Annotation of /sml/trunk/HISTORY

Parent Directory | Revision Log

---

Revision 1395 - (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	1384	Name:
11 :	george	1003	Date: yyyy/mm/dd
12 :	dbm	570	Tag: <post-commit CVS tag>
13 :			Description:
14 :	leunga	1142
15 :	george	1136	----------------------------------------------------------------------
16 :	mblume	1384	Name: Matthias Blume (blume (at) tti - c (dot) org)
17 :	mblume	1395	Date: 2003/09/26 16:05:00 CDT
18 :			Tag: blume-20030926-wrappriv
19 :			Description:
20 :
21 :			Added missing wrapper for privilege "primitive" in $smlnj/viscomp/core.cm.
22 :
23 :			----------------------------------------------------------------------
24 :			Name: Matthias Blume (blume (at) tti - c (dot) org)
25 :	mblume	1394	Date: 2003/09/26 15:00:00 CDT
26 :			Tag: blume-20030926-110_43_3
27 :			Description:
28 :
29 :			- additional cleanup
30 :			- version number bump, NEW BOOTFILES
31 :
32 :			----------------------------------------------------------------------
33 :			Name: Matthias Blume (blume (at) tti - c (dot) org)
34 :	mblume	1393	Date: 2003/09/26 12:00:00 CDT
35 :			Tag: blume-20030926-ppautoload
36 :			Description:
37 :
38 :			I modified the read-eval-print loop so that the autoloader gets
39 :			invoked whenever the prettyprinter tries to look up a symbol that
40 :			is not currently defined in the toplevel environment but which
41 :			appears in CM's autoload registry. As a result, we see far fewer of
42 :			those ?.Foo.Bar.xxx names in the prettyprinter's output.
43 :
44 :			In addition to this I tried to clean up some pieces of the Basis
45 :			implementation (e.g., Socket, Word8Array) in order to prevent other
46 :			instances of these ?.Foo.Bar.xxx names from being printed.
47 :
48 :			The mechanism that picks names for types still needs some work, though.
49 :			(Right now it seems that if there is a type A.t which is defined to
50 :	mblume	1394	be B.u, but B is unavailable at toplevel, then A.t gets printed as
51 :			"?.B.u" although the perhaps more sensible solution would be to use
52 :	mblume	1393	"A.t" in this case. In other words, the prettyprinter should follow
53 :			a chain of DEFtycs not farther than there are corresponding toplevel
54 :			names in the current environment.)
55 :
56 :			----------------------------------------------------------------------
57 :			Name: Matthias Blume (blume (at) tti - c (dot) org)
58 :	mblume	1391	Date: 2003/09/24 16:31:00 CDT
59 :			Tag: blume-20030924-installer
60 :			Description:
61 :
62 :			Another installer tweak: All the ML code for the installer is now
63 :			compiled during CMB.make and put into a little library called
64 :			$smlnj/installer.cm. The installation then simply invokes
65 :
66 :			sml -m $smlnj/installer.cm
67 :
68 :			and everything happens automagically.
69 :
70 :			Win32: ML code senses value of environment variable SMLNJ_HOME.
71 :			Unix: ML code senses values of environment variables ROOT, CONFIGDIR,
72 :			and BINDIR.
73 :
74 :			The new scheme guarantees that the ML code responsible for the installation
75 :			is in sync with the APIs of the main system. Also, the installer is
76 :			somewhat faster because the installer script is precompiled.
77 :
78 :			----------------------------------------------------------------------
79 :			Name: Matthias Blume (blume (at) tti - c (dot) org)
80 :	mblume	1390	Date: 2003/09/24 15:35:00 CDT
81 :			Tag: blume-20030924-synsock
82 :			Description:
83 :
84 :			Added a signature SYNCHRONOUS_SOCKET to basis.cm. This is like SOCKET
85 :			but excludes all non-blocking operations. Defined SOCKET (in Basis)
86 :			and CML_SOCKET in terms of SYNCHRONOUS_SOCKET. Removed superfluous
87 :			implementations of non-blocking operations from CML's Socket
88 :			structure.
89 :
90 :			----------------------------------------------------------------------
91 :			Name: Matthias Blume (blume (at) tti - c (dot) org)
92 :	mblume	1389	Date: 2003/09/24 15:10:05 CDT
93 :			Tag: blume-20030924-sockets
94 :			Description:
95 :
96 :			1. Fixed SOCKET API and implementation to match Basis spec.
97 :			This required changing the internal representation of sockets to one
98 :			that remembers (for each socket file descriptor) whether it is currently
99 :			blocking or non-blocking. This state is maintained lazily (i.e., a system
100 :			call is made only if the state actually needs to change).
101 :
102 :			2. OS-specific details of sockets were moved into separate files, thus
103 :			making it possible to unify the bulk of the socket implementations
104 :			between Unix and Win32.
105 :
106 :			3. CML's socket API changed accordingly.
107 :			(Note that we need to remove non-blocking functions from this API
108 :			since they are redundant in the case of CML!)
109 :
110 :			4. CML's socket implementation now makes use of non-blocking functions
111 :			provided by Basis, thus removing all OS-dependent code from this part
112 :			of CML.
113 :
114 :			5. Changed Real64.precision from 52 to 53. Minor cleanup in Real64 code.
115 :
116 :			----------------------------------------------------------------------
117 :			Name: Matthias Blume (blume (at) tti - c (dot) org)
118 :	mblume	1387	Date: 2003/09/22 12:10:00 CDT
119 :	mblume	1388	Tag: blume-20030922-110_43_2
120 :	mblume	1387	Description:
121 :
122 :			Made a new interim version and bootfiles for developer's bootstrapping
123 :			convenience.
124 :
125 :			110.43.2 -- NEW BOOTFILES
126 :
127 :			----------------------------------------------------------------------
128 :			Name: Matthias Blume (blume (at) tti - c (dot) org)
129 :	mblume	1385	Date: 2003/09/19 15:55:00 CDT
130 :			Tag: blume-20030919-cmdir
131 :			Description:
132 :
133 :			1. new-install.sh -> install.sh
134 :			2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
135 :			3. tweaked installer so that another name instead of .cm can be chosen
136 :			at install time (by setting the CM_DIR_ARC environment variable
137 :			during installation); once installation is complete, the name is
138 :			fixed
139 :
140 :			----------------------------------------------------------------------
141 :			Name: Matthias Blume (blume (at) tti - c (dot) org)
142 :	mblume	1384	Date: 2003/09/18 16:00:00 CDT
143 :			Tag: blume-20030918-110_43_1
144 :			Description:
145 :
146 :			Made a new interim version and bootfiles for developer's bootstrapping
147 :			convenience.
148 :
149 :			110.43.1 -- NEW BOOTFILES
150 :
151 :			----------------------------------------------------------------------
152 :			Name: Matthias Blume (blume (at) tti - c (dot) org)
153 :	mblume	1383	Date: 2003/09/18 15:20:00 CDT
154 :			Tag: blume-20030918-misc
155 :			Description:
156 :
157 :			1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
158 :			this seems to be controversial at the moment) and used those in
159 :			Posix.ProcEnv.times.
160 :
161 :			2. Added Time.{from,to}Nanoseconds to Time.
162 :
163 :			3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
164 :			For example, fromLargeInt never needs to look at more than 3 "big
165 :			digits" to get its 53 bits of precision.
166 :
167 :			----------------------------------------------------------------------
168 :	mblume	1384	Name: Matthias Blume (blume (at) tti - c (dot) org)
169 :	mblume	1381	Date: 2003/09/17 16:30:00 CDT
170 :			Tag: blume-20030917-real32-slices
171 :			Description:
172 :
173 :			Added an entry to the primitive environment
174 :			(compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
175 :			added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
176 :
177 :			Removed some of the "magic" constants in real64.sml and replaced them
178 :			with code that generates these values from their corresponding
179 :			integer counterparts.
180 :
181 :			Made all(?) the slice-related changes to the Basis and made everything
182 :			compile again...
183 :
184 :			----------------------------------------------------------------------
185 :	mblume	1368	Name: Matthias Blume (blume (at) tti - c (dot) org)
186 :	mblume	1375	Date: 2003/09/15 17:45:00 CDT
187 :			Tag: blume-20030915-rbase
188 :			Description:
189 :
190 :			Fixed bug in Real.fromLargeInt.
191 :
192 :			----------------------------------------------------------------------
193 :			Name: Matthias Blume (blume (at) tti - c (dot) org)
194 :	mblume	1369	Date: 2003/09/13 18:11:00 CDT
195 :			Tag: blume-20030913-libinstall
196 :			Description:
197 :
198 :			Minor bugfix in config/libinstall (set anchor with path to
199 :			standalone tool after installing it, otherwise libraries that
200 :			need ml-lex or ml-yacc won't compile the first time the installer
201 :			runs).
202 :
203 :			----------------------------------------------------------------------
204 :			Name: Matthias Blume (blume (at) tti - c (dot) org)
205 :	mblume	1368	Date: 2003/09/12 11:45:00 CDT
206 :			Tag: blume-20030912-various
207 :			Description:
208 :
209 :			- fixed bug in Real.toLargeInt
210 :			- fixed bug in Posix.ProcEnv.times
211 :			- changed inputLine functions to return an option
212 :			- minor installer improvements / bugfixes
213 :			- changed default @SMLalloc parameter for x86/celeron to 64k
214 :
215 :			----------------------------------------------------------------------
216 :			Name: Matthias Blume (blume (at) tti - c (dot) org)
217 :	mblume	1356	Date: 2003/09/09 22:00:00 CDT
218 :			Tag: Release_110_43
219 :			Description:
220 :
221 :			New working release 110.43. New bootfiles.
222 :
223 :			----------------------------------------------------------------------
224 :	mblume	1368	Name: Matthias Blume (blume (at) tti - c (dot) org)
225 :	mblume	1354	Date: 2003/09/09 19:20:00 CDT
226 :			Tag: blume-20030909-installer
227 :			Description:
228 :
229 :			Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
230 :			Modified config/install.bat to take advantage of it. Also modified
231 :			config/install.sh (and called it config/new-install.sh) to take advantage
232 :			of it on Unix systems. (The SML code is (supposed to be) platform-
233 :			independent.)
234 :
235 :			The installer can now install everything under Win32
236 :			as well as under *nix as long as it compiles.
237 :
238 :			Other changes:
239 :
240 :			- made CML compile again under Win32
241 :			- made eXene compile under Win32 (by providing a fake structure UnixSock
242 :			and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
243 :			- fixed a bug in nowhere: it assumed that type OS.Process.status is the
244 :			same as type int; under Win32 it isn't
245 :			- fixed some slice-related problems in the win32-specific parts of CML
246 :			- added a functor argument "sameVol" to os-path-fn.sml in the Basis
247 :			(under Win32, the volume name is case-insensitive, and the
248 :			OS.Path code compares volume names for equality)
249 :
250 :			----------------------------------------------------------------------
251 :	mblume	1368	Name: Matthias Blume (blume (at) tti - c (dot) org)
252 :	mblume	1353	Date: 2003/09/08 11:55:00 CDT
253 :			Tag: blume-20030908-fullpath
254 :			Description:
255 :
256 :			Made Win32 version of OS.FileSys.fullPath return current directory
257 :			when given an empty string. This is what the spec says, and incidentally,
258 :			CM depends on it. (CM otherwise goes into an infinite loop in certain
259 :			cases when presented with the name of a non-existing .cm file.)
260 :
261 :			----------------------------------------------------------------------
262 :	mblume	1368	Name: Matthias Blume (blume (at) tti - c (dot) org)
263 :	mblume	1350	Date: 2003/09/04 16:30:00 CDT
264 :	mblume	1352	Tag: blume-20030905-slices-etc
265 :	mblume	1350	Description:
266 :
267 :			1. Changed interface to vectors and arrays in Basis to match
268 :			(draft) Basis spec.
269 :			2. Added signatures and implementations of slices according to
270 :			Basis spec.
271 :			3. Edited source code throughout the system to make it compile again
272 :			under 1. and 2. (In some cases code had to be added to have it
273 :			match the new signatures.)
274 :			4. MLRISC should be backward-compatible: the copies of the originals
275 :			of files that needed to change under 3. were retained, the .cm files
276 :			check the compiler version number and use old versions when
277 :			appropriate.
278 :			5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
279 :			dirstream -> string option (in accordance with Basis spec).
280 :			6. When generating code that counts lines, ml-lex used function
281 :			CharVector.foldli, taking advantage of its old interface.
282 :			This has been replaced with the corresponding code from
283 :			CharVectorSlice. (html-lex must be re-lexed!)
284 :			7. BitArray in smlnj-lib/Util has been extended/modified to match the
285 :			new MONO_ARRAY signature. (Do we need BitArraySlice?)
286 :			8. Removed temporary additions (fromInternal, toInternal) from the
287 :			(now obsolete) IntInf in smlnj-lib/Util.
288 :			9. Cleaned up structure Byte.
289 :			10. Added localOffset, scan, and fromString to Date (according to spec).
290 :			Cleaned/corrected implementation of Date.
291 :			(Still need to check for correctness; implement better canonicalizeDate.)
292 :			11. Added "scan" to signature IEEE_REAL.
293 :			12. Some improvements to IntInf [in particular: efficiency-hack for
294 :			mod and rem when second operand is 2 (for parity checks).]
295 :			13. Changed representation of type Time.time, using a single IntInf.int
296 :			value counting microseconds. This considerably simplified the
297 :			implementation of structure Time. We now support negative time
298 :			values; scan and fromString handle signs.
299 :			14. Functor PrimIO now takes two additional arguments (VectorSlice and
300 :			ArraySlice).
301 :
302 :			----------------------------------------------------------------------
303 :	mblume	1368	Name: Matthias Blume (blume (at) tti - c (dot) org)
304 :	mblume	1347	Date: 2003/08/28 17:00:00 CDT
305 :			Tag: blume-20030828-intinf
306 :			Description:
307 :
308 :			This is a major update which comes with a version number bump
309 :			(110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
310 :			and an implementation of IntInf in the Basis.
311 :
312 :			There are a fairly large number of related changes and updates throughout
313 :			the system:
314 :
315 :			Basis:
316 :			- Implemented IntInf.
317 :			- Made LargeInt a projection of IntInf (by filtering through INTEGER).
318 :	mblume	1348	- Added some missing Real64 operations, most notably Real.toLargeInt.
319 :	mblume	1347	- Added FixedInt as a synonym for Int32.
320 :
321 :			compiler:
322 :			* Added support for a built-in intinf type.
323 :			- literals
324 :			- pattern matching
325 :			- conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
326 :			- overloading on literals and operations
327 :
328 :			This required adding a primitive type intinf, some additional
329 :			primops, and implementations for several non-trivial intinf
330 :			operations in Core. (The intinf type is completely abstract
331 :			to the compiler; all operations get delegated back to the Core.)
332 :
333 :			* Intinf equality is handled by polyequal. However, the compiler
334 :			does not print its usual warning in this case (since polyequal
335 :			is the right thing to do there).
336 :
337 :			* Improved the organization of structure InlineT.
338 :
339 :			* A word about conversion primops:
340 :			If conversions involving intinf do not cancel out during
341 :	mblume	1348	CPS contract, then the compiler must insert calls to Core functions.
342 :	mblume	1347	Since all core access must be resolved already during the FLINT
343 :			translate phase, it would be too late a the time of CPS contract
344 :	mblume	1348	to add new Core calls. For this reason, conversion primops
345 :	mblume	1347	for intinf carry two arguments: 1. the numeric argument that
346 :			they are supposed to convert, and 2. the Core function that
347 :			can help with this conversion if necessary. If CPS contract
348 :	mblume	1348	eliminates a primop, then the associated Core function becomes
349 :			dead and goes away. Intinf conversion primops that do not get
350 :			eliminated by CPS contract get rewritten into calls of their
351 :			core functions by a separate, new phase.
352 :	mblume	1347
353 :			interactive system:
354 :			- Control.Print.intinfDepth controls max length of intinf constants
355 :			being printed. (Analogous to Control.Print.stringDepth.)
356 :			- Cleanup in printutil and pputil: got rid of unused stuff and
357 :			duplicates; replaced some of the code with code that makes better
358 :			use of library functionality.
359 :
360 :			CM:
361 :			Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
362 :			are no longer silent.
363 :
364 :			CKIT:
365 :			Fixed mismatched uses of Int32 and LargeInt. I always decided
366 :			in favor of LargeInt -- which is now the same as IntInf.
367 :			CKIT-knowledgable people should check whether this is what's
368 :			intended and otherwise change things back to using Int32 or
369 :			FixedInt.
370 :
371 :			Throughout the code:
372 :	mblume	1348	Started using IntInf.int literals and built-in operations
373 :	mblume	1347	(e.g., comparison with 0) where this seems appropriate.
374 :
375 :
376 :			----------------------------------------------------------------------
377 :	macqueen	1343	Name: Dave MacQueen (dbm@cs.uchicago.edu)
378 :			Date: 2003/08/13 11:36:00 CDT
379 :			Tag: dbm-20030813-mcz-merge1
380 :			Description:
381 :
382 :			Merging changes from the mcz-branch development branch into trunk.
383 :			These changes involve replacement of the emulated old prettyprinter
384 :			interface with direct use of the SML/NJ Lib PP library, and fixing
385 :			of a couple of bugs (895, 1186) relating to error messages. A new
386 :			prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
387 :			has been added.
388 :
389 :			----------------------------------------------------------------------
390 :	mblume	1368	Name: Matthias Blume (blume (at) tti - c (dot) org)
391 :	mblume	1342	Date: 2003/08/11 15:45:00 CDT
392 :			Tag: blume-20030811-windows
393 :			Description:
394 :
395 :			Version number bumped to 110.42.9. NEW BOOTFILES!!!
396 :
397 :			http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
398 :
399 :			This patch restores SML/NJ's ability to run under win32. There are a
400 :			number of changes, including fixes for several bugs that had gone
401 :			unnoticed until now:
402 :
403 :			- uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
404 :			trying to run the win32 version from within cygwin.)
405 :
406 :			- There are a number of simple .bat scripts that substitute for their
407 :			corresponding Unix shell-scripts. (See below.)
408 :
409 :			- The internals of ml-build have been modified slightly. The main
410 :			difference is that instead of calling ".link-sml" (or link-sml.bat)
411 :			using OS.Process.system, the ML process delegates this task back
412 :			to the script. Otherwise problems arise in mixed environments such
413 :			as Cygwin where scripts look and work like Unix scripts, but
414 :			where OS.Process.system cannot run them.
415 :
416 :			- In CM, the srcpath pickler used native pathname syntax -- which
417 :			is incorrect in the case of cross-compilation. The new pickle format
418 :			is independent of platform-specific naming conventions.
419 :
420 :			- Path configuration files (such as lib/pathconfig) can now choose
421 :			between native and standard syntax. Placing a line of the form
422 :
423 :			standard!
424 :
425 :			into the file causes all subsequent paths to be interpreted using
426 :			CM standard pathname syntax (= Unix conventions); a line
427 :
428 :			native!
429 :
430 :			switches back to native style. This was needed so that
431 :			path config files can be written portably, see src/system/pathconfig.
432 :
433 :			- Runtime system:
434 :
435 :			- win32-filesys.c: get_file_time and set_file_time now
436 :			access modification time, not creation time.
437 :
438 :			- I/O code made aware of new array representation.
439 :
440 :			- Bug fixes in X86.prim.masm.
441 :
442 :			- src/system/makeml made aware of win32. (For use under cygwin
443 :			and other Unix-environments for windows.)
444 :
445 :			- In Basis, fixed off-by-one error in win32-io.sml (function vecF)
446 :			which caused BinIO.inputAll to fail consistently.
447 :
448 :			.bat scripts:
449 :
450 :			Windows .bat scripts assume that SMLNJ_HOME is defined.
451 :
452 :			- sml.bat, ml-yacc.bat, ml-lex.bat: Driver scripts for standalone
453 :			applications (sml, ml-yacc, ml-lex).
454 :			- ml-build.bat: analogous to ml-build.
455 :			- config\install.bat: Analogous to config/install.sh. This requires
456 :			that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
457 :			(nmake etc. must be on the path, and vcvars32 must have been run.)
458 :			Moreover, sources for ml-lex and ml-yacc need to exist under src,
459 :			and the bootfile hierarchy must have been unpacked under
460 :			sml.boot.x86-win32.
461 :			The script is very primitive and does a poor job at error checking.
462 :			It only installs the base system, ml-lex, and ml-yacc. No other
463 :			libraries are being installed (i.e., you get only those that
464 :			are part of the compiler.)
465 :			- link-sml.bat: analogous to .link-sml, but not currently used
466 :
467 :			Unrelated bug fixes:
468 :
469 :			- ml-nlffigen now exports structures ST_* corresponding to incomplete
470 :			types.
471 :			- Added getDevice to PP/src/pp-debug-fn.sml. (Would not compile
472 :			otherwise.)
473 :
474 :			----------------------------------------------------------------------
475 :	macqueen	1337	Name: Dave MacQueen (macqueen@cs.uchicago.edu)
476 :			Date: 2003/06/17
477 :			Tag: macqueen-20030617-bug895
478 :			Description:
479 :
480 :			Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
481 :			Tag will be used for new development branch (mcz-branch) for
482 :			use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
483 :
484 :			----------------------------------------------------------------------
485 :	mblume	1368	Name: Matthias Blume (blume (at) tti - c (dot) org)
486 :	mblume	1335	Date: 2003/05/27 16:55:00 CDT
487 :			Tag: blume-20030527-polyeq
488 :			Description:
489 :
490 :			Tried to eliminated most cases of polymorphic equality.
491 :
492 :			----------------------------------------------------------------------
493 :	mblume	1368	Name: Matthias Blume (blume (at) tti - c (dot) org)
494 :	mblume	1334	Date: 2003/05/21 17:45:00 CDT
495 :			Tag: blume-20030517-complete
496 :			Description:
497 :
498 :			Two changes:
499 :
500 :			1. Added a flag for controlling whether non-exhaustive bindings will
501 :			be treated as errors (default is false).
502 :			2. Cleaned up the *entire* source tree so that CMB.make goes through
503 :			without a single non-exhaustive match- or bind warning.
504 :
505 :			----------------------------------------------------------------------
506 :	mblume	1368	Name: Matthias Blume (blume (at) tti - c (dot) org)
507 :	mblume	1332	Date: 2003/05/17 10:20:00 CDT
508 :			Tag: blume-20030517-absyn
509 :			Description:
510 :
511 :			1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
512 :
513 :			This mainly affects the quality of error messages. However, some
514 :			of the code is now more straightforward than before. (Treatment of
515 :			the above four constructs in translate.sml is much simpler than
516 :			the "macro-expansion" that was going on before. Plus, the mach-
517 :			compiler no longer gets invoked just to be able to compile an
518 :			if-expression.)
519 :
520 :			2. The ErrorMsg.Error exception is now caught and absorbed by the
521 :			interactive loop.
