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

# SCM Repository

[smlnj] Annotation of /sml/branches/primop-branch-3/NOTES/HISTORY
 [smlnj] / sml / branches / primop-branch-3 / NOTES / HISTORY

# Annotation of /sml/branches/primop-branch-3/NOTES/HISTORY

 1 : dbm 570 This is the HISTORY file for the Yale SML/NJ CVS repository. 2 : 3 : An entry should be made for _every_ commit to the repository. 4 : The entries in this file will be used when creating the README 5 : for new versions, so keep that in mind when writing the 6 : description. 7 : 8 : The form of an entry should be: 9 : 10 : mblume 1615 Name: 11 : mblume 1480 Date: yyyy/mm/dd 12 : Tag: 13 : Description: 14 : 15 : ---------------------------------------------------------------------- 16 : mblume 1622 Name: Matthias Blume (blume (at) tti - c (dot) org) 17 : mblume 1639 Date: 2004/09/30 10:55:00 CDT 18 : Tag: blume-20040930-version 19 : Description: 20 : 21 : * Moved the "version" magic into its own little library under 22 : src/system/smlnj/internal. This avoids expensive reconstruction 23 : of a stable src/compiler/core.cm. 24 : 25 : * At the same time, structure CompilerVersion is now known as structure 26 : SMLNJVersion. 27 : 28 : * Arranged for the version tool to NOT kick in when rebuilding the system 29 : (makeml -rebuild, fixpt). Otherwise one would never reach a fixpoint. 30 : Also, loading the versiontool does not work when rebuilding the system 31 : because CM is not properly initialized at that time. 32 : 33 : ---------------------------------------------------------------------- 34 : Name: Matthias Blume (blume (at) tti - c (dot) org) 35 : mblume 1638 Date: 2004/09/29 14:00:00 CDT 36 : Tag: blume-20040929-autoversion 37 : Description: 38 : 39 : Implemented some CM magic to have 40 : file src/compiler/TopLevel/main/version.sml 41 : generated automagically. 42 : The version is taken from two files: config/version and config/release. 43 : The first is expected to contain a two-part version number such as 110.49. 44 : The second should contain a single number, but it may be missing. 45 : 46 : If the environment variable VERSIONTOOL_BUMP_RELEASE is defined at the 47 : time the version tool is loaded (which is the first time you say CMB.make), 48 : then the tool will increment the value stored in config/release every 49 : time CMB.make is invoked. 50 : 51 : The binfile format is now insensitive to anything beyond the first 52 : two components of a version number, so bumping the release does not render 53 : binfiles incompatible. Auto-bumping can be used to keep track of versions 54 : during development without invalidating existing binfiles. 55 : 56 : In any case, every CMB.make updates the date information in version.sml. 57 : (This is the date that is printed in the banner.) 58 : 59 : ---------------------------------------------------------------------- 60 : Name: Matthias Blume (blume (at) tti - c (dot) org) 61 : mblume 1634 Date: 2004/09/28 10:53:00 CDT 62 : Tag: blume-20040928-controls 63 : Description: 64 : 65 : Some cleanup of the controls code. 66 : 67 : ---------------------------------------------------------------------- 68 : Name: Matthias Blume (blume (at) tti - c (dot) org) 69 : mblume 1633 Date: 2004/09/27 22:08:00 CDT 70 : Tag: blume-20040927-controls 71 : Description: 72 : 73 : Added two pieces of functionality to the Controls interface: 74 : 75 : 1. val save'restore: 'a control -> unit -> unit 76 : 77 : grabs the current value of the control in stage 1 and restores it 78 : in stage 2. 79 : 80 : 2. val set' : 'a control * 'a -> unit -> unit 81 : 82 : stores the given value into the control in stage 2 (i.e., delayed) 83 : but does all error checking in stage 1. 84 : (This is for string controls that need to do parse their argument -- 85 : something that might fail. In some cases, notably in CM, one 86 : already knows the intended argument but wants to delay the actual 87 : assignment until a time when error recovery would be more difficult.) 88 : 89 : Changed the handling of controls in tool arguments to classes "sml" and 90 : "lazysml": 91 : - use Controls.save'restore as a more robust way of restoring the 92 : old value (in particular: without having to re-parse the string) 93 : - use controls to handle the "overload" keyword in the init group 94 : (I believe this change actually fixes a long-standing obscure bug.) 95 : 96 : ---------------------------------------------------------------------- 97 : Name: Matthias Blume (blume (at) tti - c (dot) org) 98 : mblume 1632 Date: 2004/09/27 17:00:00 CDT 99 : Tag: blume-20040927-lazysml 100 : Description: 101 : 102 : Added a new tool class called "lazysml" to CM's tool chest. The only 103 : difference to "sml" is that compilation is done with Control.lazysml 104 : set to true. A source of class "lazysml" is automatically recognized 105 : by a file name suffix of ".lml". 106 : 107 : In addition to the above feature, the original class "sml" now also 108 : supports a tool argument "lazy" which has the same effect. As a 109 : result, the following three lines are equivalent: 110 : 111 : foo.sml : lazysml 112 : foo.sml : sml (lazy) 113 : foo.sml (lazy) 114 : 115 : The setting goes into effect both during parsing and during 116 : compilation. The original setting is restored right after parsing and 117 : after compilation, respectively. 118 : 119 : In addition to all the above, there is also a general mechanism to set 120 : ANY of the "controls" that are available at the command line via 121 : "-C..." on a per-sml-file basis. The same rules that apply for "lazy" 122 : apply as well. (In fact, "lazy" is implemented as a special case of 123 : the general mechanism.) 124 : 125 : The .cm file syntax uses a new keyword tool argument called "with". 126 : There are several ways of indicating the desired settings: 127 : 128 : foo.sml (with:parser.quotations=true) 129 : foo.sml (with:(name:parser.quotations value:true)) 130 : foo.sml (with:(name:name1 value:value1 name:name2 value:value2 ...)) 131 : foo.sml (with:(name1=value1 name2=value2 ...)) 132 : foo.sml (with:(name1=value1 name:name2 value:value2 name3=value3 ...)) 133 : 134 : etc. 135 : 136 : Another possible abbreviation is to leave out the =v or value:v part 137 : if the name refers to a boolean control (in which case the value is 138 : taken to be true). Thus, one could get lazy sml also by saying: 139 : 140 : foo.sml (with:parser.lazy-keyword=true) 141 : foo.sml (with:parser.lazy-keyword) 142 : foo.sml (with:(name:parser.lazy-keyword value:true)) 143 : foo.sml (with:(name:parser.lazy-keyword)) 144 : 145 : ---------------------------------------------------------------------- 146 : Name: Matthias Blume (blume (at) tti - c (dot) org) 147 : mblume 1631 Date: 2004/09/24 16:22:00 CDT 148 : Tag: blume-20040924-ppc-long-branch 149 : Description: 150 : 151 : Turned message about "emiting long form of branch" off by default. 152 : Added a control flag to turn it back on when desired. 153 : 154 : ---------------------------------------------------------------------- 155 : Name: Matthias Blume (blume (at) tti - c (dot) org) 156 : mblume 1630 Date: 2004/09/24 16:05:00 CDT 157 : Tag: blume-20040924-rounding 158 : Description: 159 : 160 : Applied patch for setting rounding modes under Mac OS X. Thanks to 161 : Melissa O'Neill for providing the code! 162 : 163 : ---------------------------------------------------------------------- 164 : Name: Matthias Blume (blume (at) tti - c (dot) org) 165 : mblume 1629 Date: 2004/09/23 17:30:00 CDT 166 : Tag: blume-20040923-envvars 167 : Description: 168 : 169 : 1. Changed definition of type ControlRegistry.registry_tree to 170 : include control_info (i.e., the name of the controlling environment 171 : variable). 172 : 173 : 2. Added command-line flags -e and -E to print the names of environment 174 : variables that can be used to control internal settings. 175 : (This uses the new API mentioned in 1.) 176 : 177 : ---------------------------------------------------------------------- 178 : Name: Matthias Blume (blume (at) tti - c (dot) org) 179 : mblume 1622 Date: 2004/09/13 16:50:00 CDT 180 : Tag: Release_110_49 181 : Description: 182 : 183 : New working version (110.49). NEW BOOTFILES! 184 : 185 : ---------------------------------------------------------------------- 186 : mblume 1621 Name: Matthias Blume 187 : Date: 2004/09/13 16:20:00 CDT 188 : Tag: blume-20040913-config-mlrisc 189 : Description: 190 : 191 : Put target "mlrisc" back into the default list. 192 : (There is no harm in having it, and some users have expressed their 193 : wish to have "mlrisc" included by default.) 194 : 195 : ---------------------------------------------------------------------- 196 : jhr 1620 Name: John Reppy 197 : Date: 2004/09/13 198 : Tag: jhr-20040913-signals 199 : Description: 200 : 201 : Fixed the signal masking code to properly nest mask/unmask operations 202 : on a per-signal basis. 203 : 204 : ---------------------------------------------------------------------- 205 : mblume 1619 Name: Matthias Blume (blume (at) tti - c (dot) org) 206 : Date: 2004/09/08 13:20:00 CDT 207 : Tag: blume-20040908-heap-magic 208 : Description: 209 : 210 : Bumped the heap macig number to 0x09082004 to account for the changed 211 : layout of the ML frame under MacOS X. 212 : 213 : ---------------------------------------------------------------------- 214 : allenleung 1618 Name: Allen Leung (leunga (at) reservoir (dot) com) 215 : Date: 2004/09/03 11:26:00 EST 216 : Tag: leunga-20040903-cygwin-install 217 : Description: 218 : 219 : Added a patch to _arch-n-opsys to enable the Cygwin runtime. 220 : The Cygwin runtime is turned on by setting the environment 221 : variable SMLNJ_CYGWIN_RUNTIME to 1. 222 : 223 : ---------------------------------------------------------------------- 224 : mblume 1615 Name: Matthias Blume (blume (at) tti - c (dot) org) 225 : mblume 1617 Date: 2004/08/31 17:15:00 CDT 226 : Tag: blume-20040831-core 227 : Description: 228 : 229 : Added some exports to src/compiler/core.cm upon request by J. Joemann. 230 : 231 : ---------------------------------------------------------------------- 232 : Name: Matthias Blume (blume (at) tti - c (dot) org) 233 : mblume 1616 Date: 2004/08/30 17:55:00 CDT 234 : Tag: blume-20040830-installer 235 : Description: 236 : 237 : Upon request by Johannes Joemann: 238 : 239 : - improved ML code of installer to fall back to coping when renaming 240 : fails (i.e., when source and target are on different file systems); 241 : the code compiles but has yet to be tested in anger 242 : - removed mlrisc from list of default targets (config/targets) 243 : 244 : ---------------------------------------------------------------------- 245 : Name: Matthias Blume (blume (at) tti - c (dot) org) 246 : mblume 1615 Date: 2004/08/27 17:20:00 CDT 247 : Tag: blume-20040827-ptreql 248 : Description: 249 : 250 : Added ptreql primop to structure InlineT (upon request from Larry 251 : Paulson). 252 : 253 : ---------------------------------------------------------------------- 254 : allenleung 1613 Name: Allen Leung (leunga (at) reservoir (dot) com) 255 : allenleung 1614 Date: 2004/08/15 21:21:00 EST 256 : Tag: leunga-110_48-udgraph 257 : Description: 258 : 259 : Another bug fix from Carl Hauser: 260 : 261 : diff /net/niflab/smlnj48/src/MLRISC/graphs/udgraph.sml udgraph.sml 262 : > 48c48 263 : > < | rmv((e as (k,_))::es,L) = rmv(es,if k = i then es else 264 : > e::L) 265 : > --- 266 : > > | rmv((e as (k,_))::es,L) = rmv(es,if k = i then L else e::L) 267 : > Without this, any deletion of an edge in an undirected graph does severe 268 : > violence to the graph. 269 : 270 : ---------------------------------------------------------------------- 271 : Name: Allen Leung (leunga (at) reservoir (dot) com) 272 : allenleung 1613 Date: 2004/08/10 23:35:00 EST 273 : Tag: leunga-110_48-ppc 274 : Description: 275 : 276 : The IBM/MacOS syntax switch on PPC was incorrectly swapped. 277 : 278 : ---------------------------------------------------------------------- 279 : mblume 1602 Name: Matthias Blume (blume (at) tti - c (dot) org) 280 : mblume 1615 Date: 2004/08/10 12:00:00 CDT 281 : mblume 1602 Tag: Release_110_48 282 : Description: 283 : 284 : New working version (110.48). NEW BOOTFILES! 285 : 286 : ---------------------------------------------------------------------- 287 : allenleung 1599 Name: Allen Leung (leunga (at) reservoir (dot) com) 288 : Date: 2004/08/09 12:21:00 EST 289 : Tag: leunga-110_47-dijsktra 290 : Description: 291 : 292 : Bug fix from Carl Hauser: 293 : 294 : single_source_shortest_paths in dijkstra.sml was observed to get wrong 295 : answers (by comparing to single_source_shortest_paths in bellman-ford.sml). 296 : 297 : The problem is that following the expression A.update(dist,s,Num.zero) 298 : it is necessary to update the priority queue using Q.decreaseWeight(Q,s). 299 : 300 : ---------------------------------------------------------------------- 301 : mblume 1595 Name: Matthias Blume (blume (at) tti - c (dot) org) 302 : Date: 2004/08/06 18:10:00 CDT 303 : Tag: blume-20040806-cmdline 304 : Description: 305 : 306 : Fiddled with handling of command-line options: 307 : 308 : * sml now quits after processing the command line 309 : if -H, -S, -h, or -s appears as the last 310 : command-line argument 311 : * a new option -q terminates the session when encountered on 312 : the command line; subsequent arguments will be ignored 313 : * bug fixes: short (erroneous) arguments are no longer ignored 314 : completely 315 : 316 : ---------------------------------------------------------------------- 317 : allenleung 1593 Name: Allen Leung (leunga (at) reservoir (dot) com) 318 : Date: 2004/08/04 18:17:00 EST 319 : Tag: leunga-110_47-ppc-ibm-asm 320 : Description: 321 : 322 : - Added minimal IBM assembly syntax support for PowerPC. 323 : 324 : - Cygwin: manually changed the file cygwin.def. Some exported symbols have 325 : been altered in the runtime. We need an automatic way to keep the file 326 : in sync. 327 : 328 : ---------------------------------------------------------------------- 329 : mblume 1548 Name: Matthias Blume (blume (at) tti - c (dot) org) 330 : mblume 1581 Date: 2004/08/04 14:00:00 CDT 331 : Tag: Release_110_47 332 : Description: 333 : 334 : New working version (110.47). NEW BOOTFILES! 335 : 336 : ---------------------------------------------------------------------- 337 : Name: Matthias Blume (blume (at) tti - c (dot) org) 338 : mblume 1580 Date: 2004/08/03 14:25:00 CDT 339 : Tag: blume-20040803-callingconv 340 : Description: 341 : 342 : Added low-level support for choosing C calling conventions by 343 : twiddling the type of rawccall. (See 344 : src/compiler/Semant/types/cproto.sml for details.) 345 : 346 : ---------------------------------------------------------------------- 347 : Name: Matthias Blume (blume (at) tti - c (dot) org) 348 : mblume 1579 Date: 2004/08/02 15:55:00 CDT 349 : Tag: blume-20040802-backout 350 : Description: 351 : 352 : Backed out of change to win32-filesys.c. The earlier patch to 353 : get_file_time caused CM to produce files with the wrong time stamp. 354 : 355 : ---------------------------------------------------------------------- 356 : Name: Matthias Blume (blume (at) tti - c (dot) org) 357 : mblume 1573 Date: 2004/08/02 14:45:00 CDT 358 : Tag: blume-20040802-nlffi-win32 359 : Description: 360 : 361 : Added NLFFI support for Win32, adapted from a patch provided by David 362 : Hansel. This is currently completely untested. Also, the issue 363 : concerning stdcall vs. ccall is still unresolved. 364 : 365 : ---------------------------------------------------------------------- 366 : Name: Matthias Blume (blume (at) tti - c (dot) org) 367 : mblume 1572 Date: 2004/07/30 17:55:00 CDT 368 : Tag: blume-20040730-various 369 : Description: 370 : 371 : Gearing up towards 110.47... 372 : 373 : - various minor bugfixes to ml-nlffigen 374 : - a beginning of a manual for nlffi 375 : 376 : - eliminated 'export name=value' in config/install.sh as this does 377 : not work with certain versions of /bin/sh 378 : (Thanks to David King at Motorola for catching this.) 379 : 380 : - several bugfixes provided or suggested by David Hansel at Reactive Systems: 381 : - added a test for tm==NULL to gmtime.c and localtime.c 382 : - applied patch for incorrect GetFileTime under win32 383 : - toSeconds -> toMilliseconds in Win32/win32-process.sml 384 : 385 : ---------------------------------------------------------------------- 386 : Name: Matthias Blume (blume (at) tti - c (dot) org) 387 : mblume 1557 Date: 2004/07/21 18:20:00 CDT 388 : Tag: blume-20040721-nlffigen 389 : Description: 390 : 391 : - Fixed minor issue in ml-nlffigen: Now generate structure T_foo 392 : for a typedef to an incomplete type, but leave out the "typ" member. 393 : (This is just for consistency.) 394 : - Started to produce what is supposed to become better (i.e., comprehensive) 395 : documentation of what ml-nlffigen does and produces. 396 : 397 : ---------------------------------------------------------------------- 398 : Name: Matthias Blume (blume (at) tti - c (dot) org) 399 : mblume 1548 Date: 2004/07/14 16:25:00 CDT 400 : Tag: blume-20040714-union 401 : Description: 402 : 403 : Added C_UNION to c-calls/c-types.sml and updated the machinery 404 : (ml-nlffigen, cproto.sml) that conveys C function interface 405 : information to the code generator. 406 : 407 : However, the actual architecture-specific implementation of function 408 : arguments and results that are C unions is still not implemented. 409 : 410 : ---------------------------------------------------------------------- 411 : allenleung 1539 Name: Allen Leung (leunga (at) reservoir (dot) com) 412 : allenleung 1546 Date: 2004/07/14 14:38:00 EST 413 : Tag: leunga-110_46_1-ppc-lwzu 414 : Description: 415 : 416 : Added these instructions to the PowerPC architecture: 417 : LBZU(X), LHZU(X), LWZU(X), 418 : STWU(X), STFDU, STFSU 419 : 420 : etc... 421 : 422 : Note: I haven't added their instruction encoding into the description. 423 : 424 : ---------------------------------------------------------------------- 425 : Name: Allen Leung (leunga (at) reservoir (dot) com) 426 : allenleung 1539 Date: 2004/07/13 15:04:00 EST 427 : Tag: leunga-110_46_1-ppc-lwarx 428 : Description: 429 : 430 : Added the two instructions LWARX and STWCX to the PowerPC 431 : instruction set. 432 : 433 : A (untested) rewrite of loop-structure.sml. The old version 434 : is completely broken. 435 : 436 : ---------------------------------------------------------------------- 437 : mblume 1479 Name: Matthias Blume (blume (at) tti - c (dot) org) 438 : mblume 1537 Date: 2004/07/13 13:50:00 CDT 439 : Tag: blume-20040713-nlffi 440 : Description: 441 : 442 : - use paramAlloc to report c-calls with too many arguments 443 : (for PPC version where parameter area is pre-allocated) 444 : - added ccall_maxargspace to machspec (to implement the above) 445 : - made "make" commend in CM's "make" tool configurable 446 : - added option (default: on) for passing the name of the SML/NJ's "bin" 447 : directory to "make"; the call looks like this: 448 : 449 : make SMLNJ_BINDIR= 450 : 451 : This can be used by the Makefile to, e.g., pick the "right" version 452 : of ml-nlffigen. 453 : - minor code tweaks 454 : 455 : ---------------------------------------------------------------------- 456 : Name: Matthias Blume (blume (at) tti - c (dot) org) 457 : mblume 1532 Date: 2004/07/12 22:50:00 CDT 458 : Tag: blume-110_46_1-macosx-nlffi 459 : Description: 460 : 461 : NLFFI under Mac OS X now working (sort of). This is largely untested, 462 : though. 463 : 464 : Note: 1. You have to make a new, clean build of the runtime system. 465 : 2. There are new BOOTFILES, you have to use them! 466 : (Doing the bootstrap process yourself would be *very* painful! 467 : If you absolutely have to do it, build the system under 468 : a different architecture and then cross-compile.) 469 : 470 : Version bumped to 110.46.1 to account for runtime data format changes. 471 : 472 : ---------------------------------------------------------------------- 473 : Name: Matthias Blume (blume (at) tti - c (dot) org) 474 : mblume 1515 Date: 2004/06/18 14:30:00 CDT 475 : Tag: blume-20040618-unix 476 : Description: 477 : 478 : Changed the implementation of structure Unix so that the same stream 479 : is returned every time one of the {text,bin}{In,Out}streamOf functions 480 : is invoked on the same proc. This is not what the spec currently 481 : says -- although IMO it arguably should. (See discussion below.) 482 : 483 : ---------------------------------------------------------------------- 484 : Name: Matthias Blume (blume (at) tti - c (dot) org) 485 : mblume 1507 Date: 2004/06/17 18:15:00 CDT 486 : Tag: Release_110_46 487 : Description: 488 : 489 : New working version (110.46). NEW BOOTFILES! 490 : 491 : ---------------------------------------------------------------------- 492 : Name: Matthias Blume (blume (at) tti - c (dot) org) 493 : mblume 1506 Date: 2004/06/17 17:20:00 CDT 494 : Tag: blume-20040617-timer-unix 495 : Description: 496 : 497 : Changed the interface of structures Timer and Unix to match the most 498 : recent Basis spec. 499 : 500 : In the case of Unix there still seems to be an open/weird issue: 501 : 502 : The {text,bin}{In,Out}streamOf functions are supposed to create 503 : fresh streams whenever they are called -- as opposed to have them 504 : return the same stream every time. This design is supposed to 505 : prevent space leaks caused by proc values hanging on to streams. 506 : 507 : The reap function, on the other hand, is supposed to close the 508 : streams. This cannot be done without having a handle on the 509 : stream in proc after all... 510 : 511 : I took the liberty to implement the following stopgap solution: 512 : 513 : The proc value hangs on to the most recently created stream(s). 514 : Reap closes those. If either or both of the two streams hadn't 515 : been created at all yet, then reap will close the corresponding 516 : file descriptors directly. 517 : 518 : PS: I don't understand the original space leak argument anymore. If 519 : a proc hangs on to the imperative stream, then I/O operations on those 520 : will advance the state of the cached stream and avoid the space leak. 521 : 522 : ---------------------------------------------------------------------- 523 : Name: Matthias Blume (blume (at) tti - c (dot) org) 524 : mblume 1500 Date: 2004/05/28 16:45:00 CDT 525 : Tag: blume-20040528-basis 526 : Description: 527 : 528 : Added signature PACK_REAL and exported functor PrimIO. 