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

# SCM Repository

[smlnj] Diff of /sml/trunk/HISTORY
 [smlnj] / sml / trunk / HISTORY

# Diff of /sml/trunk/HISTORY

revision 749, Fri Dec 8 15:16:36 2000 UTC revision 1180, Tue Mar 26 22:24:24 2002 UTC
# Line 8  Line 8
8  The form of an entry should be:  The form of an entry should be:
9
10  Name:  Name:
11  Date:  Date: yyyy/mm/dd
12  Tag: <post-commit CVS tag>  Tag: <post-commit CVS tag>
13  Description:  Description:
14
15    ----------------------------------------------------------------------
16    Name: Matthias Blume
17    Date: 2002/03/25 17:25:00 EST
18    Tag: blume-20020325-divmod
19    Description:
20
21    I improved (hopefully without breaking them) the implementation of Int.div,
22    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
23    advantage of the following observations:
24
25      Let  q = x quot y      r = x rem y
26           d = x div  y      m = x mod y
27
28    where "quot" is the round-to-zero version of integer division that
29    hardware usually provides.  Then we have:
30
31         r = x - q * y        where neither the * nor the - will overflow
32         d = if q >= 0 orelse x = q * y then q else q - 1
33                              where neither the * nor the - will overflow
34         m = if q >= 0 orelse r = 0 then r else r + y
35                              where the + will not overflow
36
37    This results in substantial simplification of the generated code.
38    The following table shows the number of CFG nodes and edges generated
39    for
40            fun f (x, y) = x OPER y
41            (* with OPER \in div, mod, quot, rem *)
42
43
44        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
45        --------------------------------------------------------
46         div |         24 |         39 |         12 |         16
47         mod |         41 |         71 |         12 |         16
48        quot |          8 |         10 |          8 |         10
49         rem |         10 |         14 |          8 |         10
50
51
52    ----------------------------------------------------------------------
53    Name: Matthias Blume
54    Date: 2002/03/25 22:06:00 EST
55    Tag: blume-20020325-cprotobug
56    Description:
57
58    Fixed a bug in cproto (c prototype decoder).
59
60    ----------------------------------------------------------------------
61    Name: Matthias Blume
62    Date: 2002/03/25 16:00:00 EST
63    Tag: blume-20020325-raw-primops
64    Description:
65
66    I did some cleanup to Allen's new primop code and
67    replaced yesterday's bootfiles with new ones.
68    (But they are stored in the same place.)
69
70    ----------------------------------------------------------------------
71    Name: Matthias Blume
72    Date: 2002/03/24 22:40:00 EST
73    Tag: blume-20020324-bootfiles
74    Description:
75
77
78    ----------------------------------------------------------------------
79    Name: Allen Leung
80    Date: 2002/03/23 15:50:00 EST
81    Tag: leunga-20020323-flint-cps-rcc-primops
82    Description:
83
84      1. Changes to FLINT primops:
85
86        (* make a call to a C-function;
87         * The primop carries C function prototype information and specifies
88         * which of its (ML-) arguments are floating point. C prototype
89         * information is for use by the backend, ML information is for
90         * use by the CPS converter. *)
91      | RAW_CCALL of { c_proto: CTypes.c_proto,
92                       ml_args: ccall_type list,
93                       ml_res_opt: ccall_type option,
94                       reentrant : bool
95                     } option
96       (* Allocate uninitialized storage on the heap.
97        * The record is meant to hold short-lived C objects, i.e., they
98        * are not ML pointers.  With the tag, the representation is
99        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
100        *)
101      | RAW_RECORD of {tag:bool,sz:int}
102      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
103
104      2.  These CPS primops are now overloaded:
105
107           rawstore of {kind:numkind}
108
109          The one argument form is:
110
112
113          The two argument form is:
114
115             rawload {kind} [ml object, byte-offset]
116
117      3. RAW_CCALL/RCC now takes two extra arguments:
118
119         a. The first is whether the C call is reentrant, i.e., whether
120            ML state should be saved and restored.
121         b. The second argument is a string argument specifying the name of
122            library and the C function.
123
124         These things are currently not handled in the code generator, yet.
125
126      4. In CProto,
127
128         An encoding type of "bool" means "ml object" and is mapped into
129         C prototype of PTR.  Note that "bool" is different than "string",
130         even though "string" is also mapped into PTR, because "bool"
131         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
132
133      5. Pickler/unpicker
134
135         Changed to handle RAW_RECORD and newest RAW_CCALL
136
137      6. MLRiscGen,
138
139         1. Changed to handle the new rawload/rawstore/rawrecord operators.
140         2. Code for handling C Calls has been moved to a new module CPSCCalls,
141            in the file CodeGen/cpscompile/cps-c-calls.sml
142
143      7. Added the conditional move operator
144
145             condmove of branch
146
147         to cps.  Generation of this is still buggy so it is currently
148         disabled.
149
150    ----------------------------------------------------------------------
151    Name: Lal George
152    Date: 2002/03/22 14:18:25 EST
153    Tag: george-20020322-cps-branch-prob
154    Description:
155
156    Implemented the Ball-Larus branch prediction-heuristics, and
157    incorporated graphical viewers for control flow graphs.
158
159    Ball-Larus Heuristics:
160    ---------------------
161    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
162
163    By design it uses the Dempster-Shafer theory for combining
164    probabilities.  For example, in the function:
165
166        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
167
168    the ball-larus heuristics predicts that the n=0 is unlikely
169    (OH-heuristic), and the 'then' branch is unlikely because of the
170    RH-heuristic -- giving the 'then' branch an even lower combined
171    probability using the Dempster-Shafer theory.
172
173    Finally, John Reppy's loop analysis in MLRISC, further lowers the
174    probability of the 'then' branch because of the loop in the else
175    branch.
176
177
178    Graphical Viewing:
179    ------------------
180    I merely plugged in Allen's graphical viewers into the compiler. The
181    additional code is not much. At the top level, saying:
182
183            Control.MLRISC.getFlag "cfg-graphical-view" := true;
184
185    will display the graphical view of the control flow graph just before
186    back-patching.  daVinci must be in your path for this to work. If
187    daVinci is not available, then the default viewer can be changed
188    using:
189
190            Control.MLRISC.getString "viewer"
191
192    which can be set to "dot" or "vcg" for the corresponding viewers. Of
193    course, these viewers must be in your path.
194
195    The above will display the compilation unit at the level of clusters,
196    many of which are small, boring, and un-interesting. Also setting:
197
198            Control.MLRISC.getInt "cfg-graphical-view_size"
199
200    will display clusters that are larger than the value set by the above.
201
202
203    ----------------------------------------------------------------------
204    Name: Matthias Blume
205    Date: 2002/03/21 22:20:00 EST
206    Tag: blume-20020321-kmp-bugfix
207    Description:
208
209    Changed the interface to the KMP routine in PreString and fixed
210    a minor bug in one place where it was used.
211
212    ----------------------------------------------------------------------
213    Name: Allen Leung
214    Date: 2002/03/21 20:30:00 EST
215    Tag: leunga-20020321-cfg
216    Description:
217
218      Fixed a potential problem in cfg edge splitting.
219
220    ----------------------------------------------------------------------
221    Name: Allen Leung
222    Date: 2002/03/21 17:15:00 EST
223    Tag: leunga-20020321-x86-fp-cfg
224    Description:
225
226      1. Recoded the buggy parts of x86-fp.
227
228         a. All the block reordering code has been removed.
229            We now depend on the block placement phases to do this work.
230
231         b. Critical edge splitting code has been simplified and moved into the
232            CFG modules, as where they belong.
233
234         Both of these were quite buggy and complex.  The code is now much, much
235         simpler.
236
237      2. X86 backend.
238
239         a. Added instructions for 64-bit support.  Instruction selection for
240            64-bit has not been committed, however, since that
241            requires changes to MLTREE which haven't been approved by
242            Lal and John.
243
244         b. Added support for FUCOMI and FUCOMIP when generating code for
245            PentiumPro and above.  We only generate these instructions in
246            the fast-fp mode.
247
248         c. Added cases for JP and JNP in X86FreqProps.
249
250      3. CFG
251
252         CFG now has a bunch of methods for edge splitting and merging.
253
254      4. Machine description.
255
256         John's simplification of MLTREE_BASIS.fcond broke a few machine
257         description things:
258
259         rtl-build.{sig,sml} and hppa.mdl fixed.
260
261         NOTE: the machine description stuff in the repository is still broken.
262               Again, I can't put my fixes in because that involves
263               changes to MLTREE.
264
265    ----------------------------------------------------------------------
266    Name: Matthias Blume
267    Date: 2002/03/20 15:55:00 EST
268    Tag: blume-20020320-kmp
269    Description:
270
271    Implemented Knuth-Morris-Pratt string matching in PreString and used
272    it for String.isSubstring, Substring.isSubstring, and
273    Substring.position.
274
275    (Might need some stress-testing.  Simple examples worked fine.)
276
277    ----------------------------------------------------------------------
278    Name: Matthias Blume
279    Date: 2002/03/19 16:37:00 EST
280    Tag: blume-20020319-witnesses
281    Description:
282
283    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
284
285    This implements a generic mechanism for changing constness qualifiers
286    anywhere within big C types without resorting to outright "casts".
287    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
288    modify the constness at the outermost level.)
289    The implementation of "convert" is based on the idea of "witness"
290    values -- values that are not used by the operation but whose types
291    "testify" to their applicability.  On the implementation side, "convert"
292    is simply a projection (returning its second curried argument).  With
293    cross-module inlining, it should not result in any machine code being
294    generated.