522 :
523 :			----------------------------------------------------------------------
524 :	allenleung	1327	Name: Allen Leung
525 :			Date: 2003/05/16 13:05:00 CDT
526 :			Tag: leunga-20030516-cygwin-runtime
527 :			Description:
528 :
529 :			Ported the runtime system to cygwin, which uses the unix
530 :			x86-unix bin files. Missing/buggy features:
531 :
532 :			o getnetbyname, getnetbyaddr: these functions seem to be missing in
533 :			the Cygwin library.
534 :			o Ctrl-C handling may be flaky.
535 :			o Windows system calls and Windows I/O are not supported.
536 :
537 :			A new set of binfiles is located at:
538 :
539 :			http://www.dorsai.org/~leunga/boot.x86-unix.tgz
540 :
541 :			This is only needed for bootstrapping the cygwin version of smlnj.
542 :			Other x86 versions can use the existing binfiles.
543 :
544 :			----------------------------------------------------------------------
545 :	mblume	1317	Name: Matthias Blume
546 :			Date: 2003/04/08 15:42:00 CDT
547 :			Tag: blume-20030408-listpair
548 :			Description:
549 :
550 :			1. Added a target 'mlrisc' to installer.
551 :
552 :			2. Added missing elements to structure ListPair.
553 :
554 :			----------------------------------------------------------------------
555 :	leunga	1296	Name: Allen Leung
556 :			Date: 2003/01/07 10:40:00 EST
557 :			Tag: leunga-20030107-int-rem
558 :			Description:
559 :
560 :			Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
561 :			The arguments to the SUBL instruction were swapped.
562 :
563 :			----------------------------------------------------------------------
564 :	blume	1229	Name: Matthias Blume
565 :	blume	1294	Date: 2002/12/12 16:25:00 EST
566 :			Tag: blume-20021212-risc-ra
567 :			Description:
568 :
569 :			Fixed a serious bug in the rewrite code for FP spilling/reloading that
570 :			sent the RA into an infinite loop when floating point registers get
571 :			spilled. (Because of this bug, e.g., nucleic stopped compiling between
572 :			110.37 and 110.38.)
573 :			There was another set of potential problems related to the handling of
574 :			MLRISC annotations (but those did not yet cause real problems, apparently).
575 :
576 :			----------------------------------------------------------------------
577 :			Name: Matthias Blume
578 :	blume	1291	Date: 2002/12/06 22:40:00 EST
579 :			Tag: blume-20021206-cm-fileid
580 :			Description:
581 :
582 :			Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
583 :			This fixes the problem of CM getting confused by files that suddenly
584 :			change their identity (e.g., by getting unlinked and recreated by some
585 :			text editor such as vi). There might be a better/cheaper/cleaner way
586 :			of doing this, but for now this will have to do.
587 :
588 :			----------------------------------------------------------------------
589 :			Name: Matthias Blume
590 :	blume	1288	Date: 2002/10/28 09:50:00 EST
591 :			Tag: blume-20021028-typecheck
592 :			Description:
593 :
594 :			Exported structure Typecheck from $smlnj/viscomp/core.cm.
595 :
596 :			----------------------------------------------------------------------
597 :			Name: Matthias Blume
598 :	blume	1282	Date: 2002/10/17 09:10:00 EDT
599 :	blume	1279	Tag: Release_110_42
600 :			Description:
601 :
602 :	blume	1282	In good old tradition, there has been a slight hiccup so that we have
603 :			to patch 110.42 after the fact. The old release tag has been replaced
604 :			(see below).
605 :
606 :			The change solves a problem with two competing approaches the
607 :			configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
608 :			each other's way.
609 :
610 :			This change only affects the runtime system code and the installer script.
611 :			(No new bootfiles.)
612 :
613 :			----------------------------------------------------------------------
614 :			Name: Matthias Blume
615 :			Date: 2002/10/16 12:00:00 EDT
616 :			Tag: Release_110_42_removed
617 :			Description:
618 :
619 :	blume	1279	New working release. New bootfiles.
620 :
621 :			----------------------------------------------------------------------
622 :			Name: Matthias Blume
623 :	blume	1273	Date: 2002/10/10 13:10:00 EDT
624 :			Tag: blume-20021010-ppc-divs
625 :			Description:
626 :
627 :			The mltree operator DIVS must be implemented with an overflow check on
628 :			the PPC because the hardware indicates divide-by-zero using "overflow" as
629 :			well.
630 :
631 :			----------------------------------------------------------------------
632 :			Name: Matthias Blume
633 :	blume	1261	Date: 2002/07/23 11:45:00
634 :			Tag: blume-20020723-smlnj-home
635 :			Description:
636 :
637 :			Sml now senses the SMLNJ_HOME environment variable. If this is set,
638 :			then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
639 :			CM_PATHCONFIG is also set), the path configuration file is assumed
640 :			to be in $SMLNJ_HOME/lib/pathconfig. This way one can easily move
641 :			the entire tree to some other place and everything will "just work".
642 :
643 :			(Companion commands such as ml-build and ml-makedepend also sense this
644 :			variable.)
645 :
646 :			----------------------------------------------------------------------
647 :			Name: Matthias Blume
648 :	blume	1259	Date: 2002/07/12 21:19:00 EDT
649 :			Tag: blume-20020712-liveness
650 :			Description:
651 :
652 :			Exported two useful "step" functions from liveness module (MLRISC).
653 :
654 :			----------------------------------------------------------------------
655 :			Name: Matthias Blume
656 :	blume	1253	Date: 2002/07/05 16:00 EDT
657 :			Tag: Release_110_41
658 :			Description:
659 :
660 :			New working release. New bootfiles.
661 :
662 :			----------------------------------------------------------------------
663 :			Name: Matthias Blume
664 :	blume	1250	Date: 2002/07/05 10:25:00 EDT
665 :			Tag: blume-20020705-btimp
666 :			Description:
667 :
668 :			Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
669 :			other clients can set up backtracing support.
670 :
671 :			----------------------------------------------------------------------
672 :			Name: Matthias Blume
673 :	blume	1249	Date: 2002/06/25 17:23:00 EDT
674 :			Tag: blume-20020625-fpmax
675 :			Description:
676 :
677 :			Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
678 :			case. (The sense of the isNaN test was reversed -- which made min and
679 :			max always return their first argument.)
680 :
681 :			----------------------------------------------------------------------
682 :			Name: Matthias Blume
683 :	blume	1240	Date: 2002/06/11
684 :			Tag: blume-20020611-unixpath
685 :			Description:
686 :
687 :			Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
688 :
689 :			----------------------------------------------------------------------
690 :			Name: Matthias Blume
691 :	blume	1238	Date: 2002/06/10 16:35:00 EDT
692 :			Tag: blume-20020610-ieeereal
693 :			Description:
694 :
695 :			I back-ported my implementation of IEEEReal.fromString from the
696 :			idlbasis-devel branch so that we can test it.
697 :
698 :			Another small change is that ppDec tries to give more information
699 :			than just "<sig>" in the case of functors. However, this code is
700 :			broken in some mysterious way if the functor's body's signature
701 :			has not been declared by ascription but gets inferred from the
702 :			implementation. This needs fixing...
703 :
704 :			----------------------------------------------------------------------
705 :			Name: Matthias Blume
706 :	blume	1229	Date: 2002/05/31
707 :			Tag: blume-20020531-btrace-mode
708 :			Description:
709 :
710 :			Resurrected SMLofNJ.Internals.BTrace.mode. (It accidentally fell by
711 :			the wayside when I switched over to using Controls everywhere.)
712 :
713 :			----------------------------------------------------------------------
714 :	george	1222	Name: Lal George
715 :			Date: 2002/05/23 12:21:40 EDT
716 :			Tag: george-20020523-visual-labels
717 :			Description:
718 :
719 :			Labels are now displayed in the graphical output to make
720 :			the fall-through and target blocks obvious.
721 :
722 :			----------------------------------------------------------------------
723 :	blume	1204	Name: Matthias Blume
724 :	blume	1220	Date: 2002/05/22 11:03:00 EDT
725 :			Tag: blume-20020522-shrink
726 :			Description:
727 :
728 :			John tweaked yesterday's fix for 1131 to handle an out-of-memory
729 :			situation that comes up when allocating huge arrays.
730 :
731 :			----------------------------------------------------------------------
732 :			Name: Matthias Blume
733 :	blume	1215	Date: 2002/05/21 16:00:00 EDT
734 :			Tag: Release_110_40
735 :			Description:
736 :
737 :			New working release (110.40). New bootfiles.
738 :
739 :			[Also: John Reppy fixed GC bug 1131.]
740 :
741 :			----------------------------------------------------------------------
742 :			Name: Matthias Blume
743 :	blume	1212	Date: 2002/05/21 12:35:00 EDT
744 :			Tag: blume-20020521-cmdoc
745 :			Description:
746 :
747 :			CM documentation update.
748 :
749 :			----------------------------------------------------------------------
750 :			Name: Matthias Blume
751 :	blume	1211	Date: 2002/05/21 10:55:00 EDT
752 :			Tag: blume-20020521-misc
753 :			Description:
754 :
755 :			- John tweaked runtime to be silent on heap export (except when
756 :			GC messages are on).
757 :			- I added a few more things (cross-compiling versions of CMB) to
758 :			config/preloads (as suggestions).
759 :
760 :			----------------------------------------------------------------------
761 :			Name: Matthias Blume
762 :	blume	1208	Date: 2002/05/20 22:25:00 EDT
763 :			Tag: blume-20020520-controls
764 :			Description:
765 :
766 :			- Added ControlUtil structure to control-lib.cm.
767 :			- Use it throughout.
768 :			- Used Controls facility to define MLRISC controls (as opposed to
769 :			registering MLRISC control ref cells with Controls after the fact)
770 :			- Fixed messed-up controls priorities.
771 :
772 :			* Removed again all the stuff from config/preloads that one wouldn't
773 :			be able to preload at the time the initial heap image is built.
774 :			(Many libraries, e.g., CML, do not exist yet at this time. The
775 :			only libraries that can be preloaded via config/preloads are those
776 :			that come bundled with the bootfiles.)
777 :
778 :			----------------------------------------------------------------------
779 :			Name: Matthias Blume
780 :	blume	1204	Date: 2002/05/20 10:59:00 EDT
781 :			Tag: blume-20020520-preloads
782 :			Description:
783 :
784 :			Added a lot of commented-out suggestions for things to be included
785 :			in config/preloads.
786 :
787 :			----------------------------------------------------------------------
788 :	leunga	1203	Name: Allen Leung
789 :			Date: 2002/05/18 14:20:00 EDT
790 :			Tag: leunga-20020518-mdl
791 :			Description:
792 :
793 :			o Made the mdl tool stuff compile and run again.
794 :			o I've disabled all the stuff that depends on RTL specifications; they
795 :			are all badly broken anyway.
796 :
797 :			----------------------------------------------------------------------
798 :	blume	1201	Name: Matthias Blume
799 :			Date: 2002/05/17 16:49:00 EDT
800 :			Tag: blume-20020517-controls
801 :			Description:
802 :
803 :			0. John Reppy made several modifications to the SML/NJ library.
804 :			In particular, there is a shiny new controls-lib.cm.
805 :
806 :			1. Pushed new controls interface through compiler so that everything
807 :			compiles again.
808 :
809 :			2. Added FormatComb and FORMAT_COMB to the CML version of the
810 :			SML/NJ library (so that CML compiles again).
811 :
812 :			3. Modified init scripts because XXX_DEFAULT environment variables
813 :			are no longer with us. (Boot-time initialization is now done
814 :			using the same environment variables that are also used for
815 :			startup-time initialization of controls.)
816 :
817 :			----------------------------------------------------------------------
818 :	george	1192	Name: Lal George
819 :			Date: 2002/05/15 09:20:10 EDT
820 :			Tag: george-20020515-pseudo-op-decls
821 :			Description:
822 :
823 :			All pseudo-ops emitted before the first segment declaration
824 :			such as TEXT, DATA, and BSS directives are assumed to be global
825 :			declarations and are emitted first in the assembly file. This is
826 :			useful in a number of situations where one has pseudo-ops that are not
827 :			specific to any segment, and also works around the constraint that one
828 :			cannot have client pseudo-ops in the TEXT segment.
829 :
830 :			Because no segment is associated with these declarations it is
831 :			an error to allocate any space or objects before the first segment
832 :			directive and an exception will be raised. However, we cannot make
833 :			this check for client pseudo-ops.
834 :
835 :			These top level declarations are a field in the CFG graph_info.
836 :			In theory you can continue to add to this field after the CFG has been
837 :			built -- provided you know what you are doing;-)
838 :
839 :			----------------------------------------------------------------------
840 :	blume	1175	Name: Matthias Blume
841 :	blume	1190	Date: 2002/05/13 16:40:00 EDT
842 :			Tag: blume-20020513-pp-etc
843 :			Description:
844 :
845 :			A few minor bugfixes:
846 :
847 :			- Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
848 :			(Bogus printouts for redefined bindings still occur. Compiler
849 :			bug should no longer occur now. We need to redo the prettyprinter
850 :			from scratch.)
851 :
852 :			- CM pathname printer now also adds escape sequences for ( and )
853 :
854 :			- commend and docu fixes for ml-nlffi
855 :
856 :			----------------------------------------------------------------------
857 :			Name: Matthias Blume
858 :	blume	1189	Date: 2002/05/10 16:40:00 EDT
859 :			Tag: blume-20020510-erg-textio
860 :			Description:
861 :
862 :			Applied the following bugfix provided by Emden Gansner:
863 :
864 :			Output is corrupted when outputSubstr is used rather than output.
865 :
866 :			The problem occurs when a substring
867 :
868 :			ss = (s, dataStart, dataLen)
869 :
870 :			where dataStart > 0, fills a stream buffer with avail bytes left.
871 :			avail bytes of s, starting at index dataStart, are copied into the
872 :			buffer, the buffer is flushed, and then the remaining dataLen-avail
873 :			bytes of ss are copied into the beginning of the buffer. Instead of
874 :			starting this copy at index dataStart+avail in s, the current code
875 :			starts the copy at index avail.
876 :
877 :			Fix:
878 :			In text-io-fn.sml, change line 695 from
879 :			val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
880 :			to
881 :			val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
882 :
883 :			----------------------------------------------------------------------
884 :			Name: Matthias Blume
885 :	blume	1186	Date: 2002/04/12 13:55:00 EDT
886 :			Tag: blume-20020412-assyntax
887 :			Description:
888 :
889 :			1. Grabbed newer assyntax.h from the XFree86 project.
890 :			2. Fiddled with how to compile X86.prim.asm without warnings.
891 :			3. (Very) Minor cleanup in CM.
892 :
893 :			----------------------------------------------------------------------
894 :			Name: Matthias Blume
895 :	blume	1185	Date: 2002/04/01 (no joke!) 17:07:00 EST
896 :			Tag: blume-20020401-x86div
897 :			Description:
898 :
899 :			Added full support for div/mod/rem/quot on the x86, using the machine
900 :			instruction's two results (without clumsily recomputing the remainder)
901 :			directly where appropriate.
902 :
903 :			Some more extensive power-of-two support was added to the x86 instruction
904 :			selector (avoiding expensive divs, mods, and muls where they can be
905 :			replaced with cheaper shifts and masks). However, this sort of thing
906 :			ought to be done earlier, e.g., within the CPS optimizer so that
907 :			all architectures benefit from it.
908 :
909 :			The compiler compiles to a fixed point, but changes might be somewhat
910 :			fragile nevertheless. Please, report any strange things that you might
911 :			see wrt. div/mod/quot/rem...
912 :
913 :			----------------------------------------------------------------------
914 :			Name: Matthias Blume
915 :	blume	1184	Date: 2002/03/29 17:22:00
916 :			Tag: blume-20020329-div
917 :			Description:
918 :
919 :			Fixed my broken div/mod logic. Unfortunately, this means that the
920 :			inline code for div/mod now has one more comparison than before.
921 :			Fast paths (quotient > 0 or remainder = 0) are not affected, though.
922 :			The problem was with quotient = 0, because that alone does not tell
923 :			us which way the rounding went. One then has to look at whether
924 :			remainder and divisor have the same sign... :(
925 :
926 :			Anyway, I replaced the bootfiles with fresh ones...
927 :
928 :			----------------------------------------------------------------------
929 :			Name: Matthias Blume
930 :	blume	1183	Date: 2002/03/29 14:10:00 EST
931 :			Tag: blume-20020329-inlprims
932 :			Description:
933 :
934 :			NEW BOOTFILES!!! Version number bumped to 110.39.3.
935 :
936 :			Primops have changed. This means that the bin/boot-file formats have
937 :			changed as well.
938 :
939 :			To make sure that there is no confusion, I made a new version.
940 :
941 :
942 :			CHANGES:
943 :
944 :			* removed REMT from mltree (remainder should never overflow).
945 :
946 :			* added primops to deal with divisions of all flavors to the frontend
947 :
948 :			* handled these primops all the way through so they map to their respective
949 :			MLRISC support
950 :
951 :			* used these primops in the implementation of Int, Int32, Word, Word32
952 :
953 :			* removed INLDIV, INLMOD, and INLREM as they are no longer necessary
954 :
955 :			* parameterized INLMIN, INLMAX, and INLABS by a numkind
956 :
957 :			* translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
958 :			floating point
959 :
960 :			* used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
961 :			and Real (but Real.abs maps to a separate floating-point-only primop)
962 :
963 :
964 :			TODO items:
965 :
966 :			* Hacked Alpha32 instruction selection, disabling the selection of REMx
967 :			instructions because the machine instruction encoder cannot handle
968 :			them. (Hppa, PPC, and Sparc instruction selection did not handle
969 :			REM in the first place, and REM is supported by the x86 machine coder.)
970 :
971 :			* Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
972 :			selection phase. (The two can be streamlined because the hardware
973 :			delivers both quotient and remainder at the same time anyway.)
974 :
975 :			* Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
976 :			(Currently the behavior is inconsistent both across architectures and
977 :			wrt. the draft Basis spec.)
978 :
979 :			* Word8 should eventually be handled natively, too.
980 :
981 :			* There seems to be one serious bug in mltree-gen.sml. It appears, though,
982 :			as if there currently is no execution path that could trigger it in
983 :			SML/NJ. (The assumptions underlying functions arith and promotable do not
984 :			hold for things like multiplication and division.)
985 :
986 :			----------------------------------------------------------------------
987 :			Name: Matthias Blume
988 :	blume	1181	Date: 2002/03/27 16:27:00 EST
989 :			Tag: blume-20020327-mlrisc-divisions
990 :			Description:
991 :
992 :			Added support for all four division operations (ML's div, mod, quot,
993 :			and rem) to MLRISC. In the course of doing so, I also rationalized
994 :			the naming (no more annoying switch-around of DIV and QUOT), by
995 :			parameterizing the operation by div_rounding_mode (which can be either
996 :			DIV_TO_ZERO or DIV_TO_NEGINF).
997 :
998 :			The generic MLTreeGen functor takes care of compiling all four
999 :			operations down to only round-to-zero div.
1000 :
1001 :			Missing pieces:
1002 :
1003 :			* Doing something smarter than relying on MLTreeGen on architectures
1004 :			like, e.g., the x86 where hardware division delivers both quotient and
1005 :			remainder at the same time. With this, the implementation of the
1006 :			round-to-neginf operations could be further streamlined.
1007 :
1008 :			* Remove inlining support for div/mod/rem from the frontend and replace it
1009 :			with primops that get carried through to the backend. Do this for all
1010 :			int and word types.
1011 :
1012 :			----------------------------------------------------------------------
1013 :			Name: Matthias Blume
1014 :	blume	1180	Date: 2002/03/25 17:25:00 EST
1015 :			Tag: blume-20020325-divmod
1016 :			Description:
1017 :
1018 :			I improved (hopefully without breaking them) the implementation of Int.div,
1019 :			Int.mod, and Int.rem. For this, the code in translate.sml now takes
1020 :			advantage of the following observations:
1021 :
1022 :			Let q = x quot y r = x rem y
1023 :			d = x div y m = x mod y
1024 :
1025 :			where "quot" is the round-to-zero version of integer division that
1026 :			hardware usually provides. Then we have:
1027 :
1028 :			r = x - q * y where neither the * nor the - will overflow
1029 :			d = if q >= 0 orelse x = q * y then q else q - 1
1030 :			where neither the * nor the - will overflow
1031 :			m = if q >= 0 orelse r = 0 then r else r + y
1032 :			where the + will not overflow
1033 :
1034 :			This results in substantial simplification of the generated code.
1035 :			The following table shows the number of CFG nodes and edges generated
1036 :			for
1037 :			fun f (x, y) = x OPER y
1038 :			(* with OPER \in div, mod, quot, rem *)
1039 :
1040 :
1041 :			OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
1042 :			--------------------------------------------------------
1043 :			div | 24 | 39 | 12 | 16
1044 :			mod | 41 | 71 | 12 | 16
1045 :			quot | 8 | 10 | 8 | 10
1046 :			rem | 10 | 14 | 8 | 10
1047 :
1048 :
1049 :			----------------------------------------------------------------------
1050 :			Name: Matthias Blume
1051 :	blume	1179	Date: 2002/03/25 22:06:00 EST
1052 :			Tag: blume-20020325-cprotobug
1053 :			Description:
1054 :
1055 :			Fixed a bug in cproto (c prototype decoder).
1056 :
1057 :			----------------------------------------------------------------------
1058 :			Name: Matthias Blume
1059 :	blume	1178	Date: 2002/03/25 16:00:00 EST
1060 :			Tag: blume-20020325-raw-primops
1061 :			Description:
1062 :
1063 :			I did some cleanup to Allen's new primop code and
1064 :			replaced yesterday's bootfiles with new ones.
1065 :			(But they are stored in the same place.)
1066 :
1067 :			----------------------------------------------------------------------
1068 :			Name: Matthias Blume
1069 :	blume	1175	Date: 2002/03/24 22:40:00 EST
1070 :			Tag: blume-20020324-bootfiles
1071 :			Description:
1072 :
1073 :			Made the bootfiles that Allen asked for.
1074 :
1075 :			----------------------------------------------------------------------
1076 :	leunga	1174	Name: Allen Leung
1077 :			Date: 2002/03/23 15:50:00 EST
1078 :			Tag: leunga-20020323-flint-cps-rcc-primops
1079 :			Description:
1080 :
1081 :			1. Changes to FLINT primops:
1082 :
1083 :			(* make a call to a C-function;
1084 :			* The primop carries C function prototype information and specifies
1085 :			* which of its (ML-) arguments are floating point. C prototype
1086 :			* information is for use by the backend, ML information is for
1087 :			* use by the CPS converter. *)
1088 :			| RAW_CCALL of { c_proto: CTypes.c_proto,
1089 :			ml_args: ccall_type list,
1090 :			ml_res_opt: ccall_type option,
1091 :			reentrant : bool
1092 :			} option
1093 :			(* Allocate uninitialized storage on the heap.
1094 :			* The record is meant to hold short-lived C objects, i.e., they
1095 :			* are not ML pointers. With the tag, the representation is
1096 :			* the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
1097 :			*)
1098 :			| RAW_RECORD of {tag:bool,sz:int}
1099 :			and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
1100 :
1101 :			2. These CPS primops are now overloaded:
1102 :
1103 :			rawload of {kind:numkind}
1104 :			rawstore of {kind:numkind}
1105 :
1106 :			The one argument form is:
1107 :
1108 :			rawload {kind} address
1109 :
1110 :			The two argument form is:
1111 :
1112 :			rawload {kind} [ml object, byte-offset]
1113 :
1114 :			3. RAW_CCALL/RCC now takes two extra arguments:
1115 :
1116 :			a. The first is whether the C call is reentrant, i.e., whether
1117 :			ML state should be saved and restored.
