| 8 |
The form of an entry should be: |
The form of an entry should be: |
| 9 |
|
|
| 10 |
Name: |
Name: |
| 11 |
Date: |
Date: yyyy/mm/dd |
| 12 |
Tag: <post-commit CVS tag> |
Tag: <post-commit CVS tag> |
| 13 |
Description: |
Description: |
| 14 |
|
|
| 15 |
---------------------------------------------------------------------- |
---------------------------------------------------------------------- |
| 16 |
Name: Matthias Blume |
Name: Matthias Blume |
| 17 |
|
Date: 2002/01/31 16:05:00 EST |
| 18 |
|
Tag: blume-20020131-sparc-ccalls |
| 19 |
|
Description: |
| 20 |
|
|
| 21 |
|
1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes) |
| 22 |
|
of extra stack space by mistake. Fixed. |
| 23 |
|
|
| 24 |
|
2. Bug in logic of handling of command-line options in ml-nlffigen fixed. |
| 25 |
|
|
| 26 |
|
---------------------------------------------------------------------- |
| 27 |
|
Name: Allen Leung |
| 28 |
|
Date: 2002/01/30 |
| 29 |
|
Tag: leunga-20020130-nowhere-bug-fix |
| 30 |
|
Description: |
| 31 |
|
|
| 32 |
|
MLRISC bug fixes: |
| 33 |
|
1. Fixed a bindings computation bug in the 'nowhere' program generator tool. |
| 34 |
|
2. MachineInt.fromString was negating its value. |
| 35 |
|
|
| 36 |
|
---------------------------------------------------------------------- |
| 37 |
|
Name: Matthias Blume |
| 38 |
|
Date: 2002/01/29 |
| 39 |
|
Tag: blume-20020129-INSTALL |
| 40 |
|
Description: |
| 41 |
|
|
| 42 |
|
- Added somewhat detailed installation instructions (file INSTALL). |
| 43 |
|
- Fixed curl-detection bug in config/install.sh. |
| 44 |
|
- It is now possible to select the URL getter using the URLGETTER |
| 45 |
|
environment variable: |
| 46 |
|
|
| 47 |
|
not set / "unknown" --> automatic detection (script tries wget, |
| 48 |
|
curl, and lynx) |
| 49 |
|
"wget" / "curl" / "lynx" --> use the specified program (script "knows" |
| 50 |
|
how to properly invoke them) |
| 51 |
|
other --> use $URLGETTER directly, it must take |
| 52 |
|
precisely two command-line arguments |
| 53 |
|
(source URL and destination file name) |
| 54 |
|
|
| 55 |
|
---------------------------------------------------------------------- |
| 56 |
|
Name: Matthias Blume |
| 57 |
|
Date: 2002/01/28 |
| 58 |
|
Tag: blume-20020128-sparc-ccalls |
| 59 |
|
Description: |
| 60 |
|
|
| 61 |
|
- Fixed problem with calculation of "used" registers in sparc-c-calls. |
| 62 |
|
- Make use of the allocParam argument in sparc-c-calls. |
| 63 |
|
|
| 64 |
|
---------------------------------------------------------------------- |
| 65 |
|
Name: Matthias Blume |
| 66 |
|
Date: 2002/01/28 |
| 67 |
|
Tag: blume-20020128-allocParam |
| 68 |
|
Description: |
| 69 |
|
|
| 70 |
|
John Reppy: Changes c-calls API to accept client-callback for |
| 71 |
|
allocating extra stack space. |
| 72 |
|
me: Corresponding changes to mlriscGen (using a dummy argument that |
| 73 |
|
does not change the current behavior). |
| 74 |
|
|
| 75 |
|
---------------------------------------------------------------------- |
| 76 |
|
Name: Matthias Blume |
| 77 |
|
Date: 2002/01/28 12:00:00 |
| 78 |
|
Tag: Release_110_38 |
| 79 |
|
Description: |
| 80 |
|
|
| 81 |
|
This time for real!!! |
| 82 |
|
|
| 83 |
|
---------------------------------------------------------------------- |
| 84 |
|
Name: Matthias Blume |
| 85 |
|
Date: 2002/01/28 10:56:00 EST |
| 86 |
|
Tag: blume-20020128-retraction |
| 87 |
|
Description: |
| 88 |
|
|
| 89 |
|
0. Retracted earlier 110.38. (The Release_110_38 tag has been replaced |