529 : 530 : ---------------------------------------------------------------------- 531 : Name: Matthias Blume (blume (at) tti - c (dot) org) 532 : mblume 1498 Date: 2004/05/25 16:00:00 CDT 533 : Tag: blume-20040525-group-owner 534 : Description: 535 : 536 : CM now ignores (but still accepts) the "owner" information in group 537 : descriptions. The owner of a group is its next enclosing 538 : library. Each group must have a unique owner. (There is a virtual 539 : mblume 1505 "toplevel" library that owns groups which are not nested within a real 540 : mblume 1498 library.) Previously, each group had to explicitly declare its owner, 541 : and CM would check that such a declaration is correct. The new scheme 542 : is to have CM check that for each group there is precisely one owning 543 : library. 544 : 545 : The advantage of the new scheme is that the programmer no longer needs 546 : to maintain the somewhat annoying owner information. The downside is 547 : that CM cannot enforce the ownership rule across multiple runs of 548 : CM.make. Fortunately, enclosing the same group in two different 549 : libraries A and B which are not part of the same program does not 550 : cause real problems. 551 : 552 : ---------------------------------------------------------------------- 553 : Name: Matthias Blume (blume (at) tti - c (dot) org) 554 : mblume 1496 Date: 2004/05/20 16:00:00 CDT 555 : Tag: blume-20040520-win32 556 : Description: 557 : 558 : Made the win32 version work again. (Strangely, a misplaced comma had 559 : slipped into win32-process.c which prevented the runtime from being 560 : compiled correctly.) 561 : 562 : Also, included a minor addition to ml-build.bat analogous to what was 563 : done in blume-20040519-ml-build. 564 : 565 : ---------------------------------------------------------------------- 566 : Name: Matthias Blume (blume (at) tti - c (dot) org) 567 : mblume 1494 Date: 2004/05/19 22:10:00 CDT 568 : Tag: blume-20040519-ml-build 569 : Description: 570 : 571 : Arranged for ml-build to clean up after itself a little bit better. 572 : The script generates a temporary SML source file and compiles it using 573 : CM, so CM generates metadata (GUID, SKEL, objectfile) for it. It now 574 : gets rid of those at the end, so they don't accumulate under .cm. 575 : 576 : This required a minor change to install.sh because the name of the 577 : metadata directory (default: .cm) is actually configurable at 578 : installation time. 579 : 580 : ---------------------------------------------------------------------- 581 : Name: Matthias Blume (blume (at) tti - c (dot) org) 582 : mblume 1491 Date: 2004/05/18 15:50:00 CDT 583 : Tag: blume-20040518-mkreader 584 : Description: 585 : 586 : Added Posix.IO.mk{Bin,Text}{Reader,Writer} by lifting their respective 587 : implementations from internal modules PosixBinPrimIO and PosixTextPrimIO. 588 : 589 : ---------------------------------------------------------------------- 590 : Name: Matthias Blume (blume (at) tti - c (dot) org) 591 : mblume 1489 Date: 2004/05/11 14:35:00 CDT 592 : Tag: blume-20040511-win32sock 593 : Description: 594 : 595 : Added previously missing support for many socket-related functions 596 : under win32. Thanks to David Hansel 597 : for the voluminous patch! 598 : 599 : (I have not tested this patch under win32 yet.) 600 : 601 : mblume 1490 Here is David's e-mail: 602 : 603 : Hi, 604 : 605 : Attached to this email you find a diff against sml/nj 110.45 606 : that will enable socket support under Windows. 607 : 608 : To apply the patch (using unix or cygwin) 609 : 1) gunzip runtime.diff.gz 610 : 2) "cd" into "src/runtime" in the source tree of a fresh 611 : 110.45 installation. 612 : 3) patch -p 1 < [your/path/to]runtime.diff 613 : 614 : The code compiles fine but has NOT yet been extensively tested. 615 : I only ran a few tests for basic socket client functionality 616 : (which worked fine). Especially the functions that use ioctl 617 : are not tested at all and might not work (see below). 618 : 619 : I implemented this since we want to move to a newer version of sml/nj 620 : but need socket support in order to use it. This is the first time I 621 : even had a look at the sml/nj source, so please review my changes 622 : before making this part of the distribution! Here are a few issues 623 : that I think might be better for someone to solve who is more 624 : familiar with the sml/nj source (and socket programming): 625 : 626 : - getnetbyaddr.c and getnetbyname.c will raise a "not implemented" 627 : exception since I could not figure out what the windows equivalent 628 : of these functions is 629 : 630 : - In sockets-osdep.h there are a some #include statements that are 631 : only used in a few files that include sockets-osdep.h 632 : 633 : - In smlnj-sock-lib.c, function init_fn() calls WSAStartup() but 634 : does not process its return value since I don't know how to 635 : report an error upwards. 636 : 637 : - It would probably be good to have a call to WSACleanup() when 638 : the library is unloaded (if there is such a possibility). 639 : Otherwise I think Windows will take care of this automatically 640 : when the process finishes. 641 : 642 : - I used ioctlsocket() as a replacement for ioctl() but I have 643 : no idea if that is actually the proper replacement on Windows. 644 : 645 : - All these issues are marked in the code by "FIXME" comments. 646 : 647 : We use sml/nj extensively in our products and are quite happy 648 : with it. I hope this contribution will help you. 649 : 650 : Keep up the good work! 651 : 652 : David 653 : 654 : mblume 1489 ---------------------------------------------------------------------- 655 : Name: Matthias Blume (blume (at) tti - c (dot) org) 656 : mblume 1488 Date: 2004/05/11 14:20:00 CDT 657 : Tag: blume-20040511-installml 658 : Description: 659 : 660 : Fixed two bugs in installml script. (Thanks to Vesa A. Norrman for 661 : the patch.) 662 : 663 : ---------------------------------------------------------------------- 664 : Name: Matthias Blume (blume (at) tti - c (dot) org) 665 : mblume 1487 Date: 2004/05/11 14:05:00 CDT 666 : Tag: blume-20040511-nlffi-netbsd 667 : Description: 668 : 669 : Added support for nlffi under netbsd. (Thanks to Vesa A. Norrman for 670 : the patch.) 671 : 672 : ---------------------------------------------------------------------- 673 : Name: Matthias Blume (blume (at) tti - c (dot) org) 674 : mblume 1486 Date: 2004/05/11 12:05:00 CDT 675 : Tag: blume-20040511-exports 676 : Description: 677 : 678 : As per request by Adam Chlipala , extended various 679 : export lists in compiler-related .cm-files. 680 : 681 : ---------------------------------------------------------------------- 682 : Name: Matthias Blume (blume (at) tti - c (dot) org) 683 : mblume 1485 Date: 2004/05/11 11:35:00 CDT 684 : Tag: blume-20040511-allsource 685 : Description: 686 : 687 : The installer now honors the "src-smlnj" target again, although its meaning 688 : has changed from "all sources required for the compiler" to "all sources 689 : the installer knows about". In other words, if you enable "src-smlnj" 690 : in the "targets" file, then the installer will pull in sources for 691 : everything. (Notice that this refers to source code only. Compiled 692 : code is still only installed for modules that were requested explicitly 693 : or which are required for other modules that were requested explicitly.) 694 : 695 : ---------------------------------------------------------------------- 696 : Name: Matthias Blume (blume (at) tti - c (dot) org) 697 : mblume 1482 Date: 2004/04/23 17:40:00 CDT 698 : Tag: blume-20040423-ieee-scan 699 : Description: 700 : 701 : Fixed IEEEReal.scan (and .fromString) so that if there is an overflow 702 : in the exponent calculation we get INF or ZERO (depending on the mantissa 703 : and the sign of the exponent). 704 : 705 : ---------------------------------------------------------------------- 706 : Name: Matthias Blume (blume (at) tti - c (dot) org) 707 : mblume 1480 Date: 2004/04/23 10:40:00 CDT 708 : Tag: blume-20040423-ml-build 709 : Description: 710 : 711 : The ml-build script now terminates with a non-0 status when something 712 : goes wrong. 713 : 714 : ---------------------------------------------------------------------- 715 : Name: Matthias Blume (blume (at) tti - c (dot) org) 716 : mblume 1479 Date: 2004/04/22 16:35:00 CDT 717 : Tag: blume-20040422-Option 718 : Description: 719 : 720 : Made exception Option to be the same as exception Option.Option 721 : (as it should be). 722 : 723 : ---------------------------------------------------------------------- 724 : allenleung 1468 Name: Allen Leung (leunga (at) reservoir (dot) com) 725 : Date: 2004/03/19 14:40:00 EST 726 : Tag: leunga-20040319-cygwin-nlffi 727 : Description: 728 : 729 : Fixed the runtime so that ml-nlffi-lib runs on the cygwin version 730 : of SML/NJ. The problem is that 731 : 732 : lib = dlopen(NULL, ...) 733 : f = dlsym(lib, "malloc"); 734 : 735 : does not work on Windows unless we explicitly export symbols 736 : such as 'malloc' during linking. We fixed this by explicitly 737 : exporting the required symbols with the magic gcc incantation: 738 : 739 : -Wl,--export-all cygwin.def 740 : 741 : where cygwin.def is a file containing all the symbols that 742 : we wish to export. 743 : 744 : I suspect this is a Windows problem and we'll have to 745 : do the same (somehow with windows compilers) when 746 : we build the native win32 version with the system 747 : calls LoadLibrary/GetProcAddress. 748 : 749 : ---------------------------------------------------------------------- 750 : mblume 1384 Name: Matthias Blume (blume (at) tti - c (dot) org) 751 : mblume 1465 Date: 2004/03/04 16:35:00 CST 752 : Tag: blume-20040304-intinf-fmt 753 : Description: 754 : 755 : Fixed problem with IntInf.fmt (sign would show up on the right instead 756 : of on the left for BIN, OCT, and HEX). 757 : 758 : ---------------------------------------------------------------------- 759 : Name: Matthias Blume (blume (at) tti - c (dot) org) 760 : mblume 1464 Date: 2004/03/04 11:25:00 CST 761 : Tag: blume-20040304-symlinks 762 : Description: 763 : 764 : Fixed problem with installer script (unix only) where bin/ml-yacc and 765 : friends pointed (via symlinks) to absolute locations instead of just 766 : .run-sml. This was reported by Vesa A Norrman. 767 : 768 : ---------------------------------------------------------------------- 769 : Name: Matthias Blume (blume (at) tti - c (dot) org) 770 : mblume 1455 Date: 2004/02/13 14:50:00 CST 771 : Tag: Release_110_45 772 : Description: 773 : 774 : New working version (110.45). New bootfiles. 775 : 776 : ---------------------------------------------------------------------- 777 : Name: Matthias Blume (blume (at) tti - c (dot) org) 778 : mblume 1448 Date: 2004/01/26 15:15:15 CST 779 : Tag: blume-20040126-toplevel 780 : Description: 781 : 782 : Improved handling of exceptions at the interactive toplevel. 783 : 784 : ---------------------------------------------------------------------- 785 : Name: Matthias Blume (blume (at) tti - c (dot) org) 786 : Date: 2004/01/26 11:25:00 787 : mblume 1446 Tag: blume-20040126-app 788 : Description: 789 : 790 : Type of top-level "app" corrected. 791 : mblume 1447 Added code for setting vp_limitPtrMask to Win32-specific runtime. 792 : mblume 1446 793 : ---------------------------------------------------------------------- 794 : Name: Matthias Blume (blume (at) tti - c (dot) org) 795 : mblume 1427 Date: 2003/11/18 17:10 CST 796 : Tag: blume-20031118-basis-fiddle 797 : Description: 798 : 799 : - changed Timer interface to what might become the spec 800 : - POSIX_FLAGS -> BIT_FLAGS according to spec 801 : - some other minor discrepancies wrt. spec eliminated 802 : 803 : ---------------------------------------------------------------------- 804 : Name: Matthias Blume (blume (at) tti - c (dot) org) 805 : mblume 1414 Date: 2003/11/06 12:00:00 CST 806 : Tag: Release_110_44 807 : Description: 808 : 809 : New working version (110.44). New bootfiles. 810 : 811 : ---------------------------------------------------------------------- 812 : Name: Matthias Blume (blume (at) tti - c (dot) org) 813 : mblume 1411 Date: 2003/11/04 11:50:00 CST 814 : Tag: blume-20031104-move-libraries 815 : Description: 816 : 817 : Eliminated the "dont_move_libraries" directive in config/targets. 818 : (The mechanism was broken and could not be fixed easily. Moreover, 819 : there does not seem to be any reason not to move all libraries into 820 : lib during installation. I originally implemented this directive as a 821 : backward-compatibility feature when I first introduced the new CM. 822 : Now that things have been stable for a long time and going back to the 823 : old CM is not an option, there is no reason to keep it around.) 824 : 825 : ---------------------------------------------------------------------- 826 : Name: Matthias Blume (blume (at) tti - c (dot) org) 827 : mblume 1409 Date: 2003/11/03 16:00:00 CST 828 : Tag: blume-20031103-installdir 829 : Description: 830 : 831 : Made installer honor INSTALLDIR variable again. (Thanks to Chris 832 : Richards for pointing out the problem and providing the solution.) 833 : 834 : ---------------------------------------------------------------------- 835 : Name: Matthias Blume (blume (at) tti - c (dot) org) 836 : mblume 1400 Date: 2003/10/01 17:05:00 CDT 837 : Tag: blume-20031001-lal-mlrisc 838 : Description: 839 : 840 : MLRISC bug fix from Lal. 841 : 842 : ---------------------------------------------------------------------- 843 : Name: Matthias Blume (blume (at) tti - c (dot) org) 844 : mblume 1399 Date: 2003/09/30 16:10:00 CDT 845 : Tag: blume-20030930-primio-bat 846 : mblume 1398 Description: 847 : 848 : 1. Added openVector, nullRd, and nullWr to PRIM_IO. 849 : 2. Improved .bat files (for Win32 port) to make things work under Win95. 850 : (thanks to Aaron S. Hawley for this one) 851 : 852 : ---------------------------------------------------------------------- 853 : Name: Matthias Blume (blume (at) tti - c (dot) org) 854 : mblume 1395 Date: 2003/09/26 16:05:00 CDT 855 : Tag: blume-20030926-wrappriv 856 : Description: 857 : 858 : Added missing wrapper for privilege "primitive" in $smlnj/viscomp/core.cm. 859 : 860 : ---------------------------------------------------------------------- 861 : Name: Matthias Blume (blume (at) tti - c (dot) org) 862 : mblume 1394 Date: 2003/09/26 15:00:00 CDT 863 : Tag: blume-20030926-110_43_3 864 : Description: 865 : 866 : - additional cleanup 867 : - version number bump, NEW BOOTFILES 868 : 869 : ---------------------------------------------------------------------- 870 : Name: Matthias Blume (blume (at) tti - c (dot) org) 871 : mblume 1393 Date: 2003/09/26 12:00:00 CDT 872 : Tag: blume-20030926-ppautoload 873 : Description: 874 : 875 : I modified the read-eval-print loop so that the autoloader gets 876 : invoked whenever the prettyprinter tries to look up a symbol that 877 : is not currently defined in the toplevel environment but which 878 : appears in CM's autoload registry. As a result, we see far fewer of 879 : those ?.Foo.Bar.xxx names in the prettyprinter's output. 880 : 881 : In addition to this I tried to clean up some pieces of the Basis 882 : implementation (e.g., Socket, Word8Array) in order to prevent other 883 : instances of these ?.Foo.Bar.xxx names from being printed. 884 : 885 : The mechanism that picks names for types still needs some work, though. 886 : (Right now it seems that if there is a type A.t which is defined to 887 : mblume 1394 be B.u, but B is unavailable at toplevel, then A.t gets printed as 888 : "?.B.u" although the perhaps more sensible solution would be to use 889 : mblume 1393 "A.t" in this case. In other words, the prettyprinter should follow 890 : a chain of DEFtycs not farther than there are corresponding toplevel 891 : names in the current environment.) 892 : 893 : ---------------------------------------------------------------------- 894 : Name: Matthias Blume (blume (at) tti - c (dot) org) 895 : mblume 1391 Date: 2003/09/24 16:31:00 CDT 896 : Tag: blume-20030924-installer 897 : Description: 898 : 899 : Another installer tweak: All the ML code for the installer is now 900 : compiled during CMB.make and put into a little library called 901 :$smlnj/installer.cm. The installation then simply invokes 902 : 903 : sml -m $smlnj/installer.cm 904 : 905 : and everything happens automagically. 906 : 907 : Win32: ML code senses value of environment variable SMLNJ_HOME. 908 : Unix: ML code senses values of environment variables ROOT, CONFIGDIR, 909 : and BINDIR. 910 : 911 : The new scheme guarantees that the ML code responsible for the installation 912 : is in sync with the APIs of the main system. Also, the installer is 913 : somewhat faster because the installer script is precompiled. 914 : 915 : ---------------------------------------------------------------------- 916 : Name: Matthias Blume (blume (at) tti - c (dot) org) 917 : mblume 1390 Date: 2003/09/24 15:35:00 CDT 918 : Tag: blume-20030924-synsock 919 : Description: 920 : 921 : Added a signature SYNCHRONOUS_SOCKET to basis.cm. This is like SOCKET 922 : but excludes all non-blocking operations. Defined SOCKET (in Basis) 923 : and CML_SOCKET in terms of SYNCHRONOUS_SOCKET. Removed superfluous 924 : implementations of non-blocking operations from CML's Socket 925 : structure. 926 : 927 : ---------------------------------------------------------------------- 928 : Name: Matthias Blume (blume (at) tti - c (dot) org) 929 : mblume 1389 Date: 2003/09/24 15:10:05 CDT 930 : Tag: blume-20030924-sockets 931 : Description: 932 : 933 : 1. Fixed SOCKET API and implementation to match Basis spec. 934 : This required changing the internal representation of sockets to one 935 : that remembers (for each socket file descriptor) whether it is currently 936 : blocking or non-blocking. This state is maintained lazily (i.e., a system 937 : call is made only if the state actually needs to change). 938 : 939 : 2. OS-specific details of sockets were moved into separate files, thus 940 : making it possible to unify the bulk of the socket implementations 941 : between Unix and Win32. 942 : 943 : 3. CML's socket API changed accordingly. 944 : (Note that we need to remove non-blocking functions from this API 945 : since they are redundant in the case of CML!) 946 : 947 : 4. CML's socket implementation now makes use of non-blocking functions 948 : provided by Basis, thus removing all OS-dependent code from this part 949 : of CML. 950 : 951 : 5. Changed Real64.precision from 52 to 53. Minor cleanup in Real64 code. 952 : 953 : ---------------------------------------------------------------------- 954 : Name: Matthias Blume (blume (at) tti - c (dot) org) 955 : mblume 1387 Date: 2003/09/22 12:10:00 CDT 956 : mblume 1388 Tag: blume-20030922-110_43_2 957 : mblume 1387 Description: 958 : 959 : Made a new interim version and bootfiles for developer's bootstrapping 960 : convenience. 961 : 962 : 110.43.2 -- NEW BOOTFILES 963 : 964 : ---------------------------------------------------------------------- 965 : Name: Matthias Blume (blume (at) tti - c (dot) org) 966 : mblume 1385 Date: 2003/09/19 15:55:00 CDT 967 : Tag: blume-20030919-cmdir 968 : Description: 969 : 970 : 1. new-install.sh -> install.sh 971 : 2. changed default CM "metadata" directory name to ".cm" (instead of "CM") 972 : 3. tweaked installer so that another name instead of .cm can be chosen 973 : at install time (by setting the CM_DIR_ARC environment variable 974 : during installation); once installation is complete, the name is 975 : fixed 976 : 977 : ---------------------------------------------------------------------- 978 : Name: Matthias Blume (blume (at) tti - c (dot) org) 979 : mblume 1384 Date: 2003/09/18 16:00:00 CDT 980 : Tag: blume-20030918-110_43_1 981 : Description: 982 : 983 : Made a new interim version and bootfiles for developer's bootstrapping 984 : convenience. 985 : 986 : 110.43.1 -- NEW BOOTFILES 987 : 988 : ---------------------------------------------------------------------- 989 : Name: Matthias Blume (blume (at) tti - c (dot) org) 990 : mblume 1383 Date: 2003/09/18 15:20:00 CDT 991 : Tag: blume-20030918-misc 992 : Description: 993 : 994 : 1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as 995 : this seems to be controversial at the moment) and used those in 996 : Posix.ProcEnv.times. 997 : 998 : 2. Added Time.{from,to}Nanoseconds to Time. 999 : 1000 : 3. Improved Real.{from,to}LargeInt by avoiding needless calculations. 1001 : For example, fromLargeInt never needs to look at more than 3 "big 1002 : digits" to get its 53 bits of precision. 1003 : 1004 : ---------------------------------------------------------------------- 1005 : mblume 1384 Name: Matthias Blume (blume (at) tti - c (dot) org) 1006 : mblume 1381 Date: 2003/09/17 16:30:00 CDT 1007 : Tag: blume-20030917-real32-slices 1008 : Description: 1009 : 1010 : Added an entry to the primitive environment 1011 : (compiler/Semant/statenv/prim.sml) for int32->real64 conversion and 1012 : added code to compiler/CodeGen/main/mlriscGen.sml to implement it. 1013 : 1014 : Removed some of the "magic" constants in real64.sml and replaced them 1015 : with code that generates these values from their corresponding 1016 : integer counterparts. 1017 : 1018 : Made all(?) the slice-related changes to the Basis and made everything 1019 : compile again... 1020 : 1021 : ---------------------------------------------------------------------- 1022 : mblume 1368 Name: Matthias Blume (blume (at) tti - c (dot) org) 1023 : mblume 1375 Date: 2003/09/15 17:45:00 CDT 1024 : Tag: blume-20030915-rbase 1025 : Description: 1026 : 1027 : Fixed bug in Real.fromLargeInt. 1028 : 1029 : ---------------------------------------------------------------------- 1030 : Name: Matthias Blume (blume (at) tti - c (dot) org) 1031 : mblume 1369 Date: 2003/09/13 18:11:00 CDT 1032 : Tag: blume-20030913-libinstall 1033 : Description: 1034 : 1035 : Minor bugfix in config/libinstall (set anchor with path to 1036 : standalone tool after installing it, otherwise libraries that 1037 : need ml-lex or ml-yacc won't compile the first time the installer 1038 : runs). 1039 : 1040 : ---------------------------------------------------------------------- 1041 : Name: Matthias Blume (blume (at) tti - c (dot) org) 1042 : mblume 1368 Date: 2003/09/12 11:45:00 CDT 1043 : Tag: blume-20030912-various 1044 : Description: 1045 : 1046 : - fixed bug in Real.toLargeInt 1047 : - fixed bug in Posix.ProcEnv.times 1048 : - changed inputLine functions to return an option 1049 : - minor installer improvements / bugfixes 1050 : - changed default @SMLalloc parameter for x86/celeron to 64k 1051 : 1052 : ---------------------------------------------------------------------- 1053 : Name: Matthias Blume (blume (at) tti - c (dot) org) 1054 : mblume 1356 Date: 2003/09/09 22:00:00 CDT 1055 : Tag: Release_110_43 1056 : Description: 1057 : 1058 : New working release 110.43. New bootfiles. 