1118 :			b. The second argument is a string argument specifying the name of
1119 :			library and the C function.
1120 :
1121 :			These things are currently not handled in the code generator, yet.
1122 :
1123 :			4. In CProto,
1124 :
1125 :			An encoding type of "bool" means "ml object" and is mapped into
1126 :			C prototype of PTR. Note that "bool" is different than "string",
1127 :			even though "string" is also mapped into PTR, because "bool"
1128 :			is assigned an CPS type of BOGt, while "string" is assigned INT32t.
1129 :
1130 :			5. Pickler/unpicker
1131 :
1132 :			Changed to handle RAW_RECORD and newest RAW_CCALL
1133 :
1134 :			6. MLRiscGen,
1135 :
1136 :			1. Changed to handle the new rawload/rawstore/rawrecord operators.
1137 :			2. Code for handling C Calls has been moved to a new module CPSCCalls,
1138 :			in the file CodeGen/cpscompile/cps-c-calls.sml
1139 :
1140 :			7. Added the conditional move operator
1141 :
1142 :			condmove of branch
1143 :
1144 :			to cps. Generation of this is still buggy so it is currently
1145 :			disabled.
1146 :
1147 :			----------------------------------------------------------------------
1148 :	george	1168	Name: Lal George
1149 :			Date: 2002/03/22 14:18:25 EST
1150 :	george	1171	Tag: george-20020322-cps-branch-prob
1151 :	george	1168	Description:
1152 :
1153 :	george	1171	Implemented the Ball-Larus branch prediction-heuristics, and
1154 :	george	1168	incorporated graphical viewers for control flow graphs.
1155 :
1156 :	george	1171	Ball-Larus Heuristics:
1157 :	george	1168	---------------------
1158 :			See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
1159 :
1160 :			By design it uses the Dempster-Shafer theory for combining
1161 :			probabilities. For example, in the function:
1162 :
1163 :			fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
1164 :
1165 :			the ball-larus heuristics predicts that the n=0 is unlikely
1166 :			(OH-heuristic), and the 'then' branch is unlikely because of the
1167 :			RH-heuristic -- giving the 'then' branch an even lower combined
1168 :	george	1171	probability using the Dempster-Shafer theory.
1169 :	george	1168
1170 :			Finally, John Reppy's loop analysis in MLRISC, further lowers the
1171 :			probability of the 'then' branch because of the loop in the else
1172 :			branch.
1173 :
1174 :
1175 :			Graphical Viewing:
1176 :			------------------
1177 :			I merely plugged in Allen's graphical viewers into the compiler. The
1178 :			additional code is not much. At the top level, saying:
1179 :
1180 :			Control.MLRISC.getFlag "cfg-graphical-view" := true;
1181 :
1182 :			will display the graphical view of the control flow graph just before
1183 :			back-patching. daVinci must be in your path for this to work. If
1184 :			daVinci is not available, then the default viewer can be changed
1185 :			using:
1186 :
1187 :			Control.MLRISC.getString "viewer"
1188 :
1189 :			which can be set to "dot" or "vcg" for the corresponding viewers. Of
1190 :			course, these viewers must be in your path.
1191 :
1192 :			The above will display the compilation unit at the level of clusters,
1193 :			many of which are small, boring, and un-interesting. Also setting:
1194 :
1195 :			Control.MLRISC.getInt "cfg-graphical-view_size"
1196 :
1197 :			will display clusters that are larger than the value set by the above.
1198 :
1199 :
1200 :			----------------------------------------------------------------------
1201 :	blume	1159	Name: Matthias Blume
1202 :			Date: 2002/03/21 22:20:00 EST
1203 :			Tag: blume-20020321-kmp-bugfix
1204 :			Description:
1205 :
1206 :			Changed the interface to the KMP routine in PreString and fixed
1207 :			a minor bug in one place where it was used.
1208 :
1209 :			----------------------------------------------------------------------
1210 :	leunga	1156	Name: Allen Leung
1211 :	leunga	1158	Date: 2002/03/21 20:30:00 EST
1212 :			Tag: leunga-20020321-cfg
1213 :			Description:
1214 :
1215 :			Fixed a potential problem in cfg edge splitting.
1216 :
1217 :			----------------------------------------------------------------------
1218 :			Name: Allen Leung
1219 :	leunga	1156	Date: 2002/03/21 17:15:00 EST
1220 :			Tag: leunga-20020321-x86-fp-cfg
1221 :			Description:
1222 :
1223 :			1. Recoded the buggy parts of x86-fp.
1224 :
1225 :			a. All the block reordering code has been removed.
1226 :			We now depend on the block placement phases to do this work.
1227 :
1228 :			b. Critical edge splitting code has been simplified and moved into the
1229 :			CFG modules, as where they belong.
1230 :
1231 :			Both of these were quite buggy and complex. The code is now much, much
1232 :			simpler.
1233 :
1234 :			2. X86 backend.
1235 :
1236 :			a. Added instructions for 64-bit support. Instruction selection for
1237 :			64-bit has not been committed, however, since that
1238 :			requires changes to MLTREE which haven't been approved by
1239 :			Lal and John.
1240 :
1241 :			b. Added support for FUCOMI and FUCOMIP when generating code for
1242 :			PentiumPro and above. We only generate these instructions in
1243 :			the fast-fp mode.
1244 :
1245 :			c. Added cases for JP and JNP in X86FreqProps.
1246 :
1247 :			3. CFG
1248 :
1249 :			CFG now has a bunch of methods for edge splitting and merging.
1250 :
1251 :			4. Machine description.
1252 :
1253 :			John's simplification of MLTREE_BASIS.fcond broke a few machine
1254 :			description things:
1255 :
1256 :			rtl-build.{sig,sml} and hppa.mdl fixed.
1257 :
1258 :			NOTE: the machine description stuff in the repository is still broken.
1259 :			Again, I can't put my fixes in because that involves
1260 :			changes to MLTREE.
1261 :
1262 :			----------------------------------------------------------------------
1263 :	blume	1145	Name: Matthias Blume
1264 :	blume	1155	Date: 2002/03/20 15:55:00 EST
1265 :			Tag: blume-20020320-kmp
1266 :			Description:
1267 :
1268 :			Implemented Knuth-Morris-Pratt string matching in PreString and used
1269 :			it for String.isSubstring, Substring.isSubstring, and
1270 :			Substring.position.
1271 :
1272 :			(Might need some stress-testing. Simple examples worked fine.)
1273 :
1274 :			----------------------------------------------------------------------
1275 :			Name: Matthias Blume
1276 :	blume	1152	Date: 2002/03/19 16:37:00 EST
1277 :			Tag: blume-20020319-witnesses
1278 :			Description:
1279 :
1280 :			Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
1281 :
1282 :			This implements a generic mechanism for changing constness qualifiers
1283 :			anywhere within big C types without resorting to outright "casts".
1284 :			(So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
1285 :			modify the constness at the outermost level.)
1286 :			The implementation of "convert" is based on the idea of "witness"
1287 :			values -- values that are not used by the operation but whose types
1288 :			"testify" to their applicability. On the implementation side, "convert"
1289 :			is simply a projection (returning its second curried argument). With
1290 :			cross-module inlining, it should not result in any machine code being
1291 :			generated.
1292 :
1293 :			----------------------------------------------------------------------
1294 :			Name: Matthias Blume
1295 :	blume	1148	Date: 2002/03/15 16:40:00 EST
1296 :			Tag: blume-20020315-basis
1297 :			Description:
1298 :
1299 :			Provided (preliminary?) implementations for
1300 :
1301 :			{String,Substring}.{concatWith,isSuffix,isSubstring}
1302 :
1303 :			and
1304 :
1305 :			Substring.full
1306 :
1307 :			Those are in the Basis spec but they were missing in SML/NJ.
1308 :
1309 :			----------------------------------------------------------------------
1310 :			Name: Matthias Blume
1311 :	blume	1145	Date: 2002/03/14 21:30:00 EST
1312 :			Tag: blume-20020314-controls
1313 :			Description:
1314 :
1315 :			Controls:
1316 :			---------
1317 :
1318 :			1. Factored out the recently-added Controls : CONTROLS stuff and put
1319 :			it into its own library $/controls-lib.cm. The source tree for
1320 :			this is under src/smlnj-lib/Controls.
1321 :
1322 :			2. Changed the names of types and functions in this interface, so they
1323 :			make a bit more "sense":
1324 :
1325 :			module -> registry
1326 :			'a registry -> 'a group
1327 :
1328 :			3. The interface now deals in ref cells only. The getter/setter interface
1329 :			is (mostly) gone.
1330 :
1331 :			4. Added a function that lets one register an already-existing ref cell.
1332 :
1333 :			5. Made the corresponding modifications to the rest of the code so that
1334 :			everything compiles again.
1335 :
1336 :			6. Changed the implementation of Controls.MLRISC back to something closer
1337 :			to the original. In particular, this module (and therefore MLRISC)
1338 :			does not depend on Controls. There now is some link-time code in
1339 :			int-sys.sml that registers the MLRISC controls with the Controls
1340 :			module.
1341 :
1342 :			CM:
1343 :			---
1344 :
1345 :			* One can now specify the lambda-split aggressiveness in init.cmi.
1346 :
1347 :			----------------------------------------------------------------------
1348 :	leunga	1142	Name: Allen Leung
1349 :			Date: 2002/03/13 17:30:00 EST
1350 :			Tag: leunga-20020313-x86-fp-unary
1351 :			Description:
1352 :
1353 :			Bug fix for:
1354 :
1355 :			> leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
1356 :			> Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
1357 :			> - fun f(x,(y,z)) = Real.~ y;
1358 :			> [autoloading]
1359 :			> [autoloading done]
1360 :			> fchsl (%eax), 184(%esp)
1361 :			> Error: MLRisc bug: X86MCEmitter.emitInstr
1362 :			>
1363 :			> uncaught exception Error
1364 :			> raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
1365 :
1366 :			The problem was that the code generator did not generate any fp registers
1367 :			in this case, and the ra didn't know that it needed to run the X86FP phase to
1368 :			translate the pseudo fp instruction. This only happened with unary fp
1369 :			operators in certain situations.
1370 :
1371 :			----------------------------------------------------------------------
1372 :	blume	1137	Name: Matthias Blume
1373 :	blume	1140	Date: 2002/03/13 14:00:00 EST
1374 :			Tag: blume-20020313-overload-etc
1375 :			Description:
1376 :
1377 :			1. Added _overload as a synonym for overload for backward compatibility.
1378 :			(Control.overloadKW must be true for either version to be accepted.)
1379 :
1380 :			2. Fixed bug in install script that caused more things to be installed
1381 :			than what was requested in config/targets.
1382 :
1383 :			3. Made CM aware of the (_)overload construct so that autoloading
1384 :			works.
1385 :
1386 :			----------------------------------------------------------------------
1387 :			Name: Matthias Blume
1388 :	blume	1138	Date: 2002/03/12 22:03:00 EST
1389 :			Tag: blume-20020312-url
1390 :			Description:
1391 :
1392 :			Forgot to update BOOT and srcarchiveurl.
1393 :
1394 :			----------------------------------------------------------------------
1395 :			Name: Matthias Blume
1396 :	blume	1137	Date: 2002/03/12 17:30:00 EST
1397 :			Tag: blume-20020312-version110392
1398 :			Description:
1399 :
1400 :			Yet another version number bump (because of small changes to the
1401 :			binfile format). Version number is now 110.39.2. NEW BOOTFILES!
1402 :
1403 :			Changes:
1404 :
1405 :			The new pid generation scheme described a few weeks ago was overly
1406 :			complicated. I implemented a new mechanism that is simpler and
1407 :			provides a bit more "stability": Once CM has seen a compilation
1408 :			unit, it keeps its identity constant (as long as you do not delete
1409 :			those crucial CM/GUID/* files). This means that when you change
1410 :	blume	1145	an interface, compile, then go back to the old interface, and
1411 :	blume	1137	compile again, you arrive at the original pid.
1412 :
1413 :			There now also is a mechanism that instructs CM to use the plain
1414 :			environment hash as a module's pid (effectively making its GUID
1415 :			the empty string). For this, "noguid" must be specified as an
1416 :			option to the .sml file in question within its .cm file.
1417 :			This is most useful for code that is being generated by tools such
1418 :			as ml-nlffigen (because during development programmers tend to
1419 :			erase the tool's entire output directory tree including CM's cached
1420 :			GUIDs). "noguid" is somewhat dangerous (since it can be used to locally
1421 :			revert to the old, broken behavior of SML/NJ, but in specific cases
1422 :			where there is no danger of interface confusion, its use is ok
1423 :			(I think).
1424 :
1425 :			ml-nlffigen by default generates "noguid" annotations. They can be
1426 :			turned off by specifying -guid in its command line.
1427 :
1428 :			----------------------------------------------------------------------
1429 :	george	1136	Name: Lal George
1430 :			Date: 2002/03/12 12 14:42:36 EST
1431 :			Tag: george-20020312-frequency-computation
1432 :			Description:
1433 :
1434 :			Integrated jump chaining and static block frequency into the
1435 :			compiler. More details and numbers later.
1436 :	leunga	1025
1437 :	george	1003	----------------------------------------------------------------------
1438 :	george	1133	Name: Lal George
1439 :			Date: 2002/03/11 11 22:38:53 EST
1440 :			Tag: george-20020311-jump-chain-elim
1441 :			Description:
1442 :
1443 :			Tested the jump chain elimination on all architectures (except the
1444 :			hppa). This is on by default right now and is profitable for the
1445 :			alpha and x86, however, it may not be profitable for the sparc and ppc
1446 :			when compiling the compiler.
1447 :
1448 :			The gc test will typically jump to a label at the end of the cluster,
1449 :			where there is another jump to an external cluster containing the actual
1450 :			code to invoke gc. This is to allow factoring of common gc invocation
1451 :			sequences. That is to say, we generate:
1452 :
1453 :			f:
1454 :			testgc
1455 :			ja L1 % jump if above to L1
1456 :
1457 :			L1:
1458 :			jmp L2
1459 :
1460 :
1461 :			After jump chain elimination the 'ja L1' instructions is converted to
1462 :			'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
1463 :			up being implemented in their long form (if L2 is far away) using:
1464 :
1465 :			jbe L3 % jump if below or equal to L3
1466 :			jmp L2
1467 :			L3:
1468 :			...
1469 :
1470 :
1471 :			For large compilation units L2 may be far away.
1472 :
1473 :
1474 :			----------------------------------------------------------------------
1475 :	blume	1132	Name: Matthias Blume
1476 :			Date: 2002/03/11 13:30:00 EST
1477 :			Tag: blume-20020311-mltreeeval
1478 :			Description:
1479 :
1480 :			A functor parameter was missing.
1481 :
1482 :			----------------------------------------------------------------------
1483 :	leunga	1130	Name: Allen Leung
1484 :	leunga	1131	Date: 2002/03/11 10:30:00 EST
1485 :	leunga	1142	Tag: leunga-20020311-runtime-string0
1486 :	leunga	1131	Description:
1487 :
1488 :			The representation of the empty string now points to a
1489 :			legal null terminated C string instead of unit. It is now possible
1490 :			to convert an ML string into C string with InlineT.CharVector.getData.
1491 :			This compiles into one single machine instruction.
1492 :
1493 :			----------------------------------------------------------------------
1494 :			Name: Allen Leung
1495 :	leunga	1130	Date: 2002/03/10 23:55:00 EST
1496 :			Tag: leunga-20020310-x86-call
1497 :			Description:
1498 :
1499 :			Added machine generation for CALL instruction (relative displacement mode)
1500 :
1501 :			----------------------------------------------------------------------
1502 :	blume	1128	Name: Matthias Blume
1503 :			Date: 2002/03/08 16:05:00
1504 :			Tag: blume-20020308-entrypoints
1505 :			Description:
1506 :
1507 :			Version number bumped to 110.39.1. NEW BOOTFILES!
1508 :
1509 :			Entrypoints: non-zero offset into a code object where execution should begin.
1510 :
1511 :			- Added the notion of an entrypoint to CodeObj.
1512 :			- Added reading/writing of entrypoint info to Binfile.
1513 :			- Made runtime system bootloader aware of entrypoints.
1514 :			- Use the address of the label of the first function given to mlriscGen
1515 :			as the entrypoint. This address is currently always 0, but it will
1516 :			not be 0 once we turn on block placement.
1517 :			- Removed the linkage cluster code (which was The Other Way(tm) of dealing
1518 :			with entry points) from mlriscGen.
1519 :
1520 :			----------------------------------------------------------------------
1521 :	leunga	1127	Name: Allen Leung
1522 :			Date: 2002/03/07 20:45:00 EST
1523 :			Tag: leunga-20020307-x86-cmov
1524 :			Description:
1525 :
1526 :			Bug fixes for CMOVcc on x86.
1527 :
1528 :			1. Added machine code generation for CMOVcc
1529 :			2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
1530 :			3. CMOVcc cannot have an immediate operand as argument.
1531 :
1532 :			----------------------------------------------------------------------
1533 :	blume	1126	Name: Matthias Blume
1534 :			Date: 2002/03/07 16:15:00 EST
1535 :			Tag: blume-20020307-controls
1536 :			Description:
1537 :
1538 :			This is a very large but mostly boring patch which makes (almost)
1539 :			every tuneable compiler knob (i.e., pretty much everything under
1540 :			Control.* plus a few other things) configurable via both the command
1541 :			line and environment variables in the style CM did its configuration
1542 :			until now.
1543 :
1544 :			Try starting sml with '-h' (or, if you are brave, '-H')
1545 :
1546 :			To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
1547 :			implements the underlying generic mechanism.
1548 :
1549 :			The interface to some of the existing such facilities has changed somewhat.
1550 :			For example, the MLRiscControl module now provides mkFoo instead of getFoo.
1551 :			(The getFoo interface is still there for backward-compatibility, but its
1552 :			use is deprecated.)
1553 :
1554 :			The ml-build script passes -Cxxx=yyy command-line arguments through so
1555 :			that one can now twiddle the compiler settings when using this "batch"
1556 :			compiler.
1557 :
1558 :			TODO items:
1559 :
1560 :			We should go through and throw out all controls that are no longer
1561 :			connected to anything. Moreover, we should go through and provide
1562 :			meaningful (and correct!) documentation strings for those controls
1563 :			that still are connected.
1564 :
1565 :			Currently, multiple calls to Controls.new are accepted (only the first
1566 :			has any effect). Eventually we should make sure that every control
1567 :			is being made (via Controls.new) exactly once. Future access can then
1568 :			be done using Controls.acc.
1569 :
1570 :			Finally, it would probably be a good idea to use the getter-setter
1571 :			interface to controls rather than ref cells. For the time being, both
1572 :			styles are provided by the Controls module, but getter-setter pairs are
1573 :			better if thread-safety is of any concern because they can be wrapped.
1574 :
1575 :			*****************************************
1576 :
1577 :			One bug fix: The function blockPlacement in three of the MLRISC
1578 :			backpatch files used to be hard-wired to one of two possibilities at
1579 :			link time (according to the value of the placementFlag). But (I
1580 :			think) it should rather sense the flag every time.
1581 :
1582 :			*****************************************
1583 :
1584 :			Other assorted changes (by other people who did not supply a HISTORY entry):
1585 :
1586 :			1. the cross-module inliner now works much better (Monnier)
1587 :			2. representation of weights, frequencies, and probabilities in MLRISC
1588 :			changed in preparation of using those for weighted block placement
1589 :			(Reppy, George)
1590 :
1591 :			----------------------------------------------------------------------
1592 :	george	1116	Name: Lal George
1593 :	george	1124	Date: 2002/03/07 14:44:24 EST 2002
1594 :			Tag: george-20020307-weighted-block-placement
1595 :
1596 :			Tested the weighted block placement optimization on all architectures
1597 :			(except the hppa) using AMPL to generate the block and edge frequencies.
1598 :			Changes were required in the machine properties to correctly
1599 :			categorize trap instructions. There is an MLRISC flag
1600 :			"weighted-block-placement" that can be used to enable weighted block
1601 :			placement, but this will be ineffective without block/edge
1602 :			frequencies (coming soon).
1603 :
1604 :
1605 :			----------------------------------------------------------------------
1606 :			Name: Lal George
1607 :	george	1116	Date: 2002/03/05 17:24:48 EST
1608 :			Tag: george-20020305-linkage-cluster
1609 :
1610 :	george	1124	In order to support the block placement optimization, a new cluster
1611 :			is generated as the very first cluster (called the linkage cluster).
1612 :			It contains a single jump to the 'real' entry point for the compilation
1613 :			unit. Block placement has no effect on the linkage cluster itself, but
1614 :			all the other clusters have full freedom in the manner in which they
1615 :			reorder blocks or functions.
1616 :	george	1116
1617 :			On the x86 the typical linkage code that is generated is:
1618 :			----------------------
1619 :			.align 2
1620 :			L0:
1621 :			addl $L1-L0, 72(%esp)
1622 :	george	1124	jmp L1
1623 :	george	1116
1624 :
1625 :			.align 2
1626 :			L1:
1627 :			----------------------
1628 :
1629 :			72(%esp) is the memory location for the stdlink register. This
1630 :			must contain the address of the CPS function being called. In the
1631 :	george	1124	above example, it contains the address of L0; before
1632 :	george	1116	calling L1 (the real entry point for the compilation unit), it
1633 :			must contain the address for L1, and hence
1634 :
1635 :			addl $L1-L0, 72(%esp)
1636 :
1637 :	george	1124	I have tested this on all architectures except the hppa.The increase
1638 :			in code size is of course negligible
1639 :	george	1116
1640 :			----------------------------------------------------------------------
1641 :	leunga	1115	Name: Allen Leung
1642 :			Date: 2002/03/03 13:20:00 EST
1643 :			Tag: leunga-20020303-mlrisc-tools
1644 :
1645 :			Added #[ ... ] expressions to mlrisc tools
1646 :
1647 :			----------------------------------------------------------------------
1648 :	blume	1096	Name: Matthias Blume
1649 :	blume	1098	Date: 2002/02/27 12:29:00 EST
1650 :			Tag: blume-20020227-cdebug
1651 :			Description:
1652 :
1653 :			- made types in structure C and C_Debug to be equal
1654 :			- got rid of code duplication (c-int.sml vs. c-int-debug.sml)
1655 :			- there no longer is a C_Int_Debug (C_Debug is directly derived from C)
1656 :
1657 :			----------------------------------------------------------------------
1658 :			Name: Matthias Blume
1659 :	blume	1096	Date: 2002/02/26 12:00:00 EST
1660 :			Tag: blume-20020226-ffi
1661 :			Description:
1662 :
1663 :			1. Fixed a minor bug in CM's "noweb" tool:
1664 :			If numbering is turned off, then truly don't number (i.e., do not
1665 :			supply the -L option to noweb). The previous behavior was to supply
1666 :			-L'' -- which caused noweb to use the "default" line numbering scheme.
