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

SCM Repository

[smlnj] Annotation of /sml/trunk/READMES/110.27-README
ViewVC logotype

Annotation of /sml/trunk/READMES/110.27-README

Parent Directory Parent Directory | Revision Log Revision Log


Revision 610 - (view) (download)

1 : dbm 610
2 :     S M L / N J
3 :    
4 :     1 1 0 . 2 7 N E W S
5 :    
6 :     April 10, 2000
7 :    
8 :     WARNING
9 :    
10 :     This version is intended for compiler hackers.
11 :     We are in the midst of substantial structural changes,
12 :     and this is a snapshot.
13 :    
14 :     http://cm.bell-labs.com/cm/cs/what/smlnj/index.html
15 :    
16 :     ------------------------------------------------------------------------
17 :     Summary:
18 :    
19 :     This version has some minor tweeks to FLINT (after the major merge
20 :     in 110.26). Work continues on tuning FLINT and the various optimizations
21 :     it implements.
22 :    
23 :     CM has been revised extensively, and the modmap environment mechanism
24 :     supporting stubbified pickles has been reworked completely. The pathconfig
25 :     file has been simplified. Installation scripts have been further
26 :     modified.
27 :    
28 :     MLRISC, and particularly the x86 back end have been modiefied extensively.
29 :    
30 :     There are a few updates to the SML/NJ Library
31 :    
32 :     Reported bug fixes:
33 :     1556. (jhr) signal race condition
34 :     Some CM bugs (not recorded)
35 :    
36 :    
37 :     The following change records are extracted from the CVS repository
38 :     HISTORY file, which started in early March, 2000, so this doesn't
39 :     cover changes between mid-December, 1999, when 110.26 was created, and
40 :     early March.
41 :    
42 :     ======================================================================
43 :     FLINT:
44 :     ======================================================================
45 :    
46 :     Name: Stefan
47 :     Date: 2000/04/07 10:00:00 EDT
48 :     Tag: monnier-20000406-branch-handling
49 :     Description:
50 :    
51 :     Improved handling of branches (mostly those generated from
52 :     polymorphic equality), removed switchoff and changed the
53 :     default optimization settings (more cpsopt and less flintopt).
54 :    
55 :    
56 :     ======================================================================
57 :     MLRISC:
58 :     ======================================================================
59 :    
60 :     Name: Allen Leung
61 :     Date: 2000/03/10 02:20:00
62 :     Tag: leunga-20000310-fix_x86_asm_ra
63 :     Description:
64 :    
65 :     More assembly output problems involving the indexed addressing mode
66 :     on the x86 have been found and corrected. Thanks to Fermin Reig for the
67 :     fix.
68 :    
69 :     The interface and implementation of the register allocator have been changed
70 :     slightly to accommodate the possibility to skip the register allocation
71 :     phases completely and go directly to memory allocation. This is needed
72 :     for C-- use.
73 :    
74 :     ----------------------------------------------------------------------
75 :     Name: Allen Leung
76 :     Date: 2000/03/22 01:23:00
77 :     Tag: leunga-20000322-fix_x86_hppa_ra
78 :     Description:
79 :    
80 :     1. X86 fixes/changes
81 :    
82 :     a. x86Rewrite bug with MUL3 (found by Lal)
83 :     b. Added the instructions FSTS, FSTL
84 :    
85 :     2. PA-RISC fixes/changes
86 :    
87 :     a. B label should not be a delay slot candidate! Why did this work?
88 :     b. ADDT(32, REG(32, r), LI n) now generates one instruction instead of two,
89 :     as it should be.
90 :     c. The assembly syntax for fstds and fstdd was wrong.
91 :     d. Added the composite instruction COMICLR/LDO, which is the immediate
92 :     operand variant of COMCLR/LDO.
93 :    
94 :     3. Generic MLRISC
95 :    
96 :     a. shuffle.sml rewritten to be slightly more efficient
97 :     b. DIV bug in mltree-simplify fixed (found by Fermin)
98 :    
99 :     4. Register Allocator
100 :    
101 :     a. I now release the interference graph earlier during spilling.
102 :     May improve memory usage.
