :     - CM pathname printer now also adds escape sequences for ( and )
1775 :    
1776 :     - commend and docu fixes for ml-nlffi
1777 :    
1778 :     ----------------------------------------------------------------------
1779 :     Name: Matthias Blume
1780 : blume 1189 Date: 2002/05/10 16:40:00 EDT
1781 :     Tag: blume-20020510-erg-textio
1782 :     Description:
1783 :    
1784 :     Applied the following bugfix provided by Emden Gansner:
1785 :    
1786 :     Output is corrupted when outputSubstr is used rather than output.
1787 :    
1788 :     The problem occurs when a substring
1789 :    
1790 :     ss = (s, dataStart, dataLen)
1791 :    
1792 :     where dataStart > 0, fills a stream buffer with avail bytes left.
1793 :     avail bytes of s, starting at index dataStart, are copied into the
1794 :     buffer, the buffer is flushed, and then the remaining dataLen-avail
1795 :     bytes of ss are copied into the beginning of the buffer. Instead of
1796 :     starting this copy at index dataStart+avail in s, the current code
1797 :     starts the copy at index avail.
1798 :    
1799 :     Fix:
1800 :     In text-io-fn.sml, change line 695 from
1801 :     val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
1802 :     to
1803 :     val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
1804 :    
1805 :     ----------------------------------------------------------------------
1806 :     Name: Matthias Blume
1807 : blume 1186 Date: 2002/04/12 13:55:00 EDT
1808 :     Tag: blume-20020412-assyntax
1809 :     Description:
1810 :    
1811 :     1. Grabbed newer assyntax.h from the XFree86 project.
1812 :     2. Fiddled with how to compile X86.prim.asm without warnings.
1813 :     3. (Very) Minor cleanup in CM.
1814 :    
1815 :     ----------------------------------------------------------------------
1816 :     Name: Matthias Blume
1817 : blume 1185 Date: 2002/04/01 (no joke!) 17:07:00 EST
1818 :     Tag: blume-20020401-x86div
1819 :     Description:
1820 :    
1821 :     Added full support for div/mod/rem/quot on the x86, using the machine
1822 :     instruction's two results (without clumsily recomputing the remainder)
1823 :     directly where appropriate.
1824 :    
1825 :     Some more extensive power-of-two support was added to the x86 instruction
1826 :     selector (avoiding expensive divs, mods, and muls where they can be
1827 :     replaced with cheaper shifts and masks). However, this sort of thing
1828 :     ought to be done earlier, e.g., within the CPS optimizer so that
1829 :     all architectures benefit from it.
1830 :    
1831 :     The compiler compiles to a fixed point, but changes might be somewhat
1832 :     fragile nevertheless. Please, report any strange things that you might
1833 :     see wrt. div/mod/quot/rem...
1834 :    
1835 :     ----------------------------------------------------------------------
1836 :     Name: Matthias Blume
1837 : blume 1184 Date: 2002/03/29 17:22:00
1838 :     Tag: blume-20020329-div
1839 :     Description:
1840 :    
1841 :     Fixed my broken div/mod logic. Unfortunately, this means that the
1842 :     inline code for div/mod now has one more comparison than before.
1843 :     Fast paths (quotient > 0 or remainder = 0) are not affected, though.
1844 :     The problem was with quotient = 0, because that alone does not tell
1845 :     us which way the rounding went. One then has to look at whether
1846 :     remainder and divisor have the same sign... :(
1847 :    
1848 :     Anyway, I replaced the bootfiles with fresh ones...
1849 :    
1850 :     ----------------------------------------------------------------------
1851 :     Name: Matthias Blume
1852 : blume 1183 Date: 2002/03/29 14:10:00 EST
1853 :     Tag: blume-20020329-inlprims
1854 :     Description:
1855 :    
1856 :     NEW BOOTFILES!!! Version number bumped to 110.39.3.
1857 :    
1858 :     Primops have changed. This means that the bin/boot-file formats have
1859 :     changed as well.
1860 :    
1861 :     To make sure that there is no confusion, I made a new version.
1862 :    
1863 :    
1864 :     CHANGES:
1865 :    
1866 :     * removed REMT from mltree (remainder should never overflow).
1867 :    
1868 :     * added primops to deal with divisions of all flavors to the frontend
1869 :    
1870 :     * handled these primops all the way through so they map to their respective
1871 :     MLRISC support
1872 :    
1873 :     * used these primops in the implementation of Int, Int32, Word, Word32
1874 :    
1875 :     * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
1876 :    
1877 :     * parameterized INLMIN, INLMAX, and INLABS by a numkind
1878 :    
1879 :     * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
1880 :     floating point
1881 :    
1882 :     * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
1883 :     and Real (but Real.abs maps to a separate floating-point-only primop)
1884 :    
1885 :    
1886 :     TODO items:
1887 :    
1888 :     * Hacked Alpha32 instruction selection, disabling the selection of REMx
1889 :     instructions because the machine instruction encoder cannot handle
1890 :     them. (Hppa, PPC, and Sparc instruction selection did not handle
1891 :     REM in the first place, and REM is supported by the x86 machine coder.)
1892 :    
1893 :     * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
1894 :     selection phase. (The two can be streamlined because the hardware
1895 :     delivers both quotient and remainder at the same time anyway.)
1896 :    
1897 :     * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
1898 :     (Currently the behavior is inconsistent both across architectures and
1899 :     wrt. the draft Basis spec.)
1900 :    
1901 :     * Word8 should eventually be handled natively, too.
1902 :    
1903 :     * There seems to be one serious bug in mltree-gen.sml. It appears, though,
1904 :     as if there currently is no execution path that could trigger it in
1905 :     SML/NJ. (The assumptions underlying functions arith and promotable do not
1906 :     hold for things like multiplication and division.)
1907 :    
1908 :     ----------------------------------------------------------------------
1909 :     Name: Matthias Blume
1910 : blume 1181 Date: 2002/03/27 16:27:00 EST
1911 :     Tag: blume-20020327-mlrisc-divisions
1912 :     Description:
1913 :    
1914 :     Added support for all four division operations (ML's div, mod, quot,
1915 :     and rem) to MLRISC. In the course of doing so, I also rationalized
1916 :     the naming (no more annoying switch-around of DIV and QUOT), by
1917 :     parameterizing the operation by div_rounding_mode (which can be either
1918 :     DIV_TO_ZERO or DIV_TO_NEGINF).
1919 :    
1920 :     The generic MLTreeGen functor takes care of compiling all four
1921 :     operations down to only round-to-zero div.
1922 :    
1923 :     Missing pieces:
1924 :    
1925 :     * Doing something smarter than relying on MLTreeGen on architectures
1926 :     like, e.g., the x86 where hardware division delivers both quotient and
1927 :     remainder at the same time. With this, the implementation of the
1928 :     round-to-neginf operations could be further streamlined.
1929 :    
1930 :     * Remove inlining support for div/mod/rem from the frontend and replace it
1931 :     with primops that get carried through to the backend. Do this for all
1932 :     int and word types.
1933 :    
1934 :     ----------------------------------------------------------------------
1935 :     Name: Matthias Blume
1936 : blume 1180 Date: 2002/03/25 17:25:00 EST
1937 :     Tag: blume-20020325-divmod
1938 :     Description:
1939 :    
1940 :     I improved (hopefully without breaking them) the implementation of Int.div,
1941 :     Int.mod, and Int.rem. For this, the code in translate.sml now takes
1942 :     advantage of the following observations:
1943 :    
1944 :     Let q = x quot y r = x rem y
1945 :     d = x div y m = x mod y
1946 :    
1947 :     where "quot" is the round-to-zero version of integer division that
1948 :     hardware usually provides. Then we have:
1949 :    
1950 :     r = x - q * y where neither the * nor the - will overflow
1951 :     d = if q >= 0 orelse x = q * y then q else q - 1
1952 :     where neither the * nor the - will overflow
1953 :     m = if q >= 0 orelse r = 0 then r else r + y
1954 :     where the + will not overflow
1955 :    
1956 :     This results in substantial simplification of the generated code.
1957 :     The following table shows the number of CFG nodes and edges generated
1958 :     for
1959 :     fun f (x, y) = x OPER y
1960 :     (* with OPER \in div, mod, quot, rem *)
1961 :    
1962 :    
1963 :     OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
1964 :     --------------------------------------------------------
1965 :     div | 24 | 39 | 12 | 16
1966 :     mod | 41 | 71 | 12 | 16
1967 :     quot | 8 | 10 | 8 | 10
1968 :     rem | 10 | 14 | 8 | 10
1969 :    
1970 :    
1971 :     ----------------------------------------------------------------------
1972 :     Name: Matthias Blume
1973 : blume 1179 Date: 2002/03/25 22:06:00 EST
1974 :     Tag: blume-20020325-cprotobug
1975 :     Description:
1976 :    
1977 :     Fixed a bug in cproto (c prototype decoder).
1978 :    
1979 :     ----------------------------------------------------------------------
1980 :     Name: Matthias Blume
1981 : blume 1178 Date: 2002/03/25 16:00:00 EST
1982 :     Tag: blume-20020325-raw-primops
1983 :     Description:
1984 :    
1985 :     I did some cleanup to Allen's new primop code and
1986 :     replaced yesterday's bootfiles with new ones.
1987 :     (But they are stored in the same place.)
1988 :    
1989 :     ----------------------------------------------------------------------
1990 :     Name: Matthias Blume
1991 : blume 1175 Date: 2002/03/24 22:40:00 EST
1992 :     Tag: blume-20020324-bootfiles
1993 :     Description:
1994 :    
1995 :     Made the bootfiles that Allen asked for.
1996 :    
1997 :     ----------------------------------------------------------------------
1998 : leunga 1174 Name: Allen Leung
1999 :     Date: 2002/03/23 15:50:00 EST
2000 :     Tag: leunga-20020323-flint-cps-rcc-primops
2001 :     Description:
2002 :    
2003 :     1. Changes to FLINT primops:
2004 :    
2005 :     (* make a call to a C-function;
2006 :     * The primop carries C function prototype information and specifies
2007 :     * which of its (ML-) arguments are floating point. C prototype
2008 :     * information is for use by the backend, ML information is for
2009 :     * use by the CPS converter. *)
2010 :     | RAW_CCALL of { c_proto: CTypes.c_proto,
2011 :     ml_args: ccall_type list,
2012 :     ml_res_opt: ccall_type option,
2013 :     reentrant : bool
2014 :     } option
2015 :     (* Allocate uninitialized storage on the heap.
2016 :     * The record is meant to hold short-lived C objects, i.e., they
2017 :     * are not ML pointers. With the tag, the representation is
2018 :     * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
2019 :     *)
2020 :     | RAW_RECORD of {tag:bool,sz:int}
2021 :     and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
2022 :    
2023 :     2. These CPS primops are now overloaded:
2024 :    
2025 :     rawload of {kind:numkind}
2026 :     rawstore of {kind:numkind}
2027 :    
2028 :     The one argument form is:
2029 :    
2030 :     rawload {kind} address
2031 :    
2032 :     The two argument form is:
2033 :    
2034 :     rawload {kind} [ml object, byte-offset]
2035 :    
2036 :     3. RAW_CCALL/RCC now takes two extra arguments:
2037 :    
2038 :     a. The first is whether the C call is reentrant, i.e., whether
2039 :     ML state should be saved and restored.
2040 :     b. The second argument is a string argument specifying the name of
2041 :     library and the C function.
2042 :    
2043 :     These things are currently not handled in the code generator, yet.
2044 :    
2045 :     4. In CProto,
2046 :    
2047 :     An encoding type of "bool" means "ml object" and is mapped into
2048 :     C prototype of PTR. Note that "bool" is different than "string",
2049 :     even though "string" is also mapped into PTR, because "bool"
2050 :     is assigned an CPS type of BOGt, while "string" is assigned INT32t.
2051 :    
2052 :     5. Pickler/unpicker
2053 :    
2054 :     Changed to handle RAW_RECORD and newest RAW_CCALL
2055 :    
2056 :     6. MLRiscGen,
2057 :    
2058 :     1. Changed to handle the new rawload/rawstore/rawrecord operators.
2059 :     2. Code for handling C Calls has been moved to a new module CPSCCalls,
2060 :     in the file CodeGen/cpscompile/cps-c-calls.sml
2061 :    
2062 :     7. Added the conditional move operator
2063 :    
2064 :     condmove of branch
2065 :    
2066 :     to cps. Generation of this is still buggy so it is currently
2067 :     disabled.
2068 :    
2069 :     ----------------------------------------------------------------------
2070 : george 1168 Name: Lal George
2071 :     Date: 2002/03/22 14:18:25 EST
2072 : george 1171 Tag: george-20020322-cps-branch-prob
2073 : george 1168 Description:
2074 :    
2075 : george 1171 Implemented the Ball-Larus branch prediction-heuristics, and
2076 : george 1168 incorporated graphical viewers for control flow graphs.
2077 :    
2078 : george 1171 Ball-Larus Heuristics:
2079 : george 1168 ---------------------
2080 :     See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
2081 :    
2082 :     By design it uses the Dempster-Shafer theory for combining
2083 :     probabilities. For example, in the function:
2084 :    
2085 :     fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
2086 :    
2087 :     the ball-larus heuristics predicts that the n=0 is unlikely
2088 :     (OH-heuristic), and the 'then' branch is unlikely because of the
2089 :     RH-heuristic -- giving the 'then' branch an even lower combined
2090 : george 1171 probability using the Dempster-Shafer theory.
2091 : george 1168
2092 :     Finally, John Reppy's loop analysis in MLRISC, further lowers the
2093 :     probability of the 'then' branch because of the loop in the else
2094 :     branch.