295
296    ----------------------------------------------------------------------
297    Name: Matthias Blume
298    Date: 2002/03/15 16:40:00 EST
299    Tag: blume-20020315-basis
300    Description:
301
302    Provided (preliminary?) implementations for
303
304      {String,Substring}.{concatWith,isSuffix,isSubstring}
305
306    and
307
308      Substring.full
309
310    Those are in the Basis spec but they were missing in SML/NJ.
311
312    ----------------------------------------------------------------------
313    Name: Matthias Blume
314    Date: 2002/03/14 21:30:00 EST
315    Tag: blume-20020314-controls
316    Description:
317
318    Controls:
319    ---------
320
321    1. Factored out the recently-added Controls : CONTROLS stuff and put
322       it into its own library /controls-lib.cm. The source tree for 323 this is under src/smlnj-lib/Controls. 324 325 2. Changed the names of types and functions in this interface, so they 326 make a bit more "sense": 327 328 module -> registry 329 'a registry -> 'a group 330 331 3. The interface now deals in ref cells only. The getter/setter interface 332 is (mostly) gone. 333 334 4. Added a function that lets one register an already-existing ref cell. 335 336 5. Made the corresponding modifications to the rest of the code so that 337 everything compiles again. 338 339 6. Changed the implementation of Controls.MLRISC back to something closer 340 to the original. In particular, this module (and therefore MLRISC) 341 does not depend on Controls. There now is some link-time code in 342 int-sys.sml that registers the MLRISC controls with the Controls 343 module. 344 345 CM: 346 --- 347 348 * One can now specify the lambda-split aggressiveness in init.cmi. 349 350 ---------------------------------------------------------------------- 351 Name: Allen Leung 352 Date: 2002/03/13 17:30:00 EST 353 Tag: leunga-20020313-x86-fp-unary 354 Description: 355 356 Bug fix for: 357 358 > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml 359 > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002 360 > - fun f(x,(y,z)) = Real.~ y; 361 > [autoloading] 362 > [autoloading done] 363 > fchsl (%eax), 184(%esp) 364 > Error: MLRisc bug: X86MCEmitter.emitInstr 365 > 366 > uncaught exception Error 367 > raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19 368 369 The problem was that the code generator did not generate any fp registers 370 in this case, and the ra didn't know that it needed to run the X86FP phase to 371 translate the pseudo fp instruction. This only happened with unary fp 372 operators in certain situations. 373 374 ---------------------------------------------------------------------- 375 Name: Matthias Blume 376 Date: 2002/03/13 14:00:00 EST 377 Tag: blume-20020313-overload-etc 378 Description: 379 380 1. Added _overload as a synonym for overload for backward compatibility. 381 (Control.overloadKW must be true for either version to be accepted.) 382 383 2. Fixed bug in install script that caused more things to be installed 384 than what was requested in config/targets. 385 386 3. Made CM aware of the (_)overload construct so that autoloading 387 works. 388 389 ---------------------------------------------------------------------- 390 Name: Matthias Blume 391 Date: 2002/03/12 22:03:00 EST 392 Tag: blume-20020312-url 393 Description: 394 395 Forgot to update BOOT and srcarchiveurl. 396 397 ---------------------------------------------------------------------- 398 Name: Matthias Blume 399 Date: 2002/03/12 17:30:00 EST 400 Tag: blume-20020312-version110392 401 Description: 402 403 Yet another version number bump (because of small changes to the 404 binfile format). Version number is now 110.39.2. NEW BOOTFILES! 405 406 Changes: 407 408 The new pid generation scheme described a few weeks ago was overly 409 complicated. I implemented a new mechanism that is simpler and 410 provides a bit more "stability": Once CM has seen a compilation 411 unit, it keeps its identity constant (as long as you do not delete 412 those crucial CM/GUID/* files). This means that when you change 413 an interface, compile, then go back to the old interface, and 414 compile again, you arrive at the original pid. 415 416 There now also is a mechanism that instructs CM to use the plain 417 environment hash as a module's pid (effectively making its GUID 418 the empty string). For this, "noguid" must be specified as an 419 option to the .sml file in question within its .cm file. 420 This is most useful for code that is being generated by tools such 421 as ml-nlffigen (because during development programmers tend to 422 erase the tool's entire output directory tree including CM's cached 423 GUIDs). "noguid" is somewhat dangerous (since it can be used to locally 424 revert to the old, broken behavior of SML/NJ, but in specific cases 425 where there is no danger of interface confusion, its use is ok 426 (I think). 427 428 ml-nlffigen by default generates "noguid" annotations. They can be 429 turned off by specifying -guid in its command line. 430 431 ---------------------------------------------------------------------- 432 Name: Lal George 433 Date: 2002/03/12 12 14:42:36 EST 434 Tag: george-20020312-frequency-computation 435 Description: 436 437 Integrated jump chaining and static block frequency into the 438 compiler. More details and numbers later. 439 440 ---------------------------------------------------------------------- 441 Name: Lal George 442 Date: 2002/03/11 11 22:38:53 EST 443 Tag: george-20020311-jump-chain-elim 444 Description: 445 446 Tested the jump chain elimination on all architectures (except the 447 hppa). This is on by default right now and is profitable for the 448 alpha and x86, however, it may not be profitable for the sparc and ppc 449 when compiling the compiler. 450 451 The gc test will typically jump to a label at the end of the cluster, 452 where there is another jump to an external cluster containing the actual 453 code to invoke gc. This is to allow factoring of common gc invocation 454 sequences. That is to say, we generate: 455 456 f: 457 testgc 458 ja L1 % jump if above to L1 459 460 L1: 461 jmp L2 462 463 464 After jump chain elimination the 'ja L1' instructions is converted to 465 'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end 466 up being implemented in their long form (if L2 is far away) using: 467 468 jbe L3 % jump if below or equal to L3 469 jmp L2 470 L3: 471 ... 472 473 474 For large compilation units L2 may be far away. 475 476 477 ---------------------------------------------------------------------- 478 Name: Matthias Blume 479 Date: 2002/03/11 13:30:00 EST 480 Tag: blume-20020311-mltreeeval 481 Description: 482 483 A functor parameter was missing. 484 485 ---------------------------------------------------------------------- 486 Name: Allen Leung 487 Date: 2002/03/11 10:30:00 EST 488 Tag: leunga-20020311-runtime-string0 489 Description: 490 491 The representation of the empty string now points to a 492 legal null terminated C string instead of unit. It is now possible 493 to convert an ML string into C string with InlineT.CharVector.getData. 494 This compiles into one single machine instruction. 495 496 ---------------------------------------------------------------------- 497 Name: Allen Leung 498 Date: 2002/03/10 23:55:00 EST 499 Tag: leunga-20020310-x86-call 500 Description: 501 502 Added machine generation for CALL instruction (relative displacement mode) 503 504 ---------------------------------------------------------------------- 505 Name: Matthias Blume 506 Date: 2002/03/08 16:05:00 507 Tag: blume-20020308-entrypoints 508 Description: 509 510 Version number bumped to 110.39.1. NEW BOOTFILES! 511 512 Entrypoints: non-zero offset into a code object where execution should begin. 513 514 - Added the notion of an entrypoint to CodeObj. 515 - Added reading/writing of entrypoint info to Binfile. 516 - Made runtime system bootloader aware of entrypoints. 517 - Use the address of the label of the first function given to mlriscGen 518 as the entrypoint. This address is currently always 0, but it will 519 not be 0 once we turn on block placement. 520 - Removed the linkage cluster code (which was The Other Way(tm) of dealing 521 with entry points) from mlriscGen. 522 523 ---------------------------------------------------------------------- 524 Name: Allen Leung 525 Date: 2002/03/07 20:45:00 EST 526 Tag: leunga-20020307-x86-cmov 527 Description: 528 529 Bug fixes for CMOVcc on x86. 530 531 1. Added machine code generation for CMOVcc 532 2. CMOVcc is now generated in preference over SETcc on PentiumPro or above. 533 3. CMOVcc cannot have an immediate operand as argument. 534 535 ---------------------------------------------------------------------- 536 Name: Matthias Blume 537 Date: 2002/03/07 16:15:00 EST 538 Tag: blume-20020307-controls 539 Description: 540 541 This is a very large but mostly boring patch which makes (almost) 542 every tuneable compiler knob (i.e., pretty much everything under 543 Control.* plus a few other things) configurable via both the command 544 line and environment variables in the style CM did its configuration 545 until now. 546 547 Try starting sml with '-h' (or, if you are brave, '-H') 548 549 To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which 550 implements the underlying generic mechanism. 551 552 The interface to some of the existing such facilities has changed somewhat. 553 For example, the MLRiscControl module now provides mkFoo instead of getFoo. 554 (The getFoo interface is still there for backward-compatibility, but its 555 use is deprecated.) 556 557 The ml-build script passes -Cxxx=yyy command-line arguments through so 558 that one can now twiddle the compiler settings when using this "batch" 559 compiler. 560 561 TODO items: 562 563 We should go through and throw out all controls that are no longer 564 connected to anything. Moreover, we should go through and provide 565 meaningful (and correct!) documentation strings for those controls 566 that still are connected. 567 568 Currently, multiple calls to Controls.new are accepted (only the first 569 has any effect). Eventually we should make sure that every control 570 is being made (via Controls.new) exactly once. Future access can then 571 be done using Controls.acc. 572 573 Finally, it would probably be a good idea to use the getter-setter 574 interface to controls rather than ref cells. For the time being, both 575 styles are provided by the Controls module, but getter-setter pairs are 576 better if thread-safety is of any concern because they can be wrapped. 577 578 ***************************************** 579 580 One bug fix: The function blockPlacement in three of the MLRISC 581 backpatch files used to be hard-wired to one of two possibilities at 582 link time (according to the value of the placementFlag). But (I 583 think) it should rather sense the flag every time. 584 585 ***************************************** 586 587 Other assorted changes (by other people who did not supply a HISTORY entry): 588 589 1. the cross-module inliner now works much better (Monnier) 590 2. representation of weights, frequencies, and probabilities in MLRISC 591 changed in preparation of using those for weighted block placement 592 (Reppy, George) 593 594 ---------------------------------------------------------------------- 595 Name: Lal George 596 Date: 2002/03/07 14:44:24 EST 2002 597 Tag: george-20020307-weighted-block-placement 598 599 Tested the weighted block placement optimization on all architectures 600 (except the hppa) using AMPL to generate the block and edge frequencies. 601 Changes were required in the machine properties to correctly 602 categorize trap instructions. There is an MLRISC flag 603 "weighted-block-placement" that can be used to enable weighted block 604 placement, but this will be ineffective without block/edge 605 frequencies (coming soon). 606 607 608 ---------------------------------------------------------------------- 609 Name: Lal George 610 Date: 2002/03/05 17:24:48 EST 611 Tag: george-20020305-linkage-cluster 612 613 In order to support the block placement optimization, a new cluster 614 is generated as the very first cluster (called the linkage cluster). 615 It contains a single jump to the 'real' entry point for the compilation 616 unit. Block placement has no effect on the linkage cluster itself, but 617 all the other clusters have full freedom in the manner in which they 618 reorder blocks or functions. 619 620 On the x86 the typical linkage code that is generated is: 621 ---------------------- 622 .align 2 623 L0: 624 addlL1-L0, 72(%esp)
625            jmp     L1
626
627
628            .align  2
629       L1:
630       ----------------------
631
632    72(%esp) is the memory location for the stdlink register. This
633    must contain the address of the CPS function being called. In the
634    above example, it contains the address of  L0; before
635    calling L1 (the real entry point for the compilation unit), it
636    must contain the address for L1, and hence
637
638            addl $L1-L0, 72(%esp) 639 640 I have tested this on all architectures except the hppa.The increase 641 in code size is of course negligible 642 643 ---------------------------------------------------------------------- 644 Name: Allen Leung 645 Date: 2002/03/03 13:20:00 EST 646 Tag: leunga-20020303-mlrisc-tools 647 648 Added #[ ... ] expressions to mlrisc tools 649 650 ---------------------------------------------------------------------- 651 Name: Matthias Blume 652 Date: 2002/02/27 12:29:00 EST 653 Tag: blume-20020227-cdebug 654 Description: 655 656 - made types in structure C and C_Debug to be equal 657 - got rid of code duplication (c-int.sml vs. c-int-debug.sml) 658 - there no longer is a C_Int_Debug (C_Debug is directly derived from C) 659 660 ---------------------------------------------------------------------- 661 Name: Matthias Blume 662 Date: 2002/02/26 12:00:00 EST 663 Tag: blume-20020226-ffi 664 Description: 665 666 1. Fixed a minor bug in CM's "noweb" tool: 667 If numbering is turned off, then truly don't number (i.e., do not 668 supply the -L option to noweb). The previous behavior was to supply 669 -L'' -- which caused noweb to use the "default" line numbering scheme. 670 Thanks to Chris Richards for pointing this out (and supplying the fix). 671 672 2. Once again, I reworked some aspects of the FFI: 673 674 A. The incomplete/complete type business: 675 676 - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are 677 gone! 678 - ML types representing an incomplete type are now *equal* to 679 ML types representing their corresponding complete types (just like 680 in C). This is still safe because ml-nlffigen will not generate 681 RTTI for incomplete types, nor will it generate functions that 682 require access to such RTTI. But when ML code generated from both 683 incomplete and complete versions of the C type meet, the ML types 684 are trivially interoperable. 685 686 NOTE: These changes restore the full generality of the translation 687 (which was previously lost when I eliminated functorization)! 688 689 B. Enum types: 690 691 - Structure C now has a type constructor "enum" that is similar to 692 how the "su" constructor works. However, "enum" is not a phantom 693 type because each "T enum" has values (and is isomorphic to 694 MLRep.Signed.int). 695 - There are generic access operations for enum objects (using 696 MLRep.Signed.int). 697 - ml-nlffigen will generate a structure E_foo for each "enum foo". 698 * The structure contains the definition of type "mlrep" (the ML-side 699 representation type of the enum). Normally, mlrep is the same 700 as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec", 701 then mlrep will be defined as a datatype -- thus facilitating 702 pattern matching on mlrep values. 703 ("-ec" will be suppressed if there are duplicate values in an 704 enumeration.) 705 * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep 706 will be generated for each C enum constant xxx. 707 * Conversion functions m2i and i2m convert between mlrep and 708 MLRep.Signed.int. (Without "-ec", these functions are identities.) 709 * Coversion functions c and ml convert between mlrep and "tag enum". 710 * Access functions (get/set) fetch and store mlrep values. 711 - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed 712 enumerations are merged into one single enumeration represented by 713 structure E_'. 714 715 ---------------------------------------------------------------------- 716 Name: Allen Leung 717 Date: 2002/02/25 04:45:00 EST 718 Tag: leunga-20020225-cps-spill 719 720 This is a new implementation of the CPS spill phase. 721 The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml 722 In case of problems, replace it with the old file spill.sml 723 724 The current compiler runs into some serious performance problems when 725 constructing a large record. This can happen when we try to compile a 726 structure with many items. Even a very simple structure like the following 727 makes the compiler slow down. 728 729 structure Foo = struct 730 val x_1 = 0w1 : Word32.int 731 val x_2 = 0w2 : Word32.int 732 val x_3 = 0w3 : Word32.int 733 ... 734 val x_N = 0wN : Word32.int 735 end 736 737 The following table shows the compile time, from N=1000 to N=4000, 738 with the old compiler: 739 740 N 741 1000 CPS 100 spill 0.04u 0.00s 0.00g 742 MLRISC ra 0.06u 0.00s 0.05g 743 (spills = 0 reloads = 0) 744 TOTAL 0.63u 0.07s 0.21g 745 746 1100 CPS 100 spill 8.25u 0.32s 0.64g 747 MLRISC ra 5.68u 0.59s 3.93g 748 (spills = 0 reloads = 0) 749 TOTAL 14.71u 0.99s 4.81g 750 751 1500 CPS 100 spill 58.55u 2.34s 1.74g 752 MLRISC ra 5.54u 0.65s 3.91g 753 (spills = 543 reloads = 1082) 754 TOTAL 65.40u 3.13s 6.00g 755 756 2000 CPS 100 spill 126.69u 4.84s 3.08g 757 MLRISC ra 0.80u 0.10s 0.55g 758 (spills = 42 reloads = 84) 759 TOTAL 129.42u 5.10s 4.13g 760 761 3000 CPS 100 spill 675.59u 19.03s 11.64g 762 MLRISC ra 2.69u 0.27s 1.38g 763 (spills = 62 reloads = 124) 764 TOTAL 682.48u 19.61s 13.99g 765 766 4000 CPS 100 spill 2362.82u 56.28s 43.60g 767 MLRISC ra 4.96u 0.27s 2.72g 768 (spills = 85 reloads = 170) 769 TOTAL 2375.26u 57.21s 48.00g 770 771 As you can see the old cps spill module suffers from some serious 772 performance problem. But since I cannot decipher the old code fully, 773 instead of patching the problems up, I'm reimplementing it 774 with a different algorithm. The new code is more modular, 775 smaller when compiled, and substantially faster 776 (O(n log n) time and O(n) space). Timing of the new spill module: 777 778 4000 CPS 100 spill 0.02u 0.00s 0.00g 779 MLRISC ra 0.25u 0.02s 0.15g 780 (spills=1 reloads=3) 781 TOTAL 7.74u 0.34s 1.62g 782 783 Implementation details: 784 785 As far as I can tell, the purpose of the CPS spill module is to make sure the 786 number of live variables at any program point (the bandwidth) 787 does not exceed a certain limit, which is determined by the 788 size of the spill area. 789 790 When the bandwidth is too large, we decrease the register pressure by 791 packing live variables into spill records. How we achieve this is 792 completely different than what we did in the old code. 793 794 First, there is something about the MLRiscGen code generator 795 that we should be aware of: 796 797 o MLRiscGen performs code motion! 798 799 In particular, it will move floating point computations and 800 address computations involving only the heap pointer to 801 their use sites (if there is only a single use). 802 What this means is that if we have a CPS record construction 803 statement 804 805 RECORD(k,vl,w,e) 806 807 we should never count the new record address w as live if w 808 has only one use (which is often the case). 809 810 We should do something similar to floating point, but the transformation 811 there is much more complex, so I won't deal with that. 812 813 Secondly, there are now two new cps primops at our disposal: 814 815 1. rawrecord of record_kind option 816 This pure operator allocates some uninitialized storage from the heap. 817 There are two forms: 818 819 rawrecord NONE [INT n] allocates a tagless record of length n 820 rawrecord (SOME rk) [INT n] allocates a tagged record of length n 821 and initializes the tag. 822 823 2. rawupdate of cty 824 rawupdate cty (v,i,x) 825 Assigns to x to the ith component of record v. 826 The storelist is not updated. 827 828 We use these new primops for both spilling and increment record construction. 829 830 1. Spilling. 831 832 This is implemented with a linear scan algorithm (but generalized 833 to trees). The algorithm will create a single spill record at the 834 beginning of the cps function and use rawupdate to spill to it, 835 and SELECT or SELp to reload from it. So both spills and reloads 836 are fine-grain operations. In contrast, in the old algorithm 837 "spills" have to be bundled together in records. 838 839 Ideally, we should sink the spill record construction to where 840 it is needed. We can even split the spill record into multiple ones 841 at the places where they are needed. But CPS is not a good 842 representation for global code motion, so I'll keep it simple and 843 am not attempting this. 844 845 2. Incremental record construction (aka record splitting). 846 847 Long records with many component values which are simulatenously live 848 (recall that single use record addresses are not considered to 849 be live) are constructed with rawrecord and rawupdate. 850 We allocate space on the heap with rawrecord first, then gradually 851 fill it in with rawupdate. This is the technique suggested to me 852 by Matthias. 853 854 Some restrictions on when this is applicable: 855 1. It is not a VECTOR record. The code generator currently does not handle 856 this case. VECTOR record uses double indirection like arrays. 857 2. All the record component values are defined in the same "basic block" 858 as the record constructor. This is to prevent speculative 859 record construction. 860 861 ---------------------------------------------------------------------- 862 Name: Allen Leung 863 Date: 2002/02/22 01:02:00 EST 864 Tag: leunga-20020222-mlrisc-tools 865 866 Minor bug fixes in the parser and rewriter 867 868 ---------------------------------------------------------------------- 869 Name: Allen Leung 870 Date: 2002/02/21 20:20:00 EST 871 Tag: leunga-20020221-peephole 872 873 Regenerated the peephole files. Some contained typos in the specification 874 and some didn't compile because of pretty printing bugs in the old version 875 of 'nowhere'. 876 877 ---------------------------------------------------------------------- 878 Name: Allen Leung 879 Date: 2002/02/19 20:20:00 EST 880 Tag: leunga-20020219-mlrisc-tools 881 Description: 882 883 Minor bug fixes to the mlrisc-tools library: 884 885 1. Fixed up parsing colon suffixed keywords 886 2. Added the ability to shut the error messages up 887 3. Reimplemented the pretty printer and fixed up/improved 888 the pretty printing of handle and -> types. 889 4. Fixed up generation of literal symbols in the nowhere tool. 890 5. Added some SML keywords to to sml.sty 891 892 ---------------------------------------------------------------------- 893 Name: Matthias Blume 894 Date: 2002/02/19 16:20:00 EST 895 Tag: blume-20020219-cmffi 896 Description: 897 898 A wild mix of changes, some minor, some major: 899 900 * All C FFI-related libraries are now anchored under$c:
901        $/c.cm -->$c/c.cm
902        $/c-int.cm -->$c/internals/c-int.cm
903        $/memory.cm -->$c/memory/memory.cm
904
905    * "make" tool (in CM) now treats its argument pathname slightly
906      differently:
907        1. If the native expansion is an absolute name, then before invoking
908           the "make" command on it, CM will apply OS.Path.mkRelative
909           (with relativeTo = OS.FileSys.getDir()) to it.