| 90 |
|
with blume-Release_110_38-retracted.) |
| 91 |
|
|
| 92 |
|
1. Fixed a problem with incorrect rounding modes in real64.sml. |
| 93 |
|
(Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.) |
| 94 |
|
|
| 95 |
|
2. A bug in ml-nlffigen related to the handling of unnamed structs, unions, |
| 96 |
|
and enums fixed. The naming of corresponding ML identifiers should |
| 97 |
|
now be consistent again. |
| 98 |
|
|
| 99 |
|
---------------------------------------------------------------------- |
| 100 |
|
Name: Allen Leung |
| 101 |
|
Date: 2002/01/27 |
| 102 |
|
Tag: leunga-20020127-nowhere |
| 103 |
|
Description: |
| 104 |
|
|
| 105 |
|
Added a target called nowhere in the configuration scripts. |
| 106 |
|
Enabling this will build the MLRISC 'nowhere' tool (for translating |
| 107 |
|
programs with where-clauses into legal SML code) during installation. |
| 108 |
|
|
| 109 |
|
---------------------------------------------------------------------- |
| 110 |
|
Name: Matthias Blume |
| 111 |
|
Date: 2002/01/25 21:27:00 EST |
| 112 |
|
Tag: blume-Release_110_38-retracted |
| 113 |
|
Description: |
| 114 |
|
|
| 115 |
|
Call it a (working) release! Version is 110.38. Bootfiles are ready. |
| 116 |
|
|
| 117 |
|
README will be added later. |
| 118 |
|
|
| 119 |
|
!!! NOTE: Re-tagged as blume-Release_110_38-retracted. Original tag |
| 120 |
|
(Release_110_38) removed. Reason: Last-minute bug fixes. |
| 121 |
|
|
| 122 |
|
---------------------------------------------------------------------- |
| 123 |
|
Name: Matthias Blume |
| 124 |
|
Date: 2002/01/25 |
| 125 |
|
Tag: blume-20020125-ffi |
| 126 |
|
Description: |
| 127 |
|
|
| 128 |
|
A large number of tweaks and improvements to ml-nlffi-lib and |
| 129 |
|
ml-nlffigen: |
| 130 |
|
|
| 131 |
|
- ML represenation types have been streamlined |
| 132 |
|
- getter and setter functions work with concrete values, not abstract |
| 133 |
|
ones where possible |
| 134 |
|
- ml-nlffigen command line more flexible (see README file there) |
| 135 |
|
- some bugs have been fixed (hopefully) |
| 136 |
|
|
| 137 |
|
---------------------------------------------------------------------- |
| 138 |
|
Name: Lal George |
| 139 |
|
Date: 2002/01/24 |
| 140 |
|
Tag: george-20020124-risc-ra-interface |
| 141 |
|
Description: |
| 142 |
|
|
| 143 |
|
There is a dramatic simplification in the interface to the |
| 144 |
|
register allocator for RISC architectures as a result of making |
| 145 |
|
parallel copy instructions explicit. |
| 146 |
|
|
| 147 |
|
---------------------------------------------------------------------- |
| 148 |
|
Name: Matthias Blume |
| 149 |
|
Date: 2002/01/22 |
| 150 |
|
Tag: blume-20020122-x86-ccalls |
| 151 |
|
Description: |
| 152 |
|
|
| 153 |
|
Bug fix for c-calls on x86 (having to do with how char- and |
| 154 |
|
short-arguments are being handled). |
| 155 |
|
|
| 156 |
|
---------------------------------------------------------------------- |
| 157 |
|
Name: Matthias Blume |
| 158 |
|
Date: 2002/01/21 |
| 159 |
|
Tag: blume-20020121-ff |
| 160 |
|
Description: |
| 161 |
|
|
| 162 |
|
Another day of fiddling with the FFI... |
| 163 |
|
|
| 164 |
|
1. Bug fix/workaround: CKIT does not complain about negative array |
| 165 |
|
dimensions, so ml-nlffigen has to guard itself against this possibility. |
| 166 |
|
(Otherwise a negative dimension would send it into an infinite loop.) |
| 167 |
|
|
| 168 |
|
2. Some of the abstract types (light objects, light pointers, most "base" |
| 169 |