2095 :    
2096 :    
2097 :     Graphical Viewing:
2098 :     ------------------
2099 :     I merely plugged in Allen's graphical viewers into the compiler. The
2100 :     additional code is not much. At the top level, saying:
2101 :    
2102 :     Control.MLRISC.getFlag "cfg-graphical-view" := true;
2103 :    
2104 :     will display the graphical view of the control flow graph just before
2105 :     back-patching. daVinci must be in your path for this to work. If
2106 :     daVinci is not available, then the default viewer can be changed
2107 :     using:
2108 :    
2109 :     Control.MLRISC.getString "viewer"
2110 :    
2111 :     which can be set to "dot" or "vcg" for the corresponding viewers. Of
2112 :     course, these viewers must be in your path.
2113 :    
2114 :     The above will display the compilation unit at the level of clusters,
2115 :     many of which are small, boring, and un-interesting. Also setting:
2116 :    
2117 :     Control.MLRISC.getInt "cfg-graphical-view_size"
2118 :    
2119 :     will display clusters that are larger than the value set by the above.
2120 :    
2121 :    
2122 :     ----------------------------------------------------------------------
2123 : blume 1159 Name: Matthias Blume
2124 :     Date: 2002/03/21 22:20:00 EST
2125 :     Tag: blume-20020321-kmp-bugfix
2126 :     Description:
2127 :    
2128 :     Changed the interface to the KMP routine in PreString and fixed
2129 :     a minor bug in one place where it was used.
2130 :    
2131 :     ----------------------------------------------------------------------
2132 : leunga 1156 Name: Allen Leung
2133 : leunga 1158 Date: 2002/03/21 20:30:00 EST
2134 :     Tag: leunga-20020321-cfg
2135 :     Description:
2136 :    
2137 :     Fixed a potential problem in cfg edge splitting.
2138 :    
2139 :     ----------------------------------------------------------------------
2140 :     Name: Allen Leung
2141 : leunga 1156 Date: 2002/03/21 17:15:00 EST
2142 :     Tag: leunga-20020321-x86-fp-cfg
2143 :     Description:
2144 :    
2145 :     1. Recoded the buggy parts of x86-fp.
2146 :    
2147 :     a. All the block reordering code has been removed.
2148 :     We now depend on the block placement phases to do this work.
2149 :    
2150 :     b. Critical edge splitting code has been simplified and moved into the
2151 :     CFG modules, as where they belong.
2152 :    
2153 :     Both of these were quite buggy and complex. The code is now much, much
2154 :     simpler.
2155 :    
2156 :     2. X86 backend.
2157 :    
2158 :     a. Added instructions for 64-bit support. Instruction selection for
2159 :     64-bit has not been committed, however, since that
2160 :     requires changes to MLTREE which haven't been approved by
2161 :     Lal and John.
2162 :    
2163 :     b. Added support for FUCOMI and FUCOMIP when generating code for
2164 :     PentiumPro and above. We only generate these instructions in
2165 :     the fast-fp mode.
2166 :    
2167 :     c. Added cases for JP and JNP in X86FreqProps.
2168 :    
2169 :     3. CFG
2170 :    
2171 :     CFG now has a bunch of methods for edge splitting and merging.
2172 :    
2173 :     4. Machine description.
2174 :    
2175 :     John's simplification of MLTREE_BASIS.fcond broke a few machine
2176 :     description things:
2177 :    
2178 :     rtl-build.{sig,sml} and hppa.mdl fixed.
2179 :    
2180 :     NOTE: the machine description stuff in the repository is still broken.
2181 :     Again, I can't put my fixes in because that involves
2182 :     changes to MLTREE.
2183 :    
2184 :     ----------------------------------------------------------------------
2185 : blume 1145 Name: Matthias Blume
2186 : blume 1155 Date: 2002/03/20 15:55:00 EST
2187 :     Tag: blume-20020320-kmp
2188 :     Description:
2189 :    
2190 :     Implemented Knuth-Morris-Pratt string matching in PreString and used
2191 :     it for String.isSubstring, Substring.isSubstring, and
2192 :     Substring.position.
2193 :    
2194 :     (Might need some stress-testing. Simple examples worked fine.)
2195 :    
2196 :     ----------------------------------------------------------------------
2197 :     Name: Matthias Blume
2198 : blume 1152 Date: 2002/03/19 16:37:00 EST
2199 :     Tag: blume-20020319-witnesses
2200 :     Description:
2201 :    
2202 :     Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
2203 :    
2204 :     This implements a generic mechanism for changing constness qualifiers
2205 :     anywhere within big C types without resorting to outright "casts".
2206 :     (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
2207 :     modify the constness at the outermost level.)
2208 :     The implementation of "convert" is based on the idea of "witness"
2209 :     values -- values that are not used by the operation but whose types
2210 :     "testify" to their applicability. On the implementation side, "convert"
2211 :     is simply a projection (returning its second curried argument). With
2212 :     cross-module inlining, it should not result in any machine code being
2213 :     generated.
2214 :    
2215 :     ----------------------------------------------------------------------
2216 :     Name: Matthias Blume
2217 : blume 1148 Date: 2002/03/15 16:40:00 EST
2218 :     Tag: blume-20020315-basis
2219 :     Description:
2220 :    
2221 :     Provided (preliminary?) implementations for
2222 :    
2223 :     {String,Substring}.{concatWith,isSuffix,isSubstring}
2224 :    
2225 :     and
2226 :    
2227 :     Substring.full
2228 :    
2229 :     Those are in the Basis spec but they were missing in SML/NJ.
2230 :    
2231 :     ----------------------------------------------------------------------
2232 :     Name: Matthias Blume
2233 : blume 1145 Date: 2002/03/14 21:30:00 EST
2234 :     Tag: blume-20020314-controls
2235 :     Description:
2236 :    
2237 :     Controls:
2238 :     ---------
2239 :    
2240 :     1. Factored out the recently-added Controls : CONTROLS stuff and put
2241 :     it into its own library $/controls-lib.cm. The source tree for
2242 :     this is under src/smlnj-lib/Controls.
2243 :    
2244 :     2. Changed the names of types and functions in this interface, so they
2245 :     make a bit more "sense":
2246 :    
2247 :     module -> registry
2248 :     'a registry -> 'a group
2249 :    
2250 :     3. The interface now deals in ref cells only. The getter/setter interface
2251 :     is (mostly) gone.
2252 :    
2253 :     4. Added a function that lets one register an already-existing ref cell.
2254 :    
2255 :     5. Made the corresponding modifications to the rest of the code so that
2256 :     everything compiles again.
2257 :    
2258 :     6. Changed the implementation of Controls.MLRISC back to something closer
2259 :     to the original. In particular, this module (and therefore MLRISC)
2260 :     does not depend on Controls. There now is some link-time code in
2261 :     int-sys.sml that registers the MLRISC controls with the Controls
2262 :     module.
2263 :    
2264 :     CM:
2265 :     ---
2266 :    
2267 :     * One can now specify the lambda-split aggressiveness in init.cmi.
2268 :    
2269 :     ----------------------------------------------------------------------
2270 : leunga 1142 Name: Allen Leung
2271 :     Date: 2002/03/13 17:30:00 EST
2272 :     Tag: leunga-20020313-x86-fp-unary
2273 :     Description:
2274 :    
2275 :     Bug fix for:
2276 :    
2277 :     > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
2278 :     > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
2279 :     > - fun f(x,(y,z)) = Real.~ y;
2280 :     > [autoloading]
2281 :     > [autoloading done]
2282 :     > fchsl (%eax), 184(%esp)
2283 :     > Error: MLRisc bug: X86MCEmitter.emitInstr
2284 :     >
2285 :     > uncaught exception Error
2286 :     > raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
2287 :    
2288 :     The problem was that the code generator did not generate any fp registers
2289 :     in this case, and the ra didn't know that it needed to run the X86FP phase to
2290 :     translate the pseudo fp instruction. This only happened with unary fp
2291 :     operators in certain situations.
2292 :    
2293 :     ----------------------------------------------------------------------
2294 : blume 1137 Name: Matthias Blume
2295 : blume 1140 Date: 2002/03/13 14:00:00 EST
2296 :     Tag: blume-20020313-overload-etc
2297 :     Description:
2298 :    
2299 :     1. Added _overload as a synonym for overload for backward compatibility.
2300 :     (Control.overloadKW must be true for either version to be accepted.)
2301 :    
2302 :     2. Fixed bug in install script that caused more things to be installed
2303 :     than what was requested in config/targets.
2304 :    
2305 :     3. Made CM aware of the (_)overload construct so that autoloading
2306 :     works.
2307 :    
2308 :     ----------------------------------------------------------------------
2309 :     Name: Matthias Blume
2310 : blume 1138 Date: 2002/03/12 22:03:00 EST
2311 :     Tag: blume-20020312-url
2312 :     Description:
2313 :    
2314 :     Forgot to update BOOT and srcarchiveurl.
2315 :    
2316 :     ----------------------------------------------------------------------
2317 :     Name: Matthias Blume
2318 : blume 1137 Date: 2002/03/12 17:30:00 EST
2319 :     Tag: blume-20020312-version110392
2320 :     Description:
2321 :    
2322 :     Yet another version number bump (because of small changes to the
2323 :     binfile format). Version number is now 110.39.2. NEW BOOTFILES!
2324 :    
2325 :     Changes:
2326 :    
2327 :     The new pid generation scheme described a few weeks ago was overly
2328 :     complicated. I implemented a new mechanism that is simpler and
2329 :     provides a bit more "stability": Once CM has seen a compilation
2330 :     unit, it keeps its identity constant (as long as you do not delete
2331 :     those crucial CM/GUID/* files). This means that when you change
2332 : blume 1145 an interface, compile, then go back to the old interface, and
2333 : blume 1137 compile again, you arrive at the original pid.
2334 :    
2335 :     There now also is a mechanism that instructs CM to use the plain
2336 :     environment hash as a module's pid (effectively making its GUID
2337 :     the empty string). For this, "noguid" must be specified as an
2338 :     option to the .sml file in question within its .cm file.
2339 :     This is most useful for code that is being generated by tools such
2340 :     as ml-nlffigen (because during development programmers tend to
2341 :     erase the tool's entire output directory tree including CM's cached
2342 :     GUIDs). "noguid" is somewhat dangerous (since it can be used to locally
2343 :     revert to the old, broken behavior of SML/NJ, but in specific cases
2344 :     where there is no danger of interface confusion, its use is ok
2345 :     (I think).
2346 :    
2347 :     ml-nlffigen by default generates "noguid" annotations. They can be
2348 :     turned off by specifying -guid in its command line.
2349 :    
2350 :     ----------------------------------------------------------------------
2351 : george 1136 Name: Lal George
2352 :     Date: 2002/03/12 12 14:42:36 EST
2353 :     Tag: george-20020312-frequency-computation
2354 :     Description:
2355 :    
2356 :     Integrated jump chaining and static block frequency into the
2357 :     compiler. More details and numbers later.
2358 : leunga 1025
2359 : george 1003 ----------------------------------------------------------------------
2360 : george 1133 Name: Lal George
2361 :     Date: 2002/03/11 11 22:38:53 EST
2362 :     Tag: george-20020311-jump-chain-elim
2363 :     Description:
2364 :    
2365 :     Tested the jump chain elimination on all architectures (except the
2366 :     hppa). This is on by default right now and is profitable for the
2367 :     alpha and x86, however, it may not be profitable for the sparc and ppc
2368 :     when compiling the compiler.
2369 :    
2370 :     The gc test will typically jump to a label at the end of the cluster,
2371 :     where there is another jump to an external cluster containing the actual
2372 :     code to invoke gc. This is to allow factoring of common gc invocation
2373 :     sequences. That is to say, we generate:
2374 :    
2375 :     f:
2376 :     testgc
2377 :     ja L1 % jump if above to L1
2378 :    
2379 :     L1:
2380 :     jmp L2
2381 :    
2382 :    
2383 :     After jump chain elimination the 'ja L1' instructions is converted to
2384 :     'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
2385 :     up being implemented in their long form (if L2 is far away) using:
2386 :    
2387 :     jbe L3 % jump if below or equal to L3
2388 :     jmp L2
2389 :     L3:
2390 :     ...
2391 :    
2392 :    
2393 :     For large compilation units L2 may be far away.
2394 :    
2395 :    
2396 :     ----------------------------------------------------------------------
2397 : blume 1132 Name: Matthias Blume
2398 :     Date: 2002/03/11 13:30:00 EST
2399 :     Tag: blume-20020311-mltreeeval
2400 :     Description:
2401 :    
2402 :     A functor parameter was missing.
2403 :    
2404 :     ----------------------------------------------------------------------
2405 : leunga 1130 Name: Allen Leung
2406 : leunga 1131 Date: 2002/03/11 10:30:00 EST
2407 : leunga 1142 Tag: leunga-20020311-runtime-string0
2408 : leunga 1131 Description:
2409 :    
2410 :     The representation of the empty string now points to a
2411 :     legal null terminated C string instead of unit. It is now possible
2412 :     to convert an ML string into C string with InlineT.CharVector.getData.
2413 :     This compiles into one single machine instruction.
2414 :    
2415 :     ----------------------------------------------------------------------
2416 :     Name: Allen Leung
2417 : leunga 1130 Date: 2002/03/10 23:55:00 EST
2418 :     Tag: leunga-20020310-x86-call
2419 :     Description:
2420 :    
2421 :     Added machine generation for CALL instruction (relative displacement mode)
2422 :    
2423 :     ----------------------------------------------------------------------
2424 : blume 1128 Name: Matthias Blume
2425 :     Date: 2002/03/08 16:05:00
2426 :     Tag: blume-20020308-entrypoints
2427 :     Description:
2428 :    
2429 :     Version number bumped to 110.39.1. NEW BOOTFILES!
2430 :    
2431 :     Entrypoints: non-zero offset into a code object where execution should begin.