910        2. The argument will be passed through to subsequent phases of CM
911           processing without "going native".  In particular, if the argument
912           was an anchored path, then "make" will not lose track of that anchor.
913
914    * Compiler backends now "know" their respective C calling conventions
915      instead of having to be told about it by ml-nlffigen.  This relieves
916      ml-nlffigen from one of its burdens.
917
918    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
919
920    * Export C_DEBUG and C_Debug from $c/c.cm. 921 922 * C type encoding in ml-nlffi-lib has been improved to model the conceptual 923 subtyping relationship between incomplete pointers and their complete 924 counterparts. For this, ('t, 'c) ptr has been changed to 'o ptr -- 925 with the convention of instantiating 'o with ('t, 'c) obj whenever 926 the pointer target type is complete. In the incomplete case, 'o 927 will be instantiated with some "'c iobj" -- a type obtained by 928 using one of the functors PointerToIncompleteType or PointerToCompleteType. 929 930 Operations that work on both incomplete and complete pointer types are 931 typed as taking an 'o ptr while operations that require the target to 932 be known are typed as taking some ('t, 'c) obj ptr. 933 934 voidptr is now a bit "more concrete", namely "type voidptr = void ptr'" 935 where void is an eqtype without any values. This makes it possible 936 to work on voidptr values using functions meant to operate on light 937 incomplete pointers. 938 939 * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has 940 been vastly simplified. 941 942 ---------------------------------------------------------------------- 943 Name: Matthias Blume 944 Date: 2002/02/19 10:48:00 EST 945 Tag: blume-20020219-pqfix 946 Description: 947 948 Applied Chris Okasaki's bug fix for priority queues. 949 950 ---------------------------------------------------------------------- 951 Name: Matthias Blume 952 Date: 2002/02/15 17:05:00 953 Tag: Release_110_39 954 Description: 955 956 Last-minute retagging is becoming a tradition... :-( 957 958 This is the working release 110.39. 959 960 ---------------------------------------------------------------------- 961 Name: Matthias Blume 962 Date: 2002/02/15 16:00:00 EST 963 Tag: Release_110_39-orig 964 Description: 965 966 Working release 110.39. New bootfiles. 967 968 (Update: There was a small bug in the installer so it wouldn't work 969 with all shells. So I retagged. -Matthias) 970 971 ---------------------------------------------------------------------- 972 Name: Matthias Blume 973 Date: 2002/02/15 14:17:00 EST 974 Tag: blume-20020215-showbindings 975 Description: 976 977 Added EnvRef.listBoundSymbols and CM.State.showBindings. Especially 978 the latter can be useful for exploring what bindings are available at 979 the interactive prompt. (The first function returns only the list 980 of symbols that are really bound, the second prints those but also the 981 ones that CM's autoloading mechanism knows about.) 982 983 ---------------------------------------------------------------------- 984 Name: Matthias Blume 985 Date: 2002/02/15 12:08:00 EST 986 Tag: blume-20020215-iptrs 987 Description: 988 989 Two improvements to ml-nlffigen: 990 991 1. Write files only if they do not exist or if their current contents 992 do not coincide with what's being written. (That is, avoid messing 993 with the time stamps unless absolutely necessary.) 994 995 2. Implement a "repository" mechanism for generated files related 996 to "incomplete pointer types". See the README file for details. 997 998 ---------------------------------------------------------------------- 999 Name: Matthias Blume 1000 Date: 2002/02/14 11:50:00 EST 1001 Tag: blume-20020214-quote 1002 Description: 1003 1004 Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm). This is required 1005 because of the new and improved tag generation scheme. (Thanks to Allen 1006 Leung for pointing it out.) 1007 1008 ---------------------------------------------------------------------- 1009 Name: Lal George 1010 Date: 2002/02/14 09:55:27 EST 2002 1011 Tag: george-20020214-isabelle-bug 1012 Description: 1013 1014 Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation 1015 of Isabelle on the x86. 1016 1017 From Allen: 1018 ----------- 1019 I've found the problem: 1020 1021 in ra-core.sml, I use the counter "blocked" to keep track of the 1022 true number of elements in the freeze queue. When the counter goes 1023 to zero, I skip examining the queue. But I've messed up the 1024 bookkeeping in combine(): 1025 1026 else (); 1027 case !ucol of 1028 PSEUDO => (if !cntv > 0 then 1029 (if !cntu > 0 then blocked := !blocked - 1 else (); 1030 ^^^^^^^^^^^^^^^^^^^^^^^ 1031 moveu := mergeMoveList(!movev, !moveu) 1032 ) 1033 else (); 1034 1035 combine() is called to coalesce two nodes u and v. 1036 I think I was thinking that if the move counts of u and v are both 1037 greater than zero then after they are coalesced then one node is 1038 removed from the freeze queue. Apparently I was thinking that 1039 both u and v are of low degree, but that's clearly not necessarily true. 1040 1041 1042 02/12/2002: 1043 Here's the patch. HOL now compiles. 1044 1045 I don't know how this impact on performance (compile 1046 time or runtime). This bug caused the RA (especially on the x86) 1047 to go thru the potential spill phase when there are still nodes on the 1048 freeze queue. 1049 1050 1051 1052 1053 ---------------------------------------------------------------------- 1054 Name: Matthias Blume 1055 Date: 2002/02/13 22:40:00 EST 1056 Tag: blume-20020213-fptr-rtti 1057 Description: 1058 1059 Fixed a bug in ml-nlffigen that was introduced with one of the previous 1060 updates. 1061 1062 ---------------------------------------------------------------------- 1063 Name: Matthias Blume 1064 Date: 2002/02/13 16:41:00 EST 1065 Tag: blume-20020213-cmlpq 1066 Description: 1067 1068 Added new priority queue export symbols (which have just been added to 1069 smlnj-lib.cm) to CML's version of smlnj-lib.cm. (Otherwise CML would 1070 not compile and the installer would choke.) 1071 1072 ---------------------------------------------------------------------- 1073 Name: Matthias Blume 1074 Date: 2002/02/13 16:15:00 EST 1075 Tag: blume-20020213-various 1076 Description: 1077 1078 1. More tweaks to ml-nlffigen: 1079 1080 - better internal datastructures (resulting in slight speedup) 1081 - "-match" option requires exact match 1082 - "localized" gensym counters (untagged structs/unions nested within 1083 other structs/unions or within typedefs get a fresh counter; their 1084 tag will be prefixed by a concatenation of their parents' tags) 1085 - bug fixes (related to calculation of transitive closure of types 1086 to be included in the output) 1087 1088 2. Minor Basis updates: 1089 1090 - added implementations for List.collate and Option.app 1091 1092 ---------------------------------------------------------------------- 1093 Name: Matthias Blume 1094 Date: 2002/02/11 15:55:00 EST 1095 Tag: blume-20020211-gensym 1096 Description: 1097 1098 Added a "-gensym" option to command line of ml-nlffigen. This can be 1099 used to specify a "stem" -- a string that is inserted in all "gensym'd" 1100 names (ML structure names that correspond to unnamed C structs, unions, 1101 and enums), so that separate runs of ml-nlffigen do not clash. 1102 1103 ---------------------------------------------------------------------- 1104 Name: Matthias Blume 1105 Date: 2002/02/11 12:05:00 EST 1106 Tag: blume-20020211-gensml 1107 Description: 1108 1109 A quick fix for a problem with GenSML (in the pgraph-util library): 1110 Make generation of toplevel "local" optional. (Strictly speaking, 1111 signature definitions within "local" are not legal SML.) 1112 1113 Other than that: updates to INSTALL and cm/TODO. 1114 1115 ---------------------------------------------------------------------- 1116 Name: Matthias Blume 1117 Date: 2002/02/08 15:00:00 EST 1118 Tag: blume-20020208-uniquepid 1119 Description: 1120 1121 0. Version number has been bumped to 110.38.1. NEW BOOTFILES!!! 1122 1123 1. The installer (config/install.sh) has gotten smarter: 1124 1125 - Configuration options are a bit easier to specify now 1126 (in config/targets). 1127 - Bug in recognizing .tar.bz2 files fixed. 1128 - Installer automatically resolves dependencies between 1129 configuration options (e.g., if you ask for eXene, you will 1130 also get cml -- regardless whether you asked for it or not). 1131 - Installer can run in "quieter mode" by setting the environment 1132 variable INSTALL_QUIETLY to "true". "Quieter" does not mean 1133 "completely silent", though. 1134 - Build HashCons library as part of smlnj-lib. 1135 1136 2. A new scheme for assigning persistent identifiers to compilation 1137 units (and, by extension, to types etc.) has been put into place. 1138 This fixes a long-standing bug where types and even dynamic values 1139 can get internally confused, thereby compromising type safety 1140 (abstraction) and dynamic correctness. See 1141 1142 http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz 1143 1144 for an example of how things could go wrong until now. 1145 1146 The downside of the new scheme is that pids are not quite as 1147 persistent as they used to be: CM will generate a fresh pid 1148 for every compilation unit that it thinks it sees for the first 1149 time. That means that if you compile starting from a clean, fresh 1150 source tree at two different times, you end up with different 1151 binaries. 1152 1153 Cutoff recompilation, however, has not been compromised because 1154 CM keeps pid information in special caches between runs. 1155 1156 ---------------------------------------------------------------------- 1157 Name: Lal George 1158 Date: 2002/02/07 15:34:13 EST 2002 1159 Tag: <none> 1160 Description: 1161 1162 Compilers that generate assembly code may produce global labels 1163 whose value is resolved at link time. The various peephole optimization 1164 modules did not take this in account. 1165 1166 TODO. The Labels.addrOf function should really return an option 1167 type so that clients are forced to deal with this issue, rather 1168 than an exception being raised. 1169 1170 ---------------------------------------------------------------------- 1171 Name: Lal George 1172 Date: 2002/02/06 13:55:02 EST 1173 Tag: george-20020206-ra-breakup 1174 Description: 1175 1176 1. A bug fix from Allen. 1177 1178 A typo causes extra fstp %st(0)'s to be generated at compensation 1179 edges, which might cause stack underflow traps at runtime. This 1180 occurs in fft where there are extraneous fstps right before the 'into' 1181 trap instruction (in this case they are harmless since none of the 1182 integers overflow.) 1183 1184 2. Pulled out various utility modules that were embedded in the modules 1185 of the register allocator. I need these modules for other purposes, but 1186 they are not complete enough to put into a library (just yet). 1187 ---------------------------------------------------------------------- 1188 Name: Matthias Blume 1189 Date: 2002/01/31 16:05:00 EST 1190 Tag: blume-20020131-sparc-ccalls 1191 Description: 1192 1193 1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes) 1194 of extra stack space by mistake. Fixed. 1195 1196 2. Bug in logic of handling of command-line options in ml-nlffigen fixed. 1197 1198 ---------------------------------------------------------------------- 1199 Name: Allen Leung 1200 Date: 2002/01/30 1201 Tag: leunga-20020130-nowhere-bug-fix 1202 Description: 1203 1204 MLRISC bug fixes: 1205 1. Fixed a bindings computation bug in the 'nowhere' program generator tool. 1206 2. MachineInt.fromString was negating its value. 1207 1208 ---------------------------------------------------------------------- 1209 Name: Matthias Blume 1210 Date: 2002/01/29 1211 Tag: blume-20020129-INSTALL 1212 Description: 1213 1214 - Added somewhat detailed installation instructions (file INSTALL). 1215 - Fixed curl-detection bug in config/install.sh. 1216 - It is now possible to select the URL getter using the URLGETTER 1217 environment variable: 1218 1219 not set / "unknown" --> automatic detection (script tries wget, 1220 curl, and lynx) 1221 "wget" / "curl" / "lynx" --> use the specified program (script "knows" 1222 how to properly invoke them) 1223 other --> use$URLGETTER directly, it must take
1224                                       precisely two command-line arguments
1225                                       (source URL and destination file name)
1226
1227    ----------------------------------------------------------------------
1228    Name: Matthias Blume
1229    Date: 2002/01/28
1230    Tag: blume-20020128-sparc-ccalls
1231    Description:
1232
1233    - Fixed problem with calculation of "used" registers in sparc-c-calls.