1667 :			Thanks to Chris Richards for pointing this out (and supplying the fix).
1668 :
1669 :			2. Once again, I reworked some aspects of the FFI:
1670 :
1671 :			A. The incomplete/complete type business:
1672 :
1673 :			- Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
1674 :			gone!
1675 :			- ML types representing an incomplete type are now *equal* to
1676 :			ML types representing their corresponding complete types (just like
1677 :			in C). This is still safe because ml-nlffigen will not generate
1678 :			RTTI for incomplete types, nor will it generate functions that
1679 :			require access to such RTTI. But when ML code generated from both
1680 :			incomplete and complete versions of the C type meet, the ML types
1681 :			are trivially interoperable.
1682 :
1683 :			NOTE: These changes restore the full generality of the translation
1684 :			(which was previously lost when I eliminated functorization)!
1685 :
1686 :			B. Enum types:
1687 :
1688 :			- Structure C now has a type constructor "enum" that is similar to
1689 :			how the "su" constructor works. However, "enum" is not a phantom
1690 :			type because each "T enum" has values (and is isomorphic to
1691 :			MLRep.Signed.int).
1692 :			- There are generic access operations for enum objects (using
1693 :			MLRep.Signed.int).
1694 :			- ml-nlffigen will generate a structure E_foo for each "enum foo".
1695 :			* The structure contains the definition of type "mlrep" (the ML-side
1696 :			representation type of the enum). Normally, mlrep is the same
1697 :			as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
1698 :			then mlrep will be defined as a datatype -- thus facilitating
1699 :			pattern matching on mlrep values.
1700 :			("-ec" will be suppressed if there are duplicate values in an
1701 :			enumeration.)
1702 :			* Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
1703 :			will be generated for each C enum constant xxx.
1704 :			* Conversion functions m2i and i2m convert between mlrep and
1705 :			MLRep.Signed.int. (Without "-ec", these functions are identities.)
1706 :			* Coversion functions c and ml convert between mlrep and "tag enum".
1707 :			* Access functions (get/set) fetch and store mlrep values.
1708 :			- By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
1709 :			enumerations are merged into one single enumeration represented by
1710 :			structure E_'.
1711 :
1712 :			----------------------------------------------------------------------
1713 :	leunga	1080	Name: Allen Leung
1714 :	leunga	1094	Date: 2002/02/25 04:45:00 EST
1715 :			Tag: leunga-20020225-cps-spill
1716 :
1717 :			This is a new implementation of the CPS spill phase.
1718 :			The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
1719 :			In case of problems, replace it with the old file spill.sml
1720 :
1721 :			The current compiler runs into some serious performance problems when
1722 :			constructing a large record. This can happen when we try to compile a
1723 :			structure with many items. Even a very simple structure like the following
1724 :			makes the compiler slow down.
1725 :
1726 :			structure Foo = struct
1727 :			val x_1 = 0w1 : Word32.int
1728 :			val x_2 = 0w2 : Word32.int
1729 :			val x_3 = 0w3 : Word32.int
1730 :			...
1731 :			val x_N = 0wN : Word32.int
1732 :			end
1733 :
1734 :			The following table shows the compile time, from N=1000 to N=4000,
1735 :			with the old compiler:
1736 :
1737 :			N
1738 :			1000 CPS 100 spill 0.04u 0.00s 0.00g
1739 :			MLRISC ra 0.06u 0.00s 0.05g
1740 :			(spills = 0 reloads = 0)
1741 :			TOTAL 0.63u 0.07s 0.21g
1742 :
1743 :			1100 CPS 100 spill 8.25u 0.32s 0.64g
1744 :			MLRISC ra 5.68u 0.59s 3.93g
1745 :			(spills = 0 reloads = 0)
1746 :			TOTAL 14.71u 0.99s 4.81g
1747 :
1748 :			1500 CPS 100 spill 58.55u 2.34s 1.74g
1749 :			MLRISC ra 5.54u 0.65s 3.91g
1750 :			(spills = 543 reloads = 1082)
1751 :			TOTAL 65.40u 3.13s 6.00g
1752 :
1753 :			2000 CPS 100 spill 126.69u 4.84s 3.08g
1754 :			MLRISC ra 0.80u 0.10s 0.55g
1755 :			(spills = 42 reloads = 84)
1756 :			TOTAL 129.42u 5.10s 4.13g
1757 :
1758 :			3000 CPS 100 spill 675.59u 19.03s 11.64g
1759 :			MLRISC ra 2.69u 0.27s 1.38g
1760 :			(spills = 62 reloads = 124)
1761 :			TOTAL 682.48u 19.61s 13.99g
1762 :
1763 :			4000 CPS 100 spill 2362.82u 56.28s 43.60g
1764 :			MLRISC ra 4.96u 0.27s 2.72g
1765 :			(spills = 85 reloads = 170)
1766 :			TOTAL 2375.26u 57.21s 48.00g
1767 :
1768 :			As you can see the old cps spill module suffers from some serious
1769 :	leunga	1115	performance problem. But since I cannot decipher the old code fully,
1770 :			instead of patching the problems up, I'm reimplementing it
1771 :	leunga	1094	with a different algorithm. The new code is more modular,
1772 :			smaller when compiled, and substantially faster
1773 :			(O(n log n) time and O(n) space). Timing of the new spill module:
1774 :
1775 :			4000 CPS 100 spill 0.02u 0.00s 0.00g
1776 :			MLRISC ra 0.25u 0.02s 0.15g
1777 :			(spills=1 reloads=3)
1778 :			TOTAL 7.74u 0.34s 1.62g
1779 :
1780 :			Implementation details:
1781 :
1782 :			As far as I can tell, the purpose of the CPS spill module is to make sure the
1783 :			number of live variables at any program point (the bandwidth)
1784 :			does not exceed a certain limit, which is determined by the
1785 :			size of the spill area.
1786 :
1787 :			When the bandwidth is too large, we decrease the register pressure by
1788 :			packing live variables into spill records. How we achieve this is
1789 :			completely different than what we did in the old code.
1790 :
1791 :			First, there is something about the MLRiscGen code generator
1792 :			that we should be aware of:
1793 :
1794 :			o MLRiscGen performs code motion!
1795 :
1796 :			In particular, it will move floating point computations and
1797 :			address computations involving only the heap pointer to
1798 :			their use sites (if there is only a single use).
1799 :			What this means is that if we have a CPS record construction
1800 :			statement
1801 :
1802 :			RECORD(k,vl,w,e)
1803 :
1804 :			we should never count the new record address w as live if w
1805 :			has only one use (which is often the case).
1806 :
1807 :			We should do something similar to floating point, but the transformation
1808 :			there is much more complex, so I won't deal with that.
1809 :
1810 :			Secondly, there are now two new cps primops at our disposal:
1811 :
1812 :			1. rawrecord of record_kind option
1813 :			This pure operator allocates some uninitialized storage from the heap.
1814 :			There are two forms:
1815 :
1816 :			rawrecord NONE [INT n] allocates a tagless record of length n
1817 :			rawrecord (SOME rk) [INT n] allocates a tagged record of length n
1818 :			and initializes the tag.
1819 :
1820 :			2. rawupdate of cty
1821 :			rawupdate cty (v,i,x)
1822 :			Assigns to x to the ith component of record v.
1823 :			The storelist is not updated.
1824 :
1825 :			We use these new primops for both spilling and increment record construction.
1826 :
1827 :			1. Spilling.
1828 :
1829 :			This is implemented with a linear scan algorithm (but generalized
1830 :			to trees). The algorithm will create a single spill record at the
1831 :			beginning of the cps function and use rawupdate to spill to it,
1832 :			and SELECT or SELp to reload from it. So both spills and reloads
1833 :			are fine-grain operations. In contrast, in the old algorithm
1834 :			"spills" have to be bundled together in records.
1835 :
1836 :			Ideally, we should sink the spill record construction to where
1837 :			it is needed. We can even split the spill record into multiple ones
1838 :			at the places where they are needed. But CPS is not a good
1839 :			representation for global code motion, so I'll keep it simple and
1840 :			am not attempting this.
1841 :
1842 :			2. Incremental record construction (aka record splitting).
1843 :
1844 :			Long records with many component values which are simulatenously live
1845 :			(recall that single use record addresses are not considered to
1846 :			be live) are constructed with rawrecord and rawupdate.
1847 :			We allocate space on the heap with rawrecord first, then gradually
1848 :			fill it in with rawupdate. This is the technique suggested to me
1849 :			by Matthias.
1850 :
1851 :			Some restrictions on when this is applicable:
1852 :			1. It is not a VECTOR record. The code generator currently does not handle
1853 :			this case. VECTOR record uses double indirection like arrays.
1854 :			2. All the record component values are defined in the same "basic block"
1855 :			as the record constructor. This is to prevent speculative
1856 :			record construction.
1857 :
1858 :			----------------------------------------------------------------------
1859 :			Name: Allen Leung
1860 :	leunga	1086	Date: 2002/02/22 01:02:00 EST
1861 :			Tag: leunga-20020222-mlrisc-tools
1862 :
1863 :			Minor bug fixes in the parser and rewriter
1864 :
1865 :			----------------------------------------------------------------------
1866 :			Name: Allen Leung
1867 :	leunga	1085	Date: 2002/02/21 20:20:00 EST
1868 :			Tag: leunga-20020221-peephole
1869 :
1870 :			Regenerated the peephole files. Some contained typos in the specification
1871 :			and some didn't compile because of pretty printing bugs in the old version
1872 :			of 'nowhere'.
1873 :
1874 :			----------------------------------------------------------------------
1875 :			Name: Allen Leung
1876 :	leunga	1079	Date: 2002/02/19 20:20:00 EST
1877 :			Tag: leunga-20020219-mlrisc-tools
1878 :			Description:
1879 :
1880 :			Minor bug fixes to the mlrisc-tools library:
1881 :
1882 :			1. Fixed up parsing colon suffixed keywords
1883 :			2. Added the ability to shut the error messages up
1884 :			3. Reimplemented the pretty printer and fixed up/improved
1885 :			the pretty printing of handle and -> types.
1886 :			4. Fixed up generation of literal symbols in the nowhere tool.
1887 :			5. Added some SML keywords to to sml.sty
1888 :
1889 :			----------------------------------------------------------------------
1890 :			Name: Matthias Blume
1891 :	blume	1078	Date: 2002/02/19 16:20:00 EST
1892 :			Tag: blume-20020219-cmffi
1893 :			Description:
1894 :
1895 :			A wild mix of changes, some minor, some major:
1896 :
1897 :			* All C FFI-related libraries are now anchored under $c:
1898 :			$/c.cm --> $c/c.cm
1899 :			$/c-int.cm --> $c/internals/c-int.cm
1900 :			$/memory.cm --> $c/memory/memory.cm
1901 :
1902 :			* "make" tool (in CM) now treats its argument pathname slightly
1903 :			differently:
1904 :			1. If the native expansion is an absolute name, then before invoking
1905 :			the "make" command on it, CM will apply OS.Path.mkRelative
1906 :			(with relativeTo = OS.FileSys.getDir()) to it.
1907 :			2. The argument will be passed through to subsequent phases of CM
1908 :			processing without "going native". In particular, if the argument
1909 :			was an anchored path, then "make" will not lose track of that anchor.
1910 :
1911 :			* Compiler backends now "know" their respective C calling conventions
1912 :			instead of having to be told about it by ml-nlffigen. This relieves
1913 :			ml-nlffigen from one of its burdens.
1914 :
1915 :			* The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
1916 :
1917 :			* Export C_DEBUG and C_Debug from $c/c.cm.
1918 :
1919 :			* C type encoding in ml-nlffi-lib has been improved to model the conceptual
1920 :			subtyping relationship between incomplete pointers and their complete
1921 :			counterparts. For this, ('t, 'c) ptr has been changed to 'o ptr --
1922 :			with the convention of instantiating 'o with ('t, 'c) obj whenever
1923 :			the pointer target type is complete. In the incomplete case, 'o
1924 :			will be instantiated with some "'c iobj" -- a type obtained by
1925 :			using one of the functors PointerToIncompleteType or PointerToCompleteType.
1926 :
1927 :			Operations that work on both incomplete and complete pointer types are
1928 :			typed as taking an 'o ptr while operations that require the target to
1929 :			be known are typed as taking some ('t, 'c) obj ptr.
1930 :
1931 :			voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
1932 :			where void is an eqtype without any values. This makes it possible
1933 :			to work on voidptr values using functions meant to operate on light
1934 :			incomplete pointers.
1935 :
1936 :			* As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
1937 :			been vastly simplified.
1938 :
1939 :			----------------------------------------------------------------------
1940 :			Name: Matthias Blume
1941 :	blume	1076	Date: 2002/02/19 10:48:00 EST
1942 :			Tag: blume-20020219-pqfix
1943 :			Description:
1944 :
1945 :			Applied Chris Okasaki's bug fix for priority queues.
1946 :
1947 :			----------------------------------------------------------------------
1948 :			Name: Matthias Blume
1949 :	blume	1073	Date: 2002/02/15 17:05:00
1950 :	blume	1069	Tag: Release_110_39
1951 :			Description:
1952 :
1953 :	blume	1073	Last-minute retagging is becoming a tradition... :-(
1954 :
1955 :			This is the working release 110.39.
1956 :
1957 :			----------------------------------------------------------------------
1958 :			Name: Matthias Blume
1959 :			Date: 2002/02/15 16:00:00 EST
1960 :			Tag: Release_110_39-orig
1961 :			Description:
1962 :
1963 :	blume	1069	Working release 110.39. New bootfiles.
1964 :
1965 :	blume	1073	(Update: There was a small bug in the installer so it wouldn't work
1966 :			with all shells. So I retagged. -Matthias)
1967 :
1968 :	blume	1069	----------------------------------------------------------------------
1969 :			Name: Matthias Blume
1970 :	blume	1068	Date: 2002/02/15 14:17:00 EST
1971 :			Tag: blume-20020215-showbindings
1972 :			Description:
1973 :
1974 :			Added EnvRef.listBoundSymbols and CM.State.showBindings. Especially
1975 :			the latter can be useful for exploring what bindings are available at
1976 :			the interactive prompt. (The first function returns only the list
1977 :			of symbols that are really bound, the second prints those but also the
1978 :			ones that CM's autoloading mechanism knows about.)
1979 :
1980 :			----------------------------------------------------------------------
1981 :			Name: Matthias Blume
1982 :	blume	1067	Date: 2002/02/15 12:08:00 EST
1983 :			Tag: blume-20020215-iptrs
1984 :			Description:
1985 :
1986 :			Two improvements to ml-nlffigen:
1987 :
1988 :			1. Write files only if they do not exist or if their current contents
1989 :			do not coincide with what's being written. (That is, avoid messing
1990 :			with the time stamps unless absolutely necessary.)
1991 :
1992 :			2. Implement a "repository" mechanism for generated files related
1993 :			to "incomplete pointer types". See the README file for details.
1994 :
1995 :			----------------------------------------------------------------------
1996 :			Name: Matthias Blume
1997 :	blume	1066	Date: 2002/02/14 11:50:00 EST
1998 :			Tag: blume-20020214-quote
1999 :			Description:
2000 :
2001 :			Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm). This is required
2002 :			because of the new and improved tag generation scheme. (Thanks to Allen
2003 :			Leung for pointing it out.)
2004 :
2005 :			----------------------------------------------------------------------
2006 :	george	1065	Name: Lal George
2007 :			Date: 2002/02/14 09:55:27 EST 2002
2008 :			Tag: george-20020214-isabelle-bug
2009 :			Description:
2010 :
2011 :			Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
2012 :			of Isabelle on the x86.
2013 :
2014 :			From Allen:
2015 :			-----------
2016 :			I've found the problem:
2017 :
2018 :			in ra-core.sml, I use the counter "blocked" to keep track of the
2019 :			true number of elements in the freeze queue. When the counter goes
2020 :			to zero, I skip examining the queue. But I've messed up the
2021 :			bookkeeping in combine():
2022 :
2023 :			else ();
2024 :			case !ucol of
2025 :			PSEUDO => (if !cntv > 0 then
2026 :			(if !cntu > 0 then blocked := !blocked - 1 else ();
2027 :			^^^^^^^^^^^^^^^^^^^^^^^
2028 :			moveu := mergeMoveList(!movev, !moveu)
2029 :			)
2030 :			else ();
2031 :
2032 :			combine() is called to coalesce two nodes u and v.
2033 :			I think I was thinking that if the move counts of u and v are both
2034 :			greater than zero then after they are coalesced then one node is
2035 :			removed from the freeze queue. Apparently I was thinking that
2036 :			both u and v are of low degree, but that's clearly not necessarily true.
2037 :
2038 :
2039 :			02/12/2002:
2040 :			Here's the patch. HOL now compiles.
2041 :
2042 :			I don't know how this impact on performance (compile
2043 :			time or runtime). This bug caused the RA (especially on the x86)
2044 :			to go thru the potential spill phase when there are still nodes on the
2045 :			freeze queue.
2046 :
2047 :
2048 :
2049 :
2050 :			----------------------------------------------------------------------
2051 :	blume	1058	Name: Matthias Blume
2052 :	blume	1064	Date: 2002/02/13 22:40:00 EST
2053 :			Tag: blume-20020213-fptr-rtti
2054 :			Description:
2055 :
2056 :			Fixed a bug in ml-nlffigen that was introduced with one of the previous
2057 :			updates.
2058 :
2059 :			----------------------------------------------------------------------
2060 :			Name: Matthias Blume
2061 :	blume	1063	Date: 2002/02/13 16:41:00 EST
2062 :			Tag: blume-20020213-cmlpq
2063 :			Description:
2064 :
2065 :			Added new priority queue export symbols (which have just been added to
2066 :			smlnj-lib.cm) to CML's version of smlnj-lib.cm. (Otherwise CML would
2067 :			not compile and the installer would choke.)
2068 :
2069 :			----------------------------------------------------------------------
2070 :			Name: Matthias Blume
2071 :	blume	1062	Date: 2002/02/13 16:15:00 EST
2072 :			Tag: blume-20020213-various
2073 :			Description:
2074 :
2075 :			1. More tweaks to ml-nlffigen:
2076 :
2077 :			- better internal datastructures (resulting in slight speedup)
2078 :			- "-match" option requires exact match
2079 :			- "localized" gensym counters (untagged structs/unions nested within
2080 :			other structs/unions or within typedefs get a fresh counter; their
2081 :			tag will be prefixed by a concatenation of their parents' tags)
2082 :			- bug fixes (related to calculation of transitive closure of types
2083 :			to be included in the output)
2084 :
2085 :			2. Minor Basis updates:
2086 :
2087 :			- added implementations for List.collate and Option.app
2088 :
2089 :			----------------------------------------------------------------------
2090 :			Name: Matthias Blume
2091 :	blume	1060	Date: 2002/02/11 15:55:00 EST
2092 :			Tag: blume-20020211-gensym
2093 :			Description:
2094 :
2095 :			Added a "-gensym" option to command line of ml-nlffigen. This can be
2096 :			used to specify a "stem" -- a string that is inserted in all "gensym'd"
2097 :			names (ML structure names that correspond to unnamed C structs, unions,
2098 :			and enums), so that separate runs of ml-nlffigen do not clash.
2099 :
2100 :			----------------------------------------------------------------------
2101 :			Name: Matthias Blume
2102 :	blume	1059	Date: 2002/02/11 12:05:00 EST
2103 :			Tag: blume-20020211-gensml
2104 :			Description:
2105 :
2106 :			A quick fix for a problem with GenSML (in the pgraph-util library):
2107 :			Make generation of toplevel "local" optional. (Strictly speaking,
2108 :			signature definitions within "local" are not legal SML.)
2109 :
2110 :			Other than that: updates to INSTALL and cm/TODO.
2111 :
2112 :			----------------------------------------------------------------------
2113 :			Name: Matthias Blume
2114 :	blume	1058	Date: 2002/02/08 15:00:00 EST
2115 :			Tag: blume-20020208-uniquepid
2116 :			Description:
2117 :
2118 :			0. Version number has been bumped to 110.38.1. NEW BOOTFILES!!!
2119 :
2120 :			1. The installer (config/install.sh) has gotten smarter:
2121 :
2122 :			- Configuration options are a bit easier to specify now
2123 :			(in config/targets).
2124 :			- Bug in recognizing .tar.bz2 files fixed.
2125 :			- Installer automatically resolves dependencies between
2126 :			configuration options (e.g., if you ask for eXene, you will
2127 :			also get cml -- regardless whether you asked for it or not).
2128 :			- Installer can run in "quieter mode" by setting the environment
2129 :			variable INSTALL_QUIETLY to "true". "Quieter" does not mean
2130 :			"completely silent", though.
2131 :			- Build HashCons library as part of smlnj-lib.
2132 :
2133 :			2. A new scheme for assigning persistent identifiers to compilation
2134 :			units (and, by extension, to types etc.) has been put into place.
2135 :			This fixes a long-standing bug where types and even dynamic values
2136 :			can get internally confused, thereby compromising type safety
2137 :			(abstraction) and dynamic correctness. See
2138 :
2139 :			http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
2140 :
2141 :			for an example of how things could go wrong until now.
2142 :
2143 :			The downside of the new scheme is that pids are not quite as
2144 :			persistent as they used to be: CM will generate a fresh pid
2145 :			for every compilation unit that it thinks it sees for the first
2146 :			time. That means that if you compile starting from a clean, fresh
2147 :			source tree at two different times, you end up with different
2148 :			binaries.
2149 :
2150 :			Cutoff recompilation, however, has not been compromised because
2151 :			CM keeps pid information in special caches between runs.
2152 :
2153 :			----------------------------------------------------------------------
2154 :	george	1053	Name: Lal George
2155 :	george	1055	Date: 2002/02/07 15:34:13 EST 2002
2156 :			Tag: <none>
2157 :			Description:
2158 :
2159 :			Compilers that generate assembly code may produce global labels
2160 :			whose value is resolved at link time. The various peephole optimization
2161 :			modules did not take this in account.
2162 :
2163 :			TODO. The Labels.addrOf function should really return an option
2164 :			type so that clients are forced to deal with this issue, rather
2165 :			than an exception being raised.
2166 :
2167 :			----------------------------------------------------------------------
2168 :			Name: Lal George
2169 :	george	1053	Date: 2002/02/06 13:55:02 EST
2170 :			Tag: george-20020206-ra-breakup
2171 :			Description:
2172 :
2173 :			1. A bug fix from Allen.
2174 :
2175 :			A typo causes extra fstp %st(0)'s to be generated at compensation
2176 :			edges, which might cause stack underflow traps at runtime. This
2177 :			occurs in fft where there are extraneous fstps right before the 'into'
2178 :			trap instruction (in this case they are harmless since none of the
2179 :			integers overflow.)
2180 :
2181 :	george	1054	2. Pulled out various utility modules that were embedded in the modules
2182 :	george	1053	of the register allocator. I need these modules for other purposes, but
2183 :	george	1054	they are not complete enough to put into a library (just yet).