2432 :    
2433 :     - Added the notion of an entrypoint to CodeObj.
2434 :     - Added reading/writing of entrypoint info to Binfile.
2435 :     - Made runtime system bootloader aware of entrypoints.
2436 :     - Use the address of the label of the first function given to mlriscGen
2437 :     as the entrypoint. This address is currently always 0, but it will
2438 :     not be 0 once we turn on block placement.
2439 :     - Removed the linkage cluster code (which was The Other Way(tm) of dealing
2440 :     with entry points) from mlriscGen.
2441 :    
2442 :     ----------------------------------------------------------------------
2443 : leunga 1127 Name: Allen Leung
2444 :     Date: 2002/03/07 20:45:00 EST
2445 :     Tag: leunga-20020307-x86-cmov
2446 :     Description:
2447 :    
2448 :     Bug fixes for CMOVcc on x86.
2449 :    
2450 :     1. Added machine code generation for CMOVcc
2451 :     2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
2452 :     3. CMOVcc cannot have an immediate operand as argument.
2453 :    
2454 :     ----------------------------------------------------------------------
2455 : blume 1126 Name: Matthias Blume
2456 :     Date: 2002/03/07 16:15:00 EST
2457 :     Tag: blume-20020307-controls
2458 :     Description:
2459 :    
2460 :     This is a very large but mostly boring patch which makes (almost)
2461 :     every tuneable compiler knob (i.e., pretty much everything under
2462 :     Control.* plus a few other things) configurable via both the command
2463 :     line and environment variables in the style CM did its configuration
2464 :     until now.
2465 :    
2466 :     Try starting sml with '-h' (or, if you are brave, '-H')
2467 :    
2468 :     To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
2469 :     implements the underlying generic mechanism.
2470 :    
2471 :     The interface to some of the existing such facilities has changed somewhat.
2472 :     For example, the MLRiscControl module now provides mkFoo instead of getFoo.
2473 :     (The getFoo interface is still there for backward-compatibility, but its
2474 :     use is deprecated.)
2475 :    
2476 :     The ml-build script passes -Cxxx=yyy command-line arguments through so
2477 :     that one can now twiddle the compiler settings when using this "batch"
2478 :     compiler.
2479 :    
2480 :     TODO items:
2481 :    
2482 :     We should go through and throw out all controls that are no longer
2483 :     connected to anything. Moreover, we should go through and provide
2484 :     meaningful (and correct!) documentation strings for those controls
2485 :     that still are connected.
2486 :    
2487 :     Currently, multiple calls to Controls.new are accepted (only the first
2488 :     has any effect). Eventually we should make sure that every control
2489 :     is being made (via Controls.new) exactly once. Future access can then
2490 :     be done using Controls.acc.
2491 :    
2492 :     Finally, it would probably be a good idea to use the getter-setter
2493 :     interface to controls rather than ref cells. For the time being, both
2494 :     styles are provided by the Controls module, but getter-setter pairs are
2495 :     better if thread-safety is of any concern because they can be wrapped.
2496 :    
2497 :     *****************************************
2498 :    
2499 :     One bug fix: The function blockPlacement in three of the MLRISC
2500 :     backpatch files used to be hard-wired to one of two possibilities at
2501 :     link time (according to the value of the placementFlag). But (I
2502 :     think) it should rather sense the flag every time.
2503 :    
2504 :     *****************************************
2505 :    
2506 :     Other assorted changes (by other people who did not supply a HISTORY entry):
2507 :    
2508 :     1. the cross-module inliner now works much better (Monnier)
2509 :     2. representation of weights, frequencies, and probabilities in MLRISC
2510 :     changed in preparation of using those for weighted block placement
2511 :     (Reppy, George)
2512 :    
2513 :     ----------------------------------------------------------------------
2514 : george 1116 Name: Lal George
2515 : george 1124 Date: 2002/03/07 14:44:24 EST 2002
2516 :     Tag: george-20020307-weighted-block-placement
2517 :    
2518 :     Tested the weighted block placement optimization on all architectures
2519 :     (except the hppa) using AMPL to generate the block and edge frequencies.
2520 :     Changes were required in the machine properties to correctly
2521 :     categorize trap instructions. There is an MLRISC flag
2522 :     "weighted-block-placement" that can be used to enable weighted block
2523 :     placement, but this will be ineffective without block/edge
2524 :     frequencies (coming soon).
2525 :    
2526 :    
2527 :     ----------------------------------------------------------------------
2528 :     Name: Lal George
2529 : george 1116 Date: 2002/03/05 17:24:48 EST
2530 :     Tag: george-20020305-linkage-cluster
2531 :    
2532 : george 1124 In order to support the block placement optimization, a new cluster
2533 :     is generated as the very first cluster (called the linkage cluster).
2534 :     It contains a single jump to the 'real' entry point for the compilation
2535 :     unit. Block placement has no effect on the linkage cluster itself, but
2536 :     all the other clusters have full freedom in the manner in which they
2537 :     reorder blocks or functions.
2538 : george 1116
2539 :     On the x86 the typical linkage code that is generated is:
2540 :     ----------------------
2541 :     .align 2
2542 :     L0:
2543 :     addl $L1-L0, 72(%esp)
2544 : george 1124 jmp L1
2545 : george 1116
2546 :    
2547 :     .align 2
2548 :     L1:
2549 :     ----------------------
2550 :    
2551 :     72(%esp) is the memory location for the stdlink register. This
2552 :     must contain the address of the CPS function being called. In the
2553 : george 1124 above example, it contains the address of L0; before
2554 : george 1116 calling L1 (the real entry point for the compilation unit), it
2555 :     must contain the address for L1, and hence
2556 :    
2557 :     addl $L1-L0, 72(%esp)
2558 :    
2559 : george 1124 I have tested this on all architectures except the hppa.The increase
2560 :     in code size is of course negligible
2561 : george 1116
2562 :     ----------------------------------------------------------------------
2563 : leunga 1115 Name: Allen Leung
2564 :     Date: 2002/03/03 13:20:00 EST
2565 :     Tag: leunga-20020303-mlrisc-tools
2566 :    
2567 :     Added #[ ... ] expressions to mlrisc tools
2568 :    
2569 :     ----------------------------------------------------------------------
2570 : blume 1096 Name: Matthias Blume
2571 : blume 1098 Date: 2002/02/27 12:29:00 EST
2572 :     Tag: blume-20020227-cdebug
2573 :     Description:
2574 :    
2575 :     - made types in structure C and C_Debug to be equal
2576 :     - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
2577 :     - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
2578 :    
2579 :     ----------------------------------------------------------------------
2580 :     Name: Matthias Blume
2581 : blume 1096 Date: 2002/02/26 12:00:00 EST
2582 :     Tag: blume-20020226-ffi
2583 :     Description:
2584 :    
2585 :     1. Fixed a minor bug in CM's "noweb" tool:
2586 :     If numbering is turned off, then truly don't number (i.e., do not
2587 :     supply the -L option to noweb). The previous behavior was to supply
2588 :     -L'' -- which caused noweb to use the "default" line numbering scheme.
2589 :     Thanks to Chris Richards for pointing this out (and supplying the fix).
2590 :    
2591 :     2. Once again, I reworked some aspects of the FFI:
2592 :    
2593 :     A. The incomplete/complete type business:
2594 :    
2595 :     - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
2596 :     gone!
2597 :     - ML types representing an incomplete type are now *equal* to
2598 :     ML types representing their corresponding complete types (just like
2599 :     in C). This is still safe because ml-nlffigen will not generate
2600 :     RTTI for incomplete types, nor will it generate functions that
2601 :     require access to such RTTI. But when ML code generated from both
2602 :     incomplete and complete versions of the C type meet, the ML types
2603 :     are trivially interoperable.
2604 :    
2605 :     NOTE: These changes restore the full generality of the translation
2606 :     (which was previously lost when I eliminated functorization)!
2607 :    
2608 :     B. Enum types:
2609 :    
2610 :     - Structure C now has a type constructor "enum" that is similar to
2611 :     how the "su" constructor works. However, "enum" is not a phantom
2612 :     type because each "T enum" has values (and is isomorphic to
2613 :     MLRep.Signed.int).
2614 :     - There are generic access operations for enum objects (using
2615 :     MLRep.Signed.int).
2616 :     - ml-nlffigen will generate a structure E_foo for each "enum foo".
2617 :     * The structure contains the definition of type "mlrep" (the ML-side
2618 :     representation type of the enum). Normally, mlrep is the same
2619 :     as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
2620 :     then mlrep will be defined as a datatype -- thus facilitating
2621 :     pattern matching on mlrep values.
2622 :     ("-ec" will be suppressed if there are duplicate values in an
2623 :     enumeration.)
2624 :     * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
2625 :     will be generated for each C enum constant xxx.
2626 :     * Conversion functions m2i and i2m convert between mlrep and
2627 :     MLRep.Signed.int. (Without "-ec", these functions are identities.)
2628 :     * Coversion functions c and ml convert between mlrep and "tag enum".
2629 :     * Access functions (get/set) fetch and store mlrep values.
2630 :     - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
2631 :     enumerations are merged into one single enumeration represented by
2632 :     structure E_'.
2633 :    
2634 :     ----------------------------------------------------------------------
2635 : leunga 1080 Name: Allen Leung
2636 : leunga 1094 Date: 2002/02/25 04:45:00 EST
2637 :     Tag: leunga-20020225-cps-spill
2638 :    
2639 :     This is a new implementation of the CPS spill phase.
2640 :     The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
2641 :     In case of problems, replace it with the old file spill.sml
2642 :    
2643 :     The current compiler runs into some serious performance problems when
2644 :     constructing a large record. This can happen when we try to compile a
2645 :     structure with many items. Even a very simple structure like the following
2646 :     makes the compiler slow down.
2647 :    
2648 :     structure Foo = struct
2649 :     val x_1 = 0w1 : Word32.int
2650 :     val x_2 = 0w2 : Word32.int
2651 :     val x_3 = 0w3 : Word32.int
2652 :     ...
2653 :     val x_N = 0wN : Word32.int
2654 :     end
2655 :    
2656 :     The following table shows the compile time, from N=1000 to N=4000,
2657 :     with the old compiler:
2658 :    
2659 :     N
2660 :     1000 CPS 100 spill 0.04u 0.00s 0.00g
2661 :     MLRISC ra 0.06u 0.00s 0.05g
2662 :     (spills = 0 reloads = 0)
2663 :     TOTAL 0.63u 0.07s 0.21g
2664 :    
2665 :     1100 CPS 100 spill 8.25u 0.32s 0.64g
2666 :     MLRISC ra 5.68u 0.59s 3.93g
2667 :     (spills = 0 reloads = 0)
2668 :     TOTAL 14.71u 0.99s 4.81g
2669 :    
2670 :     1500 CPS 100 spill 58.55u 2.34s 1.74g
2671 :     MLRISC ra 5.54u 0.65s 3.91g
2672 :     (spills = 543 reloads = 1082)
2673 :     TOTAL 65.40u 3.13s 6.00g
2674 :    
2675 :     2000 CPS 100 spill 126.69u 4.84s 3.08g
2676 :     MLRISC ra 0.80u 0.10s 0.55g
2677 :     (spills = 42 reloads = 84)
2678 :     TOTAL 129.42u 5.10s 4.13g
2679 :    
2680 :     3000 CPS 100 spill 675.59u 19.03s 11.64g
2681 :     MLRISC ra 2.69u 0.27s 1.38g
2682 :     (spills = 62 reloads = 124)
2683 :     TOTAL 682.48u 19.61s 13.99g
2684 :    
2685 :     4000 CPS 100 spill 2362.82u 56.28s 43.60g
2686 :     MLRISC ra 4.96u 0.27s 2.72g
2687 :     (spills = 85 reloads = 170)
2688 :     TOTAL 2375.26u 57.21s 48.00g
2689 :    
2690 :     As you can see the old cps spill module suffers from some serious
2691 : leunga 1115 performance problem. But since I cannot decipher the old code fully,
2692 :     instead of patching the problems up, I'm reimplementing it
2693 : leunga 1094 with a different algorithm. The new code is more modular,
2694 :     smaller when compiled, and substantially faster
2695 :     (O(n log n) time and O(n) space). Timing of the new spill module:
2696 :    
2697 :     4000 CPS 100 spill 0.02u 0.00s 0.00g
2698 :     MLRISC ra 0.25u 0.02s 0.15g
2699 :     (spills=1 reloads=3)
2700 :     TOTAL 7.74u 0.34s 1.62g
2701 :    
2702 :     Implementation details:
2703 :    
2704 :     As far as I can tell, the purpose of the CPS spill module is to make sure the
2705 :     number of live variables at any program point (the bandwidth)
2706 :     does not exceed a certain limit, which is determined by the
2707 :     size of the spill area.
2708 :    
2709 :     When the bandwidth is too large, we decrease the register pressure by
2710 :     packing live variables into spill records. How we achieve this is
2711 :     completely different than what we did in the old code.
2712 :    
2713 :     First, there is something about the MLRiscGen code generator
2714 :     that we should be aware of:
2715 :    
2716 :     o MLRiscGen performs code motion!
2717 :    
2718 :     In particular, it will move floating point computations and
2719 :     address computations involving only the heap pointer to
2720 :     their use sites (if there is only a single use).
2721 :     What this means is that if we have a CPS record construction
2722 :     statement
2723 :    
2724 :     RECORD(k,vl,w,e)
2725 :    
2726 :     we should never count the new record address w as live if w
2727 :     has only one use (which is often the case).
2728 :    
2729 :     We should do something similar to floating point, but the transformation
2730 :     there is much more complex, so I won't deal with that.
2731 :    
2732 :     Secondly, there are now two new cps primops at our disposal:
2733 :    
2734 :     1. rawrecord of record_kind option
2735 :     This pure operator allocates some uninitialized storage from the heap.