1059 : 1060 : ---------------------------------------------------------------------- 1061 : mblume 1368 Name: Matthias Blume (blume (at) tti - c (dot) org) 1062 : mblume 1354 Date: 2003/09/09 19:20:00 CDT 1063 : Tag: blume-20030909-installer 1064 : Description: 1065 : 1066 : Rewrote large parts of config/install.sh in SML (config/libinstall.sml). 1067 : Modified config/install.bat to take advantage of it. Also modified 1068 : config/install.sh (and called it config/new-install.sh) to take advantage 1069 : of it on Unix systems. (The SML code is (supposed to be) platform- 1070 : independent.) 1071 : 1072 : The installer can now install everything under Win32 1073 : as well as under *nix as long as it compiles. 1074 : 1075 : Other changes: 1076 : 1077 : - made CML compile again under Win32 1078 : - made eXene compile under Win32 (by providing a fake structure UnixSock 1079 : and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv) 1080 : - fixed a bug in nowhere: it assumed that type OS.Process.status is the 1081 : same as type int; under Win32 it isn't 1082 : - fixed some slice-related problems in the win32-specific parts of CML 1083 : - added a functor argument "sameVol" to os-path-fn.sml in the Basis 1084 : (under Win32, the volume name is case-insensitive, and the 1085 : OS.Path code compares volume names for equality) 1086 : 1087 : ---------------------------------------------------------------------- 1088 : mblume 1368 Name: Matthias Blume (blume (at) tti - c (dot) org) 1089 : mblume 1353 Date: 2003/09/08 11:55:00 CDT 1090 : Tag: blume-20030908-fullpath 1091 : Description: 1092 : 1093 : Made Win32 version of OS.FileSys.fullPath return current directory 1094 : when given an empty string. This is what the spec says, and incidentally, 1095 : CM depends on it. (CM otherwise goes into an infinite loop in certain 1096 : cases when presented with the name of a non-existing .cm file.) 1097 : 1098 : ---------------------------------------------------------------------- 1099 : mblume 1368 Name: Matthias Blume (blume (at) tti - c (dot) org) 1100 : mblume 1350 Date: 2003/09/04 16:30:00 CDT 1101 : mblume 1352 Tag: blume-20030905-slices-etc 1102 : mblume 1350 Description: 1103 : 1104 : 1. Changed interface to vectors and arrays in Basis to match 1105 : (draft) Basis spec. 1106 : 2. Added signatures and implementations of slices according to 1107 : Basis spec. 1108 : 3. Edited source code throughout the system to make it compile again 1109 : under 1. and 2. (In some cases code had to be added to have it 1110 : match the new signatures.) 1111 : 4. MLRISC should be backward-compatible: the copies of the originals 1112 : of files that needed to change under 3. were retained, the .cm files 1113 : check the compiler version number and use old versions when 1114 : appropriate. 1115 : 5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to 1116 : dirstream -> string option (in accordance with Basis spec). 1117 : 6. When generating code that counts lines, ml-lex used function 1118 : CharVector.foldli, taking advantage of its old interface. 1119 : This has been replaced with the corresponding code from 1120 : CharVectorSlice. (html-lex must be re-lexed!) 1121 : 7. BitArray in smlnj-lib/Util has been extended/modified to match the 1122 : new MONO_ARRAY signature. (Do we need BitArraySlice?) 1123 : 8. Removed temporary additions (fromInternal, toInternal) from the 1124 : (now obsolete) IntInf in smlnj-lib/Util. 1125 : 9. Cleaned up structure Byte. 1126 : 10. Added localOffset, scan, and fromString to Date (according to spec). 1127 : Cleaned/corrected implementation of Date. 1128 : (Still need to check for correctness; implement better canonicalizeDate.) 1129 : 11. Added "scan" to signature IEEE_REAL. 1130 : 12. Some improvements to IntInf [in particular: efficiency-hack for 1131 : mod and rem when second operand is 2 (for parity checks).] 1132 : 13. Changed representation of type Time.time, using a single IntInf.int 1133 : value counting microseconds. This considerably simplified the 1134 : implementation of structure Time. We now support negative time 1135 : values; scan and fromString handle signs. 1136 : 14. Functor PrimIO now takes two additional arguments (VectorSlice and 1137 : ArraySlice). 1138 : 1139 : ---------------------------------------------------------------------- 1140 : mblume 1368 Name: Matthias Blume (blume (at) tti - c (dot) org) 1141 : mblume 1347 Date: 2003/08/28 17:00:00 CDT 1142 : Tag: blume-20030828-intinf 1143 : Description: 1144 : 1145 : This is a major update which comes with a version number bump 1146 : (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES, 1147 : and an implementation of IntInf in the Basis. 1148 : 1149 : There are a fairly large number of related changes and updates throughout 1150 : the system: 1151 : 1152 : Basis: 1153 : - Implemented IntInf. 1154 : - Made LargeInt a projection of IntInf (by filtering through INTEGER). 1155 : mblume 1348 - Added some missing Real64 operations, most notably Real.toLargeInt. 1156 : mblume 1347 - Added FixedInt as a synonym for Int32. 1157 : 1158 : compiler: 1159 : * Added support for a built-in intinf type. 1160 : - literals 1161 : - pattern matching 1162 : - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.) 1163 : - overloading on literals and operations 1164 : 1165 : This required adding a primitive type intinf, some additional 1166 : primops, and implementations for several non-trivial intinf 1167 : operations in Core. (The intinf type is completely abstract 1168 : to the compiler; all operations get delegated back to the Core.) 1169 : 1170 : * Intinf equality is handled by polyequal. However, the compiler 1171 : does not print its usual warning in this case (since polyequal 1172 : is the right thing to do there). 1173 : 1174 : * Improved the organization of structure InlineT. 1175 : 1176 : * A word about conversion primops: 1177 : If conversions involving intinf do not cancel out during 1178 : mblume 1348 CPS contract, then the compiler must insert calls to Core functions. 1179 : mblume 1347 Since all core access must be resolved already during the FLINT 1180 : translate phase, it would be too late a the time of CPS contract 1181 : mblume 1348 to add new Core calls. For this reason, conversion primops 1182 : mblume 1347 for intinf carry two arguments: 1. the numeric argument that 1183 : they are supposed to convert, and 2. the Core function that 1184 : can help with this conversion if necessary. If CPS contract 1185 : mblume 1348 eliminates a primop, then the associated Core function becomes 1186 : dead and goes away. Intinf conversion primops that do not get 1187 : eliminated by CPS contract get rewritten into calls of their 1188 : core functions by a separate, new phase. 1189 : mblume 1347 1190 : interactive system: 1191 : - Control.Print.intinfDepth controls max length of intinf constants 1192 : being printed. (Analogous to Control.Print.stringDepth.) 1193 : - Cleanup in printutil and pputil: got rid of unused stuff and 1194 : duplicates; replaced some of the code with code that makes better 1195 : use of library functionality. 1196 : 1197 : CM: 1198 : Bugfix: parse-errors in init group (system/smlnj/init/init.cmi) 1199 : are no longer silent. 1200 : 1201 : CKIT: 1202 : Fixed mismatched uses of Int32 and LargeInt. I always decided 1203 : in favor of LargeInt -- which is now the same as IntInf. 1204 : CKIT-knowledgable people should check whether this is what's 1205 : intended and otherwise change things back to using Int32 or 1206 : FixedInt. 1207 : 1208 : Throughout the code: 1209 : mblume 1348 Started using IntInf.int literals and built-in operations 1210 : mblume 1347 (e.g., comparison with 0) where this seems appropriate. 1211 : 1212 : 1213 : ---------------------------------------------------------------------- 1214 : macqueen 1343 Name: Dave MacQueen (dbm@cs.uchicago.edu) 1215 : Date: 2003/08/13 11:36:00 CDT 1216 : Tag: dbm-20030813-mcz-merge1 1217 : Description: 1218 : 1219 : Merging changes from the mcz-branch development branch into trunk. 1220 : These changes involve replacement of the emulated old prettyprinter 1221 : interface with direct use of the SML/NJ Lib PP library, and fixing 1222 : of a couple of bugs (895, 1186) relating to error messages. A new 1223 : prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml}) 1224 : has been added. 1225 : 1226 : ---------------------------------------------------------------------- 1227 : mblume 1368 Name: Matthias Blume (blume (at) tti - c (dot) org) 1228 : mblume 1342 Date: 2003/08/11 15:45:00 CDT 1229 : Tag: blume-20030811-windows 1230 : Description: 1231 : 1232 : Version number bumped to 110.42.9. NEW BOOTFILES!!! 1233 : 1234 : http://smlnj.cs.uchicago.edu/dist/working/110.42.9/ 1235 : 1236 : This patch restores SML/NJ's ability to run under win32. There are a 1237 : number of changes, including fixes for several bugs that had gone 1238 : unnoticed until now: 1239 : 1240 : - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when 1241 : trying to run the win32 version from within cygwin.) 1242 : 1243 : - There are a number of simple .bat scripts that substitute for their 1244 : corresponding Unix shell-scripts. (See below.) 1245 : 1246 : - The internals of ml-build have been modified slightly. The main 1247 : difference is that instead of calling ".link-sml" (or link-sml.bat) 1248 : using OS.Process.system, the ML process delegates this task back 1249 : to the script. Otherwise problems arise in mixed environments such 1250 : as Cygwin where scripts look and work like Unix scripts, but 1251 : where OS.Process.system cannot run them. 1252 : 1253 : - In CM, the srcpath pickler used native pathname syntax -- which 1254 : is incorrect in the case of cross-compilation. The new pickle format 1255 : is independent of platform-specific naming conventions. 1256 : 1257 : - Path configuration files (such as lib/pathconfig) can now choose 1258 : between native and standard syntax. Placing a line of the form 1259 : 1260 : standard! 1261 : 1262 : into the file causes all subsequent paths to be interpreted using 1263 : CM standard pathname syntax (= Unix conventions); a line 1264 : 1265 : native! 1266 : 1267 : switches back to native style. This was needed so that 1268 : path config files can be written portably, see src/system/pathconfig. 1269 : 1270 : - Runtime system: 1271 : 1272 : - win32-filesys.c: get_file_time and set_file_time now 1273 : access modification time, not creation time. 1274 : 1275 : - I/O code made aware of new array representation. 1276 : 1277 : - Bug fixes in X86.prim.masm. 1278 : 1279 : - src/system/makeml made aware of win32. (For use under cygwin 1280 : and other Unix-environments for windows.) 1281 : 1282 : - In Basis, fixed off-by-one error in win32-io.sml (function vecF) 1283 : which caused BinIO.inputAll to fail consistently. 1284 : 1285 : .bat scripts: 1286 : 1287 : Windows .bat scripts assume that SMLNJ_HOME is defined. 1288 : 1289 : - sml.bat, ml-yacc.bat, ml-lex.bat: Driver scripts for standalone 1290 : applications (sml, ml-yacc, ml-lex). 1291 : - ml-build.bat: analogous to ml-build. 1292 : - config\install.bat: Analogous to config/install.sh. This requires 1293 : that SMLNJ_HOME is set and that Microsoft Visual C is ready to use. 1294 : (nmake etc. must be on the path, and vcvars32 must have been run.) 1295 : Moreover, sources for ml-lex and ml-yacc need to exist under src, 1296 : and the bootfile hierarchy must have been unpacked under 1297 : sml.boot.x86-win32. 1298 : The script is very primitive and does a poor job at error checking. 1299 : It only installs the base system, ml-lex, and ml-yacc. No other 1300 : libraries are being installed (i.e., you get only those that 1301 : are part of the compiler.) 1302 : - link-sml.bat: analogous to .link-sml, but not currently used 1303 : 1304 : Unrelated bug fixes: 1305 : 1306 : - ml-nlffigen now exports structures ST_* corresponding to incomplete 1307 : types. 1308 : - Added getDevice to PP/src/pp-debug-fn.sml. (Would not compile 1309 : otherwise.) 1310 : 1311 : ---------------------------------------------------------------------- 1312 : macqueen 1337 Name: Dave MacQueen (macqueen@cs.uchicago.edu) 1313 : Date: 2003/06/17 1314 : Tag: macqueen-20030617-bug895 1315 : Description: 1316 : 1317 : Modified compiler/Elaborator/print/pptype.sml to fix bug 895. 1318 : Tag will be used for new development branch (mcz-branch) for 1319 : use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago. 1320 : 1321 : ---------------------------------------------------------------------- 1322 : mblume 1368 Name: Matthias Blume (blume (at) tti - c (dot) org) 1323 : mblume 1335 Date: 2003/05/27 16:55:00 CDT 1324 : Tag: blume-20030527-polyeq 1325 : Description: 1326 : 1327 : Tried to eliminated most cases of polymorphic equality. 1328 : 1329 : ---------------------------------------------------------------------- 1330 : mblume 1368 Name: Matthias Blume (blume (at) tti - c (dot) org) 1331 : mblume 1334 Date: 2003/05/21 17:45:00 CDT 1332 : Tag: blume-20030517-complete 1333 : Description: 1334 : 1335 : Two changes: 1336 : 1337 : 1. Added a flag for controlling whether non-exhaustive bindings will 1338 : be treated as errors (default is false). 1339 : 2. Cleaned up the *entire* source tree so that CMB.make goes through 1340 : without a single non-exhaustive match- or bind warning. 1341 : 1342 : ---------------------------------------------------------------------- 1343 : mblume 1368 Name: Matthias Blume (blume (at) tti - c (dot) org) 1344 : mblume 1332 Date: 2003/05/17 10:20:00 CDT 1345 : Tag: blume-20030517-absyn 1346 : Description: 1347 : 1348 : 1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn. 1349 : 1350 : This mainly affects the quality of error messages. However, some 1351 : of the code is now more straightforward than before. (Treatment of 1352 : the above four constructs in translate.sml is much simpler than 1353 : the "macro-expansion" that was going on before. Plus, the mach- 1354 : compiler no longer gets invoked just to be able to compile an 1355 : if-expression.) 1356 : 1357 : 2. The ErrorMsg.Error exception is now caught and absorbed by the 1358 : interactive loop. 1359 : 1360 : ---------------------------------------------------------------------- 1361 : allenleung 1327 Name: Allen Leung 1362 : Date: 2003/05/16 13:05:00 CDT 1363 : Tag: leunga-20030516-cygwin-runtime 1364 : Description: 1365 : 1366 : Ported the runtime system to cygwin, which uses the unix 1367 : x86-unix bin files. Missing/buggy features: 1368 : 1369 : o getnetbyname, getnetbyaddr: these functions seem to be missing in 1370 : the Cygwin library. 1371 : o Ctrl-C handling may be flaky. 1372 : o Windows system calls and Windows I/O are not supported. 1373 : 1374 : A new set of binfiles is located at: 1375 : 1376 : http://www.dorsai.org/~leunga/boot.x86-unix.tgz 1377 : 1378 : This is only needed for bootstrapping the cygwin version of smlnj. 1379 : Other x86 versions can use the existing binfiles. 1380 : 1381 : ---------------------------------------------------------------------- 1382 : mblume 1317 Name: Matthias Blume 1383 : Date: 2003/04/08 15:42:00 CDT 1384 : Tag: blume-20030408-listpair 1385 : Description: 1386 : 1387 : 1. Added a target 'mlrisc' to installer. 1388 : 1389 : 2. Added missing elements to structure ListPair. 1390 : 1391 : ---------------------------------------------------------------------- 1392 : leunga 1296 Name: Allen Leung 1393 : Date: 2003/01/07 10:40:00 EST 1394 : Tag: leunga-20030107-int-rem 1395 : Description: 1396 : 1397 : Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86. 1398 : The arguments to the SUBL instruction were swapped. 1399 : 1400 : ---------------------------------------------------------------------- 1401 : blume 1229 Name: Matthias Blume 1402 : blume 1294 Date: 2002/12/12 16:25:00 EST 1403 : Tag: blume-20021212-risc-ra 1404 : Description: 1405 : 1406 : Fixed a serious bug in the rewrite code for FP spilling/reloading that 1407 : sent the RA into an infinite loop when floating point registers get 1408 : spilled. (Because of this bug, e.g., nucleic stopped compiling between 1409 : 110.37 and 110.38.) 1410 : There was another set of potential problems related to the handling of 1411 : MLRISC annotations (but those did not yet cause real problems, apparently). 1412 : 1413 : ---------------------------------------------------------------------- 1414 : Name: Matthias Blume 1415 : blume 1291 Date: 2002/12/06 22:40:00 EST 1416 : Tag: blume-20021206-cm-fileid 1417 : Description: 1418 : 1419 : Added a call of SrcPath.sync at the beginning of Parse.parse (in CM). 1420 : This fixes the problem of CM getting confused by files that suddenly 1421 : change their identity (e.g., by getting unlinked and recreated by some 1422 : text editor such as vi). There might be a better/cheaper/cleaner way 1423 : of doing this, but for now this will have to do. 1424 : 1425 : ---------------------------------------------------------------------- 1426 : Name: Matthias Blume 1427 : blume 1288 Date: 2002/10/28 09:50:00 EST 1428 : Tag: blume-20021028-typecheck 1429 : Description: 1430 : 1431 : Exported structure Typecheck from$smlnj/viscomp/core.cm. 1432 : 1433 : ---------------------------------------------------------------------- 1434 : Name: Matthias Blume 1435 : blume 1282 Date: 2002/10/17 09:10:00 EDT 1436 : blume 1279 Tag: Release_110_42 1437 : Description: 1438 : 1439 : blume 1282 In good old tradition, there has been a slight hiccup so that we have 1440 : to patch 110.42 after the fact. The old release tag has been replaced 1441 : (see below). 1442 : 1443 : The change solves a problem with two competing approaches the 1444 : configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in 1445 : each other's way. 1446 : 1447 : This change only affects the runtime system code and the installer script. 1448 : (No new bootfiles.) 1449 : 1450 : ---------------------------------------------------------------------- 1451 : Name: Matthias Blume 1452 : Date: 2002/10/16 12:00:00 EDT 1453 : Tag: Release_110_42_removed 1454 : Description: 1455 : 1456 : blume 1279 New working release. New bootfiles. 1457 : 1458 : ---------------------------------------------------------------------- 1459 : Name: Matthias Blume 1460 : blume 1273 Date: 2002/10/10 13:10:00 EDT 1461 : Tag: blume-20021010-ppc-divs 1462 : Description: 1463 : 1464 : The mltree operator DIVS must be implemented with an overflow check on 1465 : the PPC because the hardware indicates divide-by-zero using "overflow" as 1466 : well. 1467 : 1468 : ---------------------------------------------------------------------- 1469 : Name: Matthias Blume 1470 : blume 1261 Date: 2002/07/23 11:45:00 1471 : Tag: blume-20020723-smlnj-home 1472 : Description: 1473 : 1474 : Sml now senses the SMLNJ_HOME environment variable. If this is set, 1475 : then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless 1476 : CM_PATHCONFIG is also set), the path configuration file is assumed 1477 : to be in$SMLNJ_HOME/lib/pathconfig. This way one can easily move 1478 : the entire tree to some other place and everything will "just work". 1479 : 1480 : (Companion commands such as ml-build and ml-makedepend also sense this 1481 : variable.) 1482 : 1483 : ---------------------------------------------------------------------- 1484 : Name: Matthias Blume 1485 : blume 1259 Date: 2002/07/12 21:19:00 EDT 1486 : Tag: blume-20020712-liveness 1487 : Description: 1488 : 1489 : Exported two useful "step" functions from liveness module (MLRISC). 1490 : 1491 : ---------------------------------------------------------------------- 1492 : Name: Matthias Blume 1493 : blume 1253 Date: 2002/07/05 16:00 EDT 1494 : Tag: Release_110_41 1495 : Description: 1496 : 1497 : New working release. New bootfiles. 1498 : 1499 : ---------------------------------------------------------------------- 1500 : Name: Matthias Blume 1501 : blume 1250 Date: 2002/07/05 10:25:00 EDT 1502 : Tag: blume-20020705-btimp 1503 : Description: 1504 : 1505 : Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that 1506 : other clients can set up backtracing support. 1507 : 1508 : ---------------------------------------------------------------------- 1509 : Name: Matthias Blume 1510 : blume 1249 Date: 2002/06/25 17:23:00 EDT 1511 : Tag: blume-20020625-fpmax 1512 : Description: 1513 : 1514 : Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point 1515 : case. (The sense of the isNaN test was reversed -- which made min and 1516 : max always return their first argument.) 1517 : 1518 : ---------------------------------------------------------------------- 1519 : Name: Matthias Blume 1520 : blume 1240 Date: 2002/06/11 1521 : Tag: blume-20020611-unixpath 1522 : Description: 1523 : 1524 : Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch. 1525 : 1526 : ---------------------------------------------------------------------- 1527 : Name: Matthias Blume 1528 : blume 1238 Date: 2002/06/10 16:35:00 EDT 1529 : Tag: blume-20020610-ieeereal 1530 : Description: 1531 : 1532 : I back-ported my implementation of IEEEReal.fromString from the 1533 : idlbasis-devel branch so that we can test it. 1534 : 1535 : Another small change is that ppDec tries to give more information 1536 : than just "" in the case of functors. However, this code is 1537 : broken in some mysterious way if the functor's body's signature 1538 : has not been declared by ascription but gets inferred from the 1539 : implementation. This needs fixing... 1540 : 1541 : ---------------------------------------------------------------------- 1542 : Name: Matthias Blume 1543 : blume 1229 Date: 2002/05/31 1544 : Tag: blume-20020531-btrace-mode 1545 : Description: 1546 : 1547 : Resurrected SMLofNJ.Internals.BTrace.mode. (It accidentally fell by 1548 : the wayside when I switched over to using Controls everywhere.) 1549 : 1550 : ---------------------------------------------------------------------- 1551 : george 1222 Name: Lal George 1552 : Date: 2002/05/23 12:21:40 EDT 1553 : Tag: george-20020523-visual-labels 1554 : Description: 1555 : 1556 : Labels are now displayed in the graphical output to make 1557 : the fall-through and target blocks obvious. 1558 : 1559 : ---------------------------------------------------------------------- 1560 : blume 1204 Name: Matthias Blume 1561 : blume 1220 Date: 2002/05/22 11:03:00 EDT 1562 : Tag: blume-20020522-shrink 1563 : Description: 1564 : 1565 : John tweaked yesterday's fix for 1131 to handle an out-of-memory 1566 : situation that comes up when allocating huge arrays. 1567 : 1568 : ---------------------------------------------------------------------- 1569 : Name: Matthias Blume 1570 : blume 1215 Date: 2002/05/21 16:00:00 EDT 1571 : Tag: Release_110_40 1572 : Description: 1573 : 1574 : New working release (110.40). New bootfiles. 1575 : 1576 : [Also: John Reppy fixed GC bug 1131.] 