1234    - Make use of the allocParam argument in sparc-c-calls.
1235
1236    ----------------------------------------------------------------------
1237    Name: Matthias Blume
1238    Date: 2002/01/28
1239    Tag: blume-20020128-allocParam
1240    Description:
1241
1242    John Reppy:  Changes c-calls API to accept client-callback for
1243    allocating extra stack space.
1244    me: Corresponding changes to mlriscGen (using a dummy argument that
1245        does not change the current behavior).
1246
1247    ----------------------------------------------------------------------
1248    Name: Matthias Blume
1249    Date: 2002/01/28 12:00:00
1250    Tag: Release_110_38
1251    Description:
1252
1253    This time for real!!!
1254
1255    ----------------------------------------------------------------------
1256    Name: Matthias Blume
1257    Date: 2002/01/28 10:56:00 EST
1258    Tag: blume-20020128-retraction
1259    Description:
1260
1261    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
1262       with blume-Release_110_38-retracted.)
1263
1264    1. Fixed a problem with incorrect rounding modes in real64.sml.
1265       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
1266
1267    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
1268       and enums fixed.  The naming of corresponding ML identifiers should
1269       now be consistent again.
1270
1271    ----------------------------------------------------------------------
1272    Name: Allen Leung
1273    Date: 2002/01/27
1274    Tag: leunga-20020127-nowhere
1275    Description:
1276
1277       Added a target called nowhere in the configuration scripts.
1278       Enabling this will build the MLRISC 'nowhere' tool (for translating
1279       programs with where-clauses into legal SML code) during installation.
1280
1281    ----------------------------------------------------------------------
1282    Name: Matthias Blume
1283    Date: 2002/01/25 21:27:00 EST
1284    Tag: blume-Release_110_38-retracted
1285    Description:
1286
1287    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
1288
1290
1291    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
1292    (Release_110_38) removed.  Reason: Last-minute bug fixes.
1293
1294    ----------------------------------------------------------------------
1295    Name: Matthias Blume
1296    Date: 2002/01/25
1297    Tag: blume-20020125-ffi
1298    Description:
1299
1300    A large number of tweaks and improvements to ml-nlffi-lib and
1301    ml-nlffigen:
1302
1303       - ML represenation types have been streamlined
1304       - getter and setter functions work with concrete values, not abstract
1305         ones where possible
1306       - ml-nlffigen command line more flexible (see README file there)
1307       - some bugs have been fixed (hopefully)
1308
1309    ----------------------------------------------------------------------
1310    Name: Lal George
1311    Date: 2002/01/24
1312    Tag: george-20020124-risc-ra-interface
1313    Description:
1314
1315       There is a dramatic simplification in the interface to the
1316       register allocator for RISC architectures as a result of making
1317       parallel copy instructions explicit.
1318
1319    ----------------------------------------------------------------------
1320    Name: Matthias Blume
1321    Date: 2002/01/22
1322    Tag: blume-20020122-x86-ccalls
1323    Description:
1324
1325    Bug fix for c-calls on x86 (having to do with how char- and
1326    short-arguments are being handled).
1327
1328    ----------------------------------------------------------------------
1329    Name: Matthias Blume
1330    Date: 2002/01/21
1331    Tag: blume-20020121-ff
1332    Description:
1333
1334    Another day of fiddling with the FFI...
1335
1336    1. Bug fix/workaround:  CKIT does not complain about negative array
1337       dimensions, so ml-nlffigen has to guard itself against this possibility.
1338       (Otherwise a negative dimension would send it into an infinite loop.)
1339
1340    2. Some of the abstract types (light objects, light pointers, most "base"
1341       types) in structure C are now eqtypes.
1342
1343    3. Added constructors and test functions for NULL function pointers.
1344
1345    ----------------------------------------------------------------------
1346    Name: Matthias Blume
1347    Date: 2002/01/18
1349    Description:
1350
1351    Made config/srcarchiveurl point to a new place.  (Will provide boot
1352    files shortly.)
1353
1354    Maybe we christen this to be 110.38?
1355
1356    ----------------------------------------------------------------------
1357    Name: Matthias Blume
1358    Date: 2002/01/18
1359    Tag: blume-20020118-more-ffifiddle
1360    Description:
1361
1362    Today's FFI fiddling:
1363
1364      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
1365        C.Set.* that use concrete (MLRep.*) arguments and results instead
1366        of abstract ones.
1367
1368      - Provided word-style bit operations etc. for "int" representation
1369        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
1370        and Long.
1371
1372    ----------------------------------------------------------------------
1373    Name: Matthias Blume
1374    Date: 2002/01/18
1375    Tag: blume-20020118-use-x86-fp
1376    Description:
1377
1378    Now that x86-fast-fp seems to be working, I turned it back on again
1379    by default.  (Seems to work fine now, even with the FFI.)
1380
1381    Other than that, I added some documentation about the FFI to
1382    src/ml-nlffigen/README and updated the FFI test examples in
1383    src/ml-nlffi-lib/Tests/*.
1384
1385    ----------------------------------------------------------------------
1386    Name: Allen Leung
1387    Date: 2002/01/17
1388    Tag: leunga-20020117-x86-fast-fp-call
1389    Description:
1390
1391       1. Fixed a problem with handling return fp values when x86's fast fp
1392          mode is turned on.
1393
1394       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
1395          of %f32.
1396
1397       3. Added a constructor INT32lit to the ast of MLRISC tools.
1398
1399    ----------------------------------------------------------------------
1400    Name: Matthias Blume
1401    Date: 2002/01/16
1402    Tag: blume-20020116-ffifiddle
1403    Description:
1404
1405    More fiddling with the FFI interface:
1406
1407     - Make constness 'c instead of rw wherever possible.  This eliminates
1408       the need for certain explicit coercions.  (However, due to ML's
1409       value polymorphism, there will still be many cases where explicit
1410       coercions are necessary.  Phantom types are not the whole answer
1411       to modeling a subtyping relationship in ML.)
1412
1413     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
1414
1415     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
1416       clumsy workaround for varargs problem.)
1417
1418    ----------------------------------------------------------------------
1419    Name: Lal George
1420    Date: 2002/01/15
1421    Tag: <none>
1422    Description:
1423
1424    1. Since COPY instructions are no longer native to the architecture,
1425       a generic functor can be used to implement the expandCopies function.
1426
1427    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
1428       TEXT segment.
1429
1430    ----------------------------------------------------------------------
1431    Name: Matthias Blume
1432    Date: 2002/01/15
1434    Description:
1435
1436    1. Fix for bug resulting in single-precision float values being returned
1437       incorrectly from FFI calls.
1438
1439    2. Small modifications to C FFI API:
1440
1441        - memory-allocation routines return straight objects (no options)
1442          and raise an exception in out-of-memory situations
1443        - unsafe extensions to cast between function pointers and pointers
1444          from/to ints
1445        - added structure C_Debug as an alternative to structure C where
1446          pointer-dereferencing (|*| and |*!) always check for null-pointers
1448          but also takes a (possibly empty) list of existing library handles
1449          that the current library depends on
1450
1451    ----------------------------------------------------------------------
1452    Name: Matthias Blume
1453    Date: 2002/01/10
1454    Tag: blume-20020110-newffigen
1455    Description:
1456
1457    1. Updates to portable graph code.
1458
1459    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
1460       more scalable now so that even huge interfaces such as the one
1461       for GTK compile in finite time and space. :-)
1462       See src/ml-nlffigen/README for details on what's new.
1463
1464    ----------------------------------------------------------------------
1465    Name: Lal George
1466    Date: 2001/01/09 14:31:35 EST 2002
1467    Tag: george-20011206-rm-native-copy
1468    Description:
1469
1470            Removed the native COPY and FCOPY instructions
1471            from all the architectures and replaced it with the
1472            explicit COPY instruction from the previous commit.
1473
1474            It is now possible to simplify many of the optimizations
1475            modules that manipulate copies. This has not been
1476            done in this change.
1477
1478    ----------------------------------------------------------------------
1479    Name: Lal George
1480    Date: 2001/12/06 16:50:13 EST 2001
1481    Tag: george-20011206-mlrisc-instruction
1482    Description:
1483
1484    Changed the representation of instructions from being fully abstract
1485    to being partially concrete. That is to say:
1486
1487      from
1488            type instruction
1489
1490      to
1491            type instr                              (* machine instruction *)
1492
1493            datatype instruction =
1494                LIVE of {regs: C.cellset, spilled: C.cellset}
1495              | KILL of {regs: C.cellset, spilled: C.cellset}
1496              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
1497              | ANNOTATION of {i: instruction, a: Annotations.annotation}
1498              | INSTR of instr
1499
1500    This makes the handling of certain special instructions that appear on
1501    all architectures easier and uniform.
1502
1503    LIVE and KILL say that a list of registers are live or killed at the
1504    program point where they appear. No spill code is generated when an
1505    element of the 'regs' field is spilled, but the register is moved to
1506    the 'spilled' (which is present, more for debugging than anything else).
1507
1508    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
1509    We used to generate:
1510
1511            DEFFREG f1
1512            f1 := f2 + f3
1513            trapb
1514
1515    but now generate:
1516
1517            f1 := f2 + f3
1518            trapb
1519            LIVE {regs=[f1,f2,f3], spilled=[]}
1520
1521    Furthermore, the DEFFREG (hack) required that all floating point instruction
1522    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
1523    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
1524    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
1525    this architecture).
1526
1527    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
1528    all the architectures. This will result in further simplification of the
1529    register allocator that must be aware of them for coalescing purposes, and
1530    will also simplify certain aspects of the machine description that provides
1531    callbacks related to parallel copies.
1532
1533    ANNOTATION should be obvious, and now INSTR represents the honest to God
1534    machine instruction set!
1535
1536    The <arch>/instructions/<arch>Instr.sml files define certain utility
1537    functions for making porting easier -- essentially converting upper case
1538    to lower case. All machine instructions (of type instr) are in upper case,
1539    and the lower case form generates an MLRISC instruction. For example on
1540    the alpha we have:
1541
1542      datatype instr =
1543         LDA of {r:cell, b:cell, d:operand}
1544       | ...
1545
1546      val lda : {r:cell, b:cell, d:operand} -> instruction
1547        ...
1548
1549    where lda is just (INSTR o LDA), etc.
1550
1551    ----------------------------------------------------------------------
1552    Name: Matthias Blume
1553    Date: 2001/11/22 21:40:00 EST
1554    Tag: Release_110_37
1555    Description:
1556
1557    Release 110.37.  This time for real.
1558
1559    ----------------------------------------------------------------------
1560    Name: Matthias Blume
1561    Date: 2001/11/21 16:35:00 EST
1562    Tag: blume-20011121-foot-in-mouth
1563    Description:
1564
1565    Removed the "Release_110_37" tag because of a serious bug.
1566    This will be re-tagged once the bug is fixed.
1567
1568    ----------------------------------------------------------------------
1569    Name: Matthias Blume
1570    Date: 2001/11/21 16:14:00 EST
1571    Tag: blume-20011121-forgottenfile
1572    Description:
1573
1574    Forgot to add a file.  (Just a .tex-file -- part of
1575    the CM manual source.)
1576
1577    ----------------------------------------------------------------------
1578    Name: Matthias Blume
1579    Date: 2001/11/21 16:10:00 EST
1580    Tag: blume-20011121-invalid_110_37
1581    Description:
1582
1583    Note: I removed the original tag "Release_110_37" from this commit
1584          because we found a serious bug in all non-x86 backends.
1585          - Matthias
1586
1587    1. Modifications to the SML/NJ code generator and to the runtime system
1588       so that code object name strings are directly inserted into code
1589       objects at code generation time.  The only business the runtime system
1590       has with this is now to read the name strings on occasions.
1591       (The encoding of the name string has also changed somewhat.)
1592
1593    2. CM now implements a simple "set calculus" for specifying export lists.
1594       In particular, it is now possible to refer to the export lists of
1595       other libraries/groups/sources and form unions as well as differences.
1596       See the latest CM manual for details.
1597
1598    3. An separate notion of "proxy" libraries has again be eliminated from
1599       CM's model.  (Proxy libraries are now simply a special case of using
1600       the export list calculus.)
1601
1602    4. Some of the existing libraries now take advantage of the new set
1603       calculus.
1604       (Notice that not all libraries have been converted because some
1605       of the existing .cm-files are supposed to be backward compatible
1606       with 110.0.x.)
1607
1608    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
1609       "General.exnMessage"!  The former relies on a certain hook to be
1610       initialized, and that often does not happen in the stand-alone case.)