2736 :     There are two forms:
2737 :    
2738 :     rawrecord NONE [INT n] allocates a tagless record of length n
2739 :     rawrecord (SOME rk) [INT n] allocates a tagged record of length n
2740 :     and initializes the tag.
2741 :    
2742 :     2. rawupdate of cty
2743 :     rawupdate cty (v,i,x)
2744 :     Assigns to x to the ith component of record v.
2745 :     The storelist is not updated.
2746 :    
2747 :     We use these new primops for both spilling and increment record construction.
2748 :    
2749 :     1. Spilling.
2750 :    
2751 :     This is implemented with a linear scan algorithm (but generalized
2752 :     to trees). The algorithm will create a single spill record at the
2753 :     beginning of the cps function and use rawupdate to spill to it,
2754 :     and SELECT or SELp to reload from it. So both spills and reloads
2755 :     are fine-grain operations. In contrast, in the old algorithm
2756 :     "spills" have to be bundled together in records.
2757 :    
2758 :     Ideally, we should sink the spill record construction to where
2759 :     it is needed. We can even split the spill record into multiple ones
2760 :     at the places where they are needed. But CPS is not a good
2761 :     representation for global code motion, so I'll keep it simple and
2762 :     am not attempting this.
2763 :    
2764 :     2. Incremental record construction (aka record splitting).
2765 :    
2766 :     Long records with many component values which are simulatenously live
2767 :     (recall that single use record addresses are not considered to
2768 :     be live) are constructed with rawrecord and rawupdate.
2769 :     We allocate space on the heap with rawrecord first, then gradually
2770 :     fill it in with rawupdate. This is the technique suggested to me
2771 :     by Matthias.
2772 :    
2773 :     Some restrictions on when this is applicable:
2774 :     1. It is not a VECTOR record. The code generator currently does not handle
2775 :     this case. VECTOR record uses double indirection like arrays.
2776 :     2. All the record component values are defined in the same "basic block"
2777 :     as the record constructor. This is to prevent speculative
2778 :     record construction.
2779 :    
2780 :     ----------------------------------------------------------------------
2781 :     Name: Allen Leung
2782 : leunga 1086 Date: 2002/02/22 01:02:00 EST
2783 :     Tag: leunga-20020222-mlrisc-tools
2784 :    
2785 :     Minor bug fixes in the parser and rewriter
2786 :    
2787 :     ----------------------------------------------------------------------
2788 :     Name: Allen Leung
2789 : leunga 1085 Date: 2002/02/21 20:20:00 EST
2790 :     Tag: leunga-20020221-peephole
2791 :    
2792 :     Regenerated the peephole files. Some contained typos in the specification
2793 :     and some didn't compile because of pretty printing bugs in the old version
2794 :     of 'nowhere'.
2795 :    
2796 :     ----------------------------------------------------------------------
2797 :     Name: Allen Leung
2798 : leunga 1079 Date: 2002/02/19 20:20:00 EST
2799 :     Tag: leunga-20020219-mlrisc-tools
2800 :     Description:
2801 :    
2802 :     Minor bug fixes to the mlrisc-tools library:
2803 :    
2804 :     1. Fixed up parsing colon suffixed keywords
2805 :     2. Added the ability to shut the error messages up
2806 :     3. Reimplemented the pretty printer and fixed up/improved
2807 :     the pretty printing of handle and -> types.
2808 :     4. Fixed up generation of literal symbols in the nowhere tool.
2809 :     5. Added some SML keywords to to sml.sty
2810 :    
2811 :     ----------------------------------------------------------------------
2812 :     Name: Matthias Blume
2813 : blume 1078 Date: 2002/02/19 16:20:00 EST
2814 :     Tag: blume-20020219-cmffi
2815 :     Description:
2816 :    
2817 :     A wild mix of changes, some minor, some major:
2818 :    
2819 :     * All C FFI-related libraries are now anchored under $c:
2820 :     $/c.cm --> $c/c.cm
2821 :     $/c-int.cm --> $c/internals/c-int.cm
2822 :     $/memory.cm --> $c/memory/memory.cm
2823 :    
2824 :     * "make" tool (in CM) now treats its argument pathname slightly
2825 :     differently:
2826 :     1. If the native expansion is an absolute name, then before invoking
2827 :     the "make" command on it, CM will apply OS.Path.mkRelative
2828 :     (with relativeTo = OS.FileSys.getDir()) to it.
2829 :     2. The argument will be passed through to subsequent phases of CM
2830 :     processing without "going native". In particular, if the argument
2831 :     was an anchored path, then "make" will not lose track of that anchor.
2832 :    
2833 :     * Compiler backends now "know" their respective C calling conventions
2834 :     instead of having to be told about it by ml-nlffigen. This relieves
2835 :     ml-nlffigen from one of its burdens.
2836 :    
2837 :     * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
2838 :    
2839 :     * Export C_DEBUG and C_Debug from $c/c.cm.
2840 :    
2841 :     * C type encoding in ml-nlffi-lib has been improved to model the conceptual
2842 :     subtyping relationship between incomplete pointers and their complete
2843 :     counterparts. For this, ('t, 'c) ptr has been changed to 'o ptr --
2844 :     with the convention of instantiating 'o with ('t, 'c) obj whenever
2845 :     the pointer target type is complete. In the incomplete case, 'o
2846 :     will be instantiated with some "'c iobj" -- a type obtained by
2847 :     using one of the functors PointerToIncompleteType or PointerToCompleteType.
2848 :    
2849 :     Operations that work on both incomplete and complete pointer types are
2850 :     typed as taking an 'o ptr while operations that require the target to
2851 :     be known are typed as taking some ('t, 'c) obj ptr.
2852 :    
2853 :     voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
2854 :     where void is an eqtype without any values. This makes it possible
2855 :     to work on voidptr values using functions meant to operate on light
2856 :     incomplete pointers.
2857 :    
2858 :     * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
2859 :     been vastly simplified.
2860 :    
2861 :     ----------------------------------------------------------------------
2862 :     Name: Matthias Blume
2863 : blume 1076 Date: 2002/02/19 10:48:00 EST
2864 :     Tag: blume-20020219-pqfix
2865 :     Description:
2866 :    
2867 :     Applied Chris Okasaki's bug fix for priority queues.
2868 :    
2869 :     ----------------------------------------------------------------------
2870 :     Name: Matthias Blume
2871 : blume 1073 Date: 2002/02/15 17:05:00
2872 : blume 1069 Tag: Release_110_39
2873 :     Description:
2874 :    
2875 : blume 1073 Last-minute retagging is becoming a tradition... :-(
2876 :    
2877 :     This is the working release 110.39.
2878 :    
2879 :     ----------------------------------------------------------------------
2880 :     Name: Matthias Blume
2881 :     Date: 2002/02/15 16:00:00 EST
2882 :     Tag: Release_110_39-orig
2883 :     Description:
2884 :    
2885 : blume 1069 Working release 110.39. New bootfiles.
2886 :    
2887 : blume 1073 (Update: There was a small bug in the installer so it wouldn't work
2888 :     with all shells. So I retagged. -Matthias)
2889 :    
2890 : blume 1069 ----------------------------------------------------------------------
2891 :     Name: Matthias Blume
2892 : blume 1068 Date: 2002/02/15 14:17:00 EST
2893 :     Tag: blume-20020215-showbindings
2894 :     Description:
2895 :    
2896 :     Added EnvRef.listBoundSymbols and CM.State.showBindings. Especially
2897 :     the latter can be useful for exploring what bindings are available at
2898 :     the interactive prompt. (The first function returns only the list
2899 :     of symbols that are really bound, the second prints those but also the
2900 :     ones that CM's autoloading mechanism knows about.)
2901 :    
2902 :     ----------------------------------------------------------------------
2903 :     Name: Matthias Blume
2904 : blume 1067 Date: 2002/02/15 12:08:00 EST
2905 :     Tag: blume-20020215-iptrs
2906 :     Description:
2907 :    
2908 :     Two improvements to ml-nlffigen:
2909 :    
2910 :     1. Write files only if they do not exist or if their current contents
2911 :     do not coincide with what's being written. (That is, avoid messing
2912 :     with the time stamps unless absolutely necessary.)
2913 :    
2914 :     2. Implement a "repository" mechanism for generated files related
2915 :     to "incomplete pointer types". See the README file for details.
2916 :    
2917 :     ----------------------------------------------------------------------
2918 :     Name: Matthias Blume
2919 : blume 1066 Date: 2002/02/14 11:50:00 EST
2920 :     Tag: blume-20020214-quote
2921 :     Description:
2922 :    
2923 :     Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm). This is required
2924 :     because of the new and improved tag generation scheme. (Thanks to Allen
2925 :     Leung for pointing it out.)
2926 :    
2927 :     ----------------------------------------------------------------------
2928 : george 1065 Name: Lal George
2929 :     Date: 2002/02/14 09:55:27 EST 2002
2930 :     Tag: george-20020214-isabelle-bug
2931 :     Description:
2932 :    
2933 :     Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
2934 :     of Isabelle on the x86.
2935 :    
2936 :     From Allen:
2937 :     -----------
2938 :     I've found the problem:
2939 :    
2940 :     in ra-core.sml, I use the counter "blocked" to keep track of the
2941 :     true number of elements in the freeze queue. When the counter goes
2942 :     to zero, I skip examining the queue. But I've messed up the
2943 :     bookkeeping in combine():
2944 :    
2945 :     else ();
2946 :     case !ucol of
2947 :     PSEUDO => (if !cntv > 0 then
2948 :     (if !cntu > 0 then blocked := !blocked - 1 else ();
2949 :     ^^^^^^^^^^^^^^^^^^^^^^^
2950 :     moveu := mergeMoveList(!movev, !moveu)
2951 :     )
2952 :     else ();
2953 :    
2954 :     combine() is called to coalesce two nodes u and v.
2955 :     I think I was thinking that if the move counts of u and v are both
2956 :     greater than zero then after they are coalesced then one node is
2957 :     removed from the freeze queue. Apparently I was thinking that
2958 :     both u and v are of low degree, but that's clearly not necessarily true.
2959 :    
2960 :    
2961 :     02/12/2002:
2962 :     Here's the patch. HOL now compiles.
2963 :    
2964 :     I don't know how this impact on performance (compile
2965 :     time or runtime). This bug caused the RA (especially on the x86)
2966 :     to go thru the potential spill phase when there are still nodes on the
2967 :     freeze queue.
2968 :    
2969 :    
2970 :    
2971 :    
2972 :     ----------------------------------------------------------------------
2973 : blume 1058 Name: Matthias Blume
2974 : blume 1064 Date: 2002/02/13 22:40:00 EST
2975 :     Tag: blume-20020213-fptr-rtti
2976 :     Description:
2977 :    
2978 :     Fixed a bug in ml-nlffigen that was introduced with one of the previous
2979 :     updates.
2980 :    
2981 :     ----------------------------------------------------------------------
2982 :     Name: Matthias Blume
2983 : blume 1063 Date: 2002/02/13 16:41:00 EST
2984 :     Tag: blume-20020213-cmlpq
2985 :     Description:
2986 :    
2987 :     Added new priority queue export symbols (which have just been added to
2988 :     smlnj-lib.cm) to CML's version of smlnj-lib.cm. (Otherwise CML would
2989 :     not compile and the installer would choke.)
2990 :    
2991 :     ----------------------------------------------------------------------
2992 :     Name: Matthias Blume
2993 : blume 1062 Date: 2002/02/13 16:15:00 EST
2994 :     Tag: blume-20020213-various
2995 :     Description:
2996 :    
2997 :     1. More tweaks to ml-nlffigen:
2998 :    
2999 :     - better internal datastructures (resulting in slight speedup)
3000 :     - "-match" option requires exact match
3001 :     - "localized" gensym counters (untagged structs/unions nested within
3002 :     other structs/unions or within typedefs get a fresh counter; their
3003 :     tag will be prefixed by a concatenation of their parents' tags)
3004 :     - bug fixes (related to calculation of transitive closure of types
3005 :     to be included in the output)
3006 :    
3007 :     2. Minor Basis updates:
3008 :    
3009 :     - added implementations for List.collate and Option.app
3010 :    
3011 :     ----------------------------------------------------------------------
3012 :     Name: Matthias Blume
3013 : blume 1060 Date: 2002/02/11 15:55:00 EST
3014 :     Tag: blume-20020211-gensym
3015 :     Description:
3016 :    
3017 :     Added a "-gensym" option to command line of ml-nlffigen. This can be
3018 :     used to specify a "stem" -- a string that is inserted in all "gensym'd"
3019 :     names (ML structure names that correspond to unnamed C structs, unions,
3020 :     and enums), so that separate runs of ml-nlffigen do not clash.
3021 :    
3022 :     ----------------------------------------------------------------------
3023 :     Name: Matthias Blume
3024 : blume 1059 Date: 2002/02/11 12:05:00 EST
3025 :     Tag: blume-20020211-gensml
3026 :     Description:
3027 :    
3028 :     A quick fix for a problem with GenSML (in the pgraph-util library):
3029 :     Make generation of toplevel "local" optional. (Strictly speaking,
3030 :     signature definitions within "local" are not legal SML.)
3031 :    
3032 :     Other than that: updates to INSTALL and cm/TODO.
3033 :    
3034 :     ----------------------------------------------------------------------
3035 :     Name: Matthias Blume
3036 : blume 1058 Date: 2002/02/08 15:00:00 EST
3037 :     Tag: blume-20020208-uniquepid
3038 :     Description:
3039 :    
3040 :     0. Version number has been bumped to 110.38.1. NEW BOOTFILES!!!
3041 :    
3042 :     1. The installer (config/install.sh) has gotten smarter:
3043 :    
3044 :     - Configuration options are a bit easier to specify now
3045 :     (in config/targets).