1577 : 1578 : ---------------------------------------------------------------------- 1579 : Name: Matthias Blume 1580 : blume 1212 Date: 2002/05/21 12:35:00 EDT 1581 : Tag: blume-20020521-cmdoc 1582 : Description: 1583 : 1584 : CM documentation update. 1585 : 1586 : ---------------------------------------------------------------------- 1587 : Name: Matthias Blume 1588 : blume 1211 Date: 2002/05/21 10:55:00 EDT 1589 : Tag: blume-20020521-misc 1590 : Description: 1591 : 1592 : - John tweaked runtime to be silent on heap export (except when 1593 : GC messages are on). 1594 : - I added a few more things (cross-compiling versions of CMB) to 1595 : config/preloads (as suggestions). 1596 : 1597 : ---------------------------------------------------------------------- 1598 : Name: Matthias Blume 1599 : blume 1208 Date: 2002/05/20 22:25:00 EDT 1600 : Tag: blume-20020520-controls 1601 : Description: 1602 : 1603 : - Added ControlUtil structure to control-lib.cm. 1604 : - Use it throughout. 1605 : - Used Controls facility to define MLRISC controls (as opposed to 1606 : registering MLRISC control ref cells with Controls after the fact) 1607 : - Fixed messed-up controls priorities. 1608 : 1609 : * Removed again all the stuff from config/preloads that one wouldn't 1610 : be able to preload at the time the initial heap image is built. 1611 : (Many libraries, e.g., CML, do not exist yet at this time. The 1612 : only libraries that can be preloaded via config/preloads are those 1613 : that come bundled with the bootfiles.) 1614 : 1615 : ---------------------------------------------------------------------- 1616 : Name: Matthias Blume 1617 : blume 1204 Date: 2002/05/20 10:59:00 EDT 1618 : Tag: blume-20020520-preloads 1619 : Description: 1620 : 1621 : Added a lot of commented-out suggestions for things to be included 1622 : in config/preloads. 1623 : 1624 : ---------------------------------------------------------------------- 1625 : leunga 1203 Name: Allen Leung 1626 : Date: 2002/05/18 14:20:00 EDT 1627 : Tag: leunga-20020518-mdl 1628 : Description: 1629 : 1630 : o Made the mdl tool stuff compile and run again. 1631 : o I've disabled all the stuff that depends on RTL specifications; they 1632 : are all badly broken anyway. 1633 : 1634 : ---------------------------------------------------------------------- 1635 : blume 1201 Name: Matthias Blume 1636 : Date: 2002/05/17 16:49:00 EDT 1637 : Tag: blume-20020517-controls 1638 : Description: 1639 : 1640 : 0. John Reppy made several modifications to the SML/NJ library. 1641 : In particular, there is a shiny new controls-lib.cm. 1642 : 1643 : 1. Pushed new controls interface through compiler so that everything 1644 : compiles again. 1645 : 1646 : 2. Added FormatComb and FORMAT_COMB to the CML version of the 1647 : SML/NJ library (so that CML compiles again). 1648 : 1649 : 3. Modified init scripts because XXX_DEFAULT environment variables 1650 : are no longer with us. (Boot-time initialization is now done 1651 : using the same environment variables that are also used for 1652 : startup-time initialization of controls.) 1653 : 1654 : ---------------------------------------------------------------------- 1655 : george 1192 Name: Lal George 1656 : Date: 2002/05/15 09:20:10 EDT 1657 : Tag: george-20020515-pseudo-op-decls 1658 : Description: 1659 : 1660 : All pseudo-ops emitted before the first segment declaration 1661 : such as TEXT, DATA, and BSS directives are assumed to be global 1662 : declarations and are emitted first in the assembly file. This is 1663 : useful in a number of situations where one has pseudo-ops that are not 1664 : specific to any segment, and also works around the constraint that one 1665 : cannot have client pseudo-ops in the TEXT segment. 1666 : 1667 : Because no segment is associated with these declarations it is 1668 : an error to allocate any space or objects before the first segment 1669 : directive and an exception will be raised. However, we cannot make 1670 : this check for client pseudo-ops. 1671 : 1672 : These top level declarations are a field in the CFG graph_info. 1673 : In theory you can continue to add to this field after the CFG has been 1674 : built -- provided you know what you are doing;-) 1675 : 1676 : ---------------------------------------------------------------------- 1677 : blume 1175 Name: Matthias Blume 1678 : blume 1190 Date: 2002/05/13 16:40:00 EDT 1679 : Tag: blume-20020513-pp-etc 1680 : Description: 1681 : 1682 : A few minor bugfixes: 1683 : 1684 : - Stopgap measure for bug recently reported by Elsa Gunter (ppDec). 1685 : (Bogus printouts for redefined bindings still occur. Compiler 1686 : bug should no longer occur now. We need to redo the prettyprinter 1687 : from scratch.) 1688 : 1689 : - CM pathname printer now also adds escape sequences for ( and ) 1690 : 1691 : - commend and docu fixes for ml-nlffi 1692 : 1693 : ---------------------------------------------------------------------- 1694 : Name: Matthias Blume 1695 : blume 1189 Date: 2002/05/10 16:40:00 EDT 1696 : Tag: blume-20020510-erg-textio 1697 : Description: 1698 : 1699 : Applied the following bugfix provided by Emden Gansner: 1700 : 1701 : Output is corrupted when outputSubstr is used rather than output. 1702 : 1703 : The problem occurs when a substring 1704 : 1705 : ss = (s, dataStart, dataLen) 1706 : 1707 : where dataStart > 0, fills a stream buffer with avail bytes left. 1708 : avail bytes of s, starting at index dataStart, are copied into the 1709 : buffer, the buffer is flushed, and then the remaining dataLen-avail 1710 : bytes of ss are copied into the beginning of the buffer. Instead of 1711 : starting this copy at index dataStart+avail in s, the current code 1712 : starts the copy at index avail. 1713 : 1714 : Fix: 1715 : In text-io-fn.sml, change line 695 from 1716 : val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0) 1717 : to 1718 : val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0) 1719 : 1720 : ---------------------------------------------------------------------- 1721 : Name: Matthias Blume 1722 : blume 1186 Date: 2002/04/12 13:55:00 EDT 1723 : Tag: blume-20020412-assyntax 1724 : Description: 1725 : 1726 : 1. Grabbed newer assyntax.h from the XFree86 project. 1727 : 2. Fiddled with how to compile X86.prim.asm without warnings. 1728 : 3. (Very) Minor cleanup in CM. 1729 : 1730 : ---------------------------------------------------------------------- 1731 : Name: Matthias Blume 1732 : blume 1185 Date: 2002/04/01 (no joke!) 17:07:00 EST 1733 : Tag: blume-20020401-x86div 1734 : Description: 1735 : 1736 : Added full support for div/mod/rem/quot on the x86, using the machine 1737 : instruction's two results (without clumsily recomputing the remainder) 1738 : directly where appropriate. 1739 : 1740 : Some more extensive power-of-two support was added to the x86 instruction 1741 : selector (avoiding expensive divs, mods, and muls where they can be 1742 : replaced with cheaper shifts and masks). However, this sort of thing 1743 : ought to be done earlier, e.g., within the CPS optimizer so that 1744 : all architectures benefit from it. 1745 : 1746 : The compiler compiles to a fixed point, but changes might be somewhat 1747 : fragile nevertheless. Please, report any strange things that you might 1748 : see wrt. div/mod/quot/rem... 1749 : 1750 : ---------------------------------------------------------------------- 1751 : Name: Matthias Blume 1752 : blume 1184 Date: 2002/03/29 17:22:00 1753 : Tag: blume-20020329-div 1754 : Description: 1755 : 1756 : Fixed my broken div/mod logic. Unfortunately, this means that the 1757 : inline code for div/mod now has one more comparison than before. 1758 : Fast paths (quotient > 0 or remainder = 0) are not affected, though. 1759 : The problem was with quotient = 0, because that alone does not tell 1760 : us which way the rounding went. One then has to look at whether 1761 : remainder and divisor have the same sign... :( 1762 : 1763 : Anyway, I replaced the bootfiles with fresh ones... 1764 : 1765 : ---------------------------------------------------------------------- 1766 : Name: Matthias Blume 1767 : blume 1183 Date: 2002/03/29 14:10:00 EST 1768 : Tag: blume-20020329-inlprims 1769 : Description: 1770 : 1771 : NEW BOOTFILES!!! Version number bumped to 110.39.3. 1772 : 1773 : Primops have changed. This means that the bin/boot-file formats have 1774 : changed as well. 1775 : 1776 : To make sure that there is no confusion, I made a new version. 1777 : 1778 : 1779 : CHANGES: 1780 : 1781 : * removed REMT from mltree (remainder should never overflow). 1782 : 1783 : * added primops to deal with divisions of all flavors to the frontend 1784 : 1785 : * handled these primops all the way through so they map to their respective 1786 : MLRISC support 1787 : 1788 : * used these primops in the implementation of Int, Int32, Word, Word32 1789 : 1790 : * removed INLDIV, INLMOD, and INLREM as they are no longer necessary 1791 : 1792 : * parameterized INLMIN, INLMAX, and INLABS by a numkind 1793 : 1794 : * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including 1795 : floating point 1796 : 1797 : * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32, 1798 : and Real (but Real.abs maps to a separate floating-point-only primop) 1799 : 1800 : 1801 : TODO items: 1802 : 1803 : * Hacked Alpha32 instruction selection, disabling the selection of REMx 1804 : instructions because the machine instruction encoder cannot handle 1805 : them. (Hppa, PPC, and Sparc instruction selection did not handle 1806 : REM in the first place, and REM is supported by the x86 machine coder.) 1807 : 1808 : * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction 1809 : selection phase. (The two can be streamlined because the hardware 1810 : delivers both quotient and remainder at the same time anyway.) 1811 : 1812 : * Think about what to do with "valOf(Int32.minInt) div ~1" and friends. 1813 : (Currently the behavior is inconsistent both across architectures and 1814 : wrt. the draft Basis spec.) 1815 : 1816 : * Word8 should eventually be handled natively, too. 1817 : 1818 : * There seems to be one serious bug in mltree-gen.sml. It appears, though, 1819 : as if there currently is no execution path that could trigger it in 1820 : SML/NJ. (The assumptions underlying functions arith and promotable do not 1821 : hold for things like multiplication and division.) 1822 : 1823 : ---------------------------------------------------------------------- 1824 : Name: Matthias Blume 1825 : blume 1181 Date: 2002/03/27 16:27:00 EST 1826 : Tag: blume-20020327-mlrisc-divisions 1827 : Description: 1828 : 1829 : Added support for all four division operations (ML's div, mod, quot, 1830 : and rem) to MLRISC. In the course of doing so, I also rationalized 1831 : the naming (no more annoying switch-around of DIV and QUOT), by 1832 : parameterizing the operation by div_rounding_mode (which can be either 1833 : DIV_TO_ZERO or DIV_TO_NEGINF). 1834 : 1835 : The generic MLTreeGen functor takes care of compiling all four 1836 : operations down to only round-to-zero div. 1837 : 1838 : Missing pieces: 1839 : 1840 : * Doing something smarter than relying on MLTreeGen on architectures 1841 : like, e.g., the x86 where hardware division delivers both quotient and 1842 : remainder at the same time. With this, the implementation of the 1843 : round-to-neginf operations could be further streamlined. 1844 : 1845 : * Remove inlining support for div/mod/rem from the frontend and replace it 1846 : with primops that get carried through to the backend. Do this for all 1847 : int and word types. 1848 : 1849 : ---------------------------------------------------------------------- 1850 : Name: Matthias Blume 1851 : blume 1180 Date: 2002/03/25 17:25:00 EST 1852 : Tag: blume-20020325-divmod 1853 : Description: 1854 : 1855 : I improved (hopefully without breaking them) the implementation of Int.div, 1856 : Int.mod, and Int.rem. For this, the code in translate.sml now takes 1857 : advantage of the following observations: 1858 : 1859 : Let q = x quot y r = x rem y 1860 : d = x div y m = x mod y 1861 : 1862 : where "quot" is the round-to-zero version of integer division that 1863 : hardware usually provides. Then we have: 1864 : 1865 : r = x - q * y where neither the * nor the - will overflow 1866 : d = if q >= 0 orelse x = q * y then q else q - 1 1867 : where neither the * nor the - will overflow 1868 : m = if q >= 0 orelse r = 0 then r else r + y 1869 : where the + will not overflow 1870 : 1871 : This results in substantial simplification of the generated code. 1872 : The following table shows the number of CFG nodes and edges generated 1873 : for 1874 : fun f (x, y) = x OPER y 1875 : (* with OPER \in div, mod, quot, rem *) 1876 : 1877 : 1878 : OPER | nodes(old) | edges(old) | nodes(new) | edges(new) 1879 : -------------------------------------------------------- 1880 : div | 24 | 39 | 12 | 16 1881 : mod | 41 | 71 | 12 | 16 1882 : quot | 8 | 10 | 8 | 10 1883 : rem | 10 | 14 | 8 | 10 1884 : 1885 : 1886 : ---------------------------------------------------------------------- 1887 : Name: Matthias Blume 1888 : blume 1179 Date: 2002/03/25 22:06:00 EST 1889 : Tag: blume-20020325-cprotobug 1890 : Description: 1891 : 1892 : Fixed a bug in cproto (c prototype decoder). 1893 : 1894 : ---------------------------------------------------------------------- 1895 : Name: Matthias Blume 1896 : blume 1178 Date: 2002/03/25 16:00:00 EST 1897 : Tag: blume-20020325-raw-primops 1898 : Description: 1899 : 1900 : I did some cleanup to Allen's new primop code and 1901 : replaced yesterday's bootfiles with new ones. 1902 : (But they are stored in the same place.) 1903 : 1904 : ---------------------------------------------------------------------- 1905 : Name: Matthias Blume 1906 : blume 1175 Date: 2002/03/24 22:40:00 EST 1907 : Tag: blume-20020324-bootfiles 1908 : Description: 1909 : 1910 : Made the bootfiles that Allen asked for. 1911 : 1912 : ---------------------------------------------------------------------- 1913 : leunga 1174 Name: Allen Leung 1914 : Date: 2002/03/23 15:50:00 EST 1915 : Tag: leunga-20020323-flint-cps-rcc-primops 1916 : Description: 1917 : 1918 : 1. Changes to FLINT primops: 1919 : 1920 : (* make a call to a C-function; 1921 : * The primop carries C function prototype information and specifies 1922 : * which of its (ML-) arguments are floating point. C prototype 1923 : * information is for use by the backend, ML information is for 1924 : * use by the CPS converter. *) 1925 : | RAW_CCALL of { c_proto: CTypes.c_proto, 1926 : ml_args: ccall_type list, 1927 : ml_res_opt: ccall_type option, 1928 : reentrant : bool 1929 : } option 1930 : (* Allocate uninitialized storage on the heap. 1931 : * The record is meant to hold short-lived C objects, i.e., they 1932 : * are not ML pointers. With the tag, the representation is 1933 : * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8) 1934 : *) 1935 : | RAW_RECORD of {tag:bool,sz:int} 1936 : and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR 1937 : 1938 : 2. These CPS primops are now overloaded: 1939 : 1940 : rawload of {kind:numkind} 1941 : rawstore of {kind:numkind} 1942 : 1943 : The one argument form is: 1944 : 1945 : rawload {kind} address 1946 : 1947 : The two argument form is: 1948 : 1949 : rawload {kind} [ml object, byte-offset] 1950 : 1951 : 3. RAW_CCALL/RCC now takes two extra arguments: 1952 : 1953 : a. The first is whether the C call is reentrant, i.e., whether 1954 : ML state should be saved and restored. 1955 : b. The second argument is a string argument specifying the name of 1956 : library and the C function. 1957 : 1958 : These things are currently not handled in the code generator, yet. 1959 : 1960 : 4. In CProto, 1961 : 1962 : An encoding type of "bool" means "ml object" and is mapped into 1963 : C prototype of PTR. Note that "bool" is different than "string", 1964 : even though "string" is also mapped into PTR, because "bool" 1965 : is assigned an CPS type of BOGt, while "string" is assigned INT32t. 1966 : 1967 : 5. Pickler/unpicker 1968 : 1969 : Changed to handle RAW_RECORD and newest RAW_CCALL 1970 : 1971 : 6. MLRiscGen, 1972 : 1973 : 1. Changed to handle the new rawload/rawstore/rawrecord operators. 1974 : 2. Code for handling C Calls has been moved to a new module CPSCCalls, 1975 : in the file CodeGen/cpscompile/cps-c-calls.sml 1976 : 1977 : 7. Added the conditional move operator 1978 : 1979 : condmove of branch 1980 : 1981 : to cps. Generation of this is still buggy so it is currently 1982 : disabled. 1983 : 1984 : ---------------------------------------------------------------------- 1985 : george 1168 Name: Lal George 1986 : Date: 2002/03/22 14:18:25 EST 1987 : george 1171 Tag: george-20020322-cps-branch-prob 1988 : george 1168 Description: 1989 : 1990 : george 1171 Implemented the Ball-Larus branch prediction-heuristics, and 1991 : george 1168 incorporated graphical viewers for control flow graphs. 1992 : 1993 : george 1171 Ball-Larus Heuristics: 1994 : george 1168 --------------------- 1995 : See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml. 1996 : 1997 : By design it uses the Dempster-Shafer theory for combining 1998 : probabilities. For example, in the function: 1999 : 2000 : fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc) 2001 : 2002 : the ball-larus heuristics predicts that the n=0 is unlikely 2003 : (OH-heuristic), and the 'then' branch is unlikely because of the 2004 : RH-heuristic -- giving the 'then' branch an even lower combined 2005 : george 1171 probability using the Dempster-Shafer theory. 2006 : george 1168 2007 : Finally, John Reppy's loop analysis in MLRISC, further lowers the 2008 : probability of the 'then' branch because of the loop in the else 2009 : branch. 2010 : 2011 : 2012 : Graphical Viewing: 2013 : ------------------ 2014 : I merely plugged in Allen's graphical viewers into the compiler. The 2015 : additional code is not much. At the top level, saying: 2016 : 2017 : Control.MLRISC.getFlag "cfg-graphical-view" := true; 2018 : 2019 : will display the graphical view of the control flow graph just before 2020 : back-patching. daVinci must be in your path for this to work. If 2021 : daVinci is not available, then the default viewer can be changed 2022 : using: 2023 : 2024 : Control.MLRISC.getString "viewer" 2025 : 2026 : which can be set to "dot" or "vcg" for the corresponding viewers. Of 2027 : course, these viewers must be in your path. 2028 : 2029 : The above will display the compilation unit at the level of clusters, 2030 : many of which are small, boring, and un-interesting. Also setting: 2031 : 2032 : Control.MLRISC.getInt "cfg-graphical-view_size" 2033 : 2034 : will display clusters that are larger than the value set by the above. 2035 : 2036 : 2037 : ---------------------------------------------------------------------- 2038 : blume 1159 Name: Matthias Blume 2039 : Date: 2002/03/21 22:20:00 EST 2040 : Tag: blume-20020321-kmp-bugfix 2041 : Description: 2042 : 2043 : Changed the interface to the KMP routine in PreString and fixed 2044 : a minor bug in one place where it was used. 2045 : 2046 : ---------------------------------------------------------------------- 2047 : leunga 1156 Name: Allen Leung 2048 : leunga 1158 Date: 2002/03/21 20:30:00 EST 2049 : Tag: leunga-20020321-cfg 2050 : Description: 2051 : 2052 : Fixed a potential problem in cfg edge splitting. 2053 : 2054 : ---------------------------------------------------------------------- 2055 : Name: Allen Leung 2056 : leunga 1156 Date: 2002/03/21 17:15:00 EST 2057 : Tag: leunga-20020321-x86-fp-cfg 2058 : Description: 2059 : 2060 : 1. Recoded the buggy parts of x86-fp. 2061 : 2062 : a. All the block reordering code has been removed. 2063 : We now depend on the block placement phases to do this work. 2064 : 2065 : b. Critical edge splitting code has been simplified and moved into the 2066 : CFG modules, as where they belong. 2067 : 2068 : Both of these were quite buggy and complex. The code is now much, much 2069 : simpler. 2070 : 2071 : 2. X86 backend. 2072 : 2073 : a. Added instructions for 64-bit support. Instruction selection for 2074 : 64-bit has not been committed, however, since that 2075 : requires changes to MLTREE which haven't been approved by 2076 : Lal and John. 2077 : 2078 : b. Added support for FUCOMI and FUCOMIP when generating code for 2079 : PentiumPro and above. We only generate these instructions in 2080 : the fast-fp mode. 2081 : 2082 : c. Added cases for JP and JNP in X86FreqProps. 2083 : 2084 : 3. CFG 2085 : 2086 : CFG now has a bunch of methods for edge splitting and merging. 2087 : 2088 : 4. Machine description. 2089 : 2090 : John's simplification of MLTREE_BASIS.fcond broke a few machine 2091 : description things: 2092 : 2093 : rtl-build.{sig,sml} and hppa.mdl fixed. 2094 : 2095 : NOTE: the machine description stuff in the repository is still broken. 2096 : Again, I can't put my fixes in because that involves 2097 : changes to MLTREE. 2098 : 2099 : ---------------------------------------------------------------------- 2100 : blume 1145 Name: Matthias Blume 2101 : blume 1155 Date: 2002/03/20 15:55:00 EST 2102 : Tag: blume-20020320-kmp 2103 : Description: 2104 : 2105 : Implemented Knuth-Morris-Pratt string matching in PreString and used 2106 : it for String.isSubstring, Substring.isSubstring, and 2107 : Substring.position. 2108 : 2109 : (Might need some stress-testing. Simple examples worked fine.) 2110 : 2111 : ---------------------------------------------------------------------- 2112 : Name: Matthias Blume 2113 : blume 1152 Date: 2002/03/19 16:37:00 EST 2114 : Tag: blume-20020319-witnesses 2115 : Description: 2116 : 2117 : Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib. 2118 : 2119 : This implements a generic mechanism for changing constness qualifiers 2120 : anywhere within big C types without resorting to outright "casts". 2121 : (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you 2122 : modify the constness at the outermost level.) 2123 : The implementation of "convert" is based on the idea of "witness" 2124 : values -- values that are not used by the operation but whose types 2125 : "testify" to their applicability. On the implementation side, "convert" 2126 : is simply a projection (returning its second curried argument). With 2127 : cross-module inlining, it should not result in any machine code being 2128 : generated. 2129 : 2130 : ---------------------------------------------------------------------- 2131 : Name: Matthias Blume 2132 : blume 1148 Date: 2002/03/15 16:40:00 EST 2133 : Tag: blume-20020315-basis 2134 : Description: 2135 : 2136 : Provided (preliminary?) implementations for 2137 : 2138 : {String,Substring}.