103 :    
104 :     ----------------------------------------------------------------------
105 :     Name: Allen Leung
106 :     Date: 2000/03/23 16:25:00
107 :     Tag: leunga-20000323-fix_x86_alpha
108 :     Description:
109 :    
110 :     1. X86 fixes/changes
111 :    
112 :     a. The old code generated for SETcc was completely wrong.
113 :     The Intel optimization guide is VERY misleading.
114 :    
115 :     2. ALPHA fixes/changes
116 :    
117 :     a. Added the instructions LDBU, LDWU, STB, STW as per Fermin's suggestion.
118 :     b. Added a new mode byteWordLoadStores to the functor parameter to Alpha()
119 :     c. Added reassociation code for address computation.
120 :    
121 :     ----------------------------------------------------------------------
122 :     Name: Allen Leung
123 :     Date: 2000/03/29 18:00:00
124 :     Tag: leunga-20000327-mlriscGen_hppa_alpha_x86
125 :     Boot files (optional): ftp://react-ilp.cs.nyu.edu/leunga/110.26.1-sml.boot.x86-unix-20000330.tar.gz
126 :     Description:
127 :    
128 :     This update contains *MAJOR* changes to the way code is generated from CPS
129 :     in the module mlriscGen, and in various backend modules.
130 :    
131 :     CHANGES
132 :     =======
133 :    
134 :     1. MLRiscGen: forward propagation fix.
135 :    
136 :     There was a bug in forward propagation introduced at about the same time
137 :     as the MLRISC x86 backend, which prohibits coalescing to be
138 :     performed effectively in loops.
139 :    
140 :     Effect: speed up of loops in RISC architectures.
141 :     By itself, this actually slowed down certain benchmarks on the x86.
142 :    
143 :     2. MLRiscGen: forward propagating addresses from consing.
144 :    
145 :     I've changed the way consing code is generated. Basically I separated
146 :     out the initialization part:
147 :    
148 :     store tag, offset(allocptr)
149 :     store elem1, offset+4(allocptr)
150 :     store elem2, offset+8(allocptr)
151 :     ...
152 :     store elemn, offset+4n(allocptr)
153 :    
154 :     and the address computation part:
155 :    
156 :     celladdr <- offset+4+alloctpr
157 :    
158 :     and move the address computation part
159 :    
160 :     Effect: register pressure is generally lower as a result. This
161 :     makes compilation of certain expressions much faster, such as
162 :     long lists with non-trivial elements.
163 :    
164 :     [(0,0), (0,0), .... (0,0)]
165 :    
166 :     3. MLRiscGen: base pointer elimination.
167 :    
168 :     As part of the linkage mechanism, we generate the sequence:
169 :    
170 :     L: ... <- start of the code fragment
171 :    
172 :     L1:
173 :     base pointer <- linkreg - L1 + L
174 :    
175 :     The base pointer was then used for computing relocatable addresses
176 :     in the code fragment. Frequently (such as in lots of continuations)
177 :     this is not needed. We now eliminate this sequence whenever possible.
178 :    
179 :     For compile time efficiency, I'm using a very stupid local heuristic.
180 :     But in general, this should be done as a control flow analysis.
181 :    
182 :     Effect: Smaller code size. Speed up of most programs.
183 :    
184 :     4. Hppa back end
185 :    
186 :     Long jumps in span dependence resolution used to depend on the existence
187 :     of the base pointer.
188 :    
189 :     A jump to a long label L was expanded into the following sequence:
190 :    
191 :     LDIL %hi(L-8192), %r29
192 :     LDO %lo(L-8192)(%r29), %r29
193 :     ADD %r29, baseptr, %r29
194 :     BV,n %r0(%r29)
195 :    
196 :     In the presence of change (3) above, this will not work. I've changed
197 :     it so that the following sequence of instructions are generated, which
198 :     doesn't mention the base pointer at all:
199 :    
200 :     BL,n L', %r29 /* branch and link, L' + 4 -> %r29 */
201 :     L': ADDIL L-(L'+4), %r29 /* Compute address of L */
202 :     BV,n %r0(%r29) /* Jump */
203 :    
204 :     5. Alpha back end
205 :    
206 :     New alpha instructions LDB/LDW have been added, as per Fermin's
207 :     suggestions. This is unrelated to all other changes.