3046 :     - Bug in recognizing .tar.bz2 files fixed.
3047 :     - Installer automatically resolves dependencies between
3048 :     configuration options (e.g., if you ask for eXene, you will
3049 :     also get cml -- regardless whether you asked for it or not).
3050 :     - Installer can run in "quieter mode" by setting the environment
3051 :     variable INSTALL_QUIETLY to "true". "Quieter" does not mean
3052 :     "completely silent", though.
3053 :     - Build HashCons library as part of smlnj-lib.
3054 :    
3055 :     2. A new scheme for assigning persistent identifiers to compilation
3056 :     units (and, by extension, to types etc.) has been put into place.
3057 :     This fixes a long-standing bug where types and even dynamic values
3058 :     can get internally confused, thereby compromising type safety
3059 :     (abstraction) and dynamic correctness. See
3060 :    
3061 :     http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
3062 :    
3063 :     for an example of how things could go wrong until now.
3064 :    
3065 :     The downside of the new scheme is that pids are not quite as
3066 :     persistent as they used to be: CM will generate a fresh pid
3067 :     for every compilation unit that it thinks it sees for the first
3068 :     time. That means that if you compile starting from a clean, fresh
3069 :     source tree at two different times, you end up with different
3070 :     binaries.
3071 :    
3072 :     Cutoff recompilation, however, has not been compromised because
3073 :     CM keeps pid information in special caches between runs.
3074 :    
3075 :     ----------------------------------------------------------------------
3076 : george 1053 Name: Lal George
3077 : george 1055 Date: 2002/02/07 15:34:13 EST 2002
3078 :     Tag: <none>
3079 :     Description:
3080 :    
3081 :     Compilers that generate assembly code may produce global labels
3082 :     whose value is resolved at link time. The various peephole optimization
3083 :     modules did not take this in account.
3084 :    
3085 :     TODO. The Labels.addrOf function should really return an option
3086 :     type so that clients are forced to deal with this issue, rather
3087 :     than an exception being raised.
3088 :    
3089 :     ----------------------------------------------------------------------
3090 :     Name: Lal George
3091 : george 1053 Date: 2002/02/06 13:55:02 EST
3092 :     Tag: george-20020206-ra-breakup
3093 :     Description:
3094 :    
3095 :     1. A bug fix from Allen.
3096 :    
3097 :     A typo causes extra fstp %st(0)'s to be generated at compensation
3098 :     edges, which might cause stack underflow traps at runtime. This
3099 :     occurs in fft where there are extraneous fstps right before the 'into'
3100 :     trap instruction (in this case they are harmless since none of the
3101 :     integers overflow.)
3102 :    
3103 : george 1054 2. Pulled out various utility modules that were embedded in the modules
3104 : george 1053 of the register allocator. I need these modules for other purposes, but
3105 : george 1054 they are not complete enough to put into a library (just yet).
3106 : george 1053 ----------------------------------------------------------------------
3107 : blume 1049 Name: Matthias Blume
3108 :     Date: 2002/01/31 16:05:00 EST
3109 :     Tag: blume-20020131-sparc-ccalls
3110 :     Description:
3111 :    
3112 :     1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
3113 :     of extra stack space by mistake. Fixed.
3114 :    
3115 :     2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
3116 :    
3117 :     ----------------------------------------------------------------------
3118 : leunga 1048 Name: Allen Leung
3119 :     Date: 2002/01/30
3120 :     Tag: leunga-20020130-nowhere-bug-fix
3121 :     Description:
3122 :    
3123 :     MLRISC bug fixes:
3124 :     1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
3125 :     2. MachineInt.fromString was negating its value.
3126 :    
3127 :     ----------------------------------------------------------------------
3128 : blume 1039 Name: Matthias Blume
3129 : blume 1047 Date: 2002/01/29
3130 :     Tag: blume-20020129-INSTALL
3131 :     Description:
3132 :    
3133 :     - Added somewhat detailed installation instructions (file INSTALL).
3134 :     - Fixed curl-detection bug in config/install.sh.
3135 :     - It is now possible to select the URL getter using the URLGETTER
3136 :     environment variable:
3137 :    
3138 :     not set / "unknown" --> automatic detection (script tries wget,
3139 :     curl, and lynx)
3140 :     "wget" / "curl" / "lynx" --> use the specified program (script "knows"
3141 :     how to properly invoke them)
3142 :     other --> use $URLGETTER directly, it must take
3143 :     precisely two command-line arguments
3144 :     (source URL and destination file name)
3145 :    
3146 :     ----------------------------------------------------------------------
3147 :     Name: Matthias Blume
3148 : blume 1044 Date: 2002/01/28
3149 : blume 1045 Tag: blume-20020128-sparc-ccalls
3150 :     Description:
3151 :    
3152 :     - Fixed problem with calculation of "used" registers in sparc-c-calls.
3153 :     - Make use of the allocParam argument in sparc-c-calls.
3154 :    
3155 :     ----------------------------------------------------------------------
3156 :     Name: Matthias Blume
3157 :     Date: 2002/01/28
3158 : blume 1044 Tag: blume-20020128-allocParam
3159 :     Description:
3160 :    
3161 :     John Reppy: Changes c-calls API to accept client-callback for
3162 :     allocating extra stack space.
3163 :     me: Corresponding changes to mlriscGen (using a dummy argument that
3164 :     does not change the current behavior).
3165 :    
3166 :     ----------------------------------------------------------------------
3167 :     Name: Matthias Blume
3168 : blume 1040 Date: 2002/01/28 12:00:00
3169 :     Tag: Release_110_38
3170 :     Description:
3171 :    
3172 :     This time for real!!!
3173 :    
3174 :     ----------------------------------------------------------------------
3175 :     Name: Matthias Blume
3176 : blume 1039 Date: 2002/01/28 10:56:00 EST
3177 :     Tag: blume-20020128-retraction
3178 :     Description:
3179 :    
3180 :     0. Retracted earlier 110.38. (The Release_110_38 tag has been replaced
3181 :     with blume-Release_110_38-retracted.)
3182 :    
3183 :     1. Fixed a problem with incorrect rounding modes in real64.sml.
3184 :     (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
3185 :    
3186 :     2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
3187 :     and enums fixed. The naming of corresponding ML identifiers should
3188 :     now be consistent again.
3189 :    
3190 :     ----------------------------------------------------------------------
3191 : leunga 1038 Name: Allen Leung
3192 :     Date: 2002/01/27
3193 :     Tag: leunga-20020127-nowhere
3194 :     Description:
3195 :    
3196 :     Added a target called nowhere in the configuration scripts.
3197 :     Enabling this will build the MLRISC 'nowhere' tool (for translating
3198 :     programs with where-clauses into legal SML code) during installation.
3199 :    
3200 :     ----------------------------------------------------------------------
3201 : blume 1036 Name: Matthias Blume
3202 : leunga 1038 Date: 2002/01/25 21:27:00 EST
3203 : blume 1039 Tag: blume-Release_110_38-retracted
3204 : blume 1037 Description:
3205 :    
3206 :     Call it a (working) release! Version is 110.38. Bootfiles are ready.
3207 :    
3208 :     README will be added later.
3209 :    
3210 : blume 1039 !!! NOTE: Re-tagged as blume-Release_110_38-retracted. Original tag
3211 :     (Release_110_38) removed. Reason: Last-minute bug fixes.
3212 :    
3213 : blume 1037 ----------------------------------------------------------------------
3214 :     Name: Matthias Blume
3215 : blume 1036 Date: 2002/01/25
3216 :     Tag: blume-20020125-ffi
3217 :     Description:
3218 :    
3219 :     A large number of tweaks and improvements to ml-nlffi-lib and
3220 :     ml-nlffigen:
3221 :    
3222 :     - ML represenation types have been streamlined
3223 :     - getter and setter functions work with concrete values, not abstract
3224 :     ones where possible
3225 :     - ml-nlffigen command line more flexible (see README file there)
3226 :     - some bugs have been fixed (hopefully)
3227 :    
3228 :     ----------------------------------------------------------------------
3229 : george 1034 Name: Lal George
3230 :     Date: 2002/01/24
3231 :     Tag: george-20020124-risc-ra-interface
3232 :     Description:
3233 :    
3234 :     There is a dramatic simplification in the interface to the
3235 :     register allocator for RISC architectures as a result of making
3236 :     parallel copy instructions explicit.
3237 :    
3238 :     ----------------------------------------------------------------------
3239 : blume 1028 Name: Matthias Blume
3240 : blume 1032 Date: 2002/01/22
3241 :     Tag: blume-20020122-x86-ccalls
3242 :     Description:
3243 :    
3244 :     Bug fix for c-calls on x86 (having to do with how char- and
3245 :     short-arguments are being handled).
3246 :    
3247 :     ----------------------------------------------------------------------
3248 :     Name: Matthias Blume
3249 : blume 1031 Date: 2002/01/21
3250 :     Tag: blume-20020121-ff
3251 :     Description:
3252 :    
3253 :     Another day of fiddling with the FFI...
3254 :    
3255 :     1. Bug fix/workaround: CKIT does not complain about negative array
3256 :     dimensions, so ml-nlffigen has to guard itself against this possibility.
3257 :     (Otherwise a negative dimension would send it into an infinite loop.)
3258 :    
3259 :     2. Some of the abstract types (light objects, light pointers, most "base"
3260 :     types) in structure C are now eqtypes.
3261 :    
3262 :     3. Added constructors and test functions for NULL function pointers.
3263 :    
3264 :     ----------------------------------------------------------------------
3265 :     Name: Matthias Blume
3266 : blume 1028 Date: 2002/01/18
3267 : blume 1030 Tag: blume-20020118-ready-for-new-release
3268 :     Description:
3269 :    
3270 :     Made config/srcarchiveurl point to a new place. (Will provide boot
3271 :     files shortly.)
3272 :    
3273 :     Maybe we christen this to be 110.38?
3274 :    
3275 :     ----------------------------------------------------------------------
3276 :     Name: Matthias Blume
3277 :     Date: 2002/01/18
3278 : blume 1029 Tag: blume-20020118-more-ffifiddle
3279 :     Description:
3280 :    
3281 :     Today's FFI fiddling:
3282 :    
3283 :     - Provided a structure CGetSet with "convenient" versions of C.Get.* and
3284 :     C.Set.* that use concrete (MLRep.*) arguments and results instead
3285 :     of abstract ones.
3286 :    
3287 :     - Provided word-style bit operations etc. for "int" representation
3288 :     types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
3289 :     and Long.
3290 :    
3291 :     ----------------------------------------------------------------------
3292 :     Name: Matthias Blume
3293 :     Date: 2002/01/18
3294 : blume 1028 Tag: blume-20020118-use-x86-fp
3295 :     Description:
3296 :    
3297 :     Now that x86-fast-fp seems to be working, I turned it back on again
3298 :     by default. (Seems to work fine now, even with the FFI.)
3299 :    
3300 :     Other than that, I added some documentation about the FFI to
3301 :     src/ml-nlffigen/README and updated the FFI test examples in
3302 :     src/ml-nlffi-lib/Tests/*.
3303 :    
3304 :     ----------------------------------------------------------------------
3305 : leunga 1025 Name: Allen Leung
3306 :     Date: 2002/01/17
3307 :     Tag: leunga-20020117-x86-fast-fp-call
3308 :     Description:
3309 :    
3310 :     1. Fixed a problem with handling return fp values when x86's fast fp
3311 :     mode is turned on.
3312 :    
3313 :     2. Minor pretty printing fix for cellset. Print %st(0) as %st(0) instead
3314 :     of %f32.
3315 :    
3316 :     3. Added a constructor INT32lit to the ast of MLRISC tools.
3317 :    
3318 :     ----------------------------------------------------------------------
3319 : blume 1021 Name: Matthias Blume
3320 :     Date: 2002/01/16
3321 :     Tag: blume-20020116-ffifiddle
3322 :     Description:
3323 :    
3324 :     More fiddling with the FFI interface:
3325 :    
3326 :     - Make constness 'c instead of rw wherever possible. This eliminates
3327 :     the need for certain explicit coercions. (However, due to ML's
3328 :     value polymorphism, there will still be many cases where explicit
3329 :     coercions are necessary. Phantom types are not the whole answer
3330 :     to modeling a subtyping relationship in ML.)
3331 :    
3332 :     - ro/rw coersions for pointers added. (Avoids the detour through */&.)
3333 :    
3334 :     - "printf" test example added to src/ml-nlffi-lib/Tests. (Demonstrates
3335 :     clumsy workaround for varargs problem.)
3336 :    
3337 :     ----------------------------------------------------------------------
3338 : george 1016 Name: Lal George
3339 :     Date: 2002/01/15
3340 :     Tag: <none>
3341 :     Description:
3342 :    
3343 :     1. Since COPY instructions are no longer native to the architecture,
3344 :     a generic functor can be used to implement the expandCopies function.
3345 :    
3346 :     2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
3347 :     TEXT segment.
3348 :    
3349 :     ----------------------------------------------------------------------
3350 : blume 1011 Name: Matthias Blume
3351 : blume 1015 Date: 2002/01/15
3352 :     Tag: blume-20020115-ffiupdates
3353 :     Description:
3354 :    
3355 :     1. Fix for bug resulting in single-precision float values being returned
3356 :     incorrectly from FFI calls.
3357 :    
3358 :     2. Small modifications to C FFI API:
3359 :    
3360 :     - memory-allocation routines return straight objects (no options)
3361 :     and raise an exception in out-of-memory situations
3362 :     - unsafe extensions to cast between function pointers and pointers
3363 :     from/to ints
3364 :     - added structure C_Debug as an alternative to structure C where
3365 :     pointer-dereferencing (|*| and |*!) always check for null-pointers
3366 :     - added open_lib' to DynLinkage; open_lib' works like open_lib
3367 :     but also takes a (possibly empty) list of existing library handles
3368 :     that the current library depends on
3369 :    
3370 :     ----------------------------------------------------------------------
3371 :     Name: Matthias Blume
3372 : blume 1011 Date: 2002/01/10
3373 :     Tag: blume-20020110-newffigen
3374 :     Description:
3375 :    
3376 :     1. Updates to portable graph code.