1611
1612    ----------------------------------------------------------------------
1613    Name: Lal George
1614    Date: 2001/11/21  13:56:18 EST
1615    Tag: george-2001121-pseudo-ops
1616    Description:
1617
1618      Implemented a complete redesign of MLRISC pseudo-ops. Now there
1619      ought to never be any question of incompatabilities with
1620      pseudo-op syntax expected by host assemblers.
1621
1622      For now, only modules supporting GAS syntax are implemented
1623      but more should follow, such as MASM, and vendor assembler
1624      syntax, e.g. IBM as, Sun as, etc.
1625
1626    ----------------------------------------------------------------------
1627    Name: Matthias Blume
1628    Date: 2001/11/14 11:52:00 EST
1629    Tag: blume-20011114-srcname
1630    Description:
1631
1632    1. Routed the name of the current source file to mlriscgen where it
1633       should be directly emitted into the code object.  (This last part
1634       is yet to be done.)
1635
1636    2. Some cleanup of the pgraph code to make it match the proposal that
1637       I put out the other day.  (The proposal notwithstanding, things are
1638       still in flux here.)
1639
1640    ----------------------------------------------------------------------
1641    Name: Lal George
1642    Date: 2001/11/14 09:44:04 EST
1643    Tag:
1644    Description:
1645
1646      Fix for a backpatching bug reported by Allen.
1647
1648      Because the boundary between short and long span-dependent
1649      instructions is +/- 128, there are an astounding number of
1650      span-dependent instructions whose size is over estimated.
1651
1652      Allen came up with the idea of letting the size of span
1653      dependent instructions be non-monotonic, for a maxIter
1654      number of times, after which the size must be monotonically
1655      increasing.
1656
1657      This table shows the number of span-dependent instructions
1658      whose size was over-estimated as a function of maxIter, for the
1659      file Parse/parse/ml.grm.sml:
1660
1661         maxIter            # of instructions:
1662            10                      687
1663            20                      438
1664            30                      198
1665            40                        0
1666
1667      In compiling the compiler, there is no significant difference in
1668      compilation speed between maxIter=10 and maxIter=40. Actually,
1669      my measurements showed that maxIter=40 was a tad faster than
1670      maxIter=10! Also 96% of the  files in the compiler reach a fix
1671      point within 13 iterations, so fixing maxIter at 40, while high,
1672      is okay.
1673
1674    ----------------------------------------------------------------------
1675    Name: Matthias Blume
1676    Date: 2001/10/31 15:25:00 EST
1677    Tag: blume-20011031-pgraph
1678    Description:
1679
1680    CKIT:
1681    * Changed the "Function" constructor of type Ast.ctype to carry optional
1682      argument identifiers.
1683    * Changed the return type of TypeUtil.getFunction accordingly.
1684    * Type equality ignores the argument names.
1685    * TypeUtil.composite tries to preserve argument names but gives up quickly
1686      if there is a mismatch.
1687
1688    installation script:
1689    * attempts to use "curl" if available (unless "wget" is available as well)
1690
1691    CM:
1692    * has an experimental implementation of "portable graphs" which I will
1693      soon propose as an implementation-independent library format
1694    * there are also new libraries $/pgraph.cm and$/pgraph-util.cm
1695
1696    NLFFI-LIB:
1697    * some cleanup (all cosmetic)
1698
1699    NLFFIGEN:
1700    * temporarily disabled the mechanism that suppresses ML output for
1702    * generate val bindings for enum constants
1703    * user can request that only one style (light or heavy) is being used;
1704      default is to use both (command-line arguments: -heavy and -light)
1705    * fixed bug in handling of function types involving incomplete pointers
1706    * generate ML entry points that take record arguments (i.e., using
1707      named arguments) for C functions that have a prototype with named
1708      arguments
1709      (see changes to CKIT)
1710
1711    ----------------------------------------------------------------------
1712    Name: Allen Leung
1713    Date: 2001/10/27 20:34:00 EDT
1714    Tag: leunga-20011027-x86-fast-fp-call
1715    Description:
1716
1717       Fixed the bug described in blume-20010920-slowfp.
1718
1719       The fix involves
1720          1. generating FCOPYs in FSTP in ia32-svid
1721          2. marking a CALL with the appropriate annotation
1722
1723    ----------------------------------------------------------------------
1724    Name: Matthias Blume
1725    Date: 2001/10/16 11:32:00 EDT
1726    Tag: blume-20011016-netbsd
1727    Description:
1728
1729    Underscore patch from Chris Richards (fixing problem with compiling
1730    runtime system under recent NetBSD).
1731
1732    ----------------------------------------------------------------------
1733    Name: Allen Leung
1734    Date: 2001/10/12 17:18:32 EDT 2001
1735    Tag: leung-20011012-x86-printflowgraph
1736    Description:
1737
1738    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
1739
1740    ----------------------------------------------------------------------
1741    Name: Lal George
1742    Date: 2001/10/11 23:51:34 EDT
1743    Tag: george-20011011-too-many-instrs
1744    Description:
1745
1746    The representation of a program point never expected to see more
1747    than 65536 instructions in a basic block!
1748
1749    ----------------------------------------------------------------------
1750    Name: Lal George
1751    Date: 2001/10/09 09:41:37 EDT
1752    Tag: george-20011008-mlrisc-labels
1753    Description:
1754
1755    Changed the machine description files to support printing of
1756    local and global labels in assembly code, based on host assembler
1757    conventions.
1758
1759    ----------------------------------------------------------------------
1760    Name: Matthias Blume
1761    Date: 2001/09/25 15:25:00 EDT
1762    Tag: blume-20010925-exninfo
1763    Description:
1764
1765    I provided a non-hook implementation of exnName (at the toplevel) and
1766    made the "dummy" implementation of exnMessage (at the toplevel) more
1767    useful: if nothing gets "hooked in", then at least you are going to
1768    see the exception name and a message indicating why you don't see more.
1769
1770    [For the time being, programs that need exnMessage and want to use
1771    ml-build should either use General.exnMessage (strongly recommended) or
1772    refer to structure General at some other point so that CM sees a
1773    static dependency.]
1774
1775    [Similar remarks go for "print" and "use":  If you want to use their
1776    functionality in stand-alone programs generated by ml-build, then use
1777    TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).] 1778 1779 ---------------------------------------------------------------------- 1780 Name: Matthias Blume 1781 Date: 2001/09/20 17:28:00 EDT 1782 Tag: blume-20010920-slowfp 1783 Description: 1784 1785 Allen says that x86-fast-fp is not safe yet, so I turned it off again... 1786 1787 ---------------------------------------------------------------------- 1788 Name: Matthias Blume 1789 Date: 2001/09/20 17:20:00 EDT 1790 Tag: blume-20010920-canonicalpaths 1791 Description: 1792 1793 0. Updated the BOOT file (something that I forgot to do earlier). 1794 1795 1. Small internal change to CM so that it avoids "/../" in filenames 1796 as much as possible (but only where it is safe). 1797 1798 2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so 1799 that arguments that contain delimiters are passed through correctly. 1800 This change also means that all "special" arguments of the form 1801 @SMLxxx... must come first. 1802 1803 3. Changed install script to put relative anchor names for tool commands 1804 into pathconfig. 1805 1806 ---------------------------------------------------------------------- 1807 Name: Matthias Blume 1808 Date: 2001/09/18 15:35:00 EDT 1809 Tag: blume-20010918-readme11036 1810 Description: 1811 1812 Added README files. 1813 1814 ---------------------------------------------------------------------- 1815 Name: Matthias Blume 1816 Date: 2001/09/18 11:45:00 EDT 1817 Tag: Release_110_36 (retag) 1818 Description: 1819 1820 Fixed mistake in config/preloads. Retagged as 110.36. 1821 1822 ---------------------------------------------------------------------- 1823 Name: Matthias Blume 1824 Date: 2001/09/18 09:40:00 EDT 1825 Tag: Release_110_36_orig (tag changed) 1826 Description: 1827 1828 New version (110.36). New bootfiles. 1829 1830 ---------------------------------------------------------------------- 1831 Name: Matthias Blume 1832 Date: 2001/09/14 16:15:00 EDT 1833 Tag: blume-20010914-x86fastfp 1834 Description: 1835 1836 John committed some changes that Allen made, in particular a (hopefully) 1837 correctly working version of the x86-fp module. 1838 1839 I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp" 1840 flag to "true". Everything seems to compile to a fixpoint ok, and 1841 "mandelbrot" speeds up by about 15%. 1842 1843 ---------------------------------------------------------------------- 1844 Name: Matthias Blume 1845 Date: 2001/09/13 11:20:00 EDT 1846 Tag: blume-20010913-minimal 1847 Description: 1848 1849 1. Stefan Monnier's patch to fix a miscompilation problem that 1850 was brought to light by John Reppy's work on Moby. 1851 1852 2. Implemented a minimal "structure Compiler" that contains just 1853 "version" and "architecture". The minimal version will be 1854 available when the full version is not. This is for backward- 1855 compatibility with code that wants to test Compiler.version. 1856 1857 ---------------------------------------------------------------------- 1858 Name: Matthias Blume 1859 Date: 2001/08/28 14:03:00 EDT 1860 Tag: blume-20010828-ml-lex 1861 Description: 1862 1863 Fix for bug 1581, received from Neophytos Michael. 1864 1865 ---------------------------------------------------------------------- 1866 Name: Matthias Blume 1867 Date: 2001/08/27 11:20:00 EDT 1868 Tag: blume-20010827-readme11035 1869 Description: 1870 1871 Fleshed out the README file for 110.35. 1872 1873 ---------------------------------------------------------------------- 1874 Name: Matthias Blume 1875 Date: 2001/08/24 17:10:00 EDT 1876 Tag: Release_110_35 1877 Description: 1878 1879 New version number (110.35). New bootfiles. 1880 1881 ---------------------------------------------------------------------- 1882 Name: Lal George 1883 Date: 2001/08/24 13:47:18 EDT 2001 1884 Tag: george-20010824-MLRISC-graphs 1885 Description: 1886 1887 removed clusters from MLRISC completely and replaced with graphs. 1888 1889 ---------------------------------------------------------------------- 1890 Name: Matthias Blume 1891 Date: 2001/08/23 17:50:00 EDT 1892 Tag: blume-20010823-toplevel 1893 Description: 1894 1895 - some reorganization of the code that implements various kinds of 1896 environments in the compiler (static, dynamic, symbolic, combined) 1897 - re-implemented the EnvRef module so that evalStream works properly 1898 (if the stream contains references to "use", "CM.make", etc.) 1899 - cleaned up evalloop.sml and interact.sml (but they need more cleaning) 1900 1901 ---------------------------------------------------------------------- 1902 Name: Matthias Blume 1903 Date: 2001/08/20 15:50 EDT 1904 Tag: blume20010820-slipup 1905 Description: 1906 1907 I forgot to commit a few files. Here they are... 1908 1909 ---------------------------------------------------------------------- 1910 Name: Matthias Blume 1911 Date: 2001/08/20 15:35:00 EDT 1912 Tag: blume-20010820-debugprof 1913 Description: 1914 1915 !!!! NEW BOOTFILES !!!! 1916 1917 This is another round of reorganizing the compiler sources. This 1918 time the main goal was to factor out all the "instrumentation" 1919 passes (for profiling and backtracing) into their own library. 1920 The difficulty was to do it in such a way that it does not depend 1921 on elaborate.cm but only on elabdata.cm. 1922 1923 Therefore there have been further changes to both elaborate.cm and 1924 elabdata.cm -- more "generic" things have been moved from the former 1925 to the latter. As a result, I was forced to split the assignment 1926 of numbers indicating "primtyc"s into two portions: SML-generic and 1927 SML/NJ-specific. Since it would have been awkward to maintain, 1928 I bit the bullet and actually _changed_ the mapping between these 1929 numbers and primtycs. The bottom line of this is that you need 1930 a new set of bin- and bootfiles. 1931 1932 I have built new bootfiles for all architectures, so doing a fresh 1933 checkout and config/install.sh should be all you need. 1934 1935 The newly created library's name is 1936 1937$smlnj/viscomp/debugprof.cm
1938
1939    and its sources live under
1940
1941        src/compiler/DebugProf
1942
1943    ----------------------------------------------------------------------
1944    Name: Matthias Blume
1945    Date: 2001/08/15 17:15:00 EDT
1946    Tag: blume-20010815-compreorg
1947    Description:
1948
1949    This is a first cut at reorganizing the CM libraries that make up the
1950    core of the compiler.  The idea is to separate out pieces that could
1951    be used independently by tools, e.g., the parser, the typechecker, etc.
1952
1953    The current status is a step in this direction, but it is not quite
1954    satisfactory yet.  Expect more changes in the future.
1955
1956    Here is the current (new) organization...
1957
1958        What used to be $smlnj/viscomp/core.cm is now divided into 1959 six CM libraries: 1960 1961$smlnj/viscomp/basics.cm
1962                           /parser.cm
1963                           /elabdata.cm
1964                           /elaborate.cm
1965                           /execute.cm
1966                           /core.cm
1967
1968        The CM files for these libraries live under src/system/smlnj/viscomp.
1969        All these libraries are proxy libraries that contain precisely
1970        one CM library component.  Here are the locations of the components
1971        (all within the src/compiler tree):
1972
1973             Basics/basics.cm
1974             Parse/parser.cm
1975             ElabData/elabdata.cm
1976             Elaborator/elaborate.cm
1977             Execution/execute.cm
1978             core.cm
1979
1980         [This organization is the same that has been used already
1981         for a while for the architecture-specific parts of the visible
1982         compiler and for the old version of core.cm.]
1983
1984         As you will notice, many source files have been moved from their
1985         respective original locations to a new home in one of the above
1986         subtrees.
1987
1988         The division of labor between the new libraries is the following:
1989
1990             basics.cm:
1991                - Simple, basic definitions that pertain to many (or all) of
1992                  the other libraries.
1993             parser.cm:
1994                - The SML parser, producing output of type Ast.dec.
1995                - The type family for Ast is also defined and exported here.
1996             elabdata.cm:
1997                - The datatypes that describe input and output of the elaborator.
1998                  This includes types, absyn, and static environments.
1999             elaborator.cm:
2000                - The SML/NJ type checker and elaborator.