2184 :	george	1053	----------------------------------------------------------------------
2185 :	blume	1049	Name: Matthias Blume
2186 :			Date: 2002/01/31 16:05:00 EST
2187 :			Tag: blume-20020131-sparc-ccalls
2188 :			Description:
2189 :
2190 :			1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
2191 :			of extra stack space by mistake. Fixed.
2192 :
2193 :			2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
2194 :
2195 :			----------------------------------------------------------------------
2196 :	leunga	1048	Name: Allen Leung
2197 :			Date: 2002/01/30
2198 :			Tag: leunga-20020130-nowhere-bug-fix
2199 :			Description:
2200 :
2201 :			MLRISC bug fixes:
2202 :			1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
2203 :			2. MachineInt.fromString was negating its value.
2204 :
2205 :			----------------------------------------------------------------------
2206 :	blume	1039	Name: Matthias Blume
2207 :	blume	1047	Date: 2002/01/29
2208 :			Tag: blume-20020129-INSTALL
2209 :			Description:
2210 :
2211 :			- Added somewhat detailed installation instructions (file INSTALL).
2212 :			- Fixed curl-detection bug in config/install.sh.
2213 :			- It is now possible to select the URL getter using the URLGETTER
2214 :			environment variable:
2215 :
2216 :			not set / "unknown" --> automatic detection (script tries wget,
2217 :			curl, and lynx)
2218 :			"wget" / "curl" / "lynx" --> use the specified program (script "knows"
2219 :			how to properly invoke them)
2220 :			other --> use $URLGETTER directly, it must take
2221 :			precisely two command-line arguments
2222 :			(source URL and destination file name)
2223 :
2224 :			----------------------------------------------------------------------
2225 :			Name: Matthias Blume
2226 :	blume	1044	Date: 2002/01/28
2227 :	blume	1045	Tag: blume-20020128-sparc-ccalls
2228 :			Description:
2229 :
2230 :			- Fixed problem with calculation of "used" registers in sparc-c-calls.
2231 :			- Make use of the allocParam argument in sparc-c-calls.
2232 :
2233 :			----------------------------------------------------------------------
2234 :			Name: Matthias Blume
2235 :			Date: 2002/01/28
2236 :	blume	1044	Tag: blume-20020128-allocParam
2237 :			Description:
2238 :
2239 :			John Reppy: Changes c-calls API to accept client-callback for
2240 :			allocating extra stack space.
2241 :			me: Corresponding changes to mlriscGen (using a dummy argument that
2242 :			does not change the current behavior).
2243 :
2244 :			----------------------------------------------------------------------
2245 :			Name: Matthias Blume
2246 :	blume	1040	Date: 2002/01/28 12:00:00
2247 :			Tag: Release_110_38
2248 :			Description:
2249 :
2250 :			This time for real!!!
2251 :
2252 :			----------------------------------------------------------------------
2253 :			Name: Matthias Blume
2254 :	blume	1039	Date: 2002/01/28 10:56:00 EST
2255 :			Tag: blume-20020128-retraction
2256 :			Description:
2257 :
2258 :			0. Retracted earlier 110.38. (The Release_110_38 tag has been replaced
2259 :			with blume-Release_110_38-retracted.)
2260 :
2261 :			1. Fixed a problem with incorrect rounding modes in real64.sml.
2262 :			(Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
2263 :
2264 :			2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
2265 :			and enums fixed. The naming of corresponding ML identifiers should
2266 :			now be consistent again.
2267 :
2268 :			----------------------------------------------------------------------
2269 :	leunga	1038	Name: Allen Leung
2270 :			Date: 2002/01/27
2271 :			Tag: leunga-20020127-nowhere
2272 :			Description:
2273 :
2274 :			Added a target called nowhere in the configuration scripts.
2275 :			Enabling this will build the MLRISC 'nowhere' tool (for translating
2276 :			programs with where-clauses into legal SML code) during installation.
2277 :
2278 :			----------------------------------------------------------------------
2279 :	blume	1036	Name: Matthias Blume
2280 :	leunga	1038	Date: 2002/01/25 21:27:00 EST
2281 :	blume	1039	Tag: blume-Release_110_38-retracted
2282 :	blume	1037	Description:
2283 :
2284 :			Call it a (working) release! Version is 110.38. Bootfiles are ready.
2285 :
2286 :			README will be added later.
2287 :
2288 :	blume	1039	!!! NOTE: Re-tagged as blume-Release_110_38-retracted. Original tag
2289 :			(Release_110_38) removed. Reason: Last-minute bug fixes.
2290 :
2291 :	blume	1037	----------------------------------------------------------------------
2292 :			Name: Matthias Blume
2293 :	blume	1036	Date: 2002/01/25
2294 :			Tag: blume-20020125-ffi
2295 :			Description:
2296 :
2297 :			A large number of tweaks and improvements to ml-nlffi-lib and
2298 :			ml-nlffigen:
2299 :
2300 :			- ML represenation types have been streamlined
2301 :			- getter and setter functions work with concrete values, not abstract
2302 :			ones where possible
2303 :			- ml-nlffigen command line more flexible (see README file there)
2304 :			- some bugs have been fixed (hopefully)
2305 :
2306 :			----------------------------------------------------------------------
2307 :	george	1034	Name: Lal George
2308 :			Date: 2002/01/24
2309 :			Tag: george-20020124-risc-ra-interface
2310 :			Description:
2311 :
2312 :			There is a dramatic simplification in the interface to the
2313 :			register allocator for RISC architectures as a result of making
2314 :			parallel copy instructions explicit.
2315 :
2316 :			----------------------------------------------------------------------
2317 :	blume	1028	Name: Matthias Blume
2318 :	blume	1032	Date: 2002/01/22
2319 :			Tag: blume-20020122-x86-ccalls
2320 :			Description:
2321 :
2322 :			Bug fix for c-calls on x86 (having to do with how char- and
2323 :			short-arguments are being handled).
2324 :
2325 :			----------------------------------------------------------------------
2326 :			Name: Matthias Blume
2327 :	blume	1031	Date: 2002/01/21
2328 :			Tag: blume-20020121-ff
2329 :			Description:
2330 :
2331 :			Another day of fiddling with the FFI...
2332 :
2333 :			1. Bug fix/workaround: CKIT does not complain about negative array
2334 :			dimensions, so ml-nlffigen has to guard itself against this possibility.
2335 :			(Otherwise a negative dimension would send it into an infinite loop.)
2336 :
2337 :			2. Some of the abstract types (light objects, light pointers, most "base"
2338 :			types) in structure C are now eqtypes.
2339 :
2340 :			3. Added constructors and test functions for NULL function pointers.
2341 :
2342 :			----------------------------------------------------------------------
2343 :			Name: Matthias Blume
2344 :	blume	1028	Date: 2002/01/18
2345 :	blume	1030	Tag: blume-20020118-ready-for-new-release
2346 :			Description:
2347 :
2348 :			Made config/srcarchiveurl point to a new place. (Will provide boot
2349 :			files shortly.)
2350 :
2351 :			Maybe we christen this to be 110.38?
2352 :
2353 :			----------------------------------------------------------------------
2354 :			Name: Matthias Blume
2355 :			Date: 2002/01/18
2356 :	blume	1029	Tag: blume-20020118-more-ffifiddle
2357 :			Description:
2358 :
2359 :			Today's FFI fiddling:
2360 :
2361 :			- Provided a structure CGetSet with "convenient" versions of C.Get.* and
2362 :			C.Set.* that use concrete (MLRep.*) arguments and results instead
2363 :			of abstract ones.
2364 :
2365 :			- Provided word-style bit operations etc. for "int" representation
2366 :			types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
2367 :			and Long.
2368 :
2369 :			----------------------------------------------------------------------
2370 :			Name: Matthias Blume
2371 :			Date: 2002/01/18
2372 :	blume	1028	Tag: blume-20020118-use-x86-fp
2373 :			Description:
2374 :
2375 :			Now that x86-fast-fp seems to be working, I turned it back on again
2376 :			by default. (Seems to work fine now, even with the FFI.)
2377 :
2378 :			Other than that, I added some documentation about the FFI to
2379 :			src/ml-nlffigen/README and updated the FFI test examples in
2380 :			src/ml-nlffi-lib/Tests/*.
2381 :
2382 :			----------------------------------------------------------------------
2383 :	leunga	1025	Name: Allen Leung
2384 :			Date: 2002/01/17
2385 :			Tag: leunga-20020117-x86-fast-fp-call
2386 :			Description:
2387 :
2388 :			1. Fixed a problem with handling return fp values when x86's fast fp
2389 :			mode is turned on.
2390 :
2391 :			2. Minor pretty printing fix for cellset. Print %st(0) as %st(0) instead
2392 :			of %f32.
2393 :
2394 :			3. Added a constructor INT32lit to the ast of MLRISC tools.
2395 :
2396 :			----------------------------------------------------------------------
2397 :	blume	1021	Name: Matthias Blume
2398 :			Date: 2002/01/16
2399 :			Tag: blume-20020116-ffifiddle
2400 :			Description:
2401 :
2402 :			More fiddling with the FFI interface:
2403 :
2404 :			- Make constness 'c instead of rw wherever possible. This eliminates
2405 :			the need for certain explicit coercions. (However, due to ML's
2406 :			value polymorphism, there will still be many cases where explicit
2407 :			coercions are necessary. Phantom types are not the whole answer
2408 :			to modeling a subtyping relationship in ML.)
2409 :
2410 :			- ro/rw coersions for pointers added. (Avoids the detour through */&.)
2411 :
2412 :			- "printf" test example added to src/ml-nlffi-lib/Tests. (Demonstrates
2413 :			clumsy workaround for varargs problem.)
2414 :
2415 :			----------------------------------------------------------------------
2416 :	george	1016	Name: Lal George
2417 :			Date: 2002/01/15
2418 :			Tag: <none>
2419 :			Description:
2420 :
2421 :			1. Since COPY instructions are no longer native to the architecture,
2422 :			a generic functor can be used to implement the expandCopies function.
2423 :
2424 :			2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
2425 :			TEXT segment.
2426 :
2427 :			----------------------------------------------------------------------
2428 :	blume	1011	Name: Matthias Blume
2429 :	blume	1015	Date: 2002/01/15
2430 :			Tag: blume-20020115-ffiupdates
2431 :			Description:
2432 :
2433 :			1. Fix for bug resulting in single-precision float values being returned
2434 :			incorrectly from FFI calls.
2435 :
2436 :			2. Small modifications to C FFI API:
2437 :
2438 :			- memory-allocation routines return straight objects (no options)
2439 :			and raise an exception in out-of-memory situations
2440 :			- unsafe extensions to cast between function pointers and pointers
2441 :			from/to ints
2442 :			- added structure C_Debug as an alternative to structure C where
2443 :			pointer-dereferencing (|*| and |*!) always check for null-pointers
2444 :			- added open_lib' to DynLinkage; open_lib' works like open_lib
2445 :			but also takes a (possibly empty) list of existing library handles
2446 :			that the current library depends on
2447 :
2448 :			----------------------------------------------------------------------
2449 :			Name: Matthias Blume
2450 :	blume	1011	Date: 2002/01/10
2451 :			Tag: blume-20020110-newffigen
2452 :			Description:
2453 :
2454 :			1. Updates to portable graph code.
2455 :
2456 :			2. Major update to ml-nlffigen and ml-nlffi-lib. Things are much
2457 :			more scalable now so that even huge interfaces such as the one
2458 :			for GTK compile in finite time and space. :-)
2459 :			See src/ml-nlffigen/README for details on what's new.
2460 :
2461 :			----------------------------------------------------------------------
2462 :	george	1003	Name: Lal George
2463 :	george	1009	Date: 2001/01/09 14:31:35 EST 2002
2464 :			Tag: george-20011206-rm-native-copy
2465 :			Description:
2466 :
2467 :			Removed the native COPY and FCOPY instructions
2468 :			from all the architectures and replaced it with the
2469 :			explicit COPY instruction from the previous commit.
2470 :
2471 :			It is now possible to simplify many of the optimizations
2472 :			modules that manipulate copies. This has not been
2473 :			done in this change.
2474 :
2475 :			----------------------------------------------------------------------
2476 :			Name: Lal George
2477 :	george	1003	Date: 2001/12/06 16:50:13 EST 2001
2478 :			Tag: george-20011206-mlrisc-instruction
2479 :			Description:
2480 :	george	824
2481 :	george	1003	Changed the representation of instructions from being fully abstract
2482 :			to being partially concrete. That is to say:
2483 :
2484 :			from
2485 :			type instruction
2486 :
2487 :			to
2488 :			type instr (* machine instruction *)
2489 :
2490 :			datatype instruction =
2491 :			LIVE of {regs: C.cellset, spilled: C.cellset}
2492 :			| KILL of {regs: C.cellset, spilled: C.cellset}
2493 :			| COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
2494 :			| ANNOTATION of {i: instruction, a: Annotations.annotation}
2495 :			| INSTR of instr
2496 :
2497 :			This makes the handling of certain special instructions that appear on
2498 :			all architectures easier and uniform.
2499 :
2500 :			LIVE and KILL say that a list of registers are live or killed at the
2501 :			program point where they appear. No spill code is generated when an
2502 :			element of the 'regs' field is spilled, but the register is moved to
2503 :			the 'spilled' (which is present, more for debugging than anything else).
2504 :
2505 :			LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
2506 :			We used to generate:
2507 :
2508 :			DEFFREG f1
2509 :			f1 := f2 + f3
2510 :			trapb
2511 :
2512 :			but now generate:
2513 :
2514 :			f1 := f2 + f3
2515 :			trapb
2516 :			LIVE {regs=[f1,f2,f3], spilled=[]}
2517 :
2518 :			Furthermore, the DEFFREG (hack) required that all floating point instruction
2519 :			use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
2520 :			defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
2521 :			in a cleaner alpha implementation. (Hopefully, intel will not get rid of
2522 :			this architecture).
2523 :
2524 :			COPYXXX is intended to replace the parallel COPY and FCOPY available on
2525 :			all the architectures. This will result in further simplification of the
2526 :			register allocator that must be aware of them for coalescing purposes, and
2527 :			will also simplify certain aspects of the machine description that provides
2528 :			callbacks related to parallel copies.
2529 :
2530 :			ANNOTATION should be obvious, and now INSTR represents the honest to God
2531 :			machine instruction set!
2532 :
2533 :			The <arch>/instructions/<arch>Instr.sml files define certain utility
2534 :			functions for making porting easier -- essentially converting upper case
2535 :			to lower case. All machine instructions (of type instr) are in upper case,
2536 :			and the lower case form generates an MLRISC instruction. For example on
2537 :			the alpha we have:
2538 :
2539 :			datatype instr =
2540 :			LDA of {r:cell, b:cell, d:operand}
2541 :			| ...
2542 :
2543 :			val lda : {r:cell, b:cell, d:operand} -> instruction
2544 :			...
2545 :
2546 :			where lda is just (INSTR o LDA), etc.
2547 :
2548 :	leunga	815	----------------------------------------------------------------------
2549 :	blume	986	Name: Matthias Blume
2550 :	blume	993	Date: 2001/11/22 21:40:00 EST
2551 :			Tag: Release_110_37
2552 :			Description:
2553 :
2554 :			Release 110.37. This time for real.
2555 :
2556 :			----------------------------------------------------------------------
2557 :			Name: Matthias Blume
2558 :	blume	988	Date: 2001/11/21 16:35:00 EST
2559 :			Tag: blume-20011121-foot-in-mouth
2560 :			Description:
2561 :
2562 :			Removed the "Release_110_37" tag because of a serious bug.
2563 :			This will be re-tagged once the bug is fixed.
2564 :
2565 :			----------------------------------------------------------------------
2566 :			Name: Matthias Blume
2567 :	blume	987	Date: 2001/11/21 16:14:00 EST
2568 :			Tag: blume-20011121-forgottenfile
2569 :			Description:
2570 :
2571 :			Forgot to add a file. (Just a .tex-file -- part of
2572 :			the CM manual source.)
2573 :
2574 :			----------------------------------------------------------------------
2575 :			Name: Matthias Blume
2576 :	blume	986	Date: 2001/11/21 16:10:00 EST
2577 :	blume	988	Tag: blume-20011121-invalid_110_37
2578 :	blume	986	Description:
2579 :
2580 :	blume	988	Note: I removed the original tag "Release_110_37" from this commit
2581 :			because we found a serious bug in all non-x86 backends.
2582 :			- Matthias
2583 :
2584 :	blume	986	1. Modifications to the SML/NJ code generator and to the runtime system
2585 :			so that code object name strings are directly inserted into code
2586 :			objects at code generation time. The only business the runtime system
2587 :			has with this is now to read the name strings on occasions.
2588 :			(The encoding of the name string has also changed somewhat.)
2589 :
2590 :			2. CM now implements a simple "set calculus" for specifying export lists.
2591 :			In particular, it is now possible to refer to the export lists of
2592 :			other libraries/groups/sources and form unions as well as differences.
2593 :			See the latest CM manual for details.
2594 :
2595 :			3. An separate notion of "proxy" libraries has again be eliminated from
2596 :			CM's model. (Proxy libraries are now simply a special case of using
2597 :			the export list calculus.)
2598 :
2599 :			4. Some of the existing libraries now take advantage of the new set
2600 :			calculus.
2601 :			(Notice that not all libraries have been converted because some
2602 :			of the existing .cm-files are supposed to be backward compatible
2603 :			with 110.0.x.)
2604 :
2605 :			5. Some cleanup in stand-alone programs. (Don't use "exnMessage" -- use
2606 :			"General.exnMessage"! The former relies on a certain hook to be
2607 :			initialized, and that often does not happen in the stand-alone case.)
2608 :
2609 :			----------------------------------------------------------------------
2610 :	george	984	Name: Lal George
2611 :			Date: 2001/11/21 13:56:18 EST
2612 :			Tag: george-2001121-pseudo-ops
2613 :			Description:
2614 :
2615 :			Implemented a complete redesign of MLRISC pseudo-ops. Now there
2616 :			ought to never be any question of incompatabilities with
2617 :			pseudo-op syntax expected by host assemblers.
2618 :
2619 :			For now, only modules supporting GAS syntax are implemented
2620 :			but more should follow, such as MASM, and vendor assembler
2621 :			syntax, e.g. IBM as, Sun as, etc.
2622 :
2623 :			----------------------------------------------------------------------
2624 :	blume	977	Name: Matthias Blume
2625 :			Date: 2001/11/14 11:52:00 EST
2626 :			Tag: blume-20011114-srcname
2627 :			Description:
2628 :
2629 :			1. Routed the name of the current source file to mlriscgen where it
2630 :			should be directly emitted into the code object. (This last part
2631 :			is yet to be done.)
2632 :
2633 :			2. Some cleanup of the pgraph code to make it match the proposal that
2634 :			I put out the other day. (The proposal notwithstanding, things are
2635 :			still in flux here.)
2636 :
2637 :			----------------------------------------------------------------------
2638 :	george	976	Name: Lal George
2639 :			Date: 2001/11/14 09:44:04 EST
2640 :			Tag:
2641 :			Description:
2642 :
2643 :			Fix for a backpatching bug reported by Allen.
2644 :
2645 :			Because the boundary between short and long span-dependent
2646 :			instructions is +/- 128, there are an astounding number of
2647 :			span-dependent instructions whose size is over estimated.
2648 :
2649 :			Allen came up with the idea of letting the size of span
2650 :			dependent instructions be non-monotonic, for a maxIter
2651 :			number of times, after which the size must be monotonically
2652 :			increasing.
2653 :
2654 :			This table shows the number of span-dependent instructions
2655 :			whose size was over-estimated as a function of maxIter, for the
2656 :			file Parse/parse/ml.grm.sml:
2657 :
2658 :			maxIter # of instructions:
2659 :			10 687
2660 :			20 438
2661 :			30 198
2662 :			40 0
2663 :
2664 :			In compiling the compiler, there is no significant difference in
2665 :			compilation speed between maxIter=10 and maxIter=40. Actually,
2666 :			my measurements showed that maxIter=40 was a tad faster than
2667 :			maxIter=10! Also 96% of the files in the compiler reach a fix
2668 :			point within 13 iterations, so fixing maxIter at 40, while high,
2669 :			is okay.
2670 :
2671 :			----------------------------------------------------------------------
2672 :	blume	975	Name: Matthias Blume
2673 :			Date: 2001/10/31 15:25:00 EST
2674 :			Tag: blume-20011031-pgraph
2675 :			Description:
2676 :
2677 :			CKIT:
2678 :			* Changed the "Function" constructor of type Ast.ctype to carry optional
2679 :			argument identifiers.
2680 :			* Changed the return type of TypeUtil.getFunction accordingly.
2681 :			* Type equality ignores the argument names.
2682 :			* TypeUtil.composite tries to preserve argument names but gives up quickly
2683 :			if there is a mismatch.
2684 :
2685 :			installation script:
2686 :			* attempts to use "curl" if available (unless "wget" is available as well)
2687 :
2688 :			CM:
2689 :			* has an experimental implementation of "portable graphs" which I will
2690 :			soon propose as an implementation-independent library format
2691 :			* there are also new libraries $/pgraph.cm and $/pgraph-util.cm
2692 :
2693 :			NLFFI-LIB:
2694 :			* some cleanup (all cosmetic)
2695 :
2696 :			NLFFIGEN:
2697 :			* temporarily disabled the mechanism that suppresses ML output for
2698 :			C definitions whose identifiers start with an underscore character
2699 :			* generate val bindings for enum constants
2700 :			* user can request that only one style (light or heavy) is being used;
2701 :			default is to use both (command-line arguments: -heavy and -light)
2702 :			* fixed bug in handling of function types involving incomplete pointers
2703 :			* generate ML entry points that take record arguments (i.e., using
2704 :			named arguments) for C functions that have a prototype with named
2705 :			arguments
2706 :			(see changes to CKIT)
2707 :
2708 :			----------------------------------------------------------------------
2709 :	leunga	972	Name: Allen Leung
2710 :			Date: 2001/10/27 20:34:00 EDT
2711 :			Tag: leunga-20011027-x86-fast-fp-call
2712 :			Description:
2713 :
2714 :			Fixed the bug described in blume-20010920-slowfp.
2715 :
2716 :			The fix involves
2717 :			1. generating FCOPYs in FSTP in ia32-svid
2718 :			2. marking a CALL with the appropriate annotation
2719 :
2720 :			----------------------------------------------------------------------
2721 :	blume	961	Name: Matthias Blume
2722 :			Date: 2001/10/16 11:32:00 EDT
2723 :			Tag: blume-20011016-netbsd
2724 :			Description:
2725 :
2726 :			Underscore patch from Chris Richards (fixing problem with compiling
2727 :			runtime system under recent NetBSD).
2728 :
2729 :			----------------------------------------------------------------------
2730 :	george	959	Name: Allen Leung
2731 :			Date: 2001/10/12 17:18:32 EDT 2001
2732 :			Tag: leung-20011012-x86-printflowgraph
2733 :			Description:
2734 :
2735 :			X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
2736 :
2737 :			----------------------------------------------------------------------
2738 :	george	952	Name: Lal George
2739 :	george	958	Date: 2001/10/11 23:51:34 EDT
2740 :			Tag: george-20011011-too-many-instrs
2741 :			Description:
2742 :
2743 :			The representation of a program point never expected to see more
2744 :			than 65536 instructions in a basic block!