3377 :    
3378 :     2. Major update to ml-nlffigen and ml-nlffi-lib. Things are much
3379 :     more scalable now so that even huge interfaces such as the one
3380 :     for GTK compile in finite time and space. :-)
3381 :     See src/ml-nlffigen/README for details on what's new.
3382 :    
3383 :     ----------------------------------------------------------------------
3384 : george 1003 Name: Lal George
3385 : george 1009 Date: 2001/01/09 14:31:35 EST 2002
3386 :     Tag: george-20011206-rm-native-copy
3387 :     Description:
3388 :    
3389 :     Removed the native COPY and FCOPY instructions
3390 :     from all the architectures and replaced it with the
3391 :     explicit COPY instruction from the previous commit.
3392 :    
3393 :     It is now possible to simplify many of the optimizations
3394 :     modules that manipulate copies. This has not been
3395 :     done in this change.
3396 :    
3397 :     ----------------------------------------------------------------------
3398 :     Name: Lal George
3399 : george 1003 Date: 2001/12/06 16:50:13 EST 2001
3400 :     Tag: george-20011206-mlrisc-instruction
3401 :     Description:
3402 : george 824
3403 : george 1003 Changed the representation of instructions from being fully abstract
3404 :     to being partially concrete. That is to say:
3405 :    
3406 :     from
3407 :     type instruction
3408 :    
3409 :     to
3410 :     type instr (* machine instruction *)
3411 :    
3412 :     datatype instruction =
3413 :     LIVE of {regs: C.cellset, spilled: C.cellset}
3414 :     | KILL of {regs: C.cellset, spilled: C.cellset}
3415 :     | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
3416 :     | ANNOTATION of {i: instruction, a: Annotations.annotation}
3417 :     | INSTR of instr
3418 :    
3419 :     This makes the handling of certain special instructions that appear on
3420 :     all architectures easier and uniform.
3421 :    
3422 :     LIVE and KILL say that a list of registers are live or killed at the
3423 :     program point where they appear. No spill code is generated when an
3424 :     element of the 'regs' field is spilled, but the register is moved to
3425 :     the 'spilled' (which is present, more for debugging than anything else).
3426 :    
3427 :     LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
3428 :     We used to generate:
3429 :    
3430 :     DEFFREG f1
3431 :     f1 := f2 + f3
3432 :     trapb
3433 :    
3434 :     but now generate:
3435 :    
3436 :     f1 := f2 + f3
3437 :     trapb
3438 :     LIVE {regs=[f1,f2,f3], spilled=[]}
3439 :    
3440 :     Furthermore, the DEFFREG (hack) required that all floating point instruction
3441 :     use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
3442 :     defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
3443 :     in a cleaner alpha implementation. (Hopefully, intel will not get rid of
3444 :     this architecture).
3445 :    
3446 :     COPYXXX is intended to replace the parallel COPY and FCOPY available on
3447 :     all the architectures. This will result in further simplification of the
3448 :     register allocator that must be aware of them for coalescing purposes, and
3449 :     will also simplify certain aspects of the machine description that provides
3450 :     callbacks related to parallel copies.
3451 :    
3452 :     ANNOTATION should be obvious, and now INSTR represents the honest to God
3453 :     machine instruction set!
3454 :    
3455 :     The <arch>/instructions/<arch>Instr.sml files define certain utility
3456 :     functions for making porting easier -- essentially converting upper case
3457 :     to lower case. All machine instructions (of type instr) are in upper case,
3458 :     and the lower case form generates an MLRISC instruction. For example on
3459 :     the alpha we have:
3460 :    
3461 :     datatype instr =
3462 :     LDA of {r:cell, b:cell, d:operand}
3463 :     | ...
3464 :    
3465 :     val lda : {r:cell, b:cell, d:operand} -> instruction
3466 :     ...
3467 :    
3468 :     where lda is just (INSTR o LDA), etc.
3469 :    
3470 : leunga 815 ----------------------------------------------------------------------
3471 : blume 986 Name: Matthias Blume
3472 : blume 993 Date: 2001/11/22 21:40:00 EST
3473 :     Tag: Release_110_37
3474 :     Description:
3475 :    
3476 :     Release 110.37. This time for real.
3477 :    
3478 :     ----------------------------------------------------------------------
3479 :     Name: Matthias Blume
3480 : blume 988 Date: 2001/11/21 16:35:00 EST
3481 :     Tag: blume-20011121-foot-in-mouth
3482 :     Description:
3483 :    
3484 :     Removed the "Release_110_37" tag because of a serious bug.
3485 :     This will be re-tagged once the bug is fixed.
3486 :    
3487 :     ----------------------------------------------------------------------
3488 :     Name: Matthias Blume
3489 : blume 987 Date: 2001/11/21 16:14:00 EST
3490 :     Tag: blume-20011121-forgottenfile
3491 :     Description:
3492 :    
3493 :     Forgot to add a file. (Just a .tex-file -- part of
3494 :     the CM manual source.)
3495 :    
3496 :     ----------------------------------------------------------------------
3497 :     Name: Matthias Blume
3498 : blume 986 Date: 2001/11/21 16:10:00 EST
3499 : blume 988 Tag: blume-20011121-invalid_110_37
3500 : blume 986 Description:
3501 :    
3502 : blume 988 Note: I removed the original tag "Release_110_37" from this commit
3503 :     because we found a serious bug in all non-x86 backends.
3504 :     - Matthias
3505 :    
3506 : blume 986 1. Modifications to the SML/NJ code generator and to the runtime system
3507 :     so that code object name strings are directly inserted into code
3508 :     objects at code generation time. The only business the runtime system
3509 :     has with this is now to read the name strings on occasions.
3510 :     (The encoding of the name string has also changed somewhat.)
3511 :    
3512 :     2. CM now implements a simple "set calculus" for specifying export lists.
3513 :     In particular, it is now possible to refer to the export lists of
3514 :     other libraries/groups/sources and form unions as well as differences.
3515 :     See the latest CM manual for details.
3516 :    
3517 :     3. An separate notion of "proxy" libraries has again be eliminated from
3518 :     CM's model. (Proxy libraries are now simply a special case of using
3519 :     the export list calculus.)
3520 :    
3521 :     4. Some of the existing libraries now take advantage of the new set
3522 :     calculus.
3523 :     (Notice that not all libraries have been converted because some
3524 :     of the existing .cm-files are supposed to be backward compatible
3525 :     with 110.0.x.)
3526 :    
3527 :     5. Some cleanup in stand-alone programs. (Don't use "exnMessage" -- use
3528 :     "General.exnMessage"! The former relies on a certain hook to be
3529 :     initialized, and that often does not happen in the stand-alone case.)
3530 :    
3531 :     ----------------------------------------------------------------------
3532 : george 984 Name: Lal George
3533 :     Date: 2001/11/21 13:56:18 EST
3534 :     Tag: george-2001121-pseudo-ops
3535 :     Description:
3536 :    
3537 :     Implemented a complete redesign of MLRISC pseudo-ops. Now there
3538 :     ought to never be any question of incompatabilities with
3539 :     pseudo-op syntax expected by host assemblers.
3540 :    
3541 :     For now, only modules supporting GAS syntax are implemented
3542 :     but more should follow, such as MASM, and vendor assembler
3543 :     syntax, e.g. IBM as, Sun as, etc.
3544 :    
3545 :     ----------------------------------------------------------------------
3546 : blume 977 Name: Matthias Blume
3547 :     Date: 2001/11/14 11:52:00 EST
3548 :     Tag: blume-20011114-srcname
3549 :     Description:
3550 :    
3551 :     1. Routed the name of the current source file to mlriscgen where it
3552 :     should be directly emitted into the code object. (This last part
3553 :     is yet to be done.)
3554 :    
3555 :     2. Some cleanup of the pgraph code to make it match the proposal that
3556 :     I put out the other day. (The proposal notwithstanding, things are
3557 :     still in flux here.)
3558 :    
3559 :     ----------------------------------------------------------------------
3560 : george 976 Name: Lal George
3561 :     Date: 2001/11/14 09:44:04 EST
3562 :     Tag:
3563 :     Description:
3564 :    
3565 :     Fix for a backpatching bug reported by Allen.
3566 :    
3567 :     Because the boundary between short and long span-dependent
3568 :     instructions is +/- 128, there are an astounding number of
3569 :     span-dependent instructions whose size is over estimated.
3570 :    
3571 :     Allen came up with the idea of letting the size of span
3572 :     dependent instructions be non-monotonic, for a maxIter
3573 :     number of times, after which the size must be monotonically
3574 :     increasing.
3575 :    
3576 :     This table shows the number of span-dependent instructions
3577 :     whose size was over-estimated as a function of maxIter, for the
3578 :     file Parse/parse/ml.grm.sml:
3579 :    
3580 :     maxIter # of instructions:
3581 :     10 687
3582 :     20 438
3583 :     30 198
3584 :     40 0
3585 :    
3586 :     In compiling the compiler, there is no significant difference in
3587 :     compilation speed between maxIter=10 and maxIter=40. Actually,
3588 :     my measurements showed that maxIter=40 was a tad faster than
3589 :     maxIter=10! Also 96% of the files in the compiler reach a fix
3590 :     point within 13 iterations, so fixing maxIter at 40, while high,
3591 :     is okay.
3592 :    
3593 :     ----------------------------------------------------------------------
3594 : blume 975 Name: Matthias Blume
3595 :     Date: 2001/10/31 15:25:00 EST
3596 :     Tag: blume-20011031-pgraph
3597 :     Description:
3598 :    
3599 :     CKIT:
3600 :     * Changed the "Function" constructor of type Ast.ctype to carry optional
3601 :     argument identifiers.
3602 :     * Changed the return type of TypeUtil.getFunction accordingly.
3603 :     * Type equality ignores the argument names.
3604 :     * TypeUtil.composite tries to preserve argument names but gives up quickly
3605 :     if there is a mismatch.
3606 :    
3607 :     installation script:
3608 :     * attempts to use "curl" if available (unless "wget" is available as well)
3609 :    
3610 :     CM:
3611 :     * has an experimental implementation of "portable graphs" which I will
3612 :     soon propose as an implementation-independent library format
3613 :     * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
3614 :    
3615 :     NLFFI-LIB:
3616 :     * some cleanup (all cosmetic)
3617 :    
3618 :     NLFFIGEN:
3619 :     * temporarily disabled the mechanism that suppresses ML output for
3620 :     C definitions whose identifiers start with an underscore character
3621 :     * generate val bindings for enum constants
3622 :     * user can request that only one style (light or heavy) is being used;
3623 :     default is to use both (command-line arguments: -heavy and -light)
3624 :     * fixed bug in handling of function types involving incomplete pointers
3625 :     * generate ML entry points that take record arguments (i.e., using
3626 :     named arguments) for C functions that have a prototype with named
3627 :     arguments
3628 :     (see changes to CKIT)
3629 :    
3630 :     ----------------------------------------------------------------------
3631 : leunga 972 Name: Allen Leung
3632 :     Date: 2001/10/27 20:34:00 EDT
3633 :     Tag: leunga-20011027-x86-fast-fp-call
3634 :     Description:
3635 :    
3636 :     Fixed the bug described in blume-20010920-slowfp.
3637 :    
3638 :     The fix involves
3639 :     1. generating FCOPYs in FSTP in ia32-svid
3640 :     2. marking a CALL with the appropriate annotation
3641 :    
3642 :     ----------------------------------------------------------------------
3643 : blume 961 Name: Matthias Blume
3644 :     Date: 2001/10/16 11:32:00 EDT
3645 :     Tag: blume-20011016-netbsd
3646 :     Description:
3647 :    
3648 :     Underscore patch from Chris Richards (fixing problem with compiling
3649 :     runtime system under recent NetBSD).
3650 :    
3651 :     ----------------------------------------------------------------------
3652 : george 959 Name: Allen Leung
3653 :     Date: 2001/10/12 17:18:32 EDT 2001
3654 :     Tag: leung-20011012-x86-printflowgraph
3655 :     Description:
3656 :    
3657 :     X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
3658 :    
3659 :     ----------------------------------------------------------------------
3660 : george 952 Name: Lal George
3661 : george 958 Date: 2001/10/11 23:51:34 EDT
3662 :     Tag: george-20011011-too-many-instrs
3663 :     Description:
3664 :    
3665 :     The representation of a program point never expected to see more
3666 :     than 65536 instructions in a basic block!
3667 :    
3668 :     ----------------------------------------------------------------------
3669 :     Name: Lal George
3670 : george 952 Date: 2001/10/09 09:41:37 EDT
3671 :     Tag: george-20011008-mlrisc-labels
3672 :     Description:
3673 :    
3674 :     Changed the machine description files to support printing of
3675 :     local and global labels in assembly code, based on host assembler
3676 :     conventions.
3677 :    
3678 :     ----------------------------------------------------------------------
3679 : blume 916 Name: Matthias Blume
3680 : blume 939 Date: 2001/09/25 15:25:00 EDT
3681 :     Tag: blume-20010925-exninfo
3682 :     Description:
3683 :    
3684 :     I provided a non-hook implementation of exnName (at the toplevel) and
3685 :     made the "dummy" implementation of exnMessage (at the toplevel) more
3686 :     useful: if nothing gets "hooked in", then at least you are going to
3687 :     see the exception name and a message indicating why you don't see more.