{concatWith,isSuffix,isSubstring} 2139 : 2140 : and 2141 : 2142 : Substring.full 2143 : 2144 : Those are in the Basis spec but they were missing in SML/NJ. 2145 : 2146 : ---------------------------------------------------------------------- 2147 : Name: Matthias Blume 2148 : blume 1145 Date: 2002/03/14 21:30:00 EST 2149 : Tag: blume-20020314-controls 2150 : Description: 2151 : 2152 : Controls: 2153 : --------- 2154 : 2155 : 1. Factored out the recently-added Controls : CONTROLS stuff and put 2156 : it into its own library$/controls-lib.cm. The source tree for 2157 : this is under src/smlnj-lib/Controls. 2158 : 2159 : 2. Changed the names of types and functions in this interface, so they 2160 : make a bit more "sense": 2161 : 2162 : module -> registry 2163 : 'a registry -> 'a group 2164 : 2165 : 3. The interface now deals in ref cells only. The getter/setter interface 2166 : is (mostly) gone. 2167 : 2168 : 4. Added a function that lets one register an already-existing ref cell. 2169 : 2170 : 5. Made the corresponding modifications to the rest of the code so that 2171 : everything compiles again. 2172 : 2173 : 6. Changed the implementation of Controls.MLRISC back to something closer 2174 : to the original. In particular, this module (and therefore MLRISC) 2175 : does not depend on Controls. There now is some link-time code in 2176 : int-sys.sml that registers the MLRISC controls with the Controls 2177 : module. 2178 : 2179 : CM: 2180 : --- 2181 : 2182 : * One can now specify the lambda-split aggressiveness in init.cmi. 2183 : 2184 : ---------------------------------------------------------------------- 2185 : leunga 1142 Name: Allen Leung 2186 : Date: 2002/03/13 17:30:00 EST 2187 : Tag: leunga-20020313-x86-fp-unary 2188 : Description: 2189 : 2190 : Bug fix for: 2191 : 2192 : > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml 2193 : > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002 2194 : > - fun f(x,(y,z)) = Real.~ y; 2195 : > [autoloading] 2196 : > [autoloading done] 2197 : > fchsl (%eax), 184(%esp) 2198 : > Error: MLRisc bug: X86MCEmitter.emitInstr 2199 : > 2200 : > uncaught exception Error 2201 : > raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19 2202 : 2203 : The problem was that the code generator did not generate any fp registers 2204 : in this case, and the ra didn't know that it needed to run the X86FP phase to 2205 : translate the pseudo fp instruction. This only happened with unary fp 2206 : operators in certain situations. 2207 : 2208 : ---------------------------------------------------------------------- 2209 : blume 1137 Name: Matthias Blume 2210 : blume 1140 Date: 2002/03/13 14:00:00 EST 2211 : Tag: blume-20020313-overload-etc 2212 : Description: 2213 : 2214 : 1. Added _overload as a synonym for overload for backward compatibility. 2215 : (Control.overloadKW must be true for either version to be accepted.) 2216 : 2217 : 2. Fixed bug in install script that caused more things to be installed 2218 : than what was requested in config/targets. 2219 : 2220 : 3. Made CM aware of the (_)overload construct so that autoloading 2221 : works. 2222 : 2223 : ---------------------------------------------------------------------- 2224 : Name: Matthias Blume 2225 : blume 1138 Date: 2002/03/12 22:03:00 EST 2226 : Tag: blume-20020312-url 2227 : Description: 2228 : 2229 : Forgot to update BOOT and srcarchiveurl. 2230 : 2231 : ---------------------------------------------------------------------- 2232 : Name: Matthias Blume 2233 : blume 1137 Date: 2002/03/12 17:30:00 EST 2234 : Tag: blume-20020312-version110392 2235 : Description: 2236 : 2237 : Yet another version number bump (because of small changes to the 2238 : binfile format). Version number is now 110.39.2. NEW BOOTFILES! 2239 : 2240 : Changes: 2241 : 2242 : The new pid generation scheme described a few weeks ago was overly 2243 : complicated. I implemented a new mechanism that is simpler and 2244 : provides a bit more "stability": Once CM has seen a compilation 2245 : unit, it keeps its identity constant (as long as you do not delete 2246 : those crucial CM/GUID/* files). This means that when you change 2247 : blume 1145 an interface, compile, then go back to the old interface, and 2248 : blume 1137 compile again, you arrive at the original pid. 2249 : 2250 : There now also is a mechanism that instructs CM to use the plain 2251 : environment hash as a module's pid (effectively making its GUID 2252 : the empty string). For this, "noguid" must be specified as an 2253 : option to the .sml file in question within its .cm file. 2254 : This is most useful for code that is being generated by tools such 2255 : as ml-nlffigen (because during development programmers tend to 2256 : erase the tool's entire output directory tree including CM's cached 2257 : GUIDs). "noguid" is somewhat dangerous (since it can be used to locally 2258 : revert to the old, broken behavior of SML/NJ, but in specific cases 2259 : where there is no danger of interface confusion, its use is ok 2260 : (I think). 2261 : 2262 : ml-nlffigen by default generates "noguid" annotations. They can be 2263 : turned off by specifying -guid in its command line. 2264 : 2265 : ---------------------------------------------------------------------- 2266 : george 1136 Name: Lal George 2267 : Date: 2002/03/12 12 14:42:36 EST 2268 : Tag: george-20020312-frequency-computation 2269 : Description: 2270 : 2271 : Integrated jump chaining and static block frequency into the 2272 : compiler. More details and numbers later. 2273 : leunga 1025 2274 : george 1003 ---------------------------------------------------------------------- 2275 : george 1133 Name: Lal George 2276 : Date: 2002/03/11 11 22:38:53 EST 2277 : Tag: george-20020311-jump-chain-elim 2278 : Description: 2279 : 2280 : Tested the jump chain elimination on all architectures (except the 2281 : hppa). This is on by default right now and is profitable for the 2282 : alpha and x86, however, it may not be profitable for the sparc and ppc 2283 : when compiling the compiler. 2284 : 2285 : The gc test will typically jump to a label at the end of the cluster, 2286 : where there is another jump to an external cluster containing the actual 2287 : code to invoke gc. This is to allow factoring of common gc invocation 2288 : sequences. That is to say, we generate: 2289 : 2290 : f: 2291 : testgc 2292 : ja L1 % jump if above to L1 2293 : 2294 : L1: 2295 : jmp L2 2296 : 2297 : 2298 : After jump chain elimination the 'ja L1' instructions is converted to 2299 : 'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end 2300 : up being implemented in their long form (if L2 is far away) using: 2301 : 2302 : jbe L3 % jump if below or equal to L3 2303 : jmp L2 2304 : L3: 2305 : ... 2306 : 2307 : 2308 : For large compilation units L2 may be far away. 2309 : 2310 : 2311 : ---------------------------------------------------------------------- 2312 : blume 1132 Name: Matthias Blume 2313 : Date: 2002/03/11 13:30:00 EST 2314 : Tag: blume-20020311-mltreeeval 2315 : Description: 2316 : 2317 : A functor parameter was missing. 2318 : 2319 : ---------------------------------------------------------------------- 2320 : leunga 1130 Name: Allen Leung 2321 : leunga 1131 Date: 2002/03/11 10:30:00 EST 2322 : leunga 1142 Tag: leunga-20020311-runtime-string0 2323 : leunga 1131 Description: 2324 : 2325 : The representation of the empty string now points to a 2326 : legal null terminated C string instead of unit. It is now possible 2327 : to convert an ML string into C string with InlineT.CharVector.getData. 2328 : This compiles into one single machine instruction. 2329 : 2330 : ---------------------------------------------------------------------- 2331 : Name: Allen Leung 2332 : leunga 1130 Date: 2002/03/10 23:55:00 EST 2333 : Tag: leunga-20020310-x86-call 2334 : Description: 2335 : 2336 : Added machine generation for CALL instruction (relative displacement mode) 2337 : 2338 : ---------------------------------------------------------------------- 2339 : blume 1128 Name: Matthias Blume 2340 : Date: 2002/03/08 16:05:00 2341 : Tag: blume-20020308-entrypoints 2342 : Description: 2343 : 2344 : Version number bumped to 110.39.1. NEW BOOTFILES! 2345 : 2346 : Entrypoints: non-zero offset into a code object where execution should begin. 2347 : 2348 : - Added the notion of an entrypoint to CodeObj. 2349 : - Added reading/writing of entrypoint info to Binfile. 2350 : - Made runtime system bootloader aware of entrypoints. 2351 : - Use the address of the label of the first function given to mlriscGen 2352 : as the entrypoint. This address is currently always 0, but it will 2353 : not be 0 once we turn on block placement. 2354 : - Removed the linkage cluster code (which was The Other Way(tm) of dealing 2355 : with entry points) from mlriscGen. 2356 : 2357 : ---------------------------------------------------------------------- 2358 : leunga 1127 Name: Allen Leung 2359 : Date: 2002/03/07 20:45:00 EST 2360 : Tag: leunga-20020307-x86-cmov 2361 : Description: 2362 : 2363 : Bug fixes for CMOVcc on x86. 2364 : 2365 : 1. Added machine code generation for CMOVcc 2366 : 2. CMOVcc is now generated in preference over SETcc on PentiumPro or above. 2367 : 3. CMOVcc cannot have an immediate operand as argument. 2368 : 2369 : ---------------------------------------------------------------------- 2370 : blume 1126 Name: Matthias Blume 2371 : Date: 2002/03/07 16:15:00 EST 2372 : Tag: blume-20020307-controls 2373 : Description: 2374 : 2375 : This is a very large but mostly boring patch which makes (almost) 2376 : every tuneable compiler knob (i.e., pretty much everything under 2377 : Control.* plus a few other things) configurable via both the command 2378 : line and environment variables in the style CM did its configuration 2379 : until now. 2380 : 2381 : Try starting sml with '-h' (or, if you are brave, '-H') 2382 : 2383 : To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which 2384 : implements the underlying generic mechanism. 2385 : 2386 : The interface to some of the existing such facilities has changed somewhat. 2387 : For example, the MLRiscControl module now provides mkFoo instead of getFoo. 2388 : (The getFoo interface is still there for backward-compatibility, but its 2389 : use is deprecated.) 2390 : 2391 : The ml-build script passes -Cxxx=yyy command-line arguments through so 2392 : that one can now twiddle the compiler settings when using this "batch" 2393 : compiler. 2394 : 2395 : TODO items: 2396 : 2397 : We should go through and throw out all controls that are no longer 2398 : connected to anything. Moreover, we should go through and provide 2399 : meaningful (and correct!) documentation strings for those controls 2400 : that still are connected. 2401 : 2402 : Currently, multiple calls to Controls.new are accepted (only the first 2403 : has any effect). Eventually we should make sure that every control 2404 : is being made (via Controls.new) exactly once. Future access can then 2405 : be done using Controls.acc. 2406 : 2407 : Finally, it would probably be a good idea to use the getter-setter 2408 : interface to controls rather than ref cells. For the time being, both 2409 : styles are provided by the Controls module, but getter-setter pairs are 2410 : better if thread-safety is of any concern because they can be wrapped. 2411 : 2412 : ***************************************** 2413 : 2414 : One bug fix: The function blockPlacement in three of the MLRISC 2415 : backpatch files used to be hard-wired to one of two possibilities at 2416 : link time (according to the value of the placementFlag). But (I 2417 : think) it should rather sense the flag every time. 2418 : 2419 : ***************************************** 2420 : 2421 : Other assorted changes (by other people who did not supply a HISTORY entry): 2422 : 2423 : 1. the cross-module inliner now works much better (Monnier) 2424 : 2. representation of weights, frequencies, and probabilities in MLRISC 2425 : changed in preparation of using those for weighted block placement 2426 : (Reppy, George) 2427 : 2428 : ---------------------------------------------------------------------- 2429 : george 1116 Name: Lal George 2430 : george 1124 Date: 2002/03/07 14:44:24 EST 2002 2431 : Tag: george-20020307-weighted-block-placement 2432 : 2433 : Tested the weighted block placement optimization on all architectures 2434 : (except the hppa) using AMPL to generate the block and edge frequencies. 2435 : Changes were required in the machine properties to correctly 2436 : categorize trap instructions. There is an MLRISC flag 2437 : "weighted-block-placement" that can be used to enable weighted block 2438 : placement, but this will be ineffective without block/edge 2439 : frequencies (coming soon). 2440 : 2441 : 2442 : ---------------------------------------------------------------------- 2443 : Name: Lal George 2444 : george 1116 Date: 2002/03/05 17:24:48 EST 2445 : Tag: george-20020305-linkage-cluster 2446 : 2447 : george 1124 In order to support the block placement optimization, a new cluster 2448 : is generated as the very first cluster (called the linkage cluster). 2449 : It contains a single jump to the 'real' entry point for the compilation 2450 : unit. Block placement has no effect on the linkage cluster itself, but 2451 : all the other clusters have full freedom in the manner in which they 2452 : reorder blocks or functions. 2453 : george 1116 2454 : On the x86 the typical linkage code that is generated is: 2455 : ---------------------- 2456 : .align 2 2457 : L0: 2458 : addl L1-L0, 72(%esp) 2459 : george 1124 jmp L1 2460 : george 1116 2461 : 2462 : .align 2 2463 : L1: 2464 : ---------------------- 2465 : 2466 : 72(%esp) is the memory location for the stdlink register. This 2467 : must contain the address of the CPS function being called. In the 2468 : george 1124 above example, it contains the address of L0; before 2469 : george 1116 calling L1 (the real entry point for the compilation unit), it 2470 : must contain the address for L1, and hence 2471 : 2472 : addlL1-L0, 72(%esp) 2473 : 2474 : george 1124 I have tested this on all architectures except the hppa.The increase 2475 : in code size is of course negligible 2476 : george 1116 2477 : ---------------------------------------------------------------------- 2478 : leunga 1115 Name: Allen Leung 2479 : Date: 2002/03/03 13:20:00 EST 2480 : Tag: leunga-20020303-mlrisc-tools 2481 : 2482 : Added #[ ... ] expressions to mlrisc tools 2483 : 2484 : ---------------------------------------------------------------------- 2485 : blume 1096 Name: Matthias Blume 2486 : blume 1098 Date: 2002/02/27 12:29:00 EST 2487 : Tag: blume-20020227-cdebug 2488 : Description: 2489 : 2490 : - made types in structure C and C_Debug to be equal 2491 : - got rid of code duplication (c-int.sml vs. c-int-debug.sml) 2492 : - there no longer is a C_Int_Debug (C_Debug is directly derived from C) 2493 : 2494 : ---------------------------------------------------------------------- 2495 : Name: Matthias Blume 2496 : blume 1096 Date: 2002/02/26 12:00:00 EST 2497 : Tag: blume-20020226-ffi 2498 : Description: 2499 : 2500 : 1. Fixed a minor bug in CM's "noweb" tool: 2501 : If numbering is turned off, then truly don't number (i.e., do not 2502 : supply the -L option to noweb). The previous behavior was to supply 2503 : -L'' -- which caused noweb to use the "default" line numbering scheme. 2504 : Thanks to Chris Richards for pointing this out (and supplying the fix). 2505 : 2506 : 2. Once again, I reworked some aspects of the FFI: 2507 : 2508 : A. The incomplete/complete type business: 2509 : 2510 : - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are 2511 : gone! 2512 : - ML types representing an incomplete type are now *equal* to 2513 : ML types representing their corresponding complete types (just like 2514 : in C). This is still safe because ml-nlffigen will not generate 2515 : RTTI for incomplete types, nor will it generate functions that 2516 : require access to such RTTI. But when ML code generated from both 2517 : incomplete and complete versions of the C type meet, the ML types 2518 : are trivially interoperable. 2519 : 2520 : NOTE: These changes restore the full generality of the translation 2521 : (which was previously lost when I eliminated functorization)! 2522 : 2523 : B. Enum types: 2524 : 2525 : - Structure C now has a type constructor "enum" that is similar to 2526 : how the "su" constructor works. However, "enum" is not a phantom 2527 : type because each "T enum" has values (and is isomorphic to 2528 : MLRep.Signed.int). 2529 : - There are generic access operations for enum objects (using 2530 : MLRep.Signed.int). 2531 : - ml-nlffigen will generate a structure E_foo for each "enum foo". 2532 : * The structure contains the definition of type "mlrep" (the ML-side 2533 : representation type of the enum). Normally, mlrep is the same 2534 : as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec", 2535 : then mlrep will be defined as a datatype -- thus facilitating 2536 : pattern matching on mlrep values. 2537 : ("-ec" will be suppressed if there are duplicate values in an 2538 : enumeration.) 2539 : * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep 2540 : will be generated for each C enum constant xxx. 2541 : * Conversion functions m2i and i2m convert between mlrep and 2542 : MLRep.Signed.int. (Without "-ec", these functions are identities.) 2543 : * Coversion functions c and ml convert between mlrep and "tag enum". 2544 : * Access functions (get/set) fetch and store mlrep values. 2545 : - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed 2546 : enumerations are merged into one single enumeration represented by 2547 : structure E_'. 2548 : 2549 : ---------------------------------------------------------------------- 2550 : leunga 1080 Name: Allen Leung 2551 : leunga 1094 Date: 2002/02/25 04:45:00 EST 2552 : Tag: leunga-20020225-cps-spill 2553 : 2554 : This is a new implementation of the CPS spill phase. 2555 : The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml 2556 : In case of problems, replace it with the old file spill.sml 2557 : 2558 : The current compiler runs into some serious performance problems when 2559 : constructing a large record. This can happen when we try to compile a 2560 : structure with many items. Even a very simple structure like the following 2561 : makes the compiler slow down. 2562 : 2563 : structure Foo = struct 2564 : val x_1 = 0w1 : Word32.int 2565 : val x_2 = 0w2 : Word32.int 2566 : val x_3 = 0w3 : Word32.int 2567 : ... 2568 : val x_N = 0wN : Word32.int 2569 : end 2570 : 2571 : The following table shows the compile time, from N=1000 to N=4000, 2572 : with the old compiler: 2573 : 2574 : N 2575 : 1000 CPS 100 spill 0.04u 0.00s 0.00g 2576 : MLRISC ra 0.06u 0.00s 0.05g 2577 : (spills = 0 reloads = 0) 2578 : TOTAL 0.63u 0.07s 0.21g 2579 : 2580 : 1100 CPS 100 spill 8.25u 0.32s 0.64g 2581 : MLRISC ra 5.68u 0.59s 3.93g 2582 : (spills = 0 reloads = 0) 2583 : TOTAL 14.71u 0.99s 4.81g 2584 : 2585 : 1500 CPS 100 spill 58.55u 2.34s 1.74g 2586 : MLRISC ra 5.54u 0.65s 3.91g 2587 : (spills = 543 reloads = 1082) 2588 : TOTAL 65.40u 3.13s 6.00g 2589 : 2590 : 2000 CPS 100 spill 126.69u 4.84s 3.08g 2591 : MLRISC ra 0.80u 0.10s 0.55g 2592 : (spills = 42 reloads = 84) 2593 : TOTAL 129.42u 5.10s 4.13g 2594 : 2595 : 3000 CPS 100 spill 675.59u 19.03s 11.64g 2596 : MLRISC ra 2.69u 0.27s 1.38g 2597 : (spills = 62 reloads = 124) 2598 : TOTAL 682.48u 19.61s 13.99g 2599 : 2600 : 4000 CPS 100 spill 2362.82u 56.28s 43.60g 2601 : MLRISC ra 4.96u 0.27s 2.72g 2602 : (spills = 85 reloads = 170) 2603 : TOTAL 2375.26u 57.21s 48.00g 2604 : 2605 : As you can see the old cps spill module suffers from some serious 2606 : leunga 1115 performance problem. But since I cannot decipher the old code fully, 2607 : instead of patching the problems up, I'm reimplementing it 2608 : leunga 1094 with a different algorithm. The new code is more modular, 2609 : smaller when compiled, and substantially faster 2610 : (O(n log n) time and O(n) space). Timing of the new spill module: 2611 : 2612 : 4000 CPS 100 spill 0.02u 0.00s 0.00g 2613 : MLRISC ra 0.25u 0.02s 0.15g 2614 : (spills=1 reloads=3) 2615 : TOTAL 7.74u 0.34s 1.62g 2616 : 2617 : Implementation details: 2618 : 2619 : As far as I can tell, the purpose of the CPS spill module is to make sure the 2620 : number of live variables at any program point (the bandwidth) 2621 : does not exceed a certain limit, which is determined by the 2622 : size of the spill area. 2623 : 2624 : When the bandwidth is too large, we decrease the register pressure by 2625 : packing live variables into spill records. How we achieve this is 2626 : completely different than what we did in the old code. 2627 : 2628 : First, there is something about the MLRiscGen code generator 2629 : that we should be aware of: 2630 : 2631 : o MLRiscGen performs code motion! 2632 : 2633 : In particular, it will move floating point computations and 2634 : address computations involving only the heap pointer to 2635 : their use sites (if there is only a single use). 2636 : What this means is that if we have a CPS record construction 2637 : statement 2638 : 2639 : RECORD(k,vl,w,e) 2640 : 2641 : we should never count the new record address w as live if w 2642 : has only one use (which is often the case). 2643 : 2644 : We should do something similar to floating point, but the transformation 2645 : there is much more complex, so I won't deal with that. 2646 : 2647 : Secondly, there are now two new cps primops at our disposal: 2648 : 2649 : 1. rawrecord of record_kind option 2650 : This pure operator allocates some uninitialized storage from the heap. 2651 : There are two forms: 2652 : 2653 : rawrecord NONE [INT n] allocates a tagless record of length n 2654 : rawrecord (SOME rk) [INT n] allocates a tagged record of length n 2655 : and initializes the tag. 2656 : 2657 : 2. rawupdate of cty 2658 : rawupdate cty (v,i,x) 2659 : Assigns to x to the ith component of record v. 2660 : The storelist is not updated. 2661 : 2662 : We use these new primops for both spilling and increment record construction. 2663 : 2664 : 1. Spilling. 2665 : 2666 : This is implemented with a linear scan algorithm (but generalized 2667 : to trees). The algorithm will create a single spill record at the 2668 : beginning of the cps function and use rawupdate to spill to it, 2669 : and SELECT or SELp to reload from it. So both spills and reloads 2670 : are fine-grain operations. In contrast, in the old algorithm 2671 : "spills" have to be bundled together in records. 2672 : 2673 : Ideally, we should sink the spill record construction to where 2674 : it is needed. We can even split the spill record into multiple ones 2675 : at the places where they are needed. But CPS is not a good 2676 : representation for global code motion, so I'll keep it simple and 2677 : am not attempting this. 2678 : 2679 : 2. Incremental record construction (aka record splitting). 