2745 :
2746 :			----------------------------------------------------------------------
2747 :			Name: Lal George
2748 :	george	952	Date: 2001/10/09 09:41:37 EDT
2749 :			Tag: george-20011008-mlrisc-labels
2750 :			Description:
2751 :
2752 :			Changed the machine description files to support printing of
2753 :			local and global labels in assembly code, based on host assembler
2754 :			conventions.
2755 :
2756 :			----------------------------------------------------------------------
2757 :	blume	916	Name: Matthias Blume
2758 :	blume	939	Date: 2001/09/25 15:25:00 EDT
2759 :			Tag: blume-20010925-exninfo
2760 :			Description:
2761 :
2762 :			I provided a non-hook implementation of exnName (at the toplevel) and
2763 :			made the "dummy" implementation of exnMessage (at the toplevel) more
2764 :			useful: if nothing gets "hooked in", then at least you are going to
2765 :			see the exception name and a message indicating why you don't see more.
2766 :
2767 :			[For the time being, programs that need exnMessage and want to use
2768 :			ml-build should either use General.exnMessage (strongly recommended) or
2769 :			refer to structure General at some other point so that CM sees a
2770 :			static dependency.]
2771 :
2772 :			[Similar remarks go for "print" and "use": If you want to use their
2773 :			functionality in stand-alone programs generated by ml-build, then use
2774 :			TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
2775 :
2776 :			----------------------------------------------------------------------
2777 :			Name: Matthias Blume
2778 :	blume	938	Date: 2001/09/20 17:28:00 EDT
2779 :			Tag: blume-20010920-slowfp
2780 :			Description:
2781 :
2782 :			Allen says that x86-fast-fp is not safe yet, so I turned it off again...
2783 :
2784 :			----------------------------------------------------------------------
2785 :			Name: Matthias Blume
2786 :	blume	937	Date: 2001/09/20 17:20:00 EDT
2787 :			Tag: blume-20010920-canonicalpaths
2788 :			Description:
2789 :
2790 :			0. Updated the BOOT file (something that I forgot to do earlier).
2791 :
2792 :			1. Small internal change to CM so that it avoids "/../" in filenames
2793 :			as much as possible (but only where it is safe).
2794 :
2795 :			2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
2796 :			that arguments that contain delimiters are passed through correctly.
2797 :			This change also means that all "special" arguments of the form
2798 :			@SMLxxx... must come first.
2799 :
2800 :			3. Changed install script to put relative anchor names for tool commands
2801 :			into pathconfig.
2802 :
2803 :			----------------------------------------------------------------------
2804 :			Name: Matthias Blume
2805 :	blume	932	Date: 2001/09/18 15:35:00 EDT
2806 :			Tag: blume-20010918-readme11036
2807 :			Description:
2808 :
2809 :			Added README files.
2810 :
2811 :			----------------------------------------------------------------------
2812 :			Name: Matthias Blume
2813 :	blume	929	Date: 2001/09/18 11:45:00 EDT
2814 :			Tag: Release_110_36 (retag)
2815 :			Description:
2816 :
2817 :			Fixed mistake in config/preloads. Retagged as 110.36.
2818 :
2819 :			----------------------------------------------------------------------
2820 :			Name: Matthias Blume
2821 :	blume	928	Date: 2001/09/18 09:40:00 EDT
2822 :	blume	929	Tag: Release_110_36_orig (tag changed)
2823 :	blume	928	Description:
2824 :
2825 :			New version (110.36). New bootfiles.
2826 :
2827 :			----------------------------------------------------------------------
2828 :			Name: Matthias Blume
2829 :	blume	927	Date: 2001/09/14 16:15:00 EDT
2830 :			Tag: blume-20010914-x86fastfp
2831 :			Description:
2832 :
2833 :			John committed some changes that Allen made, in particular a (hopefully)
2834 :			correctly working version of the x86-fp module.
2835 :
2836 :			I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
2837 :			flag to "true". Everything seems to compile to a fixpoint ok, and
2838 :			"mandelbrot" speeds up by about 15%.
2839 :
2840 :			----------------------------------------------------------------------
2841 :			Name: Matthias Blume
2842 :	blume	923	Date: 2001/09/13 11:20:00 EDT
2843 :			Tag: blume-20010913-minimal
2844 :			Description:
2845 :
2846 :			1. Stefan Monnier's patch to fix a miscompilation problem that
2847 :			was brought to light by John Reppy's work on Moby.
2848 :
2849 :			2. Implemented a minimal "structure Compiler" that contains just
2850 :			"version" and "architecture". The minimal version will be
2851 :			available when the full version is not. This is for backward-
2852 :			compatibility with code that wants to test Compiler.version.
2853 :
2854 :			----------------------------------------------------------------------
2855 :			Name: Matthias Blume
2856 :	blume	920	Date: 2001/08/28 14:03:00 EDT
2857 :			Tag: blume-20010828-ml-lex
2858 :			Description:
2859 :
2860 :			Fix for bug 1581, received from Neophytos Michael.
2861 :
2862 :			----------------------------------------------------------------------
2863 :			Name: Matthias Blume
2864 :	blume	918	Date: 2001/08/27 11:20:00 EDT
2865 :			Tag: blume-20010827-readme11035
2866 :			Description:
2867 :
2868 :			Fleshed out the README file for 110.35.
2869 :
2870 :			----------------------------------------------------------------------
2871 :			Name: Matthias Blume
2872 :	blume	916	Date: 2001/08/24 17:10:00 EDT
2873 :			Tag: Release_110_35
2874 :			Description:
2875 :
2876 :			New version number (110.35). New bootfiles.
2877 :
2878 :			----------------------------------------------------------------------
2879 :	george	915	Name: Lal George
2880 :	george	909	Date: 2001/08/24 13:47:18 EDT 2001
2881 :			Tag: george-20010824-MLRISC-graphs
2882 :			Description:
2883 :
2884 :			removed clusters from MLRISC completely and replaced with graphs.
2885 :
2886 :			----------------------------------------------------------------------
2887 :			Name: Matthias Blume
2888 :	blume	905	Date: 2001/08/23 17:50:00 EDT
2889 :			Tag: blume-20010823-toplevel
2890 :			Description:
2891 :
2892 :			- some reorganization of the code that implements various kinds of
2893 :			environments in the compiler (static, dynamic, symbolic, combined)
2894 :			- re-implemented the EnvRef module so that evalStream works properly
2895 :			(if the stream contains references to "use", "CM.make", etc.)
2896 :			- cleaned up evalloop.sml and interact.sml (but they need more cleaning)
2897 :
2898 :			----------------------------------------------------------------------
2899 :			Name: Matthias Blume
2900 :	blume	904	Date: 2001/08/20 15:50 EDT
2901 :			Tag: blume20010820-slipup
2902 :			Description:
2903 :
2904 :			I forgot to commit a few files. Here they are...
2905 :
2906 :			----------------------------------------------------------------------
2907 :			Name: Matthias Blume
2908 :	blume	903	Date: 2001/08/20 15:35:00 EDT
2909 :			Tag: blume-20010820-debugprof
2910 :			Description:
2911 :
2912 :			!!!! NEW BOOTFILES !!!!
2913 :
2914 :			This is another round of reorganizing the compiler sources. This
2915 :			time the main goal was to factor out all the "instrumentation"
2916 :			passes (for profiling and backtracing) into their own library.
2917 :			The difficulty was to do it in such a way that it does not depend
2918 :			on elaborate.cm but only on elabdata.cm.
2919 :
2920 :			Therefore there have been further changes to both elaborate.cm and
2921 :			elabdata.cm -- more "generic" things have been moved from the former
2922 :			to the latter. As a result, I was forced to split the assignment
2923 :			of numbers indicating "primtyc"s into two portions: SML-generic and
2924 :			SML/NJ-specific. Since it would have been awkward to maintain,
2925 :			I bit the bullet and actually _changed_ the mapping between these
2926 :			numbers and primtycs. The bottom line of this is that you need
2927 :			a new set of bin- and bootfiles.
2928 :
2929 :			I have built new bootfiles for all architectures, so doing a fresh
2930 :			checkout and config/install.sh should be all you need.
2931 :
2932 :			The newly created library's name is
2933 :
2934 :			$smlnj/viscomp/debugprof.cm
2935 :
2936 :			and its sources live under
2937 :
2938 :			src/compiler/DebugProf
2939 :
2940 :			----------------------------------------------------------------------
2941 :			Name: Matthias Blume
2942 :	blume	902	Date: 2001/08/15 17:15:00 EDT
2943 :			Tag: blume-20010815-compreorg
2944 :			Description:
2945 :
2946 :			This is a first cut at reorganizing the CM libraries that make up the
2947 :			core of the compiler. The idea is to separate out pieces that could
2948 :			be used independently by tools, e.g., the parser, the typechecker, etc.
2949 :
2950 :			The current status is a step in this direction, but it is not quite
2951 :			satisfactory yet. Expect more changes in the future.
2952 :
2953 :			Here is the current (new) organization...
2954 :
2955 :			What used to be $smlnj/viscomp/core.cm is now divided into
2956 :			six CM libraries:
2957 :
2958 :			$smlnj/viscomp/basics.cm
2959 :			/parser.cm
2960 :			/elabdata.cm
2961 :			/elaborate.cm
2962 :			/execute.cm
2963 :			/core.cm
2964 :
2965 :			The CM files for these libraries live under src/system/smlnj/viscomp.
2966 :			All these libraries are proxy libraries that contain precisely
2967 :			one CM library component. Here are the locations of the components
2968 :			(all within the src/compiler tree):
2969 :
2970 :			Basics/basics.cm
2971 :			Parse/parser.cm
2972 :			ElabData/elabdata.cm
2973 :			Elaborator/elaborate.cm
2974 :			Execution/execute.cm
2975 :			core.cm
2976 :
2977 :			[This organization is the same that has been used already
2978 :			for a while for the architecture-specific parts of the visible
2979 :			compiler and for the old version of core.cm.]
2980 :
2981 :			As you will notice, many source files have been moved from their
2982 :			respective original locations to a new home in one of the above
2983 :			subtrees.
2984 :
2985 :			The division of labor between the new libraries is the following:
2986 :
2987 :			basics.cm:
2988 :			- Simple, basic definitions that pertain to many (or all) of
2989 :			the other libraries.
2990 :			parser.cm:
2991 :			- The SML parser, producing output of type Ast.dec.
2992 :			- The type family for Ast is also defined and exported here.
2993 :			elabdata.cm:
2994 :			- The datatypes that describe input and output of the elaborator.
2995 :			This includes types, absyn, and static environments.
2996 :			elaborator.cm:
2997 :			- The SML/NJ type checker and elaborator.
2998 :			This maps an Ast.dec (with a given static environment) to
2999 :			an Absyn.dec (with a new static environment).
3000 :			- This libraries implements certain modules that used to be
3001 :			structures as functors (to remove dependencies on FLINT).
3002 :			execute.cm:
3003 :			- Everything having to do with executing binary code objects.
3004 :			- Dynamic environments.
3005 :			core.cm:
3006 :			- SML/NJ-specific instantiations of the elaborator and MLRISC.
3007 :			- Top-level modules.
3008 :			- FLINT (this should eventually become its own library)
3009 :
3010 :			Notes:
3011 :
3012 :			I am not 100% happy with the way I separated the elaborator (and its
3013 :			data structures) from FLINT. Two instances of the same problem:
3014 :
3015 :			1. Data structures contain certain fields that carry FLINT-specific
3016 :			information. I hacked around this using exn and the property list
3017 :			module from smlnj-lib. But the fact that there are middle-end
3018 :			specific fields around at all is a bit annoying.
3019 :
3020 :			2. The elaborator calculates certain FLINT-related information. I tried
3021 :			to make this as abstract as I could using functorization, but, again,
3022 :			the fact that the elaborator has to perform calculations on behalf
3023 :			of the middle-end at all is not nice.
3024 :
3025 :			3. Having to used exn and property lists is unfortunate because it
3026 :			weakens type checking. The other alternative (parameterizing
3027 :			nearly *everything*) is not appealing, though.
3028 :
3029 :			I removed the "rebinding =" warning hack because due to the new organization
3030 :			it was awkward to maintain it. As a result, the compiler now issues some of
3031 :			these warnings when compiling init.cmi during bootstrap compilation. On
3032 :			the plus side, you also get a warning when you do, for example:
3033 :			val op = = Int32.+
3034 :			which was not the case up to now.
3035 :
3036 :			I placed "assign" and "deref" into the _Core structure so that the
3037 :			code that deals with the "lazy" keyword can find them there. This
3038 :			removes the need for having access to the primitive environment
3039 :			during elaboration.
3040 :
3041 :			----------------------------------------------------------------------
3042 :			Name: Matthias Blume
3043 :	blume	898	Date: 2001/08/13
3044 :			Tag: blume-20010813-closures
3045 :			Description:
3046 :
3047 :			This fix was sent to us by Zhong Shao. It is supposed to improve the
3048 :			performance of certain loops by avoiding needless closure allocation.
3049 :
3050 :			----------------------------------------------------------------------
3051 :	george	890	Name: Lal George
3052 :	george	896	Date: 2001/07/31 10:03:23 EDT 2001
3053 :			Tag: george-20010731-x86-fmalloc
3054 :			Description: Fixed bug in x86 calls
3055 :
3056 :			There was a bug where call instructions would mysteriously
3057 :			vanish. The call instruction had to be one that returned
3058 :			a floating point value.
3059 :
3060 :			----------------------------------------------------------------------
3061 :			Name: Lal George
3062 :	george	890	Date: 2001/07/19 16:36:29 EDT 2001
3063 :			Tag: george-20010719-simple-cells
3064 :			Description:
3065 :
3066 :			I have dramatically simplified the interface for CELLS in MLRISC.
3067 :
3068 :			In summary, the cells interface is broken up into three parts:
3069 :
3070 :			1. CellsBasis : CELLS_BASIS
3071 :
3072 :			CellsBasis is a top level structure and common for all
3073 :			architectures. it contains the definitions of basic datatypes
3074 :			and utility functions over these types.
3075 :
3076 :			2. functor Cells() : CELLS
3077 :
3078 :			Cells generates an interface for CELLS that incorporates the
3079 :			specific resources on the target architecture, such as the
3080 :			presence of special register classes, their number and size,
3081 :			and various useful substructures.
3082 :
3083 :			3. <ARCH>CELLS
3084 :
3085 :			e.g. SparcCells: SPARCCELLS
3086 :
3087 :			<ARCH>CELLS usually contains additional bindings for special
3088 :			registers on the architecture, such as:
3089 :
3090 :			val r0 : cell (* register zero *)
3091 :			val y : cell (* Y register *)
3092 :			val psr : cell (* processor status register *)
3093 :			...
3094 :
3095 :			The structure returned by applying the Cells functor is opened
3096 :			in this interface.
3097 :
3098 :			The main implication of all this is that the datatypes for cells is
3099 :			split between CellsBasis and CELLS -- a fairly simple change for user
3100 :			code.
3101 :
3102 :			In the old scheme the CELLS interface had a definitional binding of
3103 :			the form:
3104 :
3105 :			signature CELLS = sig
3106 :
3107 :			structure CellsBasis = CellsBasis
3108 :
3109 :			...
3110 :
3111 :			end
3112 :
3113 :			With all the sharing constraints that goes on in MLRISC, this old
3114 :			design quickly leads to errors such as:
3115 :
3116 :			"structure definition spec inside of sharing ... "
3117 :
3118 :
3119 :			and appears to require an unacceptable amount of sharing and where
3120 :			constraint hackery.
3121 :
3122 :			I think this error message (the interaction of definitional specs and
3123 :			sharing) requires more explanation on our web page.
3124 :
3125 :			----------------------------------------------------------------------
3126 :	blume	826	Name: Matthias Blume
3127 :	blume	879	Date: 2001/07/19 15:00:00 EDT
3128 :			Tag: blume-20010719-libreorg
3129 :			Description:
3130 :
3131 :			This update puts together a fairly extensive but straightforward change
3132 :			to the way the libraries that implement the interactive system are
3133 :			organized:
3134 :
3135 :			The biggest change is the elimination of structure Compiler. As a
3136 :			replacement for this structure, there is now a CM library
3137 :			(known as $smlnj/compiler.cm or $smlnj/compiler/current.cm)
3138 :			that exports all the substructures of the original structure Compiler
3139 :			directly. So instead of saying Compiler.Foo.bar one now simply
3140 :			says Foo.bar. (The CM libraries actually export a collection of
3141 :			structures that is richer than the collection of substructures of
3142 :			structure Compiler.)
3143 :
3144 :			To make the transition smooth, there is a separate library called
3145 :			$smlnj/compiler/compiler.cm which puts together and exports the
3146 :			original structure Compiler (or at least something very close to it).
3147 :
3148 :			There are five members of the original structure Compiler
3149 :			that are not exported directly but which instead became members
3150 :			of a new structure Backend (described by signature BACKEND). These are:
3151 :			structure Profile (: PROFILE), structure Compile (: COMPILE), structure
3152 :			Interact (: INTERACT), structure Machine (: MACHINE), and val
3153 :			architecture (: string).
3154 :
3155 :			Structure Compiler.Version has become structure CompilerVersion.
3156 :
3157 :			Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided
3158 :			by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
3159 :			or x86, respectively.
3160 :			Each of these exports the same frontend structures that
3161 :			$smlnj/compiler.cm exports. But they do not have a structure Backend
3162 :			and instead export some structure <Arch>Backend where <Arch> is Alpha32,
3163 :			Hppa, PPC, Sparc, or X86, respectively.
3164 :
3165 :			Library $smlnj/compiler/all.cm exports the union of the exports of
3166 :			$smlnj/compiler/<arch>.cm
3167 :
3168 :			There are no structures <Arch>Compiler anymore, use
3169 :			$smlnj/compiler/<arch>.cm instead.
3170 :
3171 :			Library host-compiler-0.cm is gone. Instead, the internal library
3172 :			that instantiates CM is now called cm0.cm. Selection of the host
3173 :			compiler (backend) is no longer done here but. (Responsibility for it
3174 :			now lies with $smlnj/compiler/current.cm. This seems to be more
3175 :			logical.)
3176 :
3177 :			Many individual files have been moved or renamed. Some files have
3178 :			been split into multiple files, and some "dead" files have been deleted.
3179 :
3180 :			Aside from these changes to library organization, there are also changes
3181 :			to the way the code itself is organized:
3182 :
3183 :			Structure Binfile has been re-implemented in such a way that it no
3184 :			longer needs any knowledge of the compiler. It exclusively deals
3185 :			with the details of binfile layout. It no longer invokes the
3186 :			compiler (for the purpose of creating new prospective binfile
3187 :			content), and it no longer has any knowledge of how to interpret
3188 :			pickles.
3189 :
3190 :			Structure Compile (: COMPILE) has been stripped down to the bare
3191 :			essentials of compilation. It no longer deals with linking/execution.
3192 :			The interface has been cleaned up considerably.
3193 :
3194 :			Utility routines for dealing with linking and execution have been
3195 :			moved into their own substructures.
3196 :
3197 :			(The ultimate goal of these changes is to provide a light-weight
3198 :			binfile loader/linker (at least for, e.g., stable libraries) that
3199 :			does not require CM or the compiler to be present.)
3200 :
3201 :			CM documentation has been updated to reflect the changes to library
3202 :			organization.
3203 :
3204 :			----------------------------------------------------------------------
3205 :			Name: Matthias Blume
3206 :	blume	875	Date: 2001/07/10 17:30:00 EDT
3207 :			Tag: Release_110_34
3208 :			Description:
3209 :
3210 :	blume	876	Minor tweak to 110.34 (re-tagged):
3211 :
3212 :			- README.html file added to CVS repository
3213 :			- runtime compiles properly under FreeBSD 3.X and 4.X
3214 :
3215 :			----------------------------------------------------------------------
3216 :			Name: Matthias Blume
3217 :			Date: 2001/07/10 17:30:00 EDT
3218 :			Tag: Release_110_34
3219 :			Description:
3220 :
3221 :	blume	875	New version number (110.34). New bootfiles.
3222 :
3223 :			----------------------------------------------------------------------
3224 :			Name: Matthias Blume
3225 :	blume	874	Date: 2001/07/09 16:00:00 EDT
3226 :			Tag: blume-20010709-more-varargs
3227 :			Description:
3228 :
3229 :			I changed the handling of varargs in ml-nlffigen again:
3230 :			The ellipsis ... will now simply be ignored (with an accompanying warning).
3231 :
3232 :			The immediate effect is that you can actually call a varargs function
3233 :			from ML -- but you can't actually supply any arguments beyond the ones
3234 :			specified explicitly. (For example, you can call printf with its format
3235 :			string, but you cannot pass additional arguments.)
3236 :
3237 :			This behavior is only marginally more useful than the one before, but
3238 :			it has the advantage that a function or, more importantly, a function
3239 :			type never gets dropped on the floor, thus avoiding follow-up problems with
3240 :			other types that refer to the offending one.
3241 :
3242 :			----------------------------------------------------------------------
3243 :			Name: Matthias Blume
3244 :	blume	873	Date: 2001/07/09 11:25:00 EDT
3245 :			Tag: blume-20010709-varargs
3246 :			Description:
3247 :
3248 :			1. ckit-lib.cm now exports structure Error
3249 :			2. ml-nlffigen reports occurences of "..." (i.e., varargs function types)
3250 :			with a warning accompanied by a source location. Moreover, it
3251 :			merely skips the offending function or type and proceeds with the
3252 :			rest of its work.u As a result, one can safely feed C code containing
3253 :			"..." to ml-nlffigen.
3254 :			3. There are some internal improvements to CM, providing slightly
3255 :			more general string substitutions in the tools subsystem.
3256 :
3257 :			----------------------------------------------------------------------
3258 :			Name: Matthias Blume
3259 :	blume	854	Date: 2001/06/27 15:10:00 EDT
3260 :			Tag: blume-20010627-concur
3261 :			Description:
3262 :
3263 :			Fixed a small bug in CM's handling of parallel compilation.
3264 :			(You could observe the bug by Control-C-interrupting an ordinary
3265 :			CMB.make or CM.stabilize and then attaching some compile servers.
3266 :			The result was that all of a sudden the previously interrupted
3267 :			compilation would continue on its own. This was because of
3268 :			an over-optimization: CM did not bother to clean out certain queues
3269 :			when no servers were attached "anyway", resulting in the contents
3270 :			of these queues to grab control when new servers did get attached.)
3271 :
3272 :			There is also another minor update to the CM manual.
3273 :
3274 :			----------------------------------------------------------------------
3275 :			Name: Matthias Blume
3276 :	blume	853	Date: 2001/06/26 16:15:00 EDT
3277 :			Tag: blume-20010626-cmdoc
3278 :			Description:
3279 :
3280 :			Minor typo fixed in CM manual (syntax diagram for libraries).
3281 :
3282 :			----------------------------------------------------------------------
3283 :			Name: Matthias Blume
3284 :	blume	852	Date: 2001/06/25 22:55:00 EDT
3285 :			Tag: blume-20010625-x86pc
3286 :			Description:
3287 :
3288 :			Fixed a nasty bug in the X86 assembly code that caused signal
3289 :			handlers to fail (crash) randomly.