|
types) in structure C are now eqtypes. |
| 170 |
|
|
| 171 |
|
3. Added constructors and test functions for NULL function pointers. |
| 172 |
|
|
| 173 |
|
---------------------------------------------------------------------- |
| 174 |
|
Name: Matthias Blume |
| 175 |
|
Date: 2002/01/18 |
| 176 |
|
Tag: blume-20020118-ready-for-new-release |
| 177 |
|
Description: |
| 178 |
|
|
| 179 |
|
Made config/srcarchiveurl point to a new place. (Will provide boot |
| 180 |
|
files shortly.) |
| 181 |
|
|
| 182 |
|
Maybe we christen this to be 110.38? |
| 183 |
|
|
| 184 |
|
---------------------------------------------------------------------- |
| 185 |
|
Name: Matthias Blume |
| 186 |
|
Date: 2002/01/18 |
| 187 |
|
Tag: blume-20020118-more-ffifiddle |
| 188 |
|
Description: |
| 189 |
|
|
| 190 |
|
Today's FFI fiddling: |
| 191 |
|
|
| 192 |
|
- Provided a structure CGetSet with "convenient" versions of C.Get.* and |
| 193 |
|
C.Set.* that use concrete (MLRep.*) arguments and results instead |
| 194 |
|
of abstract ones. |
| 195 |
|
|
| 196 |
|
- Provided word-style bit operations etc. for "int" representation |
| 197 |
|
types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short, |
| 198 |
|
and Long. |
| 199 |
|
|
| 200 |
|
---------------------------------------------------------------------- |
| 201 |
|
Name: Matthias Blume |
| 202 |
|
Date: 2002/01/18 |
| 203 |
|
Tag: blume-20020118-use-x86-fp |
| 204 |
|
Description: |
| 205 |
|
|
| 206 |
|
Now that x86-fast-fp seems to be working, I turned it back on again |
| 207 |
|
by default. (Seems to work fine now, even with the FFI.) |
| 208 |
|
|
| 209 |
|
Other than that, I added some documentation about the FFI to |
| 210 |
|
src/ml-nlffigen/README and updated the FFI test examples in |
| 211 |
|
src/ml-nlffi-lib/Tests/*. |
| 212 |
|
|
| 213 |
|
---------------------------------------------------------------------- |
| 214 |
|
Name: Allen Leung |
| 215 |
|
Date: 2002/01/17 |
| 216 |
|
Tag: leunga-20020117-x86-fast-fp-call |
| 217 |
|
Description: |
| 218 |
|
|
| 219 |
|
1. Fixed a problem with handling return fp values when x86's fast fp |
| 220 |
|
mode is turned on. |
| 221 |
|
|
| 222 |
|
2. Minor pretty printing fix for cellset. Print %st(0) as %st(0) instead |
| 223 |
|
of %f32. |
| 224 |
|
|
| 225 |
|
3. Added a constructor INT32lit to the ast of MLRISC tools. |
| 226 |
|
|
| 227 |
|
---------------------------------------------------------------------- |
| 228 |
|
Name: Matthias Blume |
| 229 |
|
Date: 2002/01/16 |
| 230 |
|
Tag: blume-20020116-ffifiddle |
| 231 |
|
Description: |
| 232 |
|
|
| 233 |
|
More fiddling with the FFI interface: |
| 234 |
|
|
| 235 |
|
- Make constness 'c instead of rw wherever possible. This eliminates |
| 236 |
|
the need for certain explicit coercions. (However, due to ML's |
| 237 |
|
value polymorphism, there will still be many cases where explicit |
| 238 |
|
coercions are necessary. Phantom types are not the whole answer |
| 239 |
|
to modeling a subtyping relationship in ML.) |
| 240 |
|
|
| 241 |
|
- ro/rw coersions for pointers added. (Avoids the detour through */&.) |
| 242 |
|
|
| 243 |
|
- "printf" test example added to src/ml-nlffi-lib/Tests. (Demonstrates |
| 244 |
|
clumsy workaround for varargs problem.) |
| 245 |
|
|
| 246 |
|
---------------------------------------------------------------------- |
| 247 |
|
Name: Lal George |
| 248 |
|
Date: 2002/01/15 |
| 249 |
|
Tag: <none> |
| 250 |
|
Description: |
| 251 |
|
|
| 252 |
|