2680 : 2681 : Long records with many component values which are simulatenously live 2682 : (recall that single use record addresses are not considered to 2683 : be live) are constructed with rawrecord and rawupdate. 2684 : We allocate space on the heap with rawrecord first, then gradually 2685 : fill it in with rawupdate. This is the technique suggested to me 2686 : by Matthias. 2687 : 2688 : Some restrictions on when this is applicable: 2689 : 1. It is not a VECTOR record. The code generator currently does not handle 2690 : this case. VECTOR record uses double indirection like arrays. 2691 : 2. All the record component values are defined in the same "basic block" 2692 : as the record constructor. This is to prevent speculative 2693 : record construction. 2694 : 2695 : ---------------------------------------------------------------------- 2696 : Name: Allen Leung 2697 : leunga 1086 Date: 2002/02/22 01:02:00 EST 2698 : Tag: leunga-20020222-mlrisc-tools 2699 : 2700 : Minor bug fixes in the parser and rewriter 2701 : 2702 : ---------------------------------------------------------------------- 2703 : Name: Allen Leung 2704 : leunga 1085 Date: 2002/02/21 20:20:00 EST 2705 : Tag: leunga-20020221-peephole 2706 : 2707 : Regenerated the peephole files. Some contained typos in the specification 2708 : and some didn't compile because of pretty printing bugs in the old version 2709 : of 'nowhere'. 2710 : 2711 : ---------------------------------------------------------------------- 2712 : Name: Allen Leung 2713 : leunga 1079 Date: 2002/02/19 20:20:00 EST 2714 : Tag: leunga-20020219-mlrisc-tools 2715 : Description: 2716 : 2717 : Minor bug fixes to the mlrisc-tools library: 2718 : 2719 : 1. Fixed up parsing colon suffixed keywords 2720 : 2. Added the ability to shut the error messages up 2721 : 3. Reimplemented the pretty printer and fixed up/improved 2722 : the pretty printing of handle and -> types. 2723 : 4. Fixed up generation of literal symbols in the nowhere tool. 2724 : 5. Added some SML keywords to to sml.sty 2725 : 2726 : ---------------------------------------------------------------------- 2727 : Name: Matthias Blume 2728 : blume 1078 Date: 2002/02/19 16:20:00 EST 2729 : Tag: blume-20020219-cmffi 2730 : Description: 2731 : 2732 : A wild mix of changes, some minor, some major: 2733 : 2734 : * All C FFI-related libraries are now anchored under $c: 2735 :$/c.cm --> $c/c.cm 2736 :$/c-int.cm --> $c/internals/c-int.cm 2737 :$/memory.cm --> $c/memory/memory.cm 2738 : 2739 : * "make" tool (in CM) now treats its argument pathname slightly 2740 : differently: 2741 : 1. If the native expansion is an absolute name, then before invoking 2742 : the "make" command on it, CM will apply OS.Path.mkRelative 2743 : (with relativeTo = OS.FileSys.getDir()) to it. 2744 : 2. The argument will be passed through to subsequent phases of CM 2745 : processing without "going native". In particular, if the argument 2746 : was an anchored path, then "make" will not lose track of that anchor. 2747 : 2748 : * Compiler backends now "know" their respective C calling conventions 2749 : instead of having to be told about it by ml-nlffigen. This relieves 2750 : ml-nlffigen from one of its burdens. 2751 : 2752 : * The X86Backend has been split into X86CCallBackend and X86StdCallBackend. 2753 : 2754 : * Export C_DEBUG and C_Debug from$c/c.cm. 2755 : 2756 : * C type encoding in ml-nlffi-lib has been improved to model the conceptual 2757 : subtyping relationship between incomplete pointers and their complete 2758 : counterparts. For this, ('t, 'c) ptr has been changed to 'o ptr -- 2759 : with the convention of instantiating 'o with ('t, 'c) obj whenever 2760 : the pointer target type is complete. In the incomplete case, 'o 2761 : will be instantiated with some "'c iobj" -- a type obtained by 2762 : using one of the functors PointerToIncompleteType or PointerToCompleteType. 2763 : 2764 : Operations that work on both incomplete and complete pointer types are 2765 : typed as taking an 'o ptr while operations that require the target to 2766 : be known are typed as taking some ('t, 'c) obj ptr. 2767 : 2768 : voidptr is now a bit "more concrete", namely "type voidptr = void ptr'" 2769 : where void is an eqtype without any values. This makes it possible 2770 : to work on voidptr values using functions meant to operate on light 2771 : incomplete pointers. 2772 : 2773 : * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has 2774 : been vastly simplified. 2775 : 2776 : ---------------------------------------------------------------------- 2777 : Name: Matthias Blume 2778 : blume 1076 Date: 2002/02/19 10:48:00 EST 2779 : Tag: blume-20020219-pqfix 2780 : Description: 2781 : 2782 : Applied Chris Okasaki's bug fix for priority queues. 2783 : 2784 : ---------------------------------------------------------------------- 2785 : Name: Matthias Blume 2786 : blume 1073 Date: 2002/02/15 17:05:00 2787 : blume 1069 Tag: Release_110_39 2788 : Description: 2789 : 2790 : blume 1073 Last-minute retagging is becoming a tradition... :-( 2791 : 2792 : This is the working release 110.39. 2793 : 2794 : ---------------------------------------------------------------------- 2795 : Name: Matthias Blume 2796 : Date: 2002/02/15 16:00:00 EST 2797 : Tag: Release_110_39-orig 2798 : Description: 2799 : 2800 : blume 1069 Working release 110.39. New bootfiles. 2801 : 2802 : blume 1073 (Update: There was a small bug in the installer so it wouldn't work 2803 : with all shells. So I retagged. -Matthias) 2804 : 2805 : blume 1069 ---------------------------------------------------------------------- 2806 : Name: Matthias Blume 2807 : blume 1068 Date: 2002/02/15 14:17:00 EST 2808 : Tag: blume-20020215-showbindings 2809 : Description: 2810 : 2811 : Added EnvRef.listBoundSymbols and CM.State.showBindings. Especially 2812 : the latter can be useful for exploring what bindings are available at 2813 : the interactive prompt. (The first function returns only the list 2814 : of symbols that are really bound, the second prints those but also the 2815 : ones that CM's autoloading mechanism knows about.) 2816 : 2817 : ---------------------------------------------------------------------- 2818 : Name: Matthias Blume 2819 : blume 1067 Date: 2002/02/15 12:08:00 EST 2820 : Tag: blume-20020215-iptrs 2821 : Description: 2822 : 2823 : Two improvements to ml-nlffigen: 2824 : 2825 : 1. Write files only if they do not exist or if their current contents 2826 : do not coincide with what's being written. (That is, avoid messing 2827 : with the time stamps unless absolutely necessary.) 2828 : 2829 : 2. Implement a "repository" mechanism for generated files related 2830 : to "incomplete pointer types". See the README file for details. 2831 : 2832 : ---------------------------------------------------------------------- 2833 : Name: Matthias Blume 2834 : blume 1066 Date: 2002/02/14 11:50:00 EST 2835 : Tag: blume-20020214-quote 2836 : Description: 2837 : 2838 : Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm). This is required 2839 : because of the new and improved tag generation scheme. (Thanks to Allen 2840 : Leung for pointing it out.) 2841 : 2842 : ---------------------------------------------------------------------- 2843 : george 1065 Name: Lal George 2844 : Date: 2002/02/14 09:55:27 EST 2002 2845 : Tag: george-20020214-isabelle-bug 2846 : Description: 2847 : 2848 : Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation 2849 : of Isabelle on the x86. 2850 : 2851 : From Allen: 2852 : ----------- 2853 : I've found the problem: 2854 : 2855 : in ra-core.sml, I use the counter "blocked" to keep track of the 2856 : true number of elements in the freeze queue. When the counter goes 2857 : to zero, I skip examining the queue. But I've messed up the 2858 : bookkeeping in combine(): 2859 : 2860 : else (); 2861 : case !ucol of 2862 : PSEUDO => (if !cntv > 0 then 2863 : (if !cntu > 0 then blocked := !blocked - 1 else (); 2864 : ^^^^^^^^^^^^^^^^^^^^^^^ 2865 : moveu := mergeMoveList(!movev, !moveu) 2866 : ) 2867 : else (); 2868 : 2869 : combine() is called to coalesce two nodes u and v. 2870 : I think I was thinking that if the move counts of u and v are both 2871 : greater than zero then after they are coalesced then one node is 2872 : removed from the freeze queue. Apparently I was thinking that 2873 : both u and v are of low degree, but that's clearly not necessarily true. 2874 : 2875 : 2876 : 02/12/2002: 2877 : Here's the patch. HOL now compiles. 2878 : 2879 : I don't know how this impact on performance (compile 2880 : time or runtime). This bug caused the RA (especially on the x86) 2881 : to go thru the potential spill phase when there are still nodes on the 2882 : freeze queue. 2883 : 2884 : 2885 : 2886 : 2887 : ---------------------------------------------------------------------- 2888 : blume 1058 Name: Matthias Blume 2889 : blume 1064 Date: 2002/02/13 22:40:00 EST 2890 : Tag: blume-20020213-fptr-rtti 2891 : Description: 2892 : 2893 : Fixed a bug in ml-nlffigen that was introduced with one of the previous 2894 : updates. 2895 : 2896 : ---------------------------------------------------------------------- 2897 : Name: Matthias Blume 2898 : blume 1063 Date: 2002/02/13 16:41:00 EST 2899 : Tag: blume-20020213-cmlpq 2900 : Description: 2901 : 2902 : Added new priority queue export symbols (which have just been added to 2903 : smlnj-lib.cm) to CML's version of smlnj-lib.cm. (Otherwise CML would 2904 : not compile and the installer would choke.) 2905 : 2906 : ---------------------------------------------------------------------- 2907 : Name: Matthias Blume 2908 : blume 1062 Date: 2002/02/13 16:15:00 EST 2909 : Tag: blume-20020213-various 2910 : Description: 2911 : 2912 : 1. More tweaks to ml-nlffigen: 2913 : 2914 : - better internal datastructures (resulting in slight speedup) 2915 : - "-match" option requires exact match 2916 : - "localized" gensym counters (untagged structs/unions nested within 2917 : other structs/unions or within typedefs get a fresh counter; their 2918 : tag will be prefixed by a concatenation of their parents' tags) 2919 : - bug fixes (related to calculation of transitive closure of types 2920 : to be included in the output) 2921 : 2922 : 2. Minor Basis updates: 2923 : 2924 : - added implementations for List.collate and Option.app 2925 : 2926 : ---------------------------------------------------------------------- 2927 : Name: Matthias Blume 2928 : blume 1060 Date: 2002/02/11 15:55:00 EST 2929 : Tag: blume-20020211-gensym 2930 : Description: 2931 : 2932 : Added a "-gensym" option to command line of ml-nlffigen. This can be 2933 : used to specify a "stem" -- a string that is inserted in all "gensym'd" 2934 : names (ML structure names that correspond to unnamed C structs, unions, 2935 : and enums), so that separate runs of ml-nlffigen do not clash. 2936 : 2937 : ---------------------------------------------------------------------- 2938 : Name: Matthias Blume 2939 : blume 1059 Date: 2002/02/11 12:05:00 EST 2940 : Tag: blume-20020211-gensml 2941 : Description: 2942 : 2943 : A quick fix for a problem with GenSML (in the pgraph-util library): 2944 : Make generation of toplevel "local" optional. (Strictly speaking, 2945 : signature definitions within "local" are not legal SML.) 2946 : 2947 : Other than that: updates to INSTALL and cm/TODO. 2948 : 2949 : ---------------------------------------------------------------------- 2950 : Name: Matthias Blume 2951 : blume 1058 Date: 2002/02/08 15:00:00 EST 2952 : Tag: blume-20020208-uniquepid 2953 : Description: 2954 : 2955 : 0. Version number has been bumped to 110.38.1. NEW BOOTFILES!!! 2956 : 2957 : 1. The installer (config/install.sh) has gotten smarter: 2958 : 2959 : - Configuration options are a bit easier to specify now 2960 : (in config/targets). 2961 : - Bug in recognizing .tar.bz2 files fixed. 2962 : - Installer automatically resolves dependencies between 2963 : configuration options (e.g., if you ask for eXene, you will 2964 : also get cml -- regardless whether you asked for it or not). 2965 : - Installer can run in "quieter mode" by setting the environment 2966 : variable INSTALL_QUIETLY to "true". "Quieter" does not mean 2967 : "completely silent", though. 2968 : - Build HashCons library as part of smlnj-lib. 2969 : 2970 : 2. A new scheme for assigning persistent identifiers to compilation 2971 : units (and, by extension, to types etc.) has been put into place. 2972 : This fixes a long-standing bug where types and even dynamic values 2973 : can get internally confused, thereby compromising type safety 2974 : (abstraction) and dynamic correctness. See 2975 : 2976 : http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz 2977 : 2978 : for an example of how things could go wrong until now. 2979 : 2980 : The downside of the new scheme is that pids are not quite as 2981 : persistent as they used to be: CM will generate a fresh pid 2982 : for every compilation unit that it thinks it sees for the first 2983 : time. That means that if you compile starting from a clean, fresh 2984 : source tree at two different times, you end up with different 2985 : binaries. 2986 : 2987 : Cutoff recompilation, however, has not been compromised because 2988 : CM keeps pid information in special caches between runs. 2989 : 2990 : ---------------------------------------------------------------------- 2991 : george 1053 Name: Lal George 2992 : george 1055 Date: 2002/02/07 15:34:13 EST 2002 2993 : Tag: 2994 : Description: 2995 : 2996 : Compilers that generate assembly code may produce global labels 2997 : whose value is resolved at link time. The various peephole optimization 2998 : modules did not take this in account. 2999 : 3000 : TODO. The Labels.addrOf function should really return an option 3001 : type so that clients are forced to deal with this issue, rather 3002 : than an exception being raised. 3003 : 3004 : ---------------------------------------------------------------------- 3005 : Name: Lal George 3006 : george 1053 Date: 2002/02/06 13:55:02 EST 3007 : Tag: george-20020206-ra-breakup 3008 : Description: 3009 : 3010 : 1. A bug fix from Allen. 3011 : 3012 : A typo causes extra fstp %st(0)'s to be generated at compensation 3013 : edges, which might cause stack underflow traps at runtime. This 3014 : occurs in fft where there are extraneous fstps right before the 'into' 3015 : trap instruction (in this case they are harmless since none of the 3016 : integers overflow.) 3017 : 3018 : george 1054 2. Pulled out various utility modules that were embedded in the modules 3019 : george 1053 of the register allocator. I need these modules for other purposes, but 3020 : george 1054 they are not complete enough to put into a library (just yet). 3021 : george 1053 ---------------------------------------------------------------------- 3022 : blume 1049 Name: Matthias Blume 3023 : Date: 2002/01/31 16:05:00 EST 3024 : Tag: blume-20020131-sparc-ccalls 3025 : Description: 3026 : 3027 : 1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes) 3028 : of extra stack space by mistake. Fixed. 3029 : 3030 : 2. Bug in logic of handling of command-line options in ml-nlffigen fixed. 3031 : 3032 : ---------------------------------------------------------------------- 3033 : leunga 1048 Name: Allen Leung 3034 : Date: 2002/01/30 3035 : Tag: leunga-20020130-nowhere-bug-fix 3036 : Description: 3037 : 3038 : MLRISC bug fixes: 3039 : 1. Fixed a bindings computation bug in the 'nowhere' program generator tool. 3040 : 2. MachineInt.fromString was negating its value. 3041 : 3042 : ---------------------------------------------------------------------- 3043 : blume 1039 Name: Matthias Blume 3044 : blume 1047 Date: 2002/01/29 3045 : Tag: blume-20020129-INSTALL 3046 : Description: 3047 : 3048 : - Added somewhat detailed installation instructions (file INSTALL). 3049 : - Fixed curl-detection bug in config/install.sh. 3050 : - It is now possible to select the URL getter using the URLGETTER 3051 : environment variable: 3052 : 3053 : not set / "unknown" --> automatic detection (script tries wget, 3054 : curl, and lynx) 3055 : "wget" / "curl" / "lynx" --> use the specified program (script "knows" 3056 : how to properly invoke them) 3057 : other --> use $URLGETTER directly, it must take 3058 : precisely two command-line arguments 3059 : (source URL and destination file name) 3060 : 3061 : ---------------------------------------------------------------------- 3062 : Name: Matthias Blume 3063 : blume 1044 Date: 2002/01/28 3064 : blume 1045 Tag: blume-20020128-sparc-ccalls 3065 : Description: 3066 : 3067 : - Fixed problem with calculation of "used" registers in sparc-c-calls. 3068 : - Make use of the allocParam argument in sparc-c-calls. 3069 : 3070 : ---------------------------------------------------------------------- 3071 : Name: Matthias Blume 3072 : Date: 2002/01/28 3073 : blume 1044 Tag: blume-20020128-allocParam 3074 : Description: 3075 : 3076 : John Reppy: Changes c-calls API to accept client-callback for 3077 : allocating extra stack space. 3078 : me: Corresponding changes to mlriscGen (using a dummy argument that 3079 : does not change the current behavior). 3080 : 3081 : ---------------------------------------------------------------------- 3082 : Name: Matthias Blume 3083 : blume 1040 Date: 2002/01/28 12:00:00 3084 : Tag: Release_110_38 3085 : Description: 3086 : 3087 : This time for real!!! 3088 : 3089 : ---------------------------------------------------------------------- 3090 : Name: Matthias Blume 3091 : blume 1039 Date: 2002/01/28 10:56:00 EST 3092 : Tag: blume-20020128-retraction 3093 : Description: 3094 : 3095 : 0. Retracted earlier 110.38. (The Release_110_38 tag has been replaced 3096 : with blume-Release_110_38-retracted.) 3097 : 3098 : 1. Fixed a problem with incorrect rounding modes in real64.sml. 3099 : (Thanks to Andrew Mccreight .) 3100 : 3101 : 2. A bug in ml-nlffigen related to the handling of unnamed structs, unions, 3102 : and enums fixed. The naming of corresponding ML identifiers should 3103 : now be consistent again. 3104 : 3105 : ---------------------------------------------------------------------- 3106 : leunga 1038 Name: Allen Leung 3107 : Date: 2002/01/27 3108 : Tag: leunga-20020127-nowhere 3109 : Description: 3110 : 3111 : Added a target called nowhere in the configuration scripts. 3112 : Enabling this will build the MLRISC 'nowhere' tool (for translating 3113 : programs with where-clauses into legal SML code) during installation. 3114 : 3115 : ---------------------------------------------------------------------- 3116 : blume 1036 Name: Matthias Blume 3117 : leunga 1038 Date: 2002/01/25 21:27:00 EST 3118 : blume 1039 Tag: blume-Release_110_38-retracted 3119 : blume 1037 Description: 3120 : 3121 : Call it a (working) release! Version is 110.38. Bootfiles are ready. 3122 : 3123 : README will be added later. 3124 : 3125 : blume 1039 !!! NOTE: Re-tagged as blume-Release_110_38-retracted. Original tag 3126 : (Release_110_38) removed. Reason: Last-minute bug fixes. 3127 : 3128 : blume 1037 ---------------------------------------------------------------------- 3129 : Name: Matthias Blume 3130 : blume 1036 Date: 2002/01/25 3131 : Tag: blume-20020125-ffi 3132 : Description: 3133 : 3134 : A large number of tweaks and improvements to ml-nlffi-lib and 3135 : ml-nlffigen: 3136 : 3137 : - ML represenation types have been streamlined 3138 : - getter and setter functions work with concrete values, not abstract 3139 : ones where possible 3140 : - ml-nlffigen command line more flexible (see README file there) 3141 : - some bugs have been fixed (hopefully) 3142 : 3143 : ---------------------------------------------------------------------- 3144 : george 1034 Name: Lal George 3145 : Date: 2002/01/24 3146 : Tag: george-20020124-risc-ra-interface 3147 : Description: 3148 : 3149 : There is a dramatic simplification in the interface to the 3150 : register allocator for RISC architectures as a result of making 3151 : parallel copy instructions explicit. 3152 : 3153 : ---------------------------------------------------------------------- 3154 : blume 1028 Name: Matthias Blume 3155 : blume 1032 Date: 2002/01/22 3156 : Tag: blume-20020122-x86-ccalls 3157 : Description: 3158 : 3159 : Bug fix for c-calls on x86 (having to do with how char- and 3160 : short-arguments are being handled). 3161 : 3162 : ---------------------------------------------------------------------- 3163 : Name: Matthias Blume 3164 : blume 1031 Date: 2002/01/21 3165 : Tag: blume-20020121-ff 3166 : Description: 3167 : 3168 : Another day of fiddling with the FFI... 3169 : 3170 : 1. Bug fix/workaround: CKIT does not complain about negative array 3171 : dimensions, so ml-nlffigen has to guard itself against this possibility. 3172 : (Otherwise a negative dimension would send it into an infinite loop.) 3173 : 3174 : 2. Some of the abstract types (light objects, light pointers, most "base" 3175 : types) in structure C are now eqtypes. 3176 : 3177 : 3. Added constructors and test functions for NULL function pointers. 3178 : 3179 : ---------------------------------------------------------------------- 3180 : Name: Matthias Blume 3181 : blume 1028 Date: 2002/01/18 3182 : blume 1030 Tag: blume-20020118-ready-for-new-release 3183 : Description: 3184 : 3185 : Made config/srcarchiveurl point to a new place. (Will provide boot 3186 : files shortly.) 3187 : 3188 : Maybe we christen this to be 110.38? 3189 : 3190 : ---------------------------------------------------------------------- 3191 : Name: Matthias Blume 3192 : Date: 2002/01/18 3193 : blume 1029 Tag: blume-20020118-more-ffifiddle 3194 : Description: 3195 : 3196 : Today's FFI fiddling: 3197 : 3198 : - Provided a structure CGetSet with "convenient" versions of C.Get.* and 3199 : C.Set.* that use concrete (MLRep.*) arguments and results instead 3200 : of abstract ones. 3201 : 3202 : - Provided word-style bit operations etc. for "int" representation 3203 : types in MLRep.SBitops where ranges over Char, Int, Short, 3204 : and Long. 3205 : 3206 : ---------------------------------------------------------------------- 3207 : Name: Matthias Blume 3208 : Date: 2002/01/18 3209 : blume 1028 Tag: blume-20020118-use-x86-fp 3210 : Description: 3211 : 3212 : Now that x86-fast-fp seems to be working, I turned it back on again 3213 : by default. (Seems to work fine now, even with the FFI.) 3214 : 3215 : Other than that, I added some documentation about the FFI to 3216 : src/ml-nlffigen/README and updated the FFI test examples in 3217 : src/ml-nlffi-lib/Tests/*. 3218 : 3219 : ---------------------------------------------------------------------- 3220 : leunga 1025 Name: Allen Leung 3221 : Date: 2002/01/17 3222 : Tag: leunga-20020117-x86-fast-fp-call 3223 : Description: 3224 : 3225 : 1. Fixed a problem with handling return fp values when x86's fast fp 3226 : mode is turned on. 3227 : 3228 : 2. Minor pretty printing fix for cellset. Print %st(0) as %st(0) instead 3229 : of %f32. 