3688 :    
3689 :     [For the time being, programs that need exnMessage and want to use
3690 :     ml-build should either use General.exnMessage (strongly recommended) or
3691 :     refer to structure General at some other point so that CM sees a
3692 :     static dependency.]
3693 :    
3694 :     [Similar remarks go for "print" and "use": If you want to use their
3695 :     functionality in stand-alone programs generated by ml-build, then use
3696 :     TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
3697 :    
3698 :     ----------------------------------------------------------------------
3699 :     Name: Matthias Blume
3700 : blume 938 Date: 2001/09/20 17:28:00 EDT
3701 :     Tag: blume-20010920-slowfp
3702 :     Description:
3703 :    
3704 :     Allen says that x86-fast-fp is not safe yet, so I turned it off again...
3705 :    
3706 :     ----------------------------------------------------------------------
3707 :     Name: Matthias Blume
3708 : blume 937 Date: 2001/09/20 17:20:00 EDT
3709 :     Tag: blume-20010920-canonicalpaths
3710 :     Description:
3711 :    
3712 :     0. Updated the BOOT file (something that I forgot to do earlier).
3713 :    
3714 :     1. Small internal change to CM so that it avoids "/../" in filenames
3715 :     as much as possible (but only where it is safe).
3716 :    
3717 :     2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
3718 :     that arguments that contain delimiters are passed through correctly.
3719 :     This change also means that all "special" arguments of the form
3720 :     @SMLxxx... must come first.
3721 :    
3722 :     3. Changed install script to put relative anchor names for tool commands
3723 :     into pathconfig.
3724 :    
3725 :     ----------------------------------------------------------------------
3726 :     Name: Matthias Blume
3727 : blume 932 Date: 2001/09/18 15:35:00 EDT
3728 :     Tag: blume-20010918-readme11036
3729 :     Description:
3730 :    
3731 :     Added README files.
3732 :    
3733 :     ----------------------------------------------------------------------
3734 :     Name: Matthias Blume
3735 : blume 929 Date: 2001/09/18 11:45:00 EDT
3736 :     Tag: Release_110_36 (retag)
3737 :     Description:
3738 :    
3739 :     Fixed mistake in config/preloads. Retagged as 110.36.
3740 :    
3741 :     ----------------------------------------------------------------------
3742 :     Name: Matthias Blume
3743 : blume 928 Date: 2001/09/18 09:40:00 EDT
3744 : blume 929 Tag: Release_110_36_orig (tag changed)
3745 : blume 928 Description:
3746 :    
3747 :     New version (110.36). New bootfiles.
3748 :    
3749 :     ----------------------------------------------------------------------
3750 :     Name: Matthias Blume
3751 : blume 927 Date: 2001/09/14 16:15:00 EDT
3752 :     Tag: blume-20010914-x86fastfp
3753 :     Description:
3754 :    
3755 :     John committed some changes that Allen made, in particular a (hopefully)
3756 :     correctly working version of the x86-fp module.
3757 :    
3758 :     I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
3759 :     flag to "true". Everything seems to compile to a fixpoint ok, and
3760 :     "mandelbrot" speeds up by about 15%.
3761 :    
3762 :     ----------------------------------------------------------------------
3763 :     Name: Matthias Blume
3764 : blume 923 Date: 2001/09/13 11:20:00 EDT
3765 :     Tag: blume-20010913-minimal
3766 :     Description:
3767 :    
3768 :     1. Stefan Monnier's patch to fix a miscompilation problem that
3769 :     was brought to light by John Reppy's work on Moby.
3770 :    
3771 :     2. Implemented a minimal "structure Compiler" that contains just
3772 :     "version" and "architecture". The minimal version will be
3773 :     available when the full version is not. This is for backward-
3774 :     compatibility with code that wants to test Compiler.version.
3775 :    
3776 :     ----------------------------------------------------------------------
3777 :     Name: Matthias Blume
3778 : blume 920 Date: 2001/08/28 14:03:00 EDT
3779 :     Tag: blume-20010828-ml-lex
3780 :     Description:
3781 :    
3782 :     Fix for bug 1581, received from Neophytos Michael.
3783 :    
3784 :     ----------------------------------------------------------------------
3785 :     Name: Matthias Blume
3786 : blume 918 Date: 2001/08/27 11:20:00 EDT
3787 :     Tag: blume-20010827-readme11035
3788 :     Description:
3789 :    
3790 :     Fleshed out the README file for 110.35.
3791 :    
3792 :     ----------------------------------------------------------------------
3793 :     Name: Matthias Blume
3794 : blume 916 Date: 2001/08/24 17:10:00 EDT
3795 :     Tag: Release_110_35
3796 :     Description:
3797 :    
3798 :     New version number (110.35). New bootfiles.
3799 :    
3800 :     ----------------------------------------------------------------------
3801 : george 915 Name: Lal George
3802 : george 909 Date: 2001/08/24 13:47:18 EDT 2001
3803 :     Tag: george-20010824-MLRISC-graphs
3804 :     Description:
3805 :    
3806 :     removed clusters from MLRISC completely and replaced with graphs.
3807 :    
3808 :     ----------------------------------------------------------------------
3809 :     Name: Matthias Blume
3810 : blume 905 Date: 2001/08/23 17:50:00 EDT
3811 :     Tag: blume-20010823-toplevel
3812 :     Description:
3813 :    
3814 :     - some reorganization of the code that implements various kinds of
3815 :     environments in the compiler (static, dynamic, symbolic, combined)
3816 :     - re-implemented the EnvRef module so that evalStream works properly
3817 :     (if the stream contains references to "use", "CM.make", etc.)
3818 :     - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
3819 :    
3820 :     ----------------------------------------------------------------------
3821 :     Name: Matthias Blume
3822 : blume 904 Date: 2001/08/20 15:50 EDT
3823 :     Tag: blume20010820-slipup
3824 :     Description:
3825 :    
3826 :     I forgot to commit a few files. Here they are...
3827 :    
3828 :     ----------------------------------------------------------------------
3829 :     Name: Matthias Blume
3830 : blume 903 Date: 2001/08/20 15:35:00 EDT
3831 :     Tag: blume-20010820-debugprof
3832 :     Description:
3833 :    
3834 :     !!!! NEW BOOTFILES !!!!
3835 :    
3836 :     This is another round of reorganizing the compiler sources. This
3837 :     time the main goal was to factor out all the "instrumentation"
3838 :     passes (for profiling and backtracing) into their own library.
3839 :     The difficulty was to do it in such a way that it does not depend
3840 :     on elaborate.cm but only on elabdata.cm.
3841 :    
3842 :     Therefore there have been further changes to both elaborate.cm and
3843 :     elabdata.cm -- more "generic" things have been moved from the former
3844 :     to the latter. As a result, I was forced to split the assignment
3845 :     of numbers indicating "primtyc"s into two portions: SML-generic and
3846 :     SML/NJ-specific. Since it would have been awkward to maintain,
3847 :     I bit the bullet and actually _changed_ the mapping between these
3848 :     numbers and primtycs. The bottom line of this is that you need
3849 :     a new set of bin- and bootfiles.
3850 :    
3851 :     I have built new bootfiles for all architectures, so doing a fresh
3852 :     checkout and config/install.sh should be all you need.
3853 :    
3854 :     The newly created library's name is
3855 :    
3856 :     $smlnj/viscomp/debugprof.cm
3857 :    
3858 :     and its sources live under
3859 :    
3860 :     src/compiler/DebugProf
3861 :    
3862 :     ----------------------------------------------------------------------
3863 :     Name: Matthias Blume
3864 : blume 902 Date: 2001/08/15 17:15:00 EDT
3865 :     Tag: blume-20010815-compreorg
3866 :     Description:
3867 :    
3868 :     This is a first cut at reorganizing the CM libraries that make up the
3869 :     core of the compiler. The idea is to separate out pieces that could
3870 :     be used independently by tools, e.g., the parser, the typechecker, etc.
3871 :    
3872 :     The current status is a step in this direction, but it is not quite
3873 :     satisfactory yet. Expect more changes in the future.
3874 :    
3875 :     Here is the current (new) organization...
3876 :    
3877 :     What used to be $smlnj/viscomp/core.cm is now divided into
3878 :     six CM libraries:
3879 :    
3880 :     $smlnj/viscomp/basics.cm
3881 :     /parser.cm
3882 :     /elabdata.cm
3883 :     /elaborate.cm
3884 :     /execute.cm
3885 :     /core.cm
3886 :    
3887 :     The CM files for these libraries live under src/system/smlnj/viscomp.
3888 :     All these libraries are proxy libraries that contain precisely
3889 :     one CM library component. Here are the locations of the components
3890 :     (all within the src/compiler tree):
3891 :    
3892 :     Basics/basics.cm
3893 :     Parse/parser.cm
3894 :     ElabData/elabdata.cm
3895 :     Elaborator/elaborate.cm
3896 :     Execution/execute.cm
3897 :     core.cm
3898 :    
3899 :     [This organization is the same that has been used already
3900 :     for a while for the architecture-specific parts of the visible
3901 :     compiler and for the old version of core.cm.]
3902 :    
3903 :     As you will notice, many source files have been moved from their
3904 :     respective original locations to a new home in one of the above
3905 :     subtrees.
3906 :    
3907 :     The division of labor between the new libraries is the following:
3908 :    
3909 :     basics.cm:
3910 :     - Simple, basic definitions that pertain to many (or all) of
3911 :     the other libraries.
3912 :     parser.cm:
3913 :     - The SML parser, producing output of type Ast.dec.
3914 :     - The type family for Ast is also defined and exported here.
3915 :     elabdata.cm:
3916 :     - The datatypes that describe input and output of the elaborator.
3917 :     This includes types, absyn, and static environments.
3918 :     elaborator.cm:
3919 :     - The SML/NJ type checker and elaborator.
3920 :     This maps an Ast.dec (with a given static environment) to
3921 :     an Absyn.dec (with a new static environment).
3922 :     - This libraries implements certain modules that used to be
3923 :     structures as functors (to remove dependencies on FLINT).
3924 :     execute.cm:
3925 :     - Everything having to do with executing binary code objects.
3926 :     - Dynamic environments.
3927 :     core.cm:
3928 :     - SML/NJ-specific instantiations of the elaborator and MLRISC.
3929 :     - Top-level modules.
3930 :     - FLINT (this should eventually become its own library)
3931 :    
3932 :     Notes:
3933 :    
3934 :     I am not 100% happy with the way I separated the elaborator (and its
3935 :     data structures) from FLINT. Two instances of the same problem:
3936 :    
3937 :     1. Data structures contain certain fields that carry FLINT-specific
3938 :     information. I hacked around this using exn and the property list
3939 :     module from smlnj-lib. But the fact that there are middle-end
3940 :     specific fields around at all is a bit annoying.
3941 :    
3942 :     2. The elaborator calculates certain FLINT-related information. I tried
3943 :     to make this as abstract as I could using functorization, but, again,
3944 :     the fact that the elaborator has to perform calculations on behalf
3945 :     of the middle-end at all is not nice.
3946 :    
3947 :     3. Having to used exn and property lists is unfortunate because it
3948 :     weakens type checking. The other alternative (parameterizing
3949 :     nearly *everything*) is not appealing, though.
3950 :    
3951 :     I removed the "rebinding =" warning hack because due to the new organization
3952 :     it was awkward to maintain it. As a result, the compiler now issues some of
3953 :     these warnings when compiling init.cmi during bootstrap compilation. On
3954 :     the plus side, you also get a warning when you do, for example:
3955 :     val op = = Int32.+
3956 :     which was not the case up to now.
3957 :    
3958 :     I placed "assign" and "deref" into the _Core structure so that the
3959 :     code that deals with the "lazy" keyword can find them there. This
3960 :     removes the need for having access to the primitive environment
3961 :     during elaboration.
3962 :    
3963 :     ----------------------------------------------------------------------
3964 :     Name: Matthias Blume
3965 : blume 898 Date: 2001/08/13
3966 :     Tag: blume-20010813-closures
3967 :     Description:
3968 :    
3969 :     This fix was sent to us by Zhong Shao. It is supposed to improve the
3970 :     performance of certain loops by avoiding needless closure allocation.
3971 :    
3972 :     ----------------------------------------------------------------------
3973 : george 890 Name: Lal George
3974 : george 896 Date: 2001/07/31 10:03:23 EDT 2001
3975 :     Tag: george-20010731-x86-fmalloc
3976 :     Description: Fixed bug in x86 calls
3977 :    
3978 :     There was a bug where call instructions would mysteriously
3979 :     vanish. The call instruction had to be one that returned
3980 :     a floating point value.
3981 :    
3982 :     ----------------------------------------------------------------------
3983 :     Name: Lal George
3984 : george 890 Date: 2001/07/19 16:36:29 EDT 2001
3985 :     Tag: george-20010719-simple-cells
3986 :     Description:
3987 :    
3988 :     I have dramatically simplified the interface for CELLS in MLRISC.
3989 :    
3990 :     In summary, the cells interface is broken up into three parts:
3991 :    
3992 :     1. CellsBasis : CELLS_BASIS
3993 :    
3994 :     CellsBasis is a top level structure and common for all
3995 :     architectures. it contains the definitions of basic datatypes
3996 :     and utility functions over these types.
3997 :    
3998 :     2. functor Cells() : CELLS
3999 :    
4000 :     Cells generates an interface for CELLS that incorporates the
4001 :     specific resources on the target architecture, such as the
4002 :     presence of special register classes, their number and size,
4003 :     and various useful substructures.
4004 :    
4005 :     3. <ARCH>CELLS
4006 :    
4007 :     e.g. SparcCells: SPARCCELLS
4008 :    
4009 :     <ARCH>CELLS usually contains additional bindings for special
4010 :     registers on the architecture, such as:
4011 :    
4012 :     val r0 : cell (* register zero *)
4013 :     val y : cell (* Y register *)
4014 :     val psr : cell (* processor status register *)
4015 :     ...