2001                  This maps an Ast.dec (with a given static environment) to
2002                  an Absyn.dec (with a new static environment).
2003                - This libraries implements certain modules that used to be
2004                  structures as functors (to remove dependencies on FLINT).
2005             execute.cm:
2006                - Everything having to do with executing binary code objects.
2007                - Dynamic environments.
2008             core.cm:
2009                - SML/NJ-specific instantiations of the elaborator and MLRISC.
2010                - Top-level modules.
2011                - FLINT (this should eventually become its own library)
2012
2013    Notes:
2014
2015    I am not 100% happy with the way I separated the elaborator (and its
2016    data structures) from FLINT.  Two instances of the same problem:
2017
2018        1. Data structures contain certain fields that carry FLINT-specific
2019           information.  I hacked around this using exn and the property list
2020           module from smlnj-lib.  But the fact that there are middle-end
2021           specific fields around at all is a bit annoying.
2022
2023        2. The elaborator calculates certain FLINT-related information.  I tried
2024           to make this as abstract as I could using functorization, but, again,
2025           the fact that the elaborator has to perform calculations on behalf
2026           of the middle-end at all is not nice.
2027
2028        3. Having to used exn and property lists is unfortunate because it
2029           weakens type checking.  The other alternative (parameterizing
2030           nearly *everything*) is not appealing, though.
2031
2032    I removed the "rebinding =" warning hack because due to the new organization
2033    it was awkward to maintain it.  As a result, the compiler now issues some of
2034    these warnings when compiling init.cmi during bootstrap compilation. On
2035    the plus side, you also get a warning when you do, for example:
2036       val op = = Int32.+
2037    which was not the case up to now.
2038
2039    I placed "assign" and "deref" into the _Core structure so that the
2040    code that deals with the "lazy" keyword can find them there.  This
2042    during elaboration.
2043
2044    ----------------------------------------------------------------------
2045    Name: Matthias Blume
2046    Date: 2001/08/13
2047    Tag: blume-20010813-closures
2048    Description:
2049
2050    This fix was sent to us by Zhong Shao.  It is supposed to improve the
2051    performance of certain loops by avoiding needless closure allocation.
2052
2053    ----------------------------------------------------------------------
2054    Name: Lal George
2055    Date: 2001/07/31 10:03:23 EDT 2001
2056    Tag: george-20010731-x86-fmalloc
2057    Description: Fixed bug in x86 calls
2058
2059        There was a bug where call instructions would mysteriously
2060        vanish. The call instruction had to be one that returned
2061        a floating point value.
2062
2063    ----------------------------------------------------------------------
2064    Name: Lal George
2065    Date: 2001/07/19 16:36:29 EDT 2001
2066    Tag: george-20010719-simple-cells
2067    Description:
2068
2069    I have dramatically simplified the interface for CELLS in MLRISC.
2070
2071    In summary, the cells interface is broken up into three parts:
2072
2073      1. CellsBasis : CELLS_BASIS
2074
2075            CellsBasis is a top level structure and common for all
2076            architectures.  it contains the definitions of basic datatypes
2077            and utility  functions over these types.
2078
2079      2. functor Cells() : CELLS
2080
2081            Cells generates an interface for CELLS that incorporates the
2082            specific resources on the target architecture, such as the
2083            presence of special register classes, their number and size,
2084            and various useful substructures.
2085
2086      3. <ARCH>CELLS
2087
2088            e.g. SparcCells: SPARCCELLS
2089
2090            <ARCH>CELLS usually contains additional bindings for special
2091            registers  on the architecture, such as:
2092
2093                    val r0 : cell           (* register zero *)
2094                    val y : cell            (* Y register *)
2095                    val psr : cell          (* processor status register *)
2096                    ...
2097
2098            The structure returned by applying the Cells functor is opened
2099            in this interface.
2100
2101    The main implication of all this is that the datatypes for cells is
2102    split between CellsBasis and CELLS -- a fairly simple change for user
2103    code.
2104
2105    In the old scheme the CELLS interface had a definitional binding of
2106    the form:
2107
2108            signature CELLS = sig
2109
2110               structure CellsBasis = CellsBasis
2111
2112               ...
2113
2114            end
2115
2116    With all the sharing constraints that goes on in MLRISC, this old
2117    design  quickly leads to errors such as:
2118
2119            "structure definition spec inside of sharing ... "
2120
2121
2122    and appears to require an unacceptable amount of sharing and where
2123    constraint hackery.
2124
2125    I think this error message (the interaction of definitional specs and
2126    sharing) requires more explanation on our web page.
2127
2128    ----------------------------------------------------------------------
2129    Name: Matthias Blume
2130    Date: 2001/07/19 15:00:00 EDT
2131    Tag: blume-20010719-libreorg
2132    Description:
2133
2134    This update puts together a fairly extensive but straightforward change
2135    to the way the libraries that implement the interactive system are
2136    organized:
2137
2138       The biggest change is the elimination of structure Compiler.  As a
2139       replacement for this structure, there is now a CM library
2140       (known as $smlnj/compiler.cm or$smlnj/compiler/current.cm)
2141       that exports all the substructures of the original structure Compiler
2142       directly.  So instead of saying Compiler.Foo.bar one now simply
2143       says Foo.bar.  (The CM libraries actually export a collection of
2144       structures that is richer than the collection of substructures of
2145       structure Compiler.)
2146
2147       To make the transition smooth, there is a separate library called
2148       $smlnj/compiler/compiler.cm which puts together and exports the 2149 original structure Compiler (or at least something very close to it). 2150 2151 There are five members of the original structure Compiler 2152 that are not exported directly but which instead became members 2153 of a new structure Backend (described by signature BACKEND). These are: 2154 structure Profile (: PROFILE), structure Compile (: COMPILE), structure 2155 Interact (: INTERACT), structure Machine (: MACHINE), and val 2156 architecture (: string). 2157 2158 Structure Compiler.Version has become structure CompilerVersion. 2159 2160 Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided 2161 by$smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc,
2162       or x86, respectively.
2163       Each of these exports the same frontend structures that
2164       $smlnj/compiler.cm exports. But they do not have a structure Backend 2165 and instead export some structure <Arch>Backend where <Arch> is Alpha32, 2166 Hppa, PPC, Sparc, or X86, respectively. 2167 2168 Library$smlnj/compiler/all.cm exports the union of the exports of
2169       $smlnj/compiler/<arch>.cm 2170 2171 There are no structures <Arch>Compiler anymore, use 2172$smlnj/compiler/<arch>.cm instead.
2173
2174       Library host-compiler-0.cm is gone.  Instead, the internal library
2175       that instantiates CM is now called cm0.cm.  Selection of the host
2176       compiler (backend) is no longer done here but. (Responsibility for it
2177       now lies with $smlnj/compiler/current.cm. This seems to be more 2178 logical.) 2179 2180 Many individual files have been moved or renamed. Some files have 2181 been split into multiple files, and some "dead" files have been deleted. 2182 2183 Aside from these changes to library organization, there are also changes 2184 to the way the code itself is organized: 2185 2186 Structure Binfile has been re-implemented in such a way that it no 2187 longer needs any knowledge of the compiler. It exclusively deals 2188 with the details of binfile layout. It no longer invokes the 2189 compiler (for the purpose of creating new prospective binfile 2190 content), and it no longer has any knowledge of how to interpret 2191 pickles. 2192 2193 Structure Compile (: COMPILE) has been stripped down to the bare 2194 essentials of compilation. It no longer deals with linking/execution. 2195 The interface has been cleaned up considerably. 2196 2197 Utility routines for dealing with linking and execution have been 2198 moved into their own substructures. 2199 2200 (The ultimate goal of these changes is to provide a light-weight 2201 binfile loader/linker (at least for, e.g., stable libraries) that 2202 does not require CM or the compiler to be present.) 2203 2204 CM documentation has been updated to reflect the changes to library 2205 organization. 2206 2207 ---------------------------------------------------------------------- 2208 Name: Matthias Blume 2209 Date: 2001/07/10 17:30:00 EDT 2210 Tag: Release_110_34 2211 Description: 2212 2213 Minor tweak to 110.34 (re-tagged): 2214 2215 - README.html file added to CVS repository 2216 - runtime compiles properly under FreeBSD 3.X and 4.X 2217 2218 ---------------------------------------------------------------------- 2219 Name: Matthias Blume 2220 Date: 2001/07/10 17:30:00 EDT 2221 Tag: Release_110_34 2222 Description: 2223 2224 New version number (110.34). New bootfiles. 2225 2226 ---------------------------------------------------------------------- 2227 Name: Matthias Blume 2228 Date: 2001/07/09 16:00:00 EDT 2229 Tag: blume-20010709-more-varargs 2230 Description: 2231 2232 I changed the handling of varargs in ml-nlffigen again: 2233 The ellipsis ... will now simply be ignored (with an accompanying warning). 2234 2235 The immediate effect is that you can actually call a varargs function 2236 from ML -- but you can't actually supply any arguments beyond the ones 2237 specified explicitly. (For example, you can call printf with its format 2238 string, but you cannot pass additional arguments.) 2239 2240 This behavior is only marginally more useful than the one before, but 2241 it has the advantage that a function or, more importantly, a function 2242 type never gets dropped on the floor, thus avoiding follow-up problems with 2243 other types that refer to the offending one. 2244 2245 ---------------------------------------------------------------------- 2246 Name: Matthias Blume 2247 Date: 2001/07/09 11:25:00 EDT 2248 Tag: blume-20010709-varargs 2249 Description: 2250 2251 1. ckit-lib.cm now exports structure Error 2252 2. ml-nlffigen reports occurences of "..." (i.e., varargs function types) 2253 with a warning accompanied by a source location. Moreover, it 2254 merely skips the offending function or type and proceeds with the 2255 rest of its work.u As a result, one can safely feed C code containing 2256 "..." to ml-nlffigen. 2257 3. There are some internal improvements to CM, providing slightly 2258 more general string substitutions in the tools subsystem. 2259 2260 ---------------------------------------------------------------------- 2261 Name: Matthias Blume 2262 Date: 2001/06/27 15:10:00 EDT 2263 Tag: blume-20010627-concur 2264 Description: 2265 2266 Fixed a small bug in CM's handling of parallel compilation. 2267 (You could observe the bug by Control-C-interrupting an ordinary 2268 CMB.make or CM.stabilize and then attaching some compile servers. 2269 The result was that all of a sudden the previously interrupted 2270 compilation would continue on its own. This was because of 2271 an over-optimization: CM did not bother to clean out certain queues 2272 when no servers were attached "anyway", resulting in the contents 2273 of these queues to grab control when new servers did get attached.) 2274 2275 There is also another minor update to the CM manual. 2276 2277 ---------------------------------------------------------------------- 2278 Name: Matthias Blume 2279 Date: 2001/06/26 16:15:00 EDT 2280 Tag: blume-20010626-cmdoc 2281 Description: 2282 2283 Minor typo fixed in CM manual (syntax diagram for libraries). 2284 2285 ---------------------------------------------------------------------- 2286 Name: Matthias Blume 2287 Date: 2001/06/25 22:55:00 EDT 2288 Tag: blume-20010625-x86pc 2289 Description: 2290 2291 Fixed a nasty bug in the X86 assembly code that caused signal 2292 handlers to fail (crash) randomly. 2293 2294 ---------------------------------------------------------------------- 2295 Name: Matthias Blume 2296 Date: 2001/06/25 12:05:00 EDT 2297 Tag: blume-20010625-nlffigen 2298 Description: 2299 2300 This update fixes a number of minor bugs in ml-nlffigen as reported by 2301 Nick Carter <nbc@andrew.cmu.edu>. 2302 2303 1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted. 2304 2. Default names for generated files are now derived from the name of 2305 the C file *without its directory*. In particular, this causes generated 2306 files to be placed locally even if the C file is in some system directory. 2307 3. Default names for generated signatures and structures are also derived 2308 from the C file name without its directory. This avoids silly things 2309 like "structure GL/GL". 2310 (Other silly names are still possible because ml-nlffigen does not do 2311 a thorough check of whether generated names are legal ML identifiers. 2312 When in doubt, use command line arguments to force particular names.) 2313 2314 ---------------------------------------------------------------------- 2315 Name: Matthias Blume 2316 Date: 2001/06/21 12:25:00 EDT 2317 Tag: blume-20010621-eXene 2318 Description: 2319 2320 eXene now compiles and (sort of) works again. 2321 2322 The library name (for version > 110.33) is$/eXene.cm.
2323
2324    I also added an new example in src/eXene/examples/nbody.  See the
2325    README file there for details.
2326
2327    ----------------------------------------------------------------------
2328    Name: Matthias Blume
2329    Date: 2001/06/20 16:40:00 EDT
2330    Tag: blume-20010620-cml
2331    Description:
2332
2333    CML now compiles and works again.
2334
2335    Libraries (for version > 110.33):
2336
2337      $cml/cml.cm Main CML library. 2338$cml/basis.cm          CML's version of $/basis.cm. 2339$cml/cml-internal.cm   Internal helper library.
2340      $cml/core-cml.cm Internal helper library. 2341$cml-lib/trace-cml.cm  Tracing facility.
2342      $cml-lib/smlnj-lib.cm CML's version of$/smlnj-lib.cm
2343
2344    The installer (config/install.sh) has been taught how to properly
2345    install this stuff.
2346
2347    ----------------------------------------------------------------------
2348    Name: Matthias Blume
2349    Date: 2001/06/19 17:55:00 EDT
2350    Tag: blume-20010619-instantiate
2351    Description:
2352
2353    This un-breaks the fix for bug 1432.
2354    (The bug was originally fixed in 110.9 but I broke it again some
2355    time after that.)
2356
2357    ----------------------------------------------------------------------
2358    Name: Matthias Blume
2359    Date: 2001/06/19 17:25:00 EDT
2360    Tag: blume-20010619-signals
2361    Description:
2362
2363    This should (hopefully) fix the long-standing signal handling bug.
2364    (The runtime system was constructing a continuation record with an
2365    incorrect descriptor which would cause the GC to drop data on the floor...)