1. Since COPY instructions are no longer native to the architecture, |
| 253 |
|
a generic functor can be used to implement the expandCopies function. |
| 254 |
|
|
| 255 |
|
2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a |
| 256 |
|
TEXT segment. |
| 257 |
|
|
| 258 |
|
---------------------------------------------------------------------- |
| 259 |
|
Name: Matthias Blume |
| 260 |
|
Date: 2002/01/15 |
| 261 |
|
Tag: blume-20020115-ffiupdates |
| 262 |
|
Description: |
| 263 |
|
|
| 264 |
|
1. Fix for bug resulting in single-precision float values being returned |
| 265 |
|
incorrectly from FFI calls. |
| 266 |
|
|
| 267 |
|
2. Small modifications to C FFI API: |
| 268 |
|
|
| 269 |
|
- memory-allocation routines return straight objects (no options) |
| 270 |
|
and raise an exception in out-of-memory situations |
| 271 |
|
- unsafe extensions to cast between function pointers and pointers |
| 272 |
|
from/to ints |
| 273 |
|
- added structure C_Debug as an alternative to structure C where |
| 274 |
|
pointer-dereferencing (|*| and |*!) always check for null-pointers |
| 275 |
|
- added open_lib' to DynLinkage; open_lib' works like open_lib |
| 276 |
|
but also takes a (possibly empty) list of existing library handles |
| 277 |
|
that the current library depends on |
| 278 |
|
|
| 279 |
|
---------------------------------------------------------------------- |
| 280 |
|
Name: Matthias Blume |
| 281 |
|
Date: 2002/01/10 |
| 282 |
|
Tag: blume-20020110-newffigen |
| 283 |
|
Description: |
| 284 |
|
|
| 285 |
|
1. Updates to portable graph code. |
| 286 |
|
|
| 287 |
|
2. Major update to ml-nlffigen and ml-nlffi-lib. Things are much |
| 288 |
|
more scalable now so that even huge interfaces such as the one |
| 289 |
|
for GTK compile in finite time and space. :-) |
| 290 |
|
See src/ml-nlffigen/README for details on what's new. |
| 291 |
|
|
| 292 |
|
---------------------------------------------------------------------- |
| 293 |
|
Name: Lal George |
| 294 |
|
Date: 2001/01/09 14:31:35 EST 2002 |
| 295 |
|
Tag: george-20011206-rm-native-copy |
| 296 |
|
Description: |
| 297 |
|
|
| 298 |
|
Removed the native COPY and FCOPY instructions |
| 299 |
|
from all the architectures and replaced it with the |
| 300 |
|
explicit COPY instruction from the previous commit. |
| 301 |
|
|
| 302 |
|
It is now possible to simplify many of the optimizations |
| 303 |
|
modules that manipulate copies. This has not been |
| 304 |
|
done in this change. |
| 305 |
|
|
| 306 |
|
---------------------------------------------------------------------- |
| 307 |
|
Name: Lal George |
| 308 |
|
Date: 2001/12/06 16:50:13 EST 2001 |
| 309 |
|
Tag: george-20011206-mlrisc-instruction |
| 310 |
|
Description: |
| 311 |
|
|
| 312 |
|
Changed the representation of instructions from being fully abstract |
| 313 |
|
to being partially concrete. That is to say: |
| 314 |
|
|
| 315 |
|
from |
| 316 |
|
type instruction |
| 317 |
|
|
| 318 |
|
to |
| 319 |
|
type instr (* machine instruction *) |
| 320 |
|
|
| 321 |
|
datatype instruction = |
| 322 |
|
LIVE of {regs: C.cellset, spilled: C.cellset} |
| 323 |
|
| KILL of {regs: C.cellset, spilled: C.cellset} |
| 324 |
|
| COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list} |
| 325 |
|
| ANNOTATION of {i: instruction, a: Annotations.annotation} |
| 326 |
|
| INSTR of instr |
| 327 |
|
|
| 328 |
|
This makes the handling of certain special instructions that appear on |
| 329 |
|
all architectures easier and uniform. |
| 330 |
|
|
| 331 |
|