208 :    
209 :     6. X86 back end
210 :    
211 :     I've changed andl to testl in the floating point test sequence
212 :     whenever appropriate. The Intel optimization guide states that
213 :     testl is perferable to andl.
214 :    
215 :     7. RA (x86 only)
216 :    
217 :     I've improved the spill propagation algorithm, using an approximation
218 :     of maximal weighted independent sets. This seems to be necessary to
219 :     alleviate the negative effect in light of the slow down in (1).
220 :    
221 :     I'll write down the algorithm one of these days.
222 :    
223 :     8. MLRiscGen: frequencies
224 :    
225 :     I've added an annotation that states that all call gc blocks have zero
226 :     execution frequencies. This improves register allocation on the x86.
227 :    
228 :     BENCHMARKS
229 :     ==========
230 :    
231 :     I've only perform the comparison on 110.25.
232 :    
233 :     The platforms are:
234 :    
235 :     HPPA A four processor HP machine (E9000) with 5G of memory.
236 :     X86 A 300Hhz Pentium II with 128M of memory, and
237 :     SPARC An Ultra sparc 2 with 512M of memory.
238 :    
239 :     I used the following parameters for the SML benchmarks:
240 :    
241 :     @SMLalloc
242 :     HPPA 256k
243 :     SPARC 512k
244 :     X86 256k
245 :    
246 :     COMPILATION TIME
247 :     ----------------
248 :     Here are the numbers comparing the compilation times of the compilers.
249 :     I've only compared 110.25 compiling the new sources versus
250 :     a fixpoint version of the new compiler compiling the same.
251 :    
252 :     110.25 New
253 :     Total Time in RA Spill+Reload Total Time In RA Spill+Reload
254 :     HPPA 627s 116s 2684+3584 599s 95s 1003+1879
255 :     SPARC 892s 173s 2891+3870 708s 116s 1004+1880
256 :     X86 999s 315s 94006+130691 987s 296s 108877+141957
257 :    
258 :     110.25 New
259 :     Code Size Code Size
260 :     HPPA 8596736 8561421
261 :     SPARC 8974299 8785143
262 :     X86 9029180 8716783
263 :    
264 :     So in summary, things are at least as good as before. Dramatic
265 :     reduction in compilation is obtained on the Sparc; I can't explain it,
266 :     but it is reproducible. Perhaps someone should try to reproduce this
267 :     on their own machines.
268 :    
269 :     SML BENCHMARKS
270 :     --------------
271 :    
272 :     On the average, all benchmarks perform at least as well as before.
273 :    
274 :     HPPA Compilation Time Spill+Reload Run Time
275 :     110.25 New 110.25 New 110.25 New
276 :    
277 :     barnesHut 3.158 3.015 4.75% 1+1 0+0 2.980 2.922 2.00%
278 :     boyer 6.152 5.708 7.77% 0+0 0+0 0.218 0.213 2.34%
279 :     count-graphs 1.168 1.120 4.32% 0+0 0+0 22.705 23.073 -1.60%
280 :     fft 0.877 0.792 10.74% 1+3 1+3 0.602 0.587 2.56%
281 :     knuthBendix 3.180 2.857 11.32% 0+0 0+0 0.675 0.662 2.02%
282 :     lexgen 6.190 5.290 17.01% 0+0 0+0 0.913 0.788 15.86%
283 :     life 0.803 0.703 14.22% 25+25 0+0 0.153 0.140 9.52%
284 :     logic 2.048 2.007 2.08% 6+6 1+1 4.133 4.008 3.12%
285 :     mandelbrot 0.077 0.080 -4.17% 0+0 0+0 0.765 0.712 7.49%
286 :     mlyacc 22.932 20.937 9.53% 154+181 32+57 0.468 0.430 8.91%
287 :     nucleic 5.183 5.060 2.44% 2+2 0+0 0.125 0.120 4.17%
288 :     ratio-regions 3.357 3.142 6.84% 0+0 0+0 116.225 113.173 2.70%