2366
2367    ----------------------------------------------------------------------
2368    Name: Matthias Blume
2369    Date: 2001/06/15 15:05:00 EDT
2370    Tag: blume-20010615-moresparc
2371    Description:
2372
2373    Here is a short late-hour update related to Sparc c-calls:
2374
2375     -- made handling of double-word arguments a bit smarter
2376
2377     -- instruction selection phase tries to collapse certain clumsily
2378        constructed ML-Trees; typical example:
2379
2381
2382        This currently has no further impact on SML/NJ since mlriscGen does
2383        not seem to generate such patterns in the first place, and c-calls
2384        (which did generate them in the beginning) has meanwhile been fixed
2385        so as to avoid them as well.
2386
2387    ----------------------------------------------------------------------
2388    Name: Matthias Blume
2389    Date: 2001/06/15 15:05:00 EDT
2390    Tag: blume-20010615-sparc
2391    Description:
2392
2393    The purpose of this update is to provide an implementation of NLFFI
2394    on Sparc machines.
2395
2396    Here are the changes in detail:
2397
2398       * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing
2399       the Sparc implementation of the c-calls API.
2400       * The Sparc backend of SML/NJ has been modified to uniformely use %fp
2401       for accessing the ML frame.  Thus, we have a real frame pointer and
2402       can freely modify %sp without need for an omit-frame-ptr phase.
2403       The vfp logic in src/compiler/CodeGen/* has been changed to accomodate
2404       this case.
2405       * ml-nlffigen has been taught to produce code for different architectures
2406       and calling conventions.
2407       * In a way similar to what was done in the x86 case, the Sparc
2408       backend uses its own specific extension to mltree.  (For example,
2409       it needs to be able to generate UNIMP instructions which are part
2410       of the calling convention.)
2411       * ml-nlffi-lib was reorganized to make it more modular (in particular,
2412       to make it easier to plug in new machine- and os-dependent parts).
2413
2414    There are some other fairly unrelated bug fixes and cleanups as well:
2415
2416       * I further hacked the .cm files for MLRISC tools (like MDLGen) so
2417       that they properly share their libraries with existing SML/NJ libraries.
2418       * I fixed a minor cosmetic bug in CM, supressing certain spurious
2419       follow-up error messages.
2420       * Updates to CM/CMB documentation.
2421
2422    TODO items:
2423
2424       * MLRISC should use a different register as its asmTemp on the Sparc.
2425         (The current %o2 is a really bad choice because it is part of the
2426         calling conventions, so things might interfere in unexpected ways.)
2427
2428    ----------------------------------------------------------------------
2429    Name: Matthias Blume
2430    Date: 2001/06/07
2431    Tag: blume-20010607-calls
2432    Description:
2433
2434    A number of internal changes related to C calls and calling conventions:
2435
2436    1. ML-Tree CALL statements now carry a "pops" field.  It indicates the
2437       number of bytes popped implicitly (by the callee).  In most cases
2438       this field is 0 but on x86/win32 it is some non-zero value.  This
2439       is information provided for the benefit of the "omit-frameptr" pass.
2440    2. The CALL instruction on the x86 carries a similar "pops" field.
2441       The instruction selection phase copies its value from the ML-Tree
2442       CALL statement.
2443    3. On all other architectures, the instruction selection phase checks
2444       whether "pops=0" and complains if not.
2445    4. The c-calls implementation for x86 now accepts two calling conventions:
2446       "ccall" and "stdcall".  When "ccall" is selected, the caller cleans
2447       up after the call and pops is set to 0.  For "stdcall", the caller
2448       does nothing, leaving the cleanup to the callee; pops is set to
2449       the number of bytes that were pushed onto the stack.
2450    5. The cproto decoder (compiler/Semant/types/cproto.sml) now can
2451       distinguish between "ccall" and "stdcall".
2452    6. The UNIMP instruction has been added to the supported Sparc instruction
2453       set. (This is needed for implementing the official C calling convention
2454       on this architecture.)
2455    7. I fixed some of the .cm files under src/MLRISC/Tools to make them
2456       work with the latest CM.
2457
2458    ----------------------------------------------------------------------
2459    Name: Matthias Blume
2460    Date: 2001/06/05 15:10:00 EDT
2461    Tag: blume-20010605-cm-index
2462    Description:
2463
2464    0. The "lambdasplit" parameter for class "sml" in CM has been documented.
2465
2466    1. CM can now generate "index files".  These are human-readable files
2467       that list on a per-.cm-file basis each toplevel symbol defined or
2468       imported.  The location of the index file for
2469       <p>/<d>.cm is <p>/CM/INDEX/<d>.cm.
2470       To enable index-file generation, set CM.Control.generate_index to true
2471       or export an environment-symbol: export CM_GENERATE_INDEX=true.
2472
2473       The CM manual has been updated accordingly.
2474
2475    2. I made some slight modifications to the c-calls API in MLRISC.
2476
2477         a) There is now a callback to support saving/restoring of
2478            dedicated but caller-save registers around the actual call
2479            instruction.
2480         b) One can optionally specify a comment-annotation for the
2481            call instruction.
2482
2483    3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop.
2484       (For example, the comment annotation shows the C prototype of
2485        the function being called.)
2486
2487    ----------------------------------------------------------------------
2488    Name: Matthias Blume
2489    Date: 2001/06/01 13:30:00 EDT
2490    Tag: blume-20010601-nlffi-cleanup
2491    Description:
2492
2493    This is mostly a cleanup of MLFFI stuff:
2494
2495       - some signature files have been put into a more exposed place
2496       - the ugly 'f type parameter is gone (simplifies types tremendously!)
2497       - ml-nlffigen changed accordingly
2498       - tutorial updated
2499
2500    Other changes:
2501
2502       - author's affiliation in CM manual(s) updated
2503       - some more recognized keywords added to Allen's sml.sty
2504
2505    ----------------------------------------------------------------------
2506    Name: Matthias Blume
2507    Date: 2001/05/25 15:30:00 EDT
2508    Tag: blume-20010525-iptr
2509    Description:
2510
2511      - put the official 110.33-README (as it appears on the ftp server) under
2512        CVS
2513      - fixed a small bug related to incomplete pointer types in
2514        ml-nlffigen
2515      - small cosmetic change to the ml-nlffi-lib's "arr" type constructor
2516        (it does not need the 'f type parameter)
2517
2518    ----------------------------------------------------------------------
2519    Name: Matthias Blume
2520    Date: 2001/05/23 14:30:00 EDT
2521    Tag: Release_110_33
2522    Description:
2523
2524    New version number (110.33).  New bootfiles.
2525
2526    ----------------------------------------------------------------------
2527    Name: Matthias Blume
2528    Date: 2001/05/22 18:06:00 EDT
2529    Tag: blume-20010522-targets
2530    Description:
2531
2532    Made install.sh use file config/targets.customized if it exists, falling
2533    back to config/targets if it doesn't.  This way one can have a customized
2534    version of the targets file without touching the "real thing", thus
2535    eliminating the constant fear of accidentally checking something bogus
2536    back into the CVS repository...  (File config/targets.customized must
2537    not be added to the repository!)
2538
2539    ----------------------------------------------------------------------
2540    Name: Matthias Blume
2541    Date: 2001/05/22 16:30:00 EDT
2542    Tag: blume-20010522-minitut
2543    Description:
2544
2545    1. Bug fix in ml-nlffigen; now (hopefully) correctly handling
2546       struct returns.
2547    2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt.  This is some very
2548       incomplete, preliminary documentation for NLFFI.
2549
2550    ----------------------------------------------------------------------
2551    Name: Matthias Blume
2552    Date: 2001/05/14 11:30:00 EDT
2553    Tag: blume-20010514-script
2554    Description:
2555
2556    Some bugs in install script fixed.
2557
2558    In addition to that I also made a slight change to the NLFFI API:
2559    Functors generated by ml-nlffigen now take the dynamic library as a
2560    straight functor argument, not as a suspended one.  (The original
2561    functor code used to force the suspension right away anyway, so there
2562    was nothing gained by this complication of the interface.)
2563
2564    ----------------------------------------------------------------------
2565    Name: Matthias Blume
2566    Date: 2001/05/11 14:35:00 EDT
2567    Tag: blume-20010511-ml-nlffi
2568    Description:
2569
2570    I finally took the plunge and added my new FFI code to the main
2571    repository.  For x86-linux it is now ready for prime-time.
2572
2573    There are two new subdirectories of "src":
2574
2575      - ml-nlffi-lib:
2576           The utility library for programs using the FFI interface.
2577           Here is the implementation of $/c.cm and its associated low-level 2578 partners$/c-int.cm and $/memory.cm. 2579 - ml-nlffigen: 2580 A stand-alone program for generating ML glue code from C source 2581 code. 2582 2583 Building ml-nlffigen requires$/ckit-lib.cm.
2584
2585    The config/install.sh script has been updates to do the Right Thing
2586    (hopefully).
2587
2588    Notice that the source tree for the C-Kit will not be put under "src"
2589    but directly under the installation root directory.  (This is the
2590    structure that currently exists on the CVS server when you check out
2592
2593    Bugs: No documentation yet.
2594
2595    ----------------------------------------------------------------------
2596    Name: Matthias Blume
2597    Date: 2001/05/09 16:35:00 EDT
2598    Tag: blume-20010509-cpscontract
2599    Description:
2600
2601    Fixed a bug in the accounting code in cpsopt/contract.sml.  (The
2602    wrapper/unwrapper elimination did not decrement usage counts and some
2604
2605    ----------------------------------------------------------------------
2606    Name: Lal George
2607    Date: 2001/05/08  17:26:09 EDT
2608    Tag: george-20010508-omit-frameptr
2609    Description:
2610
2611    Changes to implement the omit-frame-pointer optimization to support
2612    raw C calls. For now, there is only support on the Intel x86, but
2613    other architectures will follow as more experience is gained with this.
2614
2615
2616    ----------------------------------------------------------------------
2617    Name: Matthias Blume
2618    Date: 2001/05/07 14:40:00 EDT
2619    Tag: blume-20010507-proxies
2620    Description:
2621
2622    I made into "proxy libraries" all libraries that qualify for such a
2623    change.  (A qualifying library is a library that has another library or
2624    groups as its sole member and repeats that member's export list
2625    verbatim.  A proxy library avoids this repetition by omitting its export
2626    list, effectively inheriting the list that its (only) member exports.
2627    See the CM manual for more explanation.)
2628    The main effect is that explicit export lists for these libraries
2629    do not have to be kepts in sync, making maintenance a bit easier.
2630
2632
2633    Last but not least, I made a new set of bootfiles.
2634
2635    ----------------------------------------------------------------------
2636    Name: Matthias Blume
2637    Date: 2001/05/04 17:00:00 EDT
2638    Tag: blume-20010504-cm-lsplit
2639    Description:
2640
2641    0. John merged pending changes to $/smlnj-lib.cm 2642 2643 1. Allen's previous change accidentally backed out of one of Lal's 2644 earlier changes. I undid this mistake (re-introducing Lal's change). 2645 2646 2. I used the new topOrder' function from graph-scc.sml (from$/smlnj-lib.cm)
2647       within the compiler where applicable.  There is some code simplification
2648       because of that.
2649
2650    3. The "split" phase (in FLINT) is now part of the default list of phases.
2651       Compiler.Control.LambdaSplitting.* can be used to globally control the
2652       lambda-splitting (cross-module-inlining) engine.  In addition to that,
2653       it can now also be controlled on a per-source basis: CM has been taught
2654       a new tool parameter applicable to ML source files.
2655
2656       - To turn lambda-splitting off completely:
2657            local open Compiler.Control.LambdaSplitting in
2658                val _ = set Off
2659            end
2660       - To make "no lambda-splitting" the global default (but allow per-source
2661         overriding); this is the initial setting:
2662            local open Compiler.Control.LambdaSplitting in
2663                val _ = set (Default NONE)
2664            end
2665       - To make "lambda-splitting with aggressiveness a" the global default
2666         (and allow per-source overriding):
2667            local open Compiler.Control.LambdaSplitting in
2668                val _ = set (Default (SOME a))
2669            end
2670
2671       - To turn lambda-splitting off for a given ML souce file (say: a.sml)
2672         write (in the respective .cm-file):
2673            a.sml (lambdasplitting:off)
2674       - To turn lambda-splitting for a.sml on with minimal aggressiveness:
2675            a.sml (lambdasplitting:on)
2676       - To turn lambda-splitting for a.sml on with aggressiveness <a> (where
2677         <a> is a decimal non-negative integer):
2678            a.sml (lambdasplitting:<a>)
2679       - To turn lambda-splitting for a.sml on with maximal aggressiveness:
2680            a.sml (lambdasplitting:infinity)
2681       - To use the global default for a.sml:
2682            a.sml (lambdasplitting:default)
2683         or simply
2684            a.sml
2685
2686    ----------------------------------------------------------------------
2687    Name: Allen Leung
2688    Date: 2001/05/04 01:57:00 EDT
2689    Tag: leunga-20010504-sync
2690    Description:
2691
2692      MLRISC features.
2693
2694      1. Fix to CMPXCHG instructions.
2695      2. Changed RA interface to allow annotations in callbacks.
2696      3. Added a new method to the stream interface to allow annotations updates.
2697
2698    ----------------------------------------------------------------------
2699    Name: Matthias Blume
2700    Date: 2001/05/01 11:45:00 EDT
2701    Tag: blume-20010501-pcedittmp
2702    Description:
2703
2704    Changed install.sh to use the current working directory instead of
2705    /usr/tmp for a temporary file (pcedittmp).  The previous choice
2706    of /usr/tmp caused trouble with MacOS X because of file premission
2707    problems.