LIVE and KILL say that a list of registers are live or killed at the |
| 332 |
|
program point where they appear. No spill code is generated when an |
| 333 |
|
element of the 'regs' field is spilled, but the register is moved to |
| 334 |
|
the 'spilled' (which is present, more for debugging than anything else). |
| 335 |
|
|
| 336 |
|
LIVE replaces the (now deprecated) DEFFREG instruction on the alpha. |
| 337 |
|
We used to generate: |
| 338 |
|
|
| 339 |
|
DEFFREG f1 |
| 340 |
|
f1 := f2 + f3 |
| 341 |
|
trapb |
| 342 |
|
|
| 343 |
|
but now generate: |
| 344 |
|
|
| 345 |
|
f1 := f2 + f3 |
| 346 |
|
trapb |
| 347 |
|
LIVE {regs=[f1,f2,f3], spilled=[]} |
| 348 |
|
|
| 349 |
|
Furthermore, the DEFFREG (hack) required that all floating point instruction |
| 350 |
|
use all registers mentioned in the instruction. Therefore f1 := f2 + f3, |
| 351 |
|
defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting |
| 352 |
|
in a cleaner alpha implementation. (Hopefully, intel will not get rid of |
| 353 |
|
this architecture). |
| 354 |
|
|
| 355 |
|
COPYXXX is intended to replace the parallel COPY and FCOPY available on |
| 356 |
|
all the architectures. This will result in further simplification of the |
| 357 |
|
register allocator that must be aware of them for coalescing purposes, and |
| 358 |
|
will also simplify certain aspects of the machine description that provides |
| 359 |
|
callbacks related to parallel copies. |
| 360 |
|
|
| 361 |
|
ANNOTATION should be obvious, and now INSTR represents the honest to God |
| 362 |
|
machine instruction set! |
| 363 |
|
|
| 364 |
|
The <arch>/instructions/<arch>Instr.sml files define certain utility |
| 365 |
|
functions for making porting easier -- essentially converting upper case |
| 366 |
|
to lower case. All machine instructions (of type instr) are in upper case, |
| 367 |
|
and the lower case form generates an MLRISC instruction. For example on |
| 368 |
|
the alpha we have: |
| 369 |
|
|
| 370 |
|
datatype instr = |
| 371 |
|
LDA of {r:cell, b:cell, d:operand} |
| 372 |
|
| ... |
| 373 |
|
|
| 374 |
|
val lda : {r:cell, b:cell, d:operand} -> instruction |
| 375 |
|
... |
| 376 |
|
|
| 377 |
|
where lda is just (INSTR o LDA), etc. |
| 378 |
|
|
| 379 |
|
---------------------------------------------------------------------- |
| 380 |
|
Name: Matthias Blume |
| 381 |
|
Date: 2001/11/22 21:40:00 EST |
| 382 |
|
Tag: Release_110_37 |
| 383 |
|
Description: |
| 384 |
|
|
| 385 |
|
Release 110.37. This time for real. |
| 386 |
|
|
| 387 |
|
---------------------------------------------------------------------- |
| 388 |
|
Name: Matthias Blume |
| 389 |
|
Date: 2001/11/21 16:35:00 EST |
| 390 |
|
Tag: blume-20011121-foot-in-mouth |
| 391 |
|
Description: |
| 392 |
|
|
| 393 |
|
Removed the "Release_110_37" tag because of a serious bug. |
| 394 |
|
This will be re-tagged once the bug is fixed. |
| 395 |
|
|
| 396 |
|
---------------------------------------------------------------------- |
| 397 |
|
Name: Matthias Blume |
| 398 |
|
Date: 2001/11/21 16:14:00 EST |
| 399 |
|
Tag: blume-20011121-forgottenfile |
| 400 |
|
Description: |
| 401 |
|
|
| 402 |
|
Forgot to add a file. (Just a .tex-file -- part of |
| 403 |
|
the CM manual source.) |
| 404 |
|
|
| 405 |
|
---------------------------------------------------------------------- |
| 406 |
|
Name: Matthias Blume |
| 407 |
|
Date: 2001/11/21 16:10:00 EST |
| 408 |
|
Tag: blume-20011121-invalid_110_37 |
| 409 |
|
Description: |
| 410 |
|
|
| 411 |
|