289 :     ray 1.283 1.290 -0.52% 0+0 0+0 2.887 2.855 1.11%
290 :     simple 6.307 6.032 4.56% 28+30 5+7 3.705 3.658 1.28%
291 :     tsp 0.888 0.862 3.09% 0+0 0+0 7.040 6.893 2.13%
292 :     vliw 24.378 23.455 3.94% 106+127 25+45 2.758 2.707 1.91%
293 :     --------------------------------------------------------------------------
294 :     Average 6.12% 4.09%
295 :    
296 :     SPARC Compilation Time Spill+Reload Run Time
297 :     110.25 New 110.25 New 110.25 New
298 :    
299 :     barnesHut 3.778 3.592 5.20% 2+2 0+0 3.648 3.453 5.65%
300 :     boyer 6.632 6.110 8.54% 0+0 0+0 0.258 0.242 6.90%
301 :     count-graphs 1.435 1.325 8.30% 0+0 0+0 33.672 34.737 -3.07%
302 :     fft 0.980 0.940 4.26% 3+9 2+6 0.838 0.827 1.41%
303 :     knuthBendix 3.590 3.138 14.39% 0+0 0+0 0.962 0.967 -0.52%
304 :     lexgen 6.593 6.072 8.59% 1+1 0+0 1.077 1.078 -0.15%
305 :     life 0.972 0.868 11.90% 26+26 0+0 0.143 0.140 2.38%
306 :     logic 2.525 2.387 5.80% 7+7 1+1 5.625 5.158 9.05%
307 :     mandelbrot 0.090 0.093 -3.57% 0+0 0+0 0.855 0.728 17.39%
308 :     mlyacc 26.732 23.827 12.19% 162+189 32+57 0.550 0.560 -1.79%
309 :     nucleic 6.233 6.197 0.59% 3+3 0+0 0.163 0.173 -5.77%
310 :     ratio-regions 3.780 3.507 7.79% 0+0 0+0 133.993 131.035 2.26%
311 :     ray 1.595 1.550 2.90% 1+1 0+0 3.440 3.418 0.63%
312 :     simple 6.972 6.487 7.48% 29+32 5+7 3.523 3.525 -0.05%
313 :     tsp 1.115 1.063 4.86% 0+0 0+0 7.393 7.265 1.77%
314 :     vliw 27.765 24.818 11.87% 110+135 25+45 2.265 2.135 6.09%
315 :     ----------------------------------------------------------------------------
316 :     Average 6.94% 2.64%
317 :    
318 :     X86 Compilation Time Spill+Reload Run Time
319 :     110.25 New 110.25 New 110.25 New
320 :    
321 :     barnesHut 5.530 5.420 2.03% 593+893 597+915 3.532 3.440 2.66%
322 :     boyer 8.768 7.747 13.19% 493+199 301+289 0.327 0.297 10.11%
323 :     count-graphs 2.040 2.010 1.49% 298+394 315+457 26.578 28.660 -7.26%
324 :     fft 1.327 1.302 1.92% 112+209 115+210 1.055 0.962 9.71%
325 :     knuthBendix 5.218 5.475 -4.69% 451+598 510+650 0.928 0.932 -0.36%
326 :     lexgen 9.970 9.623 3.60% 1014+841 1157+885 0.947 0.928 1.97%
327 :     life 1.183 1.183 0.00% 162+182 145+148 0.127 0.103 22.58%
328 :     logic 3.285 3.512 -6.45% 514+684 591+836 5.682 5.577 1.88%
329 :     mandelbrot 0.147 0.143 2.33% 38+41 33+54 0.703 0.690 1.93%
330 :     mlyacc 35.457 32.763 8.22% 3496+4564 3611+4860 0.552 0.550 0.30%
331 :     nucleic 7.100 6.888 3.07% 239+168 201+158 0.175 0.173 0.96%
332 :     ratio-regions 6.388 6.843 -6.65% 1182+257 981+300 120.142 120.345 -0.17%
333 :     ray 2.332 2.338 -0.29% 346+398 402+494 3.593 3.540 1.51%
334 :     simple 9.912 9.903 0.08% 1475+941 1579+1168 3.057 3.178 -3.83%
335 :     tsp 1.623 1.532 5.98% 266+200 250+211 8.045 7.878 2.12%
336 :     vliw 33.947 35.470 -4.29% 2629+2774 2877+3171 2.072 1.890 9.61%
337 :     ----------------------------------------------------------------------------
338 :     Average 1.22% 3.36%
339 :    
340 :     ----------------------------------------------------------------------
341 :     Name: Allen Leung
342 :     Date: 2000/03/31 21:15:00 EST
343 :     Tag: leunga-20000331-aliasing
344 :     Description:
345 :    
346 :     This update contains a rewritten (and hopefully more correct) module
347 :     for extracting aliasing information from CPS.