2708
2709    ----------------------------------------------------------------------
2710    Name: Matthias Blume
2711    Date: 2001/04/20 11:10:00 EDT
2712    Tag: blume-20010420-inMLflag
2713    Description:
2714
2716       (for use by the raw-C-calls mechanism to implement proper interrupt
2717        handling)
2718     - made the ML compiler aware of various data-structure offsets so it
2719       can generate code for accessing the vp_inML flag and vp_limitPtrMask
2720     - tweaked mlriscGen.sml to have it emit interrupt-handling code for
2721       raw C-calls
2722
2723    ----------------------------------------------------------------------
2724    Name: Lal George
2725    Date: 2001/04/20 09:15:28 EDT
2726    Tag: george-20010420-macosX
2727    Description:
2728
2729     - Changes to port to Mac OS X; Darwin.
2730
2731     - In the process I found that sqrt was broken on the PPC, because the
2732       fsqrt instruction is not implemented.
2733
2734    ----------------------------------------------------------------------
2735    Name: Matthias Blume
2736    Date: 2001/04/18 12:45:00 EDT
2737    Tag: blume-20010418-ccalls
2738    Description:
2739
2740     - fixed two off-by-4 errors in the x86-specific c-calls implementation
2741       (this bug prevented structure arguments containing pointers from being
2742        passed correctly)
2743     - changed the raw-C-call code in mlriscGen.sml in such a way that
2744       structure arguments are represented as a pointer to the beginning
2745       of the structure (instead of having a series of synthesized arguments,
2746       one for each structure member)
2747
2748     - made makeml script's verbosity level configurable via environment
2749       variable (MAKEML_VERBOSITY)
2750
2751     - eliminated placeholder implementations for f32l, w16s, i16s, and f32s
2752       in rawmem-x86.sml; we are now using the real thing
2753
2754    ----------------------------------------------------------------------
2755    Name: Matthias Blume
2756    Date: 2001/03/22 16:25:00 EST
2757    Tag: blume-20010322-bootfiles
2758    Description:
2759
2760    Created a new set of bootfiles (for your automatic installation convenience).
2761
2762    ----------------------------------------------------------------------
2763    Name: Matthias Blume
2764    Date: 2001/03/22 15:10:00 EST
2765    Tag: blume-20010322-rawmem-parcm
2766    Description:
2767
2768    1. All "raw memory access" primitives for the new FFI are implemented now
2769       (at least on the x86).
2770    2. Some further cleanup of CM's parallel make mechanism.
2771
2772    ----------------------------------------------------------------------
2773    Name: Matthias Blume
2774    Date: 2001/03/19 17:53:00 EST
2775    Tag: blume-20010319-parallel
2776    Description:
2777
2778    Parallel make (using compile servers) now works again.
2779
2780    To this end, CM.stabilize and CMB.make have been modified to work in
2781    two passes when compile servers are attached:
2782       1. Compile everything, do not perform stabilization; this pass
2783          uses compile servers
2784       2. Stabilize everything; this pass does not use compile servers
2785    If there are no compile servers, the two passes are combined into one
2786    (as before).  Splitting the passes increases the inherent parallelism
2787    in the dependency graph because the entire graph including all
2788    libraries is available at the same time.  This, in turn, improves
2789    server utilization.  The downside is that the master process will
2790    have to do some extra work after compilation is done (because for
2791    technical reasons it must re-read all the binfiles during stabilization).
2792
2793    ----------------------------------------------------------------------
2794    Name: Matthias Blume
2795    Date: 2001/03/16 12:22:00 EST
2796    Tag: blume-20010316-bootfiles
2797    Description:
2798
2799    Created a new set of bootfiles (for your automatic installation convenience).
2800
2801    ----------------------------------------------------------------------
2802    Name: Matthias Blume
2803    Date: 2001/03/16 11:00:00 EST
2804    Tag: blume-20010316-MLTREE-fixup
2805    Description:
2806
2807    This is a minor fixup for an (untagged) earlier commit by Allen.
2808    (A file was missing).
2809
2810    ----------------------------------------------------------------------
2811    Name: Allen Leung
2812    Date: Mon Mar  5 18:54:57 EST 2001
2813    Tag: leunga-20010305-cut-support
2814
2815    1. New support for alternative control-flow in MLTREE.
2816       Currently we support
2817
2818          FLOW_TO(CALL ...., [k1,...,kn])
2819
2820       This is needed for 'cuts to' in C-- and try/handle-like constructs
2821       in Moby
2822
2823       New assembler flag "asm-show-cutsto" to turn on control-flow debugging.
2824
2825    2. Register Allocator
2826
2827       Changes in interface [from Fermin, John]
2828
2829    3. Alpha 8-bit SLL support [Fermin]
2830
2831    4. All architectures
2832
2833       A new module (ClusterExpandCopies) for expanding parallel copies.
2834
2835    ----------------------------------------------------------------------
2836    Name: Allen Leung
2837    Date: 2001/02/27 23:07:00 EST
2838    Tag: leunga-20010227-minor-stuff
2839
2840    1. Alpha bug fix for CMOVNE
2841    2. Handle mltree COND(..,FCMP ...,...)
2842    3. Bug fix in simplifier
2843
2844    ----------------------------------------------------------------------
2845    Name: Matthias Blume
2846    Date: 2001/01/30 17:50:00 EST
2847    Tag: blume-20010130-sync
2848    Description:
2849
2850    This is just a minor update to sync my devel branch with the main brach.
2851    The only visible change is the addition of some README files.
2852
2853    ----------------------------------------------------------------------
2854    Name: Matthias Blume
2855    Date: 2001/01/12 23:30:00 JST
2856    Tag: blume-20010112-bootfiles
2857    Description:
2858
2859    Made a new set of bootfiles that goes with the current state of the
2860    repository.
2861
2862    ----------------------------------------------------------------------
2863    Name: Matthias Blume
2864    Date: 2001/01/12 21:20:00 JST
2865    Tag: blume-20010112-sync
2866    Description:
2867
2868    I am just flushing out some minor changes that had accumulated in
2869    my private branch in order to sync with the main tree.  (This is
2870    mainly because I had CVS trouble when trying to merge _into_ my
2871    private branch.)
2872
2873    Most people should be completely unaffected by this.
2874
2875    ----------------------------------------------------------------------
2876    Name: Allen Leung
2877    Date: Thu Jan 11 21:03:00 EST 2001
2878    Tag: leunga-20010111-labexp=mltree
2879    Description:
2880
2881    1.  Removed the type LabelExp and replace it by MLTree.
2882    2.  Rewritten mltree-simplify with the pattern matcher tool.
2883    3.  There were some bugs in alpha code generator which would break
2884        64-bit code generation.
2885    4.  Redo the tools to generate code with the
2886    5.  The CM files in MLRISC (and in src/system/smlnj/MLRISC)
2887        are now generated by perl scripts.
2888
2889    ----------------------------------------------------------------------
2890    Name: Matthias Blume
2891    Date: 2001/01/10 21:55:00 JST
2892    Tag: blume-20010110-rcc
2893    Description:
2894
2895    The RCC stuff now seems to work (but only on the x86).
2896    This required hacking of the c-calls interface (and -implementation) in
2897    MLRISC.
2898
2899    Normal compiler users should be unaffected.
2900
2901    ----------------------------------------------------------------------
2902    Name: Matthias Blume
2903    Date: 2001/01/09 01:20:00 JST
2904    Tag: blume-20010109-rcc
2905    Description:
2906
2907    This is a fairly big patch, flushing out a large number of pending
2908    changes that I made to my development copy over the last couple of days.
2909
2910    Of practical relevance at this moment is a workaround for a pickling
2911    bug that Allen ran into the other day.  The cause of the bug itself is
2912    still unknown and it might be hard to fix it properly, but the
2913    workaround has some merits of its own (namely somewhat reducing pickling
2914    overhead for certain libraries).  Therefore, I think this solution should
2915    be satisfactory at this time.
2916
2917    The rest of the changes (i.e., the vast majority) has to do with my
2918    ongoing efforts of providing direct support for C function calls from
2919    ML.  At the moment there is a new primop "RAW_CCALL", typing magic
2920    in types/cproto.sml (invoked from FLINT/trans/translate.sml), a new
2921    case in the FLINT CPS datatype (RCC), changes to cps/convert.sml to
2922    translate uses of RAW_CCALL into RCC, and changes to mlriscGen.sml to
2923    handle RCC.
2924
2925    The last part (the changes to mlriscGen.sml) are still known to be
2926    wrong on the x86 and not implemented on all other architectures.  But
2927    the infrastructure is in place. I had to change a few functor
2928    signatures in the backend to be able to route the CCalls interface
2929    from MLRISC there, and I had to specialize the mltree type (on the
2930    x86) to include the necessary extensions. (The extensions themselves
2931    were already there and redy to go in MLRISC/x86).
2932
2933    Everything should be very happy as soon as someone helps me with
2934    mlriscGen.sml...
2935
2936    In any case, nothing of this should matter to anyone as long as the
2937    new primop is not being used (which is going to be the case unless you
2938    find it where I hid it :). The rest of the compiler is completely
2939    unaffected.
2940
2941    ----------------------------------------------------------------------
2942    Name: Matthias Blume
2943    Date: 2001/01/05 00:30:00 JST
2944    Tag: blume-20010105-primops
2945    Description:
2946
2947    Added some experimental support for work that I am doing right now.
2948    These changes mostly concern added primops, but there is also a new
2949    experimental C library in the runtime system (but currently not enabled
2950    anywhere except on Linux/X86).
2951
2952    In the course of adding primops (and playing with them), I discovered that
2953    Zhong's INL_PRIM hack (no type info for certain primops) was, in fact, badly
2954    broken.  (Zhong was very right he labeled this stuff as "major gross hack".)
2955    To recover, I made type information in INL_PRIM mandatory and changed
2956    prim.sml as well as built-in.sml accordingly.  The InLine structure now
2957    has complete, correct type information (i.e., no bottom types).
2958
2959    Since all these changes mean that we need new binfiles, I also bumped the
2960    version number to 110.32.1.
2961
2962    ----------------------------------------------------------------------
2963    Name: Matthias Blume
2964    Date: 2000/12/30 22:10:00 JST
2965    Tag: blume-20001230-various
2966    Description:
2967
2968    Added proxy libraries for MLRISC and let MLRISC libraries refer
2969    to each other using path anchors.  (See CM manual for explanation.)
2970
2971    Updated CM documentation.
2972
2973    Fixed some bugs in CM.
2974
2975    Implemented "proxy" libraries (= syntactic sugar for CM).
2976
2977    Added "-quiet" option to makeml and changed runtime system accordingly.
2978
2979    Added cleanup handler for exportML to reset timers and compiler stats.
2980
2981    ----------------------------------------------------------------------
2982    Name: Lal George
2983    Date: 2000/12/22 22:22:58 EST 2000
2984    Tag: Release_110_32
2985    Description:
2986
2987            Infinite precision used throughout MLRISC.
2988            see MLRISC/mltree/machine-int.sig
2989
2990    ----------------------------------------------------------------------
2991    Name: Matthias Blume
2992    Date: 2000/12/22 23:16:00 JST
2993    Tag: blume-20001222-warn
2994    Description:
2995
2996    Corrected wording and formatting of some CM warning message which I
2997    broke in my previous patch.
2998
2999    ----------------------------------------------------------------------
3000    Name: Matthias Blume
3001    Date: 2000/12/22 21:20:00 JST
3002    Tag: blume-20001222-anchorenv
3003    Description:
3004
3005    Fixed CM's handling of anchor environments in connection with CMB.make.
3006
3007    ----------------------------------------------------------------------
3008    Name: Matthias Blume
3009    Date: 2000/12/22 13:15:00 JST
3010    Tag: blume-20001222-cleanup
3011    Description:
3012
3013    Removed src/cm/ffi which does not (and did not) belong here.
3014
3015    ----------------------------------------------------------------------
3016    Name: Matthias Blume
3017    Date: 2000/12/21 23:55:00 JST
3018    Tag: blume-20001221-exn
3019    Description:
3020
3021    Probably most important: CM no longer silently swallows all exceptions
3022    in the compiler.
3023    Plus: some other minor CM changes.  For example, CM now reports some
3024    sizes for generated binfiles (code, data, envpickle, lambdapickle).
3025
3026    ----------------------------------------------------------------------
3027    Name: Matthias Blume
3028    Date: 2000/12/15 00:01:05 JST
3029    Tag: blume-20001215-dirtool
3030    Description:
3031
3033    - improvements and cleanup to Tools structure
3035
3036    ----------------------------------------------------------------------
3037    Name: Allen Leung
3038    Date: Thu Dec 14 03:45:24 EST 2000
3039    Description:
3040    Tag:  leunga-20001214-int-inf
3041    Description:
3042
3043       In IntInf, added these standard functions, which are missing from our
3044    implementation:
3045
3046        andb : int * int -> int
3047        xorb : int * int -> int
3048        orb  : int * int -> int
3049        notb : int -> int
3050         <<   : int * word -> int
3051        ~>>  : int * word -> int
3052
3053       Not tested, I hope they are correct.
3054
3055    ----------------------------------------------------------------------
3056    Name: Allen Leung
3057    Date: Fri Dec  8 19:23:26 EST 2000
3058    Description:
3059    Tag:  leunga-20001208-nowhere
3060    Description:
3061
3062      Slight improvements to the 'nowhere' tool to handle OR-patterns,
3063    to generate better error messages etc.  Plus a brief manual.
3064
3065  ----------------------------------------------------------------------  ----------------------------------------------------------------------
3066  Name: Lal George  Name: Lal George
3067  Date: 2000/12/08 09:54:02 EST 2000  Date: 2000/12/08 09:54:02 EST 2000
# Line 1730  Line 4781
4781       elaborator). There were a lot of changes during my "linkpath" trials       elaborator). There were a lot of changes during my "linkpath" trials
4782       that could have been reverted to their original state but weren't.       that could have been reverted to their original state but weren't.
4783       Please, don't be too harsh on me for messing with this code a bit more       Please, don't be too harsh on me for messing with this code a bit more
4784       than what was strictly necessary...  (I _did_ resist the tempation       than what was strictly necessary...  (I _did_ resist the temptation
4785       of doing any "global reformatting" to avoid an untimely death at       of doing any "global reformatting" to avoid an untimely death at
4786       Dave's hands. :)       Dave's hands. :)
4787

Legend:
 Removed from v.749 changed lines Added in v.1180