3290 :
3291 :			----------------------------------------------------------------------
3292 :			Name: Matthias Blume
3293 :	blume	846	Date: 2001/06/25 12:05:00 EDT
3294 :			Tag: blume-20010625-nlffigen
3295 :			Description:
3296 :
3297 :			This update fixes a number of minor bugs in ml-nlffigen as reported by
3298 :			Nick Carter <nbc@andrew.cmu.edu>.
3299 :
3300 :			1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted.
3301 :			2. Default names for generated files are now derived from the name of
3302 :			the C file *without its directory*. In particular, this causes generated
3303 :			files to be placed locally even if the C file is in some system directory.
3304 :			3. Default names for generated signatures and structures are also derived
3305 :			from the C file name without its directory. This avoids silly things
3306 :			like "structure GL/GL".
3307 :			(Other silly names are still possible because ml-nlffigen does not do
3308 :	blume	847	a thorough check of whether generated names are legal ML identifiers.
3309 :			When in doubt, use command line arguments to force particular names.)
3310 :	blume	846
3311 :			----------------------------------------------------------------------
3312 :			Name: Matthias Blume
3313 :	blume	845	Date: 2001/06/21 12:25:00 EDT
3314 :			Tag: blume-20010621-eXene
3315 :			Description:
3316 :
3317 :			eXene now compiles and (sort of) works again.
3318 :
3319 :			The library name (for version > 110.33) is $/eXene.cm.
3320 :
3321 :			I also added an new example in src/eXene/examples/nbody. See the
3322 :			README file there for details.
3323 :
3324 :			----------------------------------------------------------------------
3325 :			Name: Matthias Blume
3326 :	blume	844	Date: 2001/06/20 16:40:00 EDT
3327 :			Tag: blume-20010620-cml
3328 :			Description:
3329 :
3330 :			CML now compiles and works again.
3331 :
3332 :			Libraries (for version > 110.33):
3333 :
3334 :			$cml/cml.cm Main CML library.
3335 :			$cml/basis.cm CML's version of $/basis.cm.
3336 :			$cml/cml-internal.cm Internal helper library.
3337 :			$cml/core-cml.cm Internal helper library.
3338 :			$cml-lib/trace-cml.cm Tracing facility.
3339 :			$cml-lib/smlnj-lib.cm CML's version of $/smlnj-lib.cm
3340 :
3341 :			The installer (config/install.sh) has been taught how to properly
3342 :			install this stuff.
3343 :
3344 :			----------------------------------------------------------------------
3345 :			Name: Matthias Blume
3346 :	blume	843	Date: 2001/06/19 17:55:00 EDT
3347 :			Tag: blume-20010619-instantiate
3348 :			Description:
3349 :
3350 :			This un-breaks the fix for bug 1432.
3351 :			(The bug was originally fixed in 110.9 but I broke it again some
3352 :			time after that.)
3353 :
3354 :			----------------------------------------------------------------------
3355 :			Name: Matthias Blume
3356 :	blume	842	Date: 2001/06/19 17:25:00 EDT
3357 :			Tag: blume-20010619-signals
3358 :			Description:
3359 :
3360 :			This should (hopefully) fix the long-standing signal handling bug.
3361 :			(The runtime system was constructing a continuation record with an
3362 :			incorrect descriptor which would cause the GC to drop data on the floor...)
3363 :
3364 :			----------------------------------------------------------------------
3365 :			Name: Matthias Blume
3366 :	blume	840	Date: 2001/06/15 15:05:00 EDT
3367 :	blume	841	Tag: blume-20010615-moresparc
3368 :			Description:
3369 :
3370 :			Here is a short late-hour update related to Sparc c-calls:
3371 :
3372 :			-- made handling of double-word arguments a bit smarter
3373 :
3374 :			-- instruction selection phase tries to collapse certain clumsily
3375 :			constructed ML-Trees; typical example:
3376 :
3377 :			ADD(ty,ADD(_,e,LI d1),LI d2) -> ADD(ty,e,LI(d1+d2))
3378 :
3379 :			This currently has no further impact on SML/NJ since mlriscGen does
3380 :			not seem to generate such patterns in the first place, and c-calls
3381 :			(which did generate them in the beginning) has meanwhile been fixed
3382 :			so as to avoid them as well.
3383 :
3384 :			----------------------------------------------------------------------
3385 :			Name: Matthias Blume
3386 :			Date: 2001/06/15 15:05:00 EDT
3387 :	blume	840	Tag: blume-20010615-sparc
3388 :			Description:
3389 :
3390 :			The purpose of this update is to provide an implementation of NLFFI
3391 :			on Sparc machines.
3392 :
3393 :			Here are the changes in detail:
3394 :
3395 :			* src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
3396 :			the Sparc implementation of the c-calls API.
3397 :			* The Sparc backend of SML/NJ has been modified to uniformely use %fp
3398 :			for accessing the ML frame. Thus, we have a real frame pointer and
3399 :			can freely modify %sp without need for an omit-frame-ptr phase.
3400 :			The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
3401 :			this case.
3402 :			* ml-nlffigen has been taught to produce code for different architectures
3403 :			and calling conventions.
3404 :			* In a way similar to what was done in the x86 case, the Sparc
3405 :			backend uses its own specific extension to mltree. (For example,
3406 :			it needs to be able to generate UNIMP instructions which are part
3407 :			of the calling convention.)
3408 :			* ml-nlffi-lib was reorganized to make it more modular (in particular,
3409 :			to make it easier to plug in new machine- and os-dependent parts).
3410 :
3411 :			There are some other fairly unrelated bug fixes and cleanups as well:
3412 :
3413 :			* I further hacked the .cm files for MLRISC tools (like MDLGen) so
3414 :			that they properly share their libraries with existing SML/NJ libraries.
3415 :			* I fixed a minor cosmetic bug in CM, supressing certain spurious
3416 :			follow-up error messages.
3417 :			* Updates to CM/CMB documentation.
3418 :
3419 :			TODO items:
3420 :
3421 :			* MLRISC should use a different register as its asmTemp on the Sparc.
3422 :			(The current %o2 is a really bad choice because it is part of the
3423 :			calling conventions, so things might interfere in unexpected ways.)
3424 :
3425 :			----------------------------------------------------------------------
3426 :			Name: Matthias Blume
3427 :	blume	839	Date: 2001/06/07
3428 :			Tag: blume-20010607-calls
3429 :			Description:
3430 :
3431 :			A number of internal changes related to C calls and calling conventions:
3432 :
3433 :			1. ML-Tree CALL statements now carry a "pops" field. It indicates the
3434 :			number of bytes popped implicitly (by the callee). In most cases
3435 :			this field is 0 but on x86/win32 it is some non-zero value. This
3436 :			is information provided for the benefit of the "omit-frameptr" pass.
3437 :			2. The CALL instruction on the x86 carries a similar "pops" field.
3438 :			The instruction selection phase copies its value from the ML-Tree
3439 :			CALL statement.
3440 :			3. On all other architectures, the instruction selection phase checks
3441 :			whether "pops=0" and complains if not.
3442 :			4. The c-calls implementation for x86 now accepts two calling conventions:
3443 :			"ccall" and "stdcall". When "ccall" is selected, the caller cleans
3444 :			up after the call and pops is set to 0. For "stdcall", the caller
3445 :			does nothing, leaving the cleanup to the callee; pops is set to
3446 :			the number of bytes that were pushed onto the stack.
3447 :			5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
3448 :			distinguish between "ccall" and "stdcall".
3449 :			6. The UNIMP instruction has been added to the supported Sparc instruction
3450 :			set. (This is needed for implementing the official C calling convention
3451 :			on this architecture.)
3452 :			7. I fixed some of the .cm files under src/MLRISC/Tools to make them
3453 :			work with the latest CM.
3454 :
3455 :			----------------------------------------------------------------------
3456 :			Name: Matthias Blume
3457 :	blume	838	Date: 2001/06/05 15:10:00 EDT
3458 :			Tag: blume-20010605-cm-index
3459 :			Description:
3460 :
3461 :			0. The "lambdasplit" parameter for class "sml" in CM has been documented.
3462 :
3463 :			1. CM can now generate "index files". These are human-readable files
3464 :			that list on a per-.cm-file basis each toplevel symbol defined or
3465 :			imported. The location of the index file for
3466 :			<p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
3467 :			To enable index-file generation, set CM.Control.generate_index to true
3468 :			or export an environment-symbol: export CM_GENERATE_INDEX=true.
3469 :
3470 :			The CM manual has been updated accordingly.
3471 :
3472 :			2. I made some slight modifications to the c-calls API in MLRISC.
3473 :
3474 :			a) There is now a callback to support saving/restoring of
3475 :			dedicated but caller-save registers around the actual call
3476 :			instruction.
3477 :			b) One can optionally specify a comment-annotation for the
3478 :			call instruction.
3479 :
3480 :			3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
3481 :			(For example, the comment annotation shows the C prototype of
3482 :			the function being called.)
3483 :
3484 :			----------------------------------------------------------------------
3485 :			Name: Matthias Blume
3486 :	blume	837	Date: 2001/06/01 13:30:00 EDT
3487 :			Tag: blume-20010601-nlffi-cleanup
3488 :			Description:
3489 :
3490 :			This is mostly a cleanup of MLFFI stuff:
3491 :
3492 :			- some signature files have been put into a more exposed place
3493 :			- the ugly 'f type parameter is gone (simplifies types tremendously!)
3494 :			- ml-nlffigen changed accordingly
3495 :			- tutorial updated
3496 :
3497 :			Other changes:
3498 :
3499 :			- author's affiliation in CM manual(s) updated
3500 :			- some more recognized keywords added to Allen's sml.sty
3501 :
3502 :			----------------------------------------------------------------------
3503 :			Name: Matthias Blume
3504 :	blume	836	Date: 2001/05/25 15:30:00 EDT
3505 :			Tag: blume-20010525-iptr
3506 :			Description:
3507 :
3508 :			- put the official 110.33-README (as it appears on the ftp server) under
3509 :			CVS
3510 :			- fixed a small bug related to incomplete pointer types in
3511 :			ml-nlffigen
3512 :			- small cosmetic change to the ml-nlffi-lib's "arr" type constructor
3513 :			(it does not need the 'f type parameter)
3514 :
3515 :			----------------------------------------------------------------------
3516 :			Name: Matthias Blume
3517 :	blume	834	Date: 2001/05/23 14:30:00 EDT
3518 :			Tag: Release_110_33
3519 :			Description:
3520 :
3521 :			New version number (110.33). New bootfiles.
3522 :
3523 :			----------------------------------------------------------------------
3524 :			Name: Matthias Blume
3525 :	blume	833	Date: 2001/05/22 18:06:00 EDT
3526 :			Tag: blume-20010522-targets
3527 :			Description:
3528 :
3529 :			Made install.sh use file config/targets.customized if it exists, falling
3530 :			back to config/targets if it doesn't. This way one can have a customized
3531 :			version of the targets file without touching the "real thing", thus
3532 :			eliminating the constant fear of accidentally checking something bogus
3533 :			back into the CVS repository... (File config/targets.customized must
3534 :			not be added to the repository!)
3535 :
3536 :			----------------------------------------------------------------------
3537 :			Name: Matthias Blume
3538 :	blume	832	Date: 2001/05/22 16:30:00 EDT
3539 :			Tag: blume-20010522-minitut
3540 :			Description:
3541 :
3542 :			1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
3543 :			struct returns.
3544 :			2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt. This is some very
3545 :			incomplete, preliminary documentation for NLFFI.
3546 :
3547 :			----------------------------------------------------------------------
3548 :			Name: Matthias Blume
3549 :	blume	829	Date: 2001/05/14 11:30:00 EDT
3550 :			Tag: blume-20010514-script
3551 :			Description:
3552 :
3553 :			Some bugs in install script fixed.
3554 :
3555 :			In addition to that I also made a slight change to the NLFFI API:
3556 :			Functors generated by ml-nlffigen now take the dynamic library as a
3557 :			straight functor argument, not as a suspended one. (The original
3558 :			functor code used to force the suspension right away anyway, so there
3559 :			was nothing gained by this complication of the interface.)
3560 :
3561 :			----------------------------------------------------------------------
3562 :			Name: Matthias Blume
3563 :	blume	828	Date: 2001/05/11 14:35:00 EDT
3564 :			Tag: blume-20010511-ml-nlffi
3565 :			Description:
3566 :
3567 :			I finally took the plunge and added my new FFI code to the main
3568 :			repository. For x86-linux it is now ready for prime-time.
3569 :
3570 :			There are two new subdirectories of "src":
3571 :
3572 :			- ml-nlffi-lib:
3573 :			The utility library for programs using the FFI interface.
3574 :			Here is the implementation of $/c.cm and its associated low-level
3575 :			partners $/c-int.cm and $/memory.cm.
3576 :			- ml-nlffigen:
3577 :			A stand-alone program for generating ML glue code from C source
3578 :			code.
3579 :
3580 :			Building ml-nlffigen requires $/ckit-lib.cm.
3581 :
3582 :			The config/install.sh script has been updates to do the Right Thing
3583 :			(hopefully).
3584 :
3585 :			Notice that the source tree for the C-Kit will not be put under "src"
3586 :			but directly under the installation root directory. (This is the
3587 :			structure that currently exists on the CVS server when you check out
3588 :			module "sml".) Fortunately, config/install.sh knows about this oddity.
3589 :
3590 :			Bugs: No documentation yet.
3591 :
3592 :			----------------------------------------------------------------------
3593 :			Name: Matthias Blume
3594 :	blume	826	Date: 2001/05/09 16:35:00 EDT
3595 :			Tag: blume-20010509-cpscontract
3596 :			Description:
3597 :
3598 :			Fixed a bug in the accounting code in cpsopt/contract.sml. (The
3599 :			wrapper/unwrapper elimination did not decrement usage counts and some
3600 :			dead variables got overlooked by the dead-up logic.)
3601 :
3602 :			----------------------------------------------------------------------
3603 :	george	824	Name: Lal George
3604 :			Date: 2001/05/08 17:26:09 EDT
3605 :			Tag: george-20010508-omit-frameptr
3606 :			Description:
3607 :
3608 :			Changes to implement the omit-frame-pointer optimization to support
3609 :			raw C calls. For now, there is only support on the Intel x86, but
3610 :			other architectures will follow as more experience is gained with this.
3611 :
3612 :
3613 :			----------------------------------------------------------------------
3614 :	blume	818	Name: Matthias Blume
3615 :	blume	819	Date: 2001/05/07 14:40:00 EDT
3616 :			Tag: blume-20010507-proxies
3617 :			Description:
3618 :
3619 :			I made into "proxy libraries" all libraries that qualify for such a
3620 :			change. (A qualifying library is a library that has another library or
3621 :			groups as its sole member and repeats that member's export list
3622 :			verbatim. A proxy library avoids this repetition by omitting its export
3623 :			list, effectively inheriting the list that its (only) member exports.
3624 :			See the CM manual for more explanation.)
3625 :			The main effect is that explicit export lists for these libraries
3626 :			do not have to be kepts in sync, making maintenance a bit easier.
3627 :
3628 :			I also added copyright notices to many .cm-files.
3629 :
3630 :			Last but not least, I made a new set of bootfiles.
3631 :
3632 :			----------------------------------------------------------------------
3633 :			Name: Matthias Blume
3634 :	blume	818	Date: 2001/05/04 17:00:00 EDT
3635 :			Tag: blume-20010504-cm-lsplit
3636 :			Description:
3637 :
3638 :			0. John merged pending changes to $/smlnj-lib.cm
3639 :
3640 :			1. Allen's previous change accidentally backed out of one of Lal's
3641 :			earlier changes. I undid this mistake (re-introducing Lal's change).
3642 :
3643 :			2. I used the new topOrder' function from graph-scc.sml (from $/smlnj-lib.cm)
3644 :			within the compiler where applicable. There is some code simplification
3645 :			because of that.
3646 :
3647 :			3. The "split" phase (in FLINT) is now part of the default list of phases.
3648 :			Compiler.Control.LambdaSplitting.* can be used to globally control the
3649 :			lambda-splitting (cross-module-inlining) engine. In addition to that,
3650 :			it can now also be controlled on a per-source basis: CM has been taught
3651 :			a new tool parameter applicable to ML source files.
3652 :
3653 :			- To turn lambda-splitting off completely:
3654 :			local open Compiler.Control.LambdaSplitting in
3655 :			val _ = set Off
3656 :			end
3657 :			- To make "no lambda-splitting" the global default (but allow per-source
3658 :			overriding); this is the initial setting:
3659 :			local open Compiler.Control.LambdaSplitting in
3660 :			val _ = set (Default NONE)
3661 :			end
3662 :			- To make "lambda-splitting with aggressiveness a" the global default
3663 :			(and allow per-source overriding):
3664 :			local open Compiler.Control.LambdaSplitting in
3665 :			val _ = set (Default (SOME a))
3666 :			end
3667 :
3668 :			- To turn lambda-splitting off for a given ML souce file (say: a.sml)
3669 :			write (in the respective .cm-file):
3670 :			a.sml (lambdasplitting:off)
3671 :			- To turn lambda-splitting for a.sml on with minimal aggressiveness:
3672 :			a.sml (lambdasplitting:on)
3673 :			- To turn lambda-splitting for a.sml on with aggressiveness <a> (where
3674 :			<a> is a decimal non-negative integer):
3675 :			a.sml (lambdasplitting:<a>)
3676 :			- To turn lambda-splitting for a.sml on with maximal aggressiveness:
3677 :			a.sml (lambdasplitting:infinity)
3678 :			- To use the global default for a.sml:
3679 :			a.sml (lambdasplitting:default)
3680 :			or simply
3681 :			a.sml
3682 :
3683 :			----------------------------------------------------------------------
3684 :	leunga	815	Name: Allen Leung
3685 :			Date: 2001/05/04 01:57:00 EDT
3686 :			Tag: leunga-20010504-sync
3687 :			Description:
3688 :	leunga	788
3689 :	leunga	815	MLRISC features.
3690 :
3691 :			1. Fix to CMPXCHG instructions.
3692 :			2. Changed RA interface to allow annotations in callbacks.
3693 :			3. Added a new method to the stream interface to allow annotations updates.
3694 :
3695 :	george	749	----------------------------------------------------------------------
3696 :	blume	812	Name: Matthias Blume
3697 :	blume	814	Date: 2001/05/01 11:45:00 EDT
3698 :			Tag: blume-20010501-pcedittmp
3699 :			Description:
3700 :
3701 :			Changed install.sh to use the current working directory instead of
3702 :			/usr/tmp for a temporary file (pcedittmp). The previous choice
3703 :			of /usr/tmp caused trouble with MacOS X because of file premission
3704 :			problems.
3705 :
3706 :			----------------------------------------------------------------------
3707 :			Name: Matthias Blume
3708 :	blume	812	Date: 2001/04/20 11:10:00 EDT
3709 :			Tag: blume-20010420-inMLflag
3710 :			Description:
3711 :
3712 :			- added vp_limitPtrMask to vproc-state.h
3713 :			(for use by the raw-C-calls mechanism to implement proper interrupt
3714 :			handling)
3715 :			- made the ML compiler aware of various data-structure offsets so it
3716 :			can generate code for accessing the vp_inML flag and vp_limitPtrMask
3717 :			- tweaked mlriscGen.sml to have it emit interrupt-handling code for
3718 :			raw C-calls
3719 :
3720 :			----------------------------------------------------------------------
3721 :	george	810	Name: Lal George
3722 :			Date: 2001/04/20 09:15:28 EDT
3723 :			Tag: george-20010420-macosX
3724 :			Description:
3725 :
3726 :			- Changes to port to Mac OS X; Darwin.
3727 :
3728 :			- In the process I found that sqrt was broken on the PPC, because the
3729 :			fsqrt instruction is not implemented.
3730 :
3731 :			----------------------------------------------------------------------
3732 :	blume	799	Name: Matthias Blume
3733 :	blume	808	Date: 2001/04/18 12:45:00 EDT
3734 :			Tag: blume-20010418-ccalls
3735 :			Description:
3736 :
3737 :			- fixed two off-by-4 errors in the x86-specific c-calls implementation
3738 :			(this bug prevented structure arguments containing pointers from being
3739 :			passed correctly)
3740 :			- changed the raw-C-call code in mlriscGen.sml in such a way that
3741 :			structure arguments are represented as a pointer to the beginning
3742 :			of the structure (instead of having a series of synthesized arguments,
3743 :			one for each structure member)
3744 :
3745 :			- made makeml script's verbosity level configurable via environment
3746 :			variable (MAKEML_VERBOSITY)
3747 :
3748 :	blume	813	- eliminated placeholder implementations for f32l, w16s, i16s, and f32s
3749 :	blume	808	in rawmem-x86.sml; we are now using the real thing
3750 :
3751 :			----------------------------------------------------------------------
3752 :			Name: Matthias Blume
3753 :	blume	806	Date: 2001/03/22 16:25:00 EST
3754 :			Tag: blume-20010322-bootfiles
3755 :			Description:
3756 :
3757 :			Created a new set of bootfiles (for your automatic installation convenience).
3758 :
3759 :			----------------------------------------------------------------------
3760 :			Name: Matthias Blume
3761 :	blume	805	Date: 2001/03/22 15:10:00 EST
3762 :			Tag: blume-20010322-rawmem-parcm
3763 :			Description:
3764 :
3765 :			1. All "raw memory access" primitives for the new FFI are implemented now
3766 :			(at least on the x86).
3767 :			2. Some further cleanup of CM's parallel make mechanism.
3768 :
3769 :			----------------------------------------------------------------------
3770 :			Name: Matthias Blume
3771 :	blume	801	Date: 2001/03/19 17:53:00 EST
3772 :			Tag: blume-20010319-parallel
3773 :			Description:
3774 :
3775 :			Parallel make (using compile servers) now works again.
3776 :
3777 :			To this end, CM.stabilize and CMB.make have been modified to work in
3778 :			two passes when compile servers are attached:
3779 :			1. Compile everything, do not perform stabilization; this pass
3780 :			uses compile servers
3781 :			2. Stabilize everything; this pass does not use compile servers
3782 :			If there are no compile servers, the two passes are combined into one
3783 :			(as before). Splitting the passes increases the inherent parallelism
3784 :			in the dependency graph because the entire graph including all
3785 :			libraries is available at the same time. This, in turn, improves
3786 :			server utilization. The downside is that the master process will
3787 :			have to do some extra work after compilation is done (because for
3788 :			technical reasons it must re-read all the binfiles during stabilization).
3789 :
3790 :			----------------------------------------------------------------------
3791 :			Name: Matthias Blume
3792 :	blume	800	Date: 2001/03/16 12:22:00 EST
3793 :			Tag: blume-20010316-bootfiles
3794 :			Description:
3795 :
3796 :			Created a new set of bootfiles (for your automatic installation convenience).
3797 :
3798 :			----------------------------------------------------------------------
3799 :			Name: Matthias Blume
3800 :	blume	799	Date: 2001/03/16 11:00:00 EST
3801 :			Tag: blume-20010316-MLTREE-fixup
3802 :			Description:
3803 :
3804 :			This is a minor fixup for an (untagged) earlier commit by Allen.