3230 : 3231 : 3. Added a constructor INT32lit to the ast of MLRISC tools. 3232 : 3233 : ---------------------------------------------------------------------- 3234 : blume 1021 Name: Matthias Blume 3235 : Date: 2002/01/16 3236 : Tag: blume-20020116-ffifiddle 3237 : Description: 3238 : 3239 : More fiddling with the FFI interface: 3240 : 3241 : - Make constness 'c instead of rw wherever possible. This eliminates 3242 : the need for certain explicit coercions. (However, due to ML's 3243 : value polymorphism, there will still be many cases where explicit 3244 : coercions are necessary. Phantom types are not the whole answer 3245 : to modeling a subtyping relationship in ML.) 3246 : 3247 : - ro/rw coersions for pointers added. (Avoids the detour through */&.) 3248 : 3249 : - "printf" test example added to src/ml-nlffi-lib/Tests. (Demonstrates 3250 : clumsy workaround for varargs problem.) 3251 : 3252 : ---------------------------------------------------------------------- 3253 : george 1016 Name: Lal George 3254 : Date: 2002/01/15 3255 : Tag: 3256 : Description: 3257 : 3258 : 1. Since COPY instructions are no longer native to the architecture, 3259 : a generic functor can be used to implement the expandCopies function. 3260 : 3261 : 2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a 3262 : TEXT segment. 3263 : 3264 : ---------------------------------------------------------------------- 3265 : blume 1011 Name: Matthias Blume 3266 : blume 1015 Date: 2002/01/15 3267 : Tag: blume-20020115-ffiupdates 3268 : Description: 3269 : 3270 : 1. Fix for bug resulting in single-precision float values being returned 3271 : incorrectly from FFI calls. 3272 : 3273 : 2. Small modifications to C FFI API: 3274 : 3275 : - memory-allocation routines return straight objects (no options) 3276 : and raise an exception in out-of-memory situations 3277 : - unsafe extensions to cast between function pointers and pointers 3278 : from/to ints 3279 : - added structure C_Debug as an alternative to structure C where 3280 : pointer-dereferencing (|*| and |*!) always check for null-pointers 3281 : - added open_lib' to DynLinkage; open_lib' works like open_lib 3282 : but also takes a (possibly empty) list of existing library handles 3283 : that the current library depends on 3284 : 3285 : ---------------------------------------------------------------------- 3286 : Name: Matthias Blume 3287 : blume 1011 Date: 2002/01/10 3288 : Tag: blume-20020110-newffigen 3289 : Description: 3290 : 3291 : 1. Updates to portable graph code. 3292 : 3293 : 2. Major update to ml-nlffigen and ml-nlffi-lib. Things are much 3294 : more scalable now so that even huge interfaces such as the one 3295 : for GTK compile in finite time and space. :-) 3296 : See src/ml-nlffigen/README for details on what's new. 3297 : 3298 : ---------------------------------------------------------------------- 3299 : george 1003 Name: Lal George 3300 : george 1009 Date: 2001/01/09 14:31:35 EST 2002 3301 : Tag: george-20011206-rm-native-copy 3302 : Description: 3303 : 3304 : Removed the native COPY and FCOPY instructions 3305 : from all the architectures and replaced it with the 3306 : explicit COPY instruction from the previous commit. 3307 : 3308 : It is now possible to simplify many of the optimizations 3309 : modules that manipulate copies. This has not been 3310 : done in this change. 3311 : 3312 : ---------------------------------------------------------------------- 3313 : Name: Lal George 3314 : george 1003 Date: 2001/12/06 16:50:13 EST 2001 3315 : Tag: george-20011206-mlrisc-instruction 3316 : Description: 3317 : george 824 3318 : george 1003 Changed the representation of instructions from being fully abstract 3319 : to being partially concrete. That is to say: 3320 : 3321 : from 3322 : type instruction 3323 : 3324 : to 3325 : type instr (* machine instruction *) 3326 : 3327 : datatype instruction = 3328 : LIVE of {regs: C.cellset, spilled: C.cellset} 3329 : | KILL of {regs: C.cellset, spilled: C.cellset} 3330 : | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list} 3331 : | ANNOTATION of {i: instruction, a: Annotations.annotation} 3332 : | INSTR of instr 3333 : 3334 : This makes the handling of certain special instructions that appear on 3335 : all architectures easier and uniform. 3336 : 3337 : LIVE and KILL say that a list of registers are live or killed at the 3338 : program point where they appear. No spill code is generated when an 3339 : element of the 'regs' field is spilled, but the register is moved to 3340 : the 'spilled' (which is present, more for debugging than anything else). 3341 : 3342 : LIVE replaces the (now deprecated) DEFFREG instruction on the alpha. 3343 : We used to generate: 3344 : 3345 : DEFFREG f1 3346 : f1 := f2 + f3 3347 : trapb 3348 : 3349 : but now generate: 3350 : 3351 : f1 := f2 + f3 3352 : trapb 3353 : LIVE {regs=[f1,f2,f3], spilled=[]} 3354 : 3355 : Furthermore, the DEFFREG (hack) required that all floating point instruction 3356 : use all registers mentioned in the instruction. Therefore f1 := f2 + f3, 3357 : defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting 3358 : in a cleaner alpha implementation. (Hopefully, intel will not get rid of 3359 : this architecture). 3360 : 3361 : COPYXXX is intended to replace the parallel COPY and FCOPY available on 3362 : all the architectures. This will result in further simplification of the 3363 : register allocator that must be aware of them for coalescing purposes, and 3364 : will also simplify certain aspects of the machine description that provides 3365 : callbacks related to parallel copies. 3366 : 3367 : ANNOTATION should be obvious, and now INSTR represents the honest to God 3368 : machine instruction set! 3369 : 3370 : The /instructions/Instr.sml files define certain utility 3371 : functions for making porting easier -- essentially converting upper case 3372 : to lower case. All machine instructions (of type instr) are in upper case, 3373 : and the lower case form generates an MLRISC instruction. For example on 3374 : the alpha we have: 3375 : 3376 : datatype instr = 3377 : LDA of {r:cell, b:cell, d:operand} 3378 : | ... 3379 : 3380 : val lda : {r:cell, b:cell, d:operand} -> instruction 3381 : ... 3382 : 3383 : where lda is just (INSTR o LDA), etc. 3384 : 3385 : leunga 815 ---------------------------------------------------------------------- 3386 : blume 986 Name: Matthias Blume 3387 : blume 993 Date: 2001/11/22 21:40:00 EST 3388 : Tag: Release_110_37 3389 : Description: 3390 : 3391 : Release 110.37. This time for real. 3392 : 3393 : ---------------------------------------------------------------------- 3394 : Name: Matthias Blume 3395 : blume 988 Date: 2001/11/21 16:35:00 EST 3396 : Tag: blume-20011121-foot-in-mouth 3397 : Description: 3398 : 3399 : Removed the "Release_110_37" tag because of a serious bug. 3400 : This will be re-tagged once the bug is fixed. 3401 : 3402 : ---------------------------------------------------------------------- 3403 : Name: Matthias Blume 3404 : blume 987 Date: 2001/11/21 16:14:00 EST 3405 : Tag: blume-20011121-forgottenfile 3406 : Description: 3407 : 3408 : Forgot to add a file. (Just a .tex-file -- part of 3409 : the CM manual source.) 3410 : 3411 : ---------------------------------------------------------------------- 3412 : Name: Matthias Blume 3413 : blume 986 Date: 2001/11/21 16:10:00 EST 3414 : blume 988 Tag: blume-20011121-invalid_110_37 3415 : blume 986 Description: 3416 : 3417 : blume 988 Note: I removed the original tag "Release_110_37" from this commit 3418 : because we found a serious bug in all non-x86 backends. 3419 : - Matthias 3420 : 3421 : blume 986 1. Modifications to the SML/NJ code generator and to the runtime system 3422 : so that code object name strings are directly inserted into code 3423 : objects at code generation time. The only business the runtime system 3424 : has with this is now to read the name strings on occasions. 3425 : (The encoding of the name string has also changed somewhat.) 3426 : 3427 : 2. CM now implements a simple "set calculus" for specifying export lists. 3428 : In particular, it is now possible to refer to the export lists of 3429 : other libraries/groups/sources and form unions as well as differences. 3430 : See the latest CM manual for details. 3431 : 3432 : 3. An separate notion of "proxy" libraries has again be eliminated from 3433 : CM's model. (Proxy libraries are now simply a special case of using 3434 : the export list calculus.) 3435 : 3436 : 4. Some of the existing libraries now take advantage of the new set 3437 : calculus. 3438 : (Notice that not all libraries have been converted because some 3439 : of the existing .cm-files are supposed to be backward compatible 3440 : with 110.0.x.) 3441 : 3442 : 5. Some cleanup in stand-alone programs. (Don't use "exnMessage" -- use 3443 : "General.exnMessage"! The former relies on a certain hook to be 3444 : initialized, and that often does not happen in the stand-alone case.) 3445 : 3446 : ---------------------------------------------------------------------- 3447 : george 984 Name: Lal George 3448 : Date: 2001/11/21 13:56:18 EST 3449 : Tag: george-2001121-pseudo-ops 3450 : Description: 3451 : 3452 : Implemented a complete redesign of MLRISC pseudo-ops. Now there 3453 : ought to never be any question of incompatabilities with 3454 : pseudo-op syntax expected by host assemblers. 3455 : 3456 : For now, only modules supporting GAS syntax are implemented 3457 : but more should follow, such as MASM, and vendor assembler 3458 : syntax, e.g. IBM as, Sun as, etc. 3459 : 3460 : ---------------------------------------------------------------------- 3461 : blume 977 Name: Matthias Blume 3462 : Date: 2001/11/14 11:52:00 EST 3463 : Tag: blume-20011114-srcname 3464 : Description: 3465 : 3466 : 1. Routed the name of the current source file to mlriscgen where it 3467 : should be directly emitted into the code object. (This last part 3468 : is yet to be done.) 3469 : 3470 : 2. Some cleanup of the pgraph code to make it match the proposal that 3471 : I put out the other day. (The proposal notwithstanding, things are 3472 : still in flux here.) 3473 : 3474 : ---------------------------------------------------------------------- 3475 : george 976 Name: Lal George 3476 : Date: 2001/11/14 09:44:04 EST 3477 : Tag: 3478 : Description: 3479 : 3480 : Fix for a backpatching bug reported by Allen. 3481 : 3482 : Because the boundary between short and long span-dependent 3483 : instructions is +/- 128, there are an astounding number of 3484 : span-dependent instructions whose size is over estimated. 3485 : 3486 : Allen came up with the idea of letting the size of span 3487 : dependent instructions be non-monotonic, for a maxIter 3488 : number of times, after which the size must be monotonically 3489 : increasing. 3490 : 3491 : This table shows the number of span-dependent instructions 3492 : whose size was over-estimated as a function of maxIter, for the 3493 : file Parse/parse/ml.grm.sml: 3494 : 3495 : maxIter # of instructions: 3496 : 10 687 3497 : 20 438 3498 : 30 198 3499 : 40 0 3500 : 3501 : In compiling the compiler, there is no significant difference in 3502 : compilation speed between maxIter=10 and maxIter=40. Actually, 3503 : my measurements showed that maxIter=40 was a tad faster than 3504 : maxIter=10! Also 96% of the files in the compiler reach a fix 3505 : point within 13 iterations, so fixing maxIter at 40, while high, 3506 : is okay. 3507 : 3508 : ---------------------------------------------------------------------- 3509 : blume 975 Name: Matthias Blume 3510 : Date: 2001/10/31 15:25:00 EST 3511 : Tag: blume-20011031-pgraph 3512 : Description: 3513 : 3514 : CKIT: 3515 : * Changed the "Function" constructor of type Ast.ctype to carry optional 3516 : argument identifiers. 3517 : * Changed the return type of TypeUtil.getFunction accordingly. 3518 : * Type equality ignores the argument names. 3519 : * TypeUtil.composite tries to preserve argument names but gives up quickly 3520 : if there is a mismatch. 3521 : 3522 : installation script: 3523 : * attempts to use "curl" if available (unless "wget" is available as well) 3524 : 3525 : CM: 3526 : * has an experimental implementation of "portable graphs" which I will 3527 : soon propose as an implementation-independent library format 3528 : * there are also new libraries$/pgraph.cm and $/pgraph-util.cm 3529 : 3530 : NLFFI-LIB: 3531 : * some cleanup (all cosmetic) 3532 : 3533 : NLFFIGEN: 3534 : * temporarily disabled the mechanism that suppresses ML output for 3535 : C definitions whose identifiers start with an underscore character 3536 : * generate val bindings for enum constants 3537 : * user can request that only one style (light or heavy) is being used; 3538 : default is to use both (command-line arguments: -heavy and -light) 3539 : * fixed bug in handling of function types involving incomplete pointers 3540 : * generate ML entry points that take record arguments (i.e., using 3541 : named arguments) for C functions that have a prototype with named 3542 : arguments 3543 : (see changes to CKIT) 3544 : 3545 : ---------------------------------------------------------------------- 3546 : leunga 972 Name: Allen Leung 3547 : Date: 2001/10/27 20:34:00 EDT 3548 : Tag: leunga-20011027-x86-fast-fp-call 3549 : Description: 3550 : 3551 : Fixed the bug described in blume-20010920-slowfp. 3552 : 3553 : The fix involves 3554 : 1. generating FCOPYs in FSTP in ia32-svid 3555 : 2. marking a CALL with the appropriate annotation 3556 : 3557 : ---------------------------------------------------------------------- 3558 : blume 961 Name: Matthias Blume 3559 : Date: 2001/10/16 11:32:00 EDT 3560 : Tag: blume-20011016-netbsd 3561 : Description: 3562 : 3563 : Underscore patch from Chris Richards (fixing problem with compiling 3564 : runtime system under recent NetBSD). 3565 : 3566 : ---------------------------------------------------------------------- 3567 : george 959 Name: Allen Leung 3568 : Date: 2001/10/12 17:18:32 EDT 2001 3569 : Tag: leung-20011012-x86-printflowgraph 3570 : Description: 3571 : 3572 : X86RA now uses a valid (instead of dummy) PrintFlowgraph module. 3573 : 3574 : ---------------------------------------------------------------------- 3575 : george 952 Name: Lal George 3576 : george 958 Date: 2001/10/11 23:51:34 EDT 3577 : Tag: george-20011011-too-many-instrs 3578 : Description: 3579 : 3580 : The representation of a program point never expected to see more 3581 : than 65536 instructions in a basic block! 3582 : 3583 : ---------------------------------------------------------------------- 3584 : Name: Lal George 3585 : george 952 Date: 2001/10/09 09:41:37 EDT 3586 : Tag: george-20011008-mlrisc-labels 3587 : Description: 3588 : 3589 : Changed the machine description files to support printing of 3590 : local and global labels in assembly code, based on host assembler 3591 : conventions. 3592 : 3593 : ---------------------------------------------------------------------- 3594 : blume 916 Name: Matthias Blume 3595 : blume 939 Date: 2001/09/25 15:25:00 EDT 3596 : Tag: blume-20010925-exninfo 3597 : Description: 3598 : 3599 : I provided a non-hook implementation of exnName (at the toplevel) and 3600 : made the "dummy" implementation of exnMessage (at the toplevel) more 3601 : useful: if nothing gets "hooked in", then at least you are going to 3602 : see the exception name and a message indicating why you don't see more. 3603 : 3604 : [For the time being, programs that need exnMessage and want to use 3605 : ml-build should either use General.exnMessage (strongly recommended) or 3606 : refer to structure General at some other point so that CM sees a 3607 : static dependency.] 3608 : 3609 : [Similar remarks go for "print" and "use": If you want to use their 3610 : functionality in stand-alone programs generated by ml-build, then use 3611 : TextIO.output and Backend.Interact.useFile (from$smlnj/compiler.cm).] 3612 : 3613 : ---------------------------------------------------------------------- 3614 : Name: Matthias Blume 3615 : blume 938 Date: 2001/09/20 17:28:00 EDT 3616 : Tag: blume-20010920-slowfp 3617 : Description: 3618 : 3619 : Allen says that x86-fast-fp is not safe yet, so I turned it off again... 3620 : 3621 : ---------------------------------------------------------------------- 3622 : Name: Matthias Blume 3623 : blume 937 Date: 2001/09/20 17:20:00 EDT 3624 : Tag: blume-20010920-canonicalpaths 3625 : Description: 3626 : 3627 : 0. Updated the BOOT file (something that I forgot to do earlier). 3628 : 3629 : 1. Small internal change to CM so that it avoids "/../" in filenames 3630 : as much as possible (but only where it is safe). 3631 : 3632 : 2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so 3633 : that arguments that contain delimiters are passed through correctly. 3634 : This change also means that all "special" arguments of the form 3635 : @SMLxxx... must come first. 3636 : 3637 : 3. Changed install script to put relative anchor names for tool commands 3638 : into pathconfig. 3639 : 3640 : ---------------------------------------------------------------------- 3641 : Name: Matthias Blume 3642 : blume 932 Date: 2001/09/18 15:35:00 EDT 3643 : Tag: blume-20010918-readme11036 3644 : Description: 3645 : 3646 : Added README files. 3647 : 3648 : ---------------------------------------------------------------------- 3649 : Name: Matthias Blume 3650 : blume 929 Date: 2001/09/18 11:45:00 EDT 3651 : Tag: Release_110_36 (retag) 3652 : Description: 3653 : 3654 : Fixed mistake in config/preloads. Retagged as 110.36. 3655 : 3656 : ---------------------------------------------------------------------- 3657 : Name: Matthias Blume 3658 : blume 928 Date: 2001/09/18 09:40:00 EDT 3659 : blume 929 Tag: Release_110_36_orig (tag changed) 3660 : blume 928 Description: 3661 : 3662 : New version (110.36). New bootfiles. 3663 : 3664 : ---------------------------------------------------------------------- 3665 : Name: Matthias Blume 3666 : blume 927 Date: 2001/09/14 16:15:00 EDT 3667 : Tag: blume-20010914-x86fastfp 3668 : Description: 3669 : 3670 : John committed some changes that Allen made, in particular a (hopefully) 3671 : correctly working version of the x86-fp module. 3672 : 3673 : I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp" 3674 : flag to "true". Everything seems to compile to a fixpoint ok, and 3675 : "mandelbrot" speeds up by about 15%. 3676 : 3677 : ---------------------------------------------------------------------- 3678 : Name: Matthias Blume 3679 : blume 923 Date: 2001/09/13 11:20:00 EDT 3680 : Tag: blume-20010913-minimal 3681 : Description: 3682 : 3683 : 1. Stefan Monnier's patch to fix a miscompilation problem that 3684 : was brought to light by John Reppy's work on Moby. 3685 : 3686 : 2. Implemented a minimal "structure Compiler" that contains just 3687 : "version" and "architecture". The minimal version will be 3688 : available when the full version is not. This is for backward- 3689 : compatibility with code that wants to test Compiler.version. 3690 : 3691 : ---------------------------------------------------------------------- 3692 : Name: Matthias Blume 3693 : blume 920 Date: 2001/08/28 14:03:00 EDT 3694 : Tag: blume-20010828-ml-lex 3695 : Description: 3696 : 3697 : Fix for bug 1581, received from Neophytos Michael. 3698 : 3699 : ---------------------------------------------------------------------- 3700 : Name: Matthias Blume 3701 : blume 918 Date: 2001/08/27 11:20:00 EDT 3702 : Tag: blume-20010827-readme11035 3703 : Description: 3704 : 3705 : Fleshed out the README file for 110.35. 3706 : 3707 : ---------------------------------------------------------------------- 3708 : Name: Matthias Blume 3709 : blume 916 Date: 2001/08/24 17:10:00 EDT 3710 : Tag: Release_110_35 3711 : Description: 3712 : 3713 : New version number (110.35). New bootfiles. 3714 : 3715 : ---------------------------------------------------------------------- 3716 : george 915 Name: Lal George 3717 : george 909 Date: 2001/08/24 13:47:18 EDT 2001 3718 : Tag: george-20010824-MLRISC-graphs 3719 : Description: 3720 : 3721 : removed clusters from MLRISC completely and replaced with graphs. 3722 : 3723 : ---------------------------------------------------------------------- 3724 : Name: Matthias Blume 3725 : blume 905 Date: 2001/08/23 17:50:00 EDT 3726 : Tag: blume-20010823-toplevel 3727 : Description: 3728 : 3729 : - some reorganization of the code that implements various kinds of 3730 : environments in the compiler (static, dynamic, symbolic, combined) 3731 : - re-implemented the EnvRef module so that evalStream works properly 3732 : (if the stream contains references to "use", "CM.make", etc.) 3733 : - cleaned up evalloop.sml and interact.sml (but they need more cleaning) 3734 : 3735 : ---------------------------------------------------------------------- 3736 : Name: Matthias Blume 3737 : blume 904 Date: 2001/08/20 15:50 EDT 3738 : Tag: blume20010820-slipup 3739 : Description: 3740 : 3741 : I forgot to commit a few files. Here they are... 3742 : 3743 : ---------------------------------------------------------------------- 3744 : Name: Matthias Blume 3745 : blume 903 Date: 2001/08/20 15:35:00 EDT 3746 : Tag: blume-20010820-debugprof 3747 : Description: 3748 : 3749 : !!!! NEW BOOTFILES !!!! 3750 : 3751 : This is another round of reorganizing the compiler sources. This 3752 : time the main goal was to factor out all the "instrumentation" 3753 : passes (for profiling and backtracing) into their own library. 3754 : The difficulty was to do it in such a way that it does not depend 3755 : on elaborate.cm but only on elabdata.cm. 3756 : 3757 : Therefore there have been further changes to both elaborate.cm and 3758 : elabdata.cm -- more "generic" things have been moved from the former 3759 : to the latter. As a result, I was forced to split the assignment 3760 : of numbers indicating "primtyc"s into two portions: SML-generic and 3761 : SML/NJ-specific. Since it would have been awkward to maintain, 3762 : I bit the bullet and actually _changed_ the mapping between these 3763 : numbers and primtycs. The bottom line of this is that you need 3764 : a new set of bin- and bootfiles. 3765 : 3766 : I have built new bootfiles for all architectures, so doing a fresh 3767 : checkout and config/install.sh should be all you need. 3768 : 3769 : The newly created library's name is 3770 : 3771 : $smlnj/viscomp/debugprof.cm 3772 : 3773 : and its sources live under 3774 : 3775 : src/compiler/DebugProf 3776 : 3777 : ---------------------------------------------------------------------- 3778 : Name: Matthias Blume 3779 : blume 902 Date: 2001/08/15 17:15:00 EDT 3780 : Tag: blume-20010815-compreorg 3781 : Description: 3782 : 3783 : This is a first cut at reorganizing the CM libraries that make up the 3784 : core of the compiler. The idea is to separate out pieces that could 3785 : be used independently by tools, e.g., the parser, the typechecker, etc. 3786 : 3787 : The current status is a step in this direction, but it is not quite 3788 : satisfactory yet. Expect more changes in the future. 