Note: I removed the original tag "Release_110_37" from this commit |
| 412 |
|
because we found a serious bug in all non-x86 backends. |
| 413 |
|
- Matthias |
| 414 |
|
|
| 415 |
|
1. Modifications to the SML/NJ code generator and to the runtime system |
| 416 |
|
so that code object name strings are directly inserted into code |
| 417 |
|
objects at code generation time. The only business the runtime system |
| 418 |
|
has with this is now to read the name strings on occasions. |
| 419 |
|
(The encoding of the name string has also changed somewhat.) |
| 420 |
|
|
| 421 |
|
2. CM now implements a simple "set calculus" for specifying export lists. |
| 422 |
|
In particular, it is now possible to refer to the export lists of |
| 423 |
|
other libraries/groups/sources and form unions as well as differences. |
| 424 |
|
See the latest CM manual for details. |
| 425 |
|
|
| 426 |
|
3. An separate notion of "proxy" libraries has again be eliminated from |
| 427 |
|
CM's model. (Proxy libraries are now simply a special case of using |
| 428 |
|
the export list calculus.) |
| 429 |
|
|
| 430 |
|
4. Some of the existing libraries now take advantage of the new set |
| 431 |
|
calculus. |
| 432 |
|
(Notice that not all libraries have been converted because some |
| 433 |
|
of the existing .cm-files are supposed to be backward compatible |
| 434 |
|
with 110.0.x.) |
| 435 |
|
|
| 436 |
|
5. Some cleanup in stand-alone programs. (Don't use "exnMessage" -- use |
| 437 |
|
"General.exnMessage"! The former relies on a certain hook to be |
| 438 |
|
initialized, and that often does not happen in the stand-alone case.) |
| 439 |
|
|
| 440 |
|
---------------------------------------------------------------------- |
| 441 |
|
Name: Lal George |
| 442 |
|
Date: 2001/11/21 13:56:18 EST |
| 443 |
|
Tag: george-2001121-pseudo-ops |
| 444 |
|
Description: |
| 445 |
|
|
| 446 |
|
Implemented a complete redesign of MLRISC pseudo-ops. Now there |
| 447 |
|
ought to never be any question of incompatabilities with |
| 448 |
|
pseudo-op syntax expected by host assemblers. |
| 449 |
|
|
| 450 |
|
For now, only modules supporting GAS syntax are implemented |
| 451 |
|
but more should follow, such as MASM, and vendor assembler |
| 452 |
|
syntax, e.g. IBM as, Sun as, etc. |
| 453 |
|
|
| 454 |
|
---------------------------------------------------------------------- |
| 455 |
|
Name: Matthias Blume |
| 456 |
|
Date: 2001/11/14 11:52:00 EST |
| 457 |
|
Tag: blume-20011114-srcname |
| 458 |
|
Description: |
| 459 |
|
|
| 460 |
|
1. Routed the name of the current source file to mlriscgen where it |
| 461 |
|
should be directly emitted into the code object. (This last part |
| 462 |
|
is yet to be done.) |
| 463 |
|
|
| 464 |
|
2. Some cleanup of the pgraph code to make it match the proposal that |
| 465 |
|
I put out the other day. (The proposal notwithstanding, things are |
| 466 |
|
still in flux here.) |
| 467 |
|
|
| 468 |
|
---------------------------------------------------------------------- |
| 469 |
|
Name: Lal George |
| 470 |
|
Date: 2001/11/14 09:44:04 EST |
| 471 |
|
Tag: |
| 472 |
|
Description: |
| 473 |
|
|
| 474 |
|
Fix for a backpatching bug reported by Allen. |
| 475 |
|
|
| 476 |
|
Because the boundary between short and long span-dependent |
| 477 |
|
instructions is +/- 128, there are an astounding number of |
| 478 |
|
span-dependent instructions whose size is over estimated. |
| 479 |
|
|
| 480 |
|
Allen came up with the idea of letting the size of span |
| 481 |
|