3805 :			(A file was missing).
3806 :
3807 :			----------------------------------------------------------------------
3808 :	leunga	788	Name: Allen Leung
3809 :	leunga	796	Date: Mon Mar 5 18:54:57 EST 2001
3810 :			Tag: leunga-20010305-cut-support
3811 :
3812 :			1. New support for alternative control-flow in MLTREE.
3813 :			Currently we support
3814 :
3815 :			FLOW_TO(CALL ...., [k1,...,kn])
3816 :
3817 :			This is needed for 'cuts to' in C-- and try/handle-like constructs
3818 :			in Moby
3819 :
3820 :			New assembler flag "asm-show-cutsto" to turn on control-flow debugging.
3821 :
3822 :			2. Register Allocator
3823 :
3824 :			Changes in interface [from Fermin, John]
3825 :
3826 :			3. Alpha 8-bit SLL support [Fermin]
3827 :
3828 :			4. All architectures
3829 :
3830 :			A new module (ClusterExpandCopies) for expanding parallel copies.
3831 :
3832 :			----------------------------------------------------------------------
3833 :			Name: Allen Leung
3834 :	leunga	788	Date: 2001/02/27 23:07:00 EST
3835 :			Tag: leunga-20010227-minor-stuff
3836 :
3837 :			1. Alpha bug fix for CMOVNE
3838 :			2. Handle mltree COND(..,FCMP ...,...)
3839 :			3. Bug fix in simplifier
3840 :
3841 :			----------------------------------------------------------------------
3842 :	blume	777	Name: Matthias Blume
3843 :	blume	781	Date: 2001/01/30 17:50:00 EST
3844 :			Tag: blume-20010130-sync
3845 :			Description:
3846 :
3847 :			This is just a minor update to sync my devel branch with the main brach.
3848 :			The only visible change is the addition of some README files.
3849 :
3850 :			----------------------------------------------------------------------
3851 :			Name: Matthias Blume
3852 :	blume	778	Date: 2001/01/12 23:30:00 JST
3853 :			Tag: blume-20010112-bootfiles
3854 :			Description:
3855 :
3856 :			Made a new set of bootfiles that goes with the current state of the
3857 :			repository.
3858 :
3859 :			----------------------------------------------------------------------
3860 :			Name: Matthias Blume
3861 :	blume	777	Date: 2001/01/12 21:20:00 JST
3862 :			Tag: blume-20010112-sync
3863 :			Description:
3864 :
3865 :			I am just flushing out some minor changes that had accumulated in
3866 :			my private branch in order to sync with the main tree. (This is
3867 :			mainly because I had CVS trouble when trying to merge _into_ my
3868 :			private branch.)
3869 :
3870 :			Most people should be completely unaffected by this.
3871 :
3872 :			----------------------------------------------------------------------
3873 :	leunga	775	Name: Allen Leung
3874 :			Date: Thu Jan 11 21:03:00 EST 2001
3875 :			Tag: leunga-20010111-labexp=mltree
3876 :			Description:
3877 :
3878 :			1. Removed the type LabelExp and replace it by MLTree.
3879 :			2. Rewritten mltree-simplify with the pattern matcher tool.
3880 :			3. There were some bugs in alpha code generator which would break
3881 :			64-bit code generation.
3882 :			4. Redo the tools to generate code with the
3883 :			5. The CM files in MLRISC (and in src/system/smlnj/MLRISC)
3884 :			are now generated by perl scripts.
3885 :
3886 :			----------------------------------------------------------------------
3887 :	blume	771	Name: Matthias Blume
3888 :	blume	774	Date: 2001/01/10 21:55:00 JST
3889 :			Tag: blume-20010110-rcc
3890 :			Description:
3891 :
3892 :			The RCC stuff now seems to work (but only on the x86).
3893 :			This required hacking of the c-calls interface (and -implementation) in
3894 :			MLRISC.
3895 :
3896 :			Normal compiler users should be unaffected.
3897 :
3898 :			----------------------------------------------------------------------
3899 :			Name: Matthias Blume
3900 :	blume	773	Date: 2001/01/09 01:20:00 JST
3901 :			Tag: blume-20010109-rcc
3902 :			Description:
3903 :
3904 :			This is a fairly big patch, flushing out a large number of pending
3905 :			changes that I made to my development copy over the last couple of days.
3906 :
3907 :			Of practical relevance at this moment is a workaround for a pickling
3908 :			bug that Allen ran into the other day. The cause of the bug itself is
3909 :			still unknown and it might be hard to fix it properly, but the
3910 :			workaround has some merits of its own (namely somewhat reducing pickling
3911 :			overhead for certain libraries). Therefore, I think this solution should
3912 :			be satisfactory at this time.
3913 :
3914 :			The rest of the changes (i.e., the vast majority) has to do with my
3915 :			ongoing efforts of providing direct support for C function calls from
3916 :			ML. At the moment there is a new primop "RAW_CCALL", typing magic
3917 :			in types/cproto.sml (invoked from FLINT/trans/translate.sml), a new
3918 :			case in the FLINT CPS datatype (RCC), changes to cps/convert.sml to
3919 :			translate uses of RAW_CCALL into RCC, and changes to mlriscGen.sml to
3920 :			handle RCC.
3921 :
3922 :			The last part (the changes to mlriscGen.sml) are still known to be
3923 :			wrong on the x86 and not implemented on all other architectures. But
3924 :			the infrastructure is in place. I had to change a few functor
3925 :			signatures in the backend to be able to route the CCalls interface
3926 :			from MLRISC there, and I had to specialize the mltree type (on the
3927 :			x86) to include the necessary extensions. (The extensions themselves
3928 :			were already there and redy to go in MLRISC/x86).
3929 :
3930 :			Everything should be very happy as soon as someone helps me with
3931 :			mlriscGen.sml...
3932 :
3933 :			In any case, nothing of this should matter to anyone as long as the
3934 :			new primop is not being used (which is going to be the case unless you
3935 :			find it where I hid it :). The rest of the compiler is completely
3936 :			unaffected.
3937 :
3938 :			----------------------------------------------------------------------
3939 :			Name: Matthias Blume
3940 :	blume	772	Date: 2001/01/05 00:30:00 JST
3941 :			Tag: blume-20010105-primops
3942 :			Description:
3943 :
3944 :			Added some experimental support for work that I am doing right now.
3945 :			These changes mostly concern added primops, but there is also a new
3946 :			experimental C library in the runtime system (but currently not enabled
3947 :			anywhere except on Linux/X86).
3948 :
3949 :			In the course of adding primops (and playing with them), I discovered that
3950 :			Zhong's INL_PRIM hack (no type info for certain primops) was, in fact, badly
3951 :			broken. (Zhong was very right he labeled this stuff as "major gross hack".)
3952 :			To recover, I made type information in INL_PRIM mandatory and changed
3953 :			prim.sml as well as built-in.sml accordingly. The InLine structure now
3954 :			has complete, correct type information (i.e., no bottom types).
3955 :
3956 :			Since all these changes mean that we need new binfiles, I also bumped the
3957 :			version number to 110.32.1.
3958 :
3959 :			----------------------------------------------------------------------
3960 :			Name: Matthias Blume
3961 :	blume	771	Date: 2000/12/30 22:10:00 JST
3962 :			Tag: blume-20001230-various
3963 :			Description:
3964 :
3965 :			Added proxy libraries for MLRISC and let MLRISC libraries refer
3966 :			to each other using path anchors. (See CM manual for explanation.)
3967 :
3968 :			Updated CM documentation.
3969 :
3970 :			Fixed some bugs in CM.
3971 :
3972 :			Implemented "proxy" libraries (= syntactic sugar for CM).
3973 :
3974 :			Added "-quiet" option to makeml and changed runtime system accordingly.
3975 :
3976 :			Added cleanup handler for exportML to reset timers and compiler stats.
3977 :
3978 :			----------------------------------------------------------------------
3979 :	george	761	Name: Lal George
3980 :			Date: 2000/12/22 22:22:58 EST 2000
3981 :			Tag: Release_110_32
3982 :			Description:
3983 :
3984 :			Infinite precision used throughout MLRISC.
3985 :			see MLRISC/mltree/machine-int.sig
3986 :
3987 :			----------------------------------------------------------------------
3988 :	blume	756	Name: Matthias Blume
3989 :	blume	760	Date: 2000/12/22 23:16:00 JST
3990 :			Tag: blume-20001222-warn
3991 :			Description:
3992 :	blume	759
3993 :	blume	760	Corrected wording and formatting of some CM warning message which I
3994 :			broke in my previous patch.
3995 :	blume	759
3996 :			----------------------------------------------------------------------
3997 :			Name: Matthias Blume
3998 :	blume	763	Date: 2000/12/22 21:20:00 JST
3999 :			Tag: blume-20001222-anchorenv
4000 :			Description:
4001 :
4002 :			Fixed CM's handling of anchor environments in connection with CMB.make.
4003 :
4004 :			----------------------------------------------------------------------
4005 :			Name: Matthias Blume
4006 :	blume	758	Date: 2000/12/22 13:15:00 JST
4007 :			Tag: blume-20001222-cleanup
4008 :			Description:
4009 :
4010 :			Removed src/cm/ffi which does not (and did not) belong here.
4011 :
4012 :			----------------------------------------------------------------------
4013 :			Name: Matthias Blume
4014 :	blume	757	Date: 2000/12/21 23:55:00 JST
4015 :			Tag: blume-20001221-exn
4016 :			Description:
4017 :
4018 :			Probably most important: CM no longer silently swallows all exceptions
4019 :			in the compiler.
4020 :			Plus: some other minor CM changes. For example, CM now reports some
4021 :			sizes for generated binfiles (code, data, envpickle, lambdapickle).
4022 :
4023 :			----------------------------------------------------------------------
4024 :			Name: Matthias Blume
4025 :	blume	756	Date: 2000/12/15 00:01:05 JST
4026 :			Tag: blume-20001215-dirtool
4027 :			Description:
4028 :
4029 :			- "dir" tool added.
4030 :			- improvements and cleanup to Tools structure
4031 :			- documentation updates
4032 :
4033 :			----------------------------------------------------------------------
4034 :	leunga	752	Name: Allen Leung
4035 :	leunga	755	Date: Thu Dec 14 03:45:24 EST 2000
4036 :			Description:
4037 :			Tag: leunga-20001214-int-inf
4038 :			Description:
4039 :
4040 :			In IntInf, added these standard functions, which are missing from our
4041 :			implementation:
4042 :
4043 :			andb : int * int -> int
4044 :			xorb : int * int -> int
4045 :			orb : int * int -> int
4046 :			notb : int -> int
4047 :			<< : int * word -> int
4048 :			~>> : int * word -> int
4049 :
4050 :			Not tested, I hope they are correct.
4051 :
4052 :			----------------------------------------------------------------------
4053 :			Name: Allen Leung
4054 :	leunga	752	Date: Fri Dec 8 19:23:26 EST 2000
4055 :			Description:
4056 :			Tag: leunga-20001208-nowhere
4057 :			Description:
4058 :
4059 :			Slight improvements to the 'nowhere' tool to handle OR-patterns,
4060 :			to generate better error messages etc. Plus a brief manual.
4061 :
4062 :			----------------------------------------------------------------------
4063 :	george	749	Name: Lal George
4064 :			Date: 2000/12/08 09:54:02 EST 2000
4065 :			Tag: Release_110_31
4066 :			Description:
4067 :	george	705
4068 :	george	749	- Version 110.31
4069 :	leunga	591	----------------------------------------------------------------------
4070 :	leunga	744	Name: Allen Leung
4071 :			Date: Thu Dec 7 22:01:04 EST 2000
4072 :			Tag: leunga-20001207-cell-monster-hack
4073 :			Description:
4074 :
4075 :			Major MLRISC internal changes. Affect all clients.
4076 :			Summary:
4077 :
4078 :			1. Type CELLS.cell = int is now replaced by a datatype.
4079 :			As a result, the old regmap is now gone. Almost all interfaces
4080 :			in MLRISC change as a consequence.
4081 :
4082 :			2. A new brand version of machine description tool (v3.0) that generates
4083 :			modules expecting the new interface. The old version is removed.
4084 :
4085 :			3. The RA interface has been further abstracted into two new functors.
4086 :			RISC_RA and X86RA. These functors have much simpler interfaces.
4087 :			[See also directory MLRISC/demo.]
4088 :
4089 :			4. Some other new source->source code generation tools are available:
4090 :
4091 :			a. MLRISC/Tools/RewriteGen -- generate rewriters from rules.
4092 :			b. MLRISC/Tools/WhereGen -- expands conditional pattern matching rules.
4093 :			I use this tool to generate the peephole optimizers---with the new
4094 :			cell type changes, peephole rules are becoming difficult to write
4095 :			without conditional pattern matching.
4096 :
4097 :			5. More Intmap -> IntHashTable change. Previous changes by Matthias didn't
4098 :			cover the entire MLRISC source tree so many things broke.
4099 :
4100 :			6. CM files have been moved to the subdirectory MLRISC/cm.
4101 :			They are moved because there are a lot of them and they clutter up the
4102 :			root dir.
4103 :
4104 :			7. More detailed documentation to come...
4105 :
4106 :			NOTE: To rebuild from 110.30 (ftp distribution), you'll have to do
4107 :			a makeml -rebuild first. This is because of other other
4108 :			changes that Matthias has made (see below).
4109 :
4110 :
4111 :			----------------------------------------------------------------------
4112 :	blume	733	Name: Matthias Blume
4113 :	blume	742	Date: 2000/11/30 23:12:00 JST
4114 :			Tag: blume-20001130-filereorg
4115 :			Description:
4116 :
4117 :			Some manual updates and some file reorganizations in CM.
4118 :
4119 :			----------------------------------------------------------------------
4120 :			Name: Matthias Blume
4121 :	blume	737	Date: 2000/11/24 17:45:00 JST
4122 :			Tag: blume-20001124-link
4123 :			Description:
4124 :
4125 :			Drastically improved link traversal code for the case that the dynamic
4126 :			value was already loaded at bootstrap time. As a result, CM and CMB
4127 :			now both load blazingly fast -- even on a very slow machine. Also,
4128 :			memory consumption has been further reduced by this.
4129 :
4130 :			Warning: The format of the PIDMAP file has changed. THerefore, to
4131 :			bootstrap you have to do this:
4132 :
4133 :			1. Run CMB.make
4134 :			2. Make a symbolic link for the boot directory:
4135 :			ln -s sml.boot.ARCH-OS xxx
4136 :			3. "Rebuild" the boot directory:
4137 :			./makeml -boot xxx -rebuild sml ; rm xxx
4138 :			4. Boot normally:
4139 :			./makeml
4140 :
4141 :			----------------------------------------------------------------------
4142 :			Name: Matthias Blume
4143 :	blume	735	Date: 2000/11/21 21:20:00 JST
4144 :			Tag: blume-20001121-tools
4145 :			Description:
4146 :
4147 :			Continued hacking on autoloading problem -- with success this time.
4148 :			Also changed tool-plugin mechanism. See new CM manual.
4149 :
4150 :			----------------------------------------------------------------------
4151 :			Name: Matthias Blume
4152 :	blume	734	Date: 2000/11/19 14:30:00 JST
4153 :			Tag: blume-20001119-autoload
4154 :			Description:
4155 :
4156 :			Some hacking to make autoloading faster. Success for CMB, no success
4157 :			so far for CM. There is a reduced structure CM' that autoloads faster.
4158 :			(This is a temporary, non-documented hack to be eliminated again when
4159 :			the general problem is solved.)
4160 :
4161 :			----------------------------------------------------------------------
4162 :			Name: Matthias Blume
4163 :	blume	733	Date: 2000/11/17 14:10:00 JST
4164 :			Tag: blume-20001117-pickle-lib
4165 :			Description:
4166 :
4167 :			1. Eliminated comp-lib.cm
4168 :			2. Made pickle-lib.cm
4169 :			3. Eliminated all uses of intset.sml (from comp-lib.cm)
4170 :			4. Replaced all uses of intmap.{sig,sml} (from comp-lib.cm) with
4171 :			equivalent constructs from smlnj-lib.cm (INtHashTable).
4172 :			5. Point 4. also goes for those uses of intmap.* in MLRISC.
4173 :			Duplicated intmap modules thrown out.
4174 :			6. Hunted down all duplicated SCC code and replaced it with
4175 :			equivalent stuff (GraphSCCFn from smlnj-lib.cm).
4176 :			7. Rewrote Feedback module.
4177 :			8. Moved sortedlist.sml into viscomp-lib.cm. Eventually it
4178 :			should be thrown out and equivalent modules from smlnj-lib.cm
4179 :			should be used (IntRedBlackSet, IntListSet, ...).
4180 :
4181 :			Confirmed that compiler compiles to fixpoint.
4182 :
4183 :			----------------------------------------------------------------------
4184 :	leunga	731	Name: Allen Leung
4185 :			Date: 2000/11/10 18:00:00
4186 :			Tag: leunga-20001110-new-x86-fp
4187 :
4188 :			A new x86 floating point code generator has been added.
4189 :			By default this is turned off. To turn this on, do:
4190 :
4191 :			CM.autoload "$smlnj/compiler.cm";
4192 :			Compiler.Control.MLRISC.getFlag "x86-fast-fp" := true;
4193 :
4194 :			Changes:
4195 :
4196 :			1. Changed FTAN to FPTAN so that the assembly output is correct.
4197 :			2. Changed the extension callback for FTANGENT to generate:
4198 :
4199 :			fptan
4200 :			fstp %st(0)
4201 :			instead of
4202 :			fptan
4203 :			fstpl ftempmem
4204 :
4205 :			3. Numerous assembly fixes for x86.
4206 :
4207 :			5. Cleaned up the machine code output module x86/x86MC.sml and added
4208 :			support for a whole bunch of instructions and addressing modes:
4209 :
4210 :			fadd/fsub/fsubr/fmul/fdiv/fdivr %st, %st(n)
4211 :			faddp/fsubp/fsubrp/fmulp/fdivp/fdivrp %st, %st(n)
4212 :			fadd/fsub/fsubr/fmul/fdiv/fdivr %st(n), %st
4213 :			fiadd/fisub/fisubr/fimul/fidiv/fidivr mem
4214 :			fxch %st(n)
4215 :			fld %st(n)
4216 :			fst %st(n)
4217 :			fst mem
4218 :			fstp %st(n)
4219 :			fucom %st(n)
4220 :			fucomp %st(n)
4221 :
4222 :			All these are now generated when the fast fp mode is turned on.
4223 :
4224 :			6. Removed the dedicated registers %st(0), ..., %st(7) from X86CpsRegs
4225 :
4226 :			----------------------------------------------------------------------
4227 :	blume	729	Name: Matthias Blume
4228 :			Date: 2000/11/09 11:20:00 JST
4229 :			Tag: blume-20001109-scc
4230 :			Description:
4231 :
4232 :			Eliminated some code duplication:
4233 :
4234 :			1. Added "where" clause to GraphSCCFn in SML/NJ Library.
4235 :			(Otherwise the functor is useless.)
4236 :			2. Used GraphSCCFn where SCCUtilFun was used previously.
4237 :			3. Got rid of SCCUtilFun (in comp-lib.cm).
4238 :
4239 :			----------------------------------------------------------------------
4240 :	george	721	Name: Lal George
4241 :			Date: 2000/11/06 09:02:21 EST 2000
4242 :			Tag: Release_110_30
4243 :			Description:
4244 :
4245 :			- Version 110.30
4246 :			----------------------------------------------------------------------
4247 :	blume	715	Name: Matthias Blume
4248 :	blume	716	Date: 2000/11/04 14:45:00
4249 :			Tag: blume-20001104-mlbuild
4250 :			Description:
4251 :
4252 :			- Made ml-build faster on startup.
4253 :			- Documentation fixes.
4254 :
4255 :			----------------------------------------------------------------------
4256 :			Name: Matthias Blume
4257 :	blume	715	Date: 2000/11/02 17:00:00 JST
4258 :			Tag: blume-20001102-condcomp
4259 :			Description:
4260 :
4261 :			- Small tweaks to pickler -- new BOOTFILES!
4262 :			- Version bumped to 110.29.2.
4263 :			- Added conditional compilation facility to init.cmi (see comment there).
4264 :			----------------------------------------------------------------------
4265 :	leunga	713	Name: Allen Leung
4266 :			Date: 2000/10/23 19:31:00
4267 :			Tag: leunga-20001023-demo-ra
4268 :
4269 :			1. Minor RA changes that improves spilling on x86 (affects Moby and C-- only)
4270 :			2. Test programs for the graph library updated
4271 :			3. Some new MLRISC demo programs added
4272 :
4273 :			----------------------------------------------------------------------
4274 :	blume	710	Name: Matthias Blume
4275 :	blume	711	Date: 2000/08/31 22:15:00 JST
4276 :			Tag: blume-20001017-errmsg
4277 :			Description:
4278 :
4279 :			More error message grief: Where there used to be no messages, there
4280 :			now were some that had bogus error regions. Fixed.
4281 :
4282 :			----------------------------------------------------------------------
4283 :			Name: Matthias Blume
4284 :	blume	710	Date: 2000/08/31 17:30:00 JST
4285 :			Tag: blume-20001017-v110p29p1
4286 :			Description:
4287 :
4288 :			I made a version 110.29.1 with new bootfiles.
4289 :
4290 :			Changes: Modified pickler/unpickler for faster and leaner unpickling.
4291 :			CM documentation changes and a small bugfix in CM's error reporting.
4292 :
4293 :			----------------------------------------------------------------------
4294 :	george	705	Name: Lal George
4295 :			Date: 2000/09/27 14:42:35 EDT
4296 :			Tag: george-20000927-nodestatus
4297 :			Description:
4298 :
4299 :			Changed the type of the nodestatus, so that:
4300 :
4301 :			SPILLED(~1) is now SPILLED
4302 :			SPILLED(m) where m>=0 is now MEMREG(m)
4303 :			SPILLED(s) where s<~1 is now SPILL_LOC(~s)
4304 :
4305 :			----------------------------------------------------------------------
4306 :	blume	697	Name: Matthias Blume
4307 :	blume	703	Date: 2000/09/07 14:45:00 JST
4308 :			Tag: blume-20000907-cmerrmsg
4309 :			Description:
4310 :
4311 :			Small tweak to CM to avoid getting ML syntax error messages twice.
4312 :
4313 :			----------------------------------------------------------------------
4314 :			Name: Matthias Blume
4315 :	blume	701	Date: 2000/08/31 18:00:00 JST
4316 :			Tag: blume-20000831-cvsbootfiles
4317 :			Description:
4318 :
4319 :			New URL for boot files (because the 110.29 files on the BL server do
4320 :			now work correctly with my updated install scripts for yacc and lex).
4321 :
4322 :			----------------------------------------------------------------------
4323 :			Name: Matthias Blume
4324 :	blume	697	Date: 2000/08/08 12:33:00 JST
4325 :			Tag: blume-20000808-manual
4326 :			Description:
4327 :
4328 :			Tiny update to CM manual.
4329 :