3789 : 3790 : Here is the current (new) organization... 3791 : 3792 : What used to be$smlnj/viscomp/core.cm is now divided into 3793 : six CM libraries: 3794 : 3795 : $smlnj/viscomp/basics.cm 3796 : /parser.cm 3797 : /elabdata.cm 3798 : /elaborate.cm 3799 : /execute.cm 3800 : /core.cm 3801 : 3802 : The CM files for these libraries live under src/system/smlnj/viscomp. 3803 : All these libraries are proxy libraries that contain precisely 3804 : one CM library component. Here are the locations of the components 3805 : (all within the src/compiler tree): 3806 : 3807 : Basics/basics.cm 3808 : Parse/parser.cm 3809 : ElabData/elabdata.cm 3810 : Elaborator/elaborate.cm 3811 : Execution/execute.cm 3812 : core.cm 3813 : 3814 : [This organization is the same that has been used already 3815 : for a while for the architecture-specific parts of the visible 3816 : compiler and for the old version of core.cm.] 3817 : 3818 : As you will notice, many source files have been moved from their 3819 : respective original locations to a new home in one of the above 3820 : subtrees. 3821 : 3822 : The division of labor between the new libraries is the following: 3823 : 3824 : basics.cm: 3825 : - Simple, basic definitions that pertain to many (or all) of 3826 : the other libraries. 3827 : parser.cm: 3828 : - The SML parser, producing output of type Ast.dec. 3829 : - The type family for Ast is also defined and exported here. 3830 : elabdata.cm: 3831 : - The datatypes that describe input and output of the elaborator. 3832 : This includes types, absyn, and static environments. 3833 : elaborator.cm: 3834 : - The SML/NJ type checker and elaborator. 3835 : This maps an Ast.dec (with a given static environment) to 3836 : an Absyn.dec (with a new static environment). 3837 : - This libraries implements certain modules that used to be 3838 : structures as functors (to remove dependencies on FLINT). 3839 : execute.cm: 3840 : - Everything having to do with executing binary code objects. 3841 : - Dynamic environments. 3842 : core.cm: 3843 : - SML/NJ-specific instantiations of the elaborator and MLRISC. 3844 : - Top-level modules. 3845 : - FLINT (this should eventually become its own library) 3846 : 3847 : Notes: 3848 : 3849 : I am not 100% happy with the way I separated the elaborator (and its 3850 : data structures) from FLINT. Two instances of the same problem: 3851 : 3852 : 1. Data structures contain certain fields that carry FLINT-specific 3853 : information. I hacked around this using exn and the property list 3854 : module from smlnj-lib. But the fact that there are middle-end 3855 : specific fields around at all is a bit annoying. 3856 : 3857 : 2. The elaborator calculates certain FLINT-related information. I tried 3858 : to make this as abstract as I could using functorization, but, again, 3859 : the fact that the elaborator has to perform calculations on behalf 3860 : of the middle-end at all is not nice. 3861 : 3862 : 3. Having to used exn and property lists is unfortunate because it 3863 : weakens type checking. The other alternative (parameterizing 3864 : nearly *everything*) is not appealing, though. 3865 : 3866 : I removed the "rebinding =" warning hack because due to the new organization 3867 : it was awkward to maintain it. As a result, the compiler now issues some of 3868 : these warnings when compiling init.cmi during bootstrap compilation. On 3869 : the plus side, you also get a warning when you do, for example: 3870 : val op = = Int32.+ 3871 : which was not the case up to now. 3872 : 3873 : I placed "assign" and "deref" into the _Core structure so that the 3874 : code that deals with the "lazy" keyword can find them there. This 3875 : removes the need for having access to the primitive environment 3876 : during elaboration. 3877 : 3878 : ---------------------------------------------------------------------- 3879 : Name: Matthias Blume 3880 : blume 898 Date: 2001/08/13 3881 : Tag: blume-20010813-closures 3882 : Description: 3883 : 3884 : This fix was sent to us by Zhong Shao. It is supposed to improve the 3885 : performance of certain loops by avoiding needless closure allocation. 3886 : 3887 : ---------------------------------------------------------------------- 3888 : george 890 Name: Lal George 3889 : george 896 Date: 2001/07/31 10:03:23 EDT 2001 3890 : Tag: george-20010731-x86-fmalloc 3891 : Description: Fixed bug in x86 calls 3892 : 3893 : There was a bug where call instructions would mysteriously 3894 : vanish. The call instruction had to be one that returned 3895 : a floating point value. 3896 : 3897 : ---------------------------------------------------------------------- 3898 : Name: Lal George 3899 : george 890 Date: 2001/07/19 16:36:29 EDT 2001 3900 : Tag: george-20010719-simple-cells 3901 : Description: 3902 : 3903 : I have dramatically simplified the interface for CELLS in MLRISC. 3904 : 3905 : In summary, the cells interface is broken up into three parts: 3906 : 3907 : 1. CellsBasis : CELLS_BASIS 3908 : 3909 : CellsBasis is a top level structure and common for all 3910 : architectures. it contains the definitions of basic datatypes 3911 : and utility functions over these types. 3912 : 3913 : 2. functor Cells() : CELLS 3914 : 3915 : Cells generates an interface for CELLS that incorporates the 3916 : specific resources on the target architecture, such as the 3917 : presence of special register classes, their number and size, 3918 : and various useful substructures. 3919 : 3920 : 3. CELLS 3921 : 3922 : e.g. SparcCells: SPARCCELLS 3923 : 3924 : CELLS usually contains additional bindings for special 3925 : registers on the architecture, such as: 3926 : 3927 : val r0 : cell (* register zero *) 3928 : val y : cell (* Y register *) 3929 : val psr : cell (* processor status register *) 3930 : ... 3931 : 3932 : The structure returned by applying the Cells functor is opened 3933 : in this interface. 3934 : 3935 : The main implication of all this is that the datatypes for cells is 3936 : split between CellsBasis and CELLS -- a fairly simple change for user 3937 : code. 3938 : 3939 : In the old scheme the CELLS interface had a definitional binding of 3940 : the form: 3941 : 3942 : signature CELLS = sig 3943 : 3944 : structure CellsBasis = CellsBasis 3945 : 3946 : ... 3947 : 3948 : end 3949 : 3950 : With all the sharing constraints that goes on in MLRISC, this old 3951 : design quickly leads to errors such as: 3952 : 3953 : "structure definition spec inside of sharing ... " 3954 : 3955 : 3956 : and appears to require an unacceptable amount of sharing and where 3957 : constraint hackery. 3958 : 3959 : I think this error message (the interaction of definitional specs and 3960 : sharing) requires more explanation on our web page. 3961 : 3962 : ---------------------------------------------------------------------- 3963 : blume 826 Name: Matthias Blume 3964 : blume 879 Date: 2001/07/19 15:00:00 EDT 3965 : Tag: blume-20010719-libreorg 3966 : Description: 3967 : 3968 : This update puts together a fairly extensive but straightforward change 3969 : to the way the libraries that implement the interactive system are 3970 : organized: 3971 : 3972 : The biggest change is the elimination of structure Compiler. As a 3973 : replacement for this structure, there is now a CM library 3974 : (known as$smlnj/compiler.cm or $smlnj/compiler/current.cm) 3975 : that exports all the substructures of the original structure Compiler 3976 : directly. So instead of saying Compiler.Foo.bar one now simply 3977 : says Foo.bar. (The CM libraries actually export a collection of 3978 : structures that is richer than the collection of substructures of 3979 : structure Compiler.) 3980 : 3981 : To make the transition smooth, there is a separate library called 3982 :$smlnj/compiler/compiler.cm which puts together and exports the 3983 : original structure Compiler (or at least something very close to it). 3984 : 3985 : There are five members of the original structure Compiler 3986 : that are not exported directly but which instead became members 3987 : of a new structure Backend (described by signature BACKEND). These are: 3988 : structure Profile (: PROFILE), structure Compile (: COMPILE), structure 3989 : Interact (: INTERACT), structure Machine (: MACHINE), and val 3990 : architecture (: string). 3991 : 3992 : Structure Compiler.Version has become structure CompilerVersion. 3993 : 3994 : Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided 3995 : by $smlnj/compiler/.cm where is alpha32, hppa, ppc, sparc, 3996 : or x86, respectively. 3997 : Each of these exports the same frontend structures that 3998 :$smlnj/compiler.cm exports. But they do not have a structure Backend 3999 : and instead export some structure Backend where is Alpha32, 4000 : Hppa, PPC, Sparc, or X86, respectively. 4001 : 4002 : Library $smlnj/compiler/all.cm exports the union of the exports of 4003 :$smlnj/compiler/.cm 4004 : 4005 : There are no structures Compiler anymore, use 4006 : $smlnj/compiler/.cm instead. 4007 : 4008 : Library host-compiler-0.cm is gone. Instead, the internal library 4009 : that instantiates CM is now called cm0.cm. Selection of the host 4010 : compiler (backend) is no longer done here but. (Responsibility for it 4011 : now lies with$smlnj/compiler/current.cm. This seems to be more 4012 : logical.) 4013 : 4014 : Many individual files have been moved or renamed. Some files have 4015 : been split into multiple files, and some "dead" files have been deleted. 4016 : 4017 : Aside from these changes to library organization, there are also changes 4018 : to the way the code itself is organized: 4019 : 4020 : Structure Binfile has been re-implemented in such a way that it no 4021 : longer needs any knowledge of the compiler. It exclusively deals 4022 : with the details of binfile layout. It no longer invokes the 4023 : compiler (for the purpose of creating new prospective binfile 4024 : content), and it no longer has any knowledge of how to interpret 4025 : pickles. 4026 : 4027 : Structure Compile (: COMPILE) has been stripped down to the bare 4028 : essentials of compilation. It no longer deals with linking/execution. 4029 : The interface has been cleaned up considerably. 4030 : 4031 : Utility routines for dealing with linking and execution have been 4032 : moved into their own substructures. 4033 : 4034 : (The ultimate goal of these changes is to provide a light-weight 4035 : binfile loader/linker (at least for, e.g., stable libraries) that 4036 : does not require CM or the compiler to be present.) 4037 : 4038 : CM documentation has been updated to reflect the changes to library 4039 : organization. 4040 : 4041 : ---------------------------------------------------------------------- 4042 : Name: Matthias Blume 4043 : blume 875 Date: 2001/07/10 17:30:00 EDT 4044 : Tag: Release_110_34 4045 : Description: 4046 : 4047 : blume 876 Minor tweak to 110.34 (re-tagged): 4048 : 4049 : - README.html file added to CVS repository 4050 : - runtime compiles properly under FreeBSD 3.X and 4.X 4051 : 4052 : ---------------------------------------------------------------------- 4053 : Name: Matthias Blume 4054 : Date: 2001/07/10 17:30:00 EDT 4055 : Tag: Release_110_34 4056 : Description: 4057 : 4058 : blume 875 New version number (110.34). New bootfiles. 4059 : 4060 : ---------------------------------------------------------------------- 4061 : Name: Matthias Blume 4062 : blume 874 Date: 2001/07/09 16:00:00 EDT 4063 : Tag: blume-20010709-more-varargs 4064 : Description: 4065 : 4066 : I changed the handling of varargs in ml-nlffigen again: 4067 : The ellipsis ... will now simply be ignored (with an accompanying warning). 4068 : 4069 : The immediate effect is that you can actually call a varargs function 4070 : from ML -- but you can't actually supply any arguments beyond the ones 4071 : specified explicitly. (For example, you can call printf with its format 4072 : string, but you cannot pass additional arguments.) 4073 : 4074 : This behavior is only marginally more useful than the one before, but 4075 : it has the advantage that a function or, more importantly, a function 4076 : type never gets dropped on the floor, thus avoiding follow-up problems with 4077 : other types that refer to the offending one. 4078 : 4079 : ---------------------------------------------------------------------- 4080 : Name: Matthias Blume 4081 : blume 873 Date: 2001/07/09 11:25:00 EDT 4082 : Tag: blume-20010709-varargs 4083 : Description: 4084 : 4085 : 1. ckit-lib.cm now exports structure Error 4086 : 2. ml-nlffigen reports occurences of "..." (i.e., varargs function types) 4087 : with a warning accompanied by a source location. Moreover, it 4088 : merely skips the offending function or type and proceeds with the 4089 : rest of its work.u As a result, one can safely feed C code containing 4090 : "..." to ml-nlffigen. 4091 : 3. There are some internal improvements to CM, providing slightly 4092 : more general string substitutions in the tools subsystem. 4093 : 4094 : ---------------------------------------------------------------------- 4095 : Name: Matthias Blume 4096 : blume 854 Date: 2001/06/27 15:10:00 EDT 4097 : Tag: blume-20010627-concur 4098 : Description: 4099 : 4100 : Fixed a small bug in CM's handling of parallel compilation. 4101 : (You could observe the bug by Control-C-interrupting an ordinary 4102 : CMB.make or CM.stabilize and then attaching some compile servers. 4103 : The result was that all of a sudden the previously interrupted 4104 : compilation would continue on its own. This was because of 4105 : an over-optimization: CM did not bother to clean out certain queues 4106 : when no servers were attached "anyway", resulting in the contents 4107 : of these queues to grab control when new servers did get attached.) 4108 : 4109 : There is also another minor update to the CM manual. 4110 : 4111 : ---------------------------------------------------------------------- 4112 : Name: Matthias Blume 4113 : blume 853 Date: 2001/06/26 16:15:00 EDT 4114 : Tag: blume-20010626-cmdoc 4115 : Description: 4116 : 4117 : Minor typo fixed in CM manual (syntax diagram for libraries). 4118 : 4119 : ---------------------------------------------------------------------- 4120 : Name: Matthias Blume 4121 : blume 852 Date: 2001/06/25 22:55:00 EDT 4122 : Tag: blume-20010625-x86pc 4123 : Description: 4124 : 4125 : Fixed a nasty bug in the X86 assembly code that caused signal 4126 : handlers to fail (crash) randomly. 4127 : 4128 : ---------------------------------------------------------------------- 4129 : Name: Matthias Blume 4130 : blume 846 Date: 2001/06/25 12:05:00 EDT 4131 : Tag: blume-20010625-nlffigen 4132 : Description: 4133 : 4134 : This update fixes a number of minor bugs in ml-nlffigen as reported by 4135 : Nick Carter . 4136 : 4137 : 1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted. 4138 : 2. Default names for generated files are now derived from the name of 4139 : the C file *without its directory*. In particular, this causes generated 4140 : files to be placed locally even if the C file is in some system directory. 4141 : 3. Default names for generated signatures and structures are also derived 4142 : from the C file name without its directory. This avoids silly things 4143 : like "structure GL/GL". 4144 : (Other silly names are still possible because ml-nlffigen does not do 4145 : blume 847 a thorough check of whether generated names are legal ML identifiers. 4146 : When in doubt, use command line arguments to force particular names.) 4147 : blume 846 4148 : ---------------------------------------------------------------------- 4149 : Name: Matthias Blume 4150 : blume 845 Date: 2001/06/21 12:25:00 EDT 4151 : Tag: blume-20010621-eXene 4152 : Description: 4153 : 4154 : eXene now compiles and (sort of) works again. 4155 : 4156 : The library name (for version > 110.33) is $/eXene.cm. 4157 : 4158 : I also added an new example in src/eXene/examples/nbody. See the 4159 : README file there for details. 4160 : 4161 : ---------------------------------------------------------------------- 4162 : Name: Matthias Blume 4163 : blume 844 Date: 2001/06/20 16:40:00 EDT 4164 : Tag: blume-20010620-cml 4165 : Description: 4166 : 4167 : CML now compiles and works again. 4168 : 4169 : Libraries (for version > 110.33): 4170 : 4171 :$cml/cml.cm Main CML library. 4172 : $cml/basis.cm CML's version of$/basis.cm. 4173 : $cml/cml-internal.cm Internal helper library. 4174 :$cml/core-cml.cm Internal helper library. 4175 : $cml-lib/trace-cml.cm Tracing facility. 4176 :$cml-lib/smlnj-lib.cm CML's version of \$/smlnj-lib.cm 4177 : 4178 : The installer (config/install.sh) has been taught how to properly 4179 : install this stuff. 4180 : 4181 : ---------------------------------------------------------------------- 4182 : Name: Matthias Blume 4183 : blume 843 Date: 2001/06/19 17:55:00 EDT 4184 : Tag: blume-20010619-instantiate 4185 : Description: 4186 : 4187 : This un-breaks the fix for bug 1432. 4188 : (The bug was originally fixed in 110.9 but I broke it again some 4189 : time after that.) 4190 : 4191 : ---------------------------------------------------------------------- 4192 : Name: Matthias Blume 4193 : blume 842 Date: 2001/06/19 17:25:00 EDT 4194 : Tag: blume-20010619-signals 4195 : Description: 4196 : 4197 : This should (hopefully) fix the long-standing signal handling bug. 4198 : (The runtime system was constructing a continuation record with an 4199 : incorrect descriptor which would cause the GC to drop data on the floor...) 4200 : 4201 : ---------------------------------------------------------------------- 4202 : Name: Matthias Blume 4203 : blume 840 Date: 2001/06/15 15:05:00 EDT 4204 : blume 841 Tag: blume-20010615-moresparc 4205 : Description: 4206 : 4207 : Here is a short late-hour update related to Sparc c-calls: 4208 : 4209 : -- made handling of double-word arguments a bit smarter 4210 : 4211 : -- instruction selection phase tries to collapse certain clumsily 4212 : constructed ML-Trees; typical example: 4213 : 4214 : ADD(ty,ADD(_,e,LI d1),LI d2) -> ADD(ty,e,LI(d1+d2)) 4215 : 4216 : This currently has no further impact on SML/NJ since mlriscGen does 4217 : not seem to generate such patterns in the first place, and c-calls 4218 : (which did generate them in the beginning) has meanwhile been fixed 4219 : so as to avoid them as well. 4220 : 4221 : ---------------------------------------------------------------------- 4222 : Name: Matthias Blume 4223 : Date: 2001/06/15 15:05:00 EDT 4224 : blume 840 Tag: blume-20010615-sparc 4225 : Description: 4226 : 4227 : The purpose of this update is to provide an implementation of NLFFI 4228 : on Sparc machines. 4229 : 4230 : Here are the changes in detail: 4231 : 4232 : * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing 4233 : the Sparc implementation of the c-calls API. 4234 : * The Sparc backend of SML/NJ has been modified to uniformely use %fp 4235 : for accessing the ML frame. Thus, we have a real frame pointer and 4236 : can freely modify %sp without need for an omit-frame-ptr phase. 4237 : The vfp logic in src/compiler/CodeGen/* has been changed to accomodate 4238 : this case. 4239 : * ml-nlffigen has been taught to produce code for different architectures 4240 : and calling conventions. 4241 : * In a way similar to what was done in the x86 case, the Sparc 4242 : backend uses its own specific extension to mltree. (For example, 4243 : it needs to be able to generate UNIMP instructions which are part 4244 : of the calling convention.) 4245 : * ml-nlffi-lib was reorganized to make it more modular (in particular, 4246 : to make it easier to plug in new machine- and os-dependent parts). 4247 : 4248 : There are some other fairly unrelated bug fixes and cleanups as well: 4249 : 4250 : * I further hacked the .cm files for MLRISC tools (like MDLGen) so 4251 : that they properly share their libraries with existing SML/NJ libraries. 4252 : * I fixed a minor cosmetic bug in CM, supressing certain spurious 4253 : follow-up error messages. 4254 : * Updates to CM/CMB documentation. 4255 : 4256 : TODO items: 4257 : 4258 : * MLRISC should use a different register as its asmTemp on the Sparc. 4259 : (The current %o2 is a really bad choice because it is part of the 4260 : calling conventions, so things might interfere in unexpected ways.) 4261 : 4262 : ---------------------------------------------------------------------- 4263 : Name: Matthias Blume 4264 : blume 839 Date: 2001/06/07 4265 : Tag: blume-20010607-calls 4266 : Description: 4267 : 4268 : A number of internal changes related to C calls and calling conventions: 4269 : 4270 : 1. ML-Tree CALL statements now carry a "pops" field. It indicates the 4271 : number of bytes popped implicitly (by the callee). In most cases 4272 : this field is 0 but on x86/win32 it is some non-zero value. This 4273 : is information provided for the benefit of the "omit-frameptr" pass. 4274 : 2. The CALL instruction on the x86 carries a similar "pops" field. 4275 : The instruction selection phase copies its value from the ML-Tree 4276 : CALL statement. 4277 : 3. On all other architectures, the instruction selection phase checks 4278 : whether "pops=0" and complains if not. 4279 : 4. The c-calls implementation for x86 now accepts two calling conventions: 4280 : "ccall" and "stdcall". When "ccall" is selected, the caller cleans 4281 : up after the call and pops is set to 0. For "stdcall", the caller 4282 : does nothing, leaving the cleanup to the callee; pops is set to 4283 : the number of bytes that were pushed onto the stack. 4284 : 5. The cproto decoder (compiler/Semant/types/cproto.sml) now can 4285 : distinguish between "ccall" and "stdcall". 4286 : 6. The UNIMP instruction has been added to the supported Sparc instruction 4287 : set. (This is needed for implementing the official C calling convention 4288 : on this architecture.) 4289 : 7. I fixed some of the .cm files under src/MLRISC/Tools to make them 4290 : work with the latest CM. 4291 : 4292 : ---------------------------------------------------------------------- 4293 : Name: Matthias Blume 4294 : blume 838 Date: 2001/06/05 15:10:00 EDT 4295 : Tag: blume-20010605-cm-index 4296 : Description: 4297 : 4298 : 0. The "lambdasplit" parameter for class "sml" in CM has been documented. 4299 : 4300 : 1. CM can now generate "index files". These are human-readable files 4301 : that list on a per-.cm-file basis each toplevel symbol defined or 4302 : imported. The location of the index file for 4303 :