4016 :    
4017 :     The structure returned by applying the Cells functor is opened
4018 :     in this interface.
4019 :    
4020 :     The main implication of all this is that the datatypes for cells is
4021 :     split between CellsBasis and CELLS -- a fairly simple change for user
4022 :     code.
4023 :    
4024 :     In the old scheme the CELLS interface had a definitional binding of
4025 :     the form:
4026 :    
4027 :     signature CELLS = sig
4028 :    
4029 :     structure CellsBasis = CellsBasis
4030 :    
4031 :     ...
4032 :    
4033 :     end
4034 :    
4035 :     With all the sharing constraints that goes on in MLRISC, this old
4036 :     design quickly leads to errors such as:
4037 :    
4038 :     "structure definition spec inside of sharing ... "
4039 :    
4040 :    
4041 :     and appears to require an unacceptable amount of sharing and where
4042 :     constraint hackery.
4043 :    
4044 :     I think this error message (the interaction of definitional specs and
4045 :     sharing) requires more explanation on our web page.
4046 :    
4047 :     ----------------------------------------------------------------------
4048 : blume 826 Name: Matthias Blume
4049 : blume 879 Date: 2001/07/19 15:00:00 EDT
4050 :     Tag: blume-20010719-libreorg
4051 :     Description:
4052 :    
4053 :     This update puts together a fairly extensive but straightforward change
4054 :     to the way the libraries that implement the interactive system are
4055 :     organized:
4056 :    
4057 :     The biggest change is the elimination of structure Compiler. As a
4058 :     replacement for this structure, there is now a CM library
4059 :     (known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
4060 :     that exports all the substructures of the original structure Compiler
4061 :     directly. So instead of saying Compiler.Foo.bar one now simply
4062 :     says Foo.bar. (The CM libraries actually export a collection of
4063 :     structures that is richer than the collection of substructures of
4064 :     structure Compiler.)
4065 :    
4066 :     To make the transition smooth, there is a separate library called
4067 :     $smlnj/compiler/compiler.cm which puts together and exports the
4068 :     original structure Compiler (or at least something very close to it).
4069 :    
4070 :     There are five members of the original structure Compiler
4071 :     that are not exported directly but which instead became members
4072 :     of a new structure Backend (described by signature BACKEND). These are:
4073 :     structure Profile (: PROFILE), structure Compile (: COMPILE), structure
4074 :     Interact (: INTERACT), structure Machine (: MACHINE), and val
4075 :     architecture (: string).
4076 :    
4077 :     Structure Compiler.Version has become structure CompilerVersion.
4078 :    
4079 :     Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
4080 :     by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
4081 :     or x86, respectively.
4082 :     Each of these exports the same frontend structures that
4083 :     $smlnj/compiler.cm exports. But they do not have a structure Backend
4084 :     and instead export some structure <Arch>Backend where <Arch> is Alpha32,
4085 :     Hppa, PPC, Sparc, or X86, respectively.
4086 :    
4087 :     Library $smlnj/compiler/all.cm exports the union of the exports of
4088 :     $smlnj/compiler/<arch>.cm
4089 :    
4090 :     There are no structures <Arch>Compiler anymore, use
4091 :     $smlnj/compiler/<arch>.cm instead.
4092 :    
4093 :     Library host-compiler-0.cm is gone. Instead, the internal library
4094 :     that instantiates CM is now called cm0.cm. Selection of the host
4095 :     compiler (backend) is no longer done here but. (Responsibility for it
4096 :     now lies with $smlnj/compiler/current.cm. This seems to be more
4097 :     logical.)
4098 :    
4099 :     Many individual files have been moved or renamed. Some files have
4100 :     been split into multiple files, and some "dead" files have been deleted.
4101 :    
4102 :     Aside from these changes to library organization, there are also changes
4103 :     to the way the code itself is organized:
4104 :    
4105 :     Structure Binfile has been re-implemented in such a way that it no
4106 :     longer needs any knowledge of the compiler. It exclusively deals
4107 :     with the details of binfile layout. It no longer invokes the
4108 :     compiler (for the purpose of creating new prospective binfile
4109 :     content), and it no longer has any knowledge of how to interpret
4110 :     pickles.
4111 :    
4112 :     Structure Compile (: COMPILE) has been stripped down to the bare
4113 :     essentials of compilation. It no longer deals with linking/execution.
4114 :     The interface has been cleaned up considerably.
4115 :    
4116 :     Utility routines for dealing with linking and execution have been
4117 :     moved into their own substructures.
4118 :    
4119 :     (The ultimate goal of these changes is to provide a light-weight
4120 :     binfile loader/linker (at least for, e.g., stable libraries) that
4121 :     does not require CM or the compiler to be present.)
4122 :    
4123 :     CM documentation has been updated to reflect the changes to library
4124 :     organization.
4125 :    
4126 :     ----------------------------------------------------------------------
4127 :     Name: Matthias Blume
4128 : blume 875 Date: 2001/07/10 17:30:00 EDT
4129 :     Tag: Release_110_34
4130 :     Description:
4131 :    
4132 : blume 876 Minor tweak to 110.34 (re-tagged):
4133 :    
4134 :     - README.html file added to CVS repository
4135 :     - runtime compiles properly under FreeBSD 3.X and 4.X
4136 :    
4137 :     ----------------------------------------------------------------------
4138 :     Name: Matthias Blume
4139 :     Date: 2001/07/10 17:30:00 EDT
4140 :     Tag: Release_110_34
4141 :     Description:
4142 :    
4143 : blume 875 New version number (110.34). New bootfiles.
4144 :    
4145 :     ----------------------------------------------------------------------
4146 :     Name: Matthias Blume
4147 : blume 874 Date: 2001/07/09 16:00:00 EDT
4148 :     Tag: blume-20010709-more-varargs
4149 :     Description:
4150 :    
4151 :     I changed the handling of varargs in ml-nlffigen again:
4152 :     The ellipsis ... will now simply be ignored (with an accompanying warning).
4153 :    
4154 :     The immediate effect is that you can actually call a varargs function
4155 :     from ML -- but you can't actually supply any arguments beyond the ones
4156 :     specified explicitly. (For example, you can call printf with its format
4157 :     string, but you cannot pass additional arguments.)
4158 :    
4159 :     This behavior is only marginally more useful than the one before, but
4160 :     it has the advantage that a function or, more importantly, a function
4161 :     type never gets dropped on the floor, thus avoiding follow-up problems with
4162 :     other types that refer to the offending one.
4163 :    
4164 :     ----------------------------------------------------------------------
4165 :     Name: Matthias Blume
4166 : blume 873 Date: 2001/07/09 11:25:00 EDT
4167 :     Tag: blume-20010709-varargs
4168 :     Description:
4169 :    
4170 :     1. ckit-lib.cm now exports structure Error
4171 :     2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
4172 :     with a warning accompanied by a source location. Moreover, it
4173 :     merely skips the offending function or type and proceeds with the
4174 :     rest of its work.u As a result, one can safely feed C code containing
4175 :     "..." to ml-nlffigen.
4176 :     3. There are some internal improvements to CM, providing slightly
4177 :     more general string substitutions in the tools subsystem.
4178 :    
4179 :     ----------------------------------------------------------------------
4180 :     Name: Matthias Blume
4181 : blume 854 Date: 2001/06/27 15:10:00 EDT
4182 :     Tag: blume-20010627-concur
4183 :     Description:
4184 :    
4185 :     Fixed a small bug in CM's handling of parallel compilation.
4186 :     (You could observe the bug by Control-C-interrupting an ordinary
4187 :     CMB.make or CM.stabilize and then attaching some compile servers.
4188 :     The result was that all of a sudden the previously interrupted
4189 :     compilation would continue on its own. This was because of
4190 :     an over-optimization: CM did not bother to clean out certain queues
4191 :     when no servers were attached "anyway", resulting in the contents
4192 :     of these queues to grab control when new servers did get attached.)
4193 :    
4194 :     There is also another minor update to the CM manual.
4195 :    
4196 :     ----------------------------------------------------------------------
4197 :     Name: Matthias Blume
4198 : blume 853 Date: 2001/06/26 16:15:00 EDT
4199 :     Tag: blume-20010626-cmdoc
4200 :     Description:
4201 :    
4202 :     Minor typo fixed in CM manual (syntax diagram for libraries).
4203 :    
4204 :     ----------------------------------------------------------------------
4205 :     Name: Matthias Blume
4206 : blume 852 Date: 2001/06/25 22:55:00 EDT
4207 :     Tag: blume-20010625-x86pc
4208 :     Description:
4209 :    
4210 :     Fixed a nasty bug in the X86 assembly code that caused signal
4211 :     handlers to fail (crash) randomly.
4212 :    
4213 :     ----------------------------------------------------------------------
4214 :     Name: Matthias Blume
4215 : blume 846 Date: 2001/06/25 12:05:00 EDT
4216 :     Tag: blume-20010625-nlffigen
4217 :     Description:
4218 :    
4219 :     This update fixes a number of minor bugs in ml-nlffigen as reported by
4220 :     Nick Carter <nbc@andrew.cmu.edu>.
4221 :    
4222 :     1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
4223 :     2. Default names for generated files are now derived from the name of
4224 :     the C file *without its directory*. In particular, this causes generated
4225 :     files to be placed locally even if the C file is in some system directory.
4226 :     3. Default names for generated signatures and structures are also derived
4227 :     from the C file name without its directory. This avoids silly things
4228 :     like "structure GL/GL".
4229 :     (Other silly names are still possible because ml-nlffigen does not do
4230 : blume 847 a thorough check of whether generated names are legal ML identifiers.
4231 :     When in doubt, use command line arguments to force particular names.)
4232 : blume 846
4233 :     ----------------------------------------------------------------------
4234 :     Name: Matthias Blume
4235 : blume 845 Date: 2001/06/21 12:25:00 EDT
4236 :     Tag: blume-20010621-eXene
4237 :     Description:
4238 :    
4239 :     eXene now compiles and (sort of) works again.
4240 :    
4241 :     The library name (for version > 110.33) is $/eXene.cm.
4242 :    
4243 :     I also added an new example in src/eXene/examples/nbody. See the
4244 :     README file there for details.
4245 :    
4246 :     ----------------------------------------------------------------------
4247 :     Name: Matthias Blume
4248 : blume 844 Date: 2001/06/20 16:40:00 EDT
4249 :     Tag: blume-20010620-cml
4250 :     Description:
4251 :    
4252 :     CML now compiles and works again.
4253 :    
4254 :     Libraries (for version > 110.33):
4255 :    
4256 :     $cml/cml.cm Main CML library.
4257 :     $cml/basis.cm CML's version of $/basis.cm.
4258 :     $cml/cml-internal.cm Internal helper library.
4259 :     $cml/core-cml.cm Internal helper library.
4260 :     $cml-lib/trace-cml.cm Tracing facility.
4261 :     $cml-lib/smlnj-lib.cm CML's version of $/smlnj-lib.cm
4262 :    
4263 :     The installer (config/install.sh) has been taught how to properly
4264 :     install this stuff.
4265 :    
4266 :     ----------------------------------------------------------------------
4267 :     Name: Matthias Blume
4268 : blume 843 Date: 2001/06/19 17:55:00 EDT
4269 :     Tag: blume-20010619-instantiate
4270 :     Description:
4271 :    
4272 :     This un-breaks the fix for bug 1432.
4273 :     (The bug was originally fixed in 110.9 but I broke it again some
4274 :     time after that.)
4275 :    
4276 :     ----------------------------------------------------------------------
4277 :     Name: Matthias Blume
4278 : blume 842 Date: 2001/06/19 17:25:00 EDT
4279 :     Tag: blume-20010619-signals
4280 :     Description:
4281 :    
4282 :     This should (hopefully) fix the long-standing signal handling bug.
4283 :     (The runtime system was constructing a continuation record with an
4284 :     incorrect descriptor which would cause the GC to drop data on the floor...)
4285 :    
4286 :     ----------------------------------------------------------------------
4287 :     Name: Matthias Blume
4288 : blume 840 Date: 2001/06/15 15:05:00 EDT
4289 : blume 841 Tag: blume-20010615-moresparc
4290 :     Description:
4291 :    
4292 :     Here is a short late-hour update related to Sparc c-calls:
4293 :    
4294 :     -- made handling of double-word arguments a bit smarter
4295 :    
4296 :     -- instruction selection phase tries to collapse certain clumsily
4297 :     constructed ML-Trees; typical example:
4298 :    
4299 :     ADD(ty,ADD(_,e,LI d1),LI d2) -> ADD(ty,e,LI(d1+d2))
4300 :    
4301 :     This currently has no further impact on SML/NJ since mlriscGen does
4302 :     not seem to generate such patterns in the first place, and c-calls
4303 :     (which did generate them in the beginning) has meanwhile been fixed
4304 :     so as to avoid them as well.
4305 :    
4306 :     ----------------------------------------------------------------------
4307 :     Name: Matthias Blume
4308 :     Date: 2001/06/15 15:05:00 EDT
4309 : blume 840 Tag: blume-20010615-sparc
4310 :     Description:
4311 :    
4312 :     The purpose of this update is to provide an implementation of NLFFI
4313 :     on Sparc machines.
4314 :    
4315 :     Here are the changes in detail:
4316 :    
4317 :     * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
4318 :     the Sparc implementation of the c-calls API.
4319 :     * The Sparc backend of SML/NJ has been modified to uniformely use %fp
4320 :     for accessing the ML frame. Thus, we have a real frame pointer and
4321 :     can freely modify %sp without need for an omit-frame-ptr phase.
4322 :     The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
4323 :     this case.
4324 :     * ml-nlffigen has been taught to produce code for different architectures
4325 :