dependent instructions be non-monotonic, for a maxIter |
| 482 |
|
number of times, after which the size must be monotonically |
| 483 |
|
increasing. |
| 484 |
|
|
| 485 |
|
This table shows the number of span-dependent instructions |
| 486 |
|
whose size was over-estimated as a function of maxIter, for the |
| 487 |
|
file Parse/parse/ml.grm.sml: |
| 488 |
|
|
| 489 |
|
maxIter # of instructions: |
| 490 |
|
10 687 |
| 491 |
|
20 438 |
| 492 |
|
30 198 |
| 493 |
|
40 0 |
| 494 |
|
|
| 495 |
|
In compiling the compiler, there is no significant difference in |
| 496 |
|
compilation speed between maxIter=10 and maxIter=40. Actually, |
| 497 |
|
my measurements showed that maxIter=40 was a tad faster than |
| 498 |
|
maxIter=10! Also 96% of the files in the compiler reach a fix |
| 499 |
|
point within 13 iterations, so fixing maxIter at 40, while high, |
| 500 |
|
is okay. |
| 501 |
|
|
| 502 |
|
---------------------------------------------------------------------- |
| 503 |
|
Name: Matthias Blume |
| 504 |
|
Date: 2001/10/31 15:25:00 EST |
| 505 |
|
Tag: blume-20011031-pgraph |
| 506 |
|
Description: |
| 507 |
|
|
| 508 |
|
CKIT: |
| 509 |
|
* Changed the "Function" constructor of type Ast.ctype to carry optional |
| 510 |
|
argument identifiers. |
| 511 |
|
* Changed the return type of TypeUtil.getFunction accordingly. |
| 512 |
|
* Type equality ignores the argument names. |
| 513 |
|
* TypeUtil.composite tries to preserve argument names but gives up quickly |
| 514 |
|
if there is a mismatch. |
| 515 |
|
|
| 516 |
|
installation script: |
| 517 |
|
* attempts to use "curl" if available (unless "wget" is available as well) |
| 518 |
|
|
| 519 |
|
CM: |
| 520 |
|
* has an experimental implementation of "portable graphs" which I will |
| 521 |
|
soon propose as an implementation-independent library format |
| 522 |
|
* there are also new libraries $/pgraph.cm and $/pgraph-util.cm |
| 523 |
|
|
| 524 |
|
NLFFI-LIB: |
| 525 |
|
* some cleanup (all cosmetic) |
| 526 |
|
|
| 527 |
|
NLFFIGEN: |
| 528 |
|
* temporarily disabled the mechanism that suppresses ML output for |
| 529 |
|
C definitions whose identifiers start with an underscore character |
| 530 |
|
* generate val bindings for enum constants |
| 531 |
|
* user can request that only one style (light or heavy) is being used; |
| 532 |
|
default is to use both (command-line arguments: -heavy and -light) |
| 533 |
|
* fixed bug in handling of function types involving incomplete pointers |
| 534 |
|
* generate ML entry points that take record arguments (i.e., using |
| 535 |
|
named arguments) for C functions that have a prototype with named |
| 536 |
|
arguments |
| 537 |
|
(see changes to CKIT) |
| 538 |
|
|
| 539 |
|
---------------------------------------------------------------------- |
| 540 |
|
Name: Allen Leung |
| 541 |
|
Date: 2001/10/27 20:34:00 EDT |
| 542 |
|
Tag: leunga-20011027-x86-fast-fp-call |
| 543 |
|
Description: |
| 544 |
|
|
| 545 |
|
Fixed the bug described in blume-20010920-slowfp. |
| 546 |
|
|
| 547 |
|
The fix involves |
| 548 |
|
1. generating FCOPYs in FSTP in ia32-svid |
| 549 |
|
2. marking a CALL with the appropriate annotation |
| 550 |
|
|
| 551 |
|
---------------------------------------------------------------------- |
| 552 |
|
Name: Matthias Blume |
| 553 |
Date: 2001/10/16 11:32:00 EDT |
Date: 2001/10/16 11:32:00 EDT |
| 554 |
Tag: blume-20011016-netbsd |
Tag: blume-20011016-netbsd |
| 555 |
Description: |
Description: |
Parent Directory
| 
Revision Log
| 
Patch