348 :    
349 :     To turn on this feature:
350 :    
351 :     Compiler.Control.CG.memDisambiguate := true
352 :    
353 :     To pretty print the region information with assembly
354 :    
355 :     Compiler.Control.MLRISC.getFlag "asm-show-region" := true;
356 :    
357 :     To control how many levels of aliasing information are printed, use:
358 :    
359 :     Compiler.Control.MLRISC.getInt "points-to-show-level" := n
360 :    
361 :     The default of n is 3.
362 :    
363 :     ----------------------------------------------------------------------
364 :     Name: Allen Leung
365 :     Date: 2000/04/02 21:17:00 EST
366 :     Tag: leunga-20000402-mltree
367 :     Description:
368 :    
369 :     1. Renamed the constructor CALL in MLTREE by popular demand.
370 :     2. Added a bunch of files from my repository. These are currently
371 :     used by other non-SMLNJ backends.
372 :    
373 :     ----------------------------------------------------------------------
374 :     Name: Allen Leung
375 :     Date: 2000/04/04 03:18:00 EST
376 :     Tag: leunga-20000404-C--Moby
377 :     Description:
378 :    
379 :     All of these fixes are related to C--, Moby, and my own optimization
380 :     stuff; so they shouldn't affect SML/NJ.
381 :    
382 :     1. X86
383 :    
384 :     Various fixes related floating point, and extensions.
385 :    
386 :     2. Alpha
387 :    
388 :     Some extra patterns related to loads with signed/zero extension
389 :     provided by Fermin.
390 :    
391 :     3. Assembly
392 :    
393 :     When generating assemby, resolve the value of client defined constants,
394 :     instead of generating symbolic values. This is controlled by the
395 :     new flag "asm-resolve-constants", which is default to true.
396 :    
397 :     4. Machine Descriptions
398 :    
399 :     a. The precedence parser was slightly broken when parsing infixr symbols.
400 :     b. The type generalizing code had the bound variables reversed, resulting
401 :     in a problem during arity raising.
402 :     c. Various fixes in machine descriptions.
403 :    
404 :     ----------------------------------------------------------------------
405 :     Name: Allen Leung
406 :     Date: 2000/04/04 19:39:00 EST
407 :     Tag: leunga-20000404-x86-asm
408 :     Description:
409 :    
410 :     1. Fixed a problem in X86 assembly.
411 :    
412 :     Things like
413 :    
414 :     jmp %eax
415 :     jmp (%eax)
416 :    
417 :     should be output as
418 :    
419 :     jmp *%eax
420 :     jmp *(%eax)
421 :    
422 :     2. Assembly output
423 :    
424 :     Added a new flag
425 :    
426 :     "asm-indent-copies" (default to false)
427 :    
428 :     When this flag is on, parallel copies will be indented an extra level.
429 :    
430 :     ----------------------------------------------------------------------
431 :     Name: Allen Leung
432 :     Date: 2000/04/06 00:36:00 EST
433 :     Tag: leunga-20000406-peephole-x86-SSA
434 :     Description:
435 :    
436 :     1. New Peephole code
437 :    
438 :     2. Minor improvement to X86 instruction selection
439 :    
440 :     3. Various fixes to SSA and machine description -> code translator
441 :    
442 :     ----------------------------------------------------------------------
443 :     Name: Allen Leung
444 :     Date: 2000/04/09 19:09:00 EST
445 :     Tag: leunga-20000409-misc
446 :     Description:
447 :    
448 :     1. Yet another fix for x86 assembly for idivl, imull, mull and friends.
449 :    
450 :     2. Miscellaneous improvements to MLRISC (unused in sml/nj)
451 :    
452 :    
453 :     ======================================================================
454 :     CM
455 :     ======================================================================
456 :    
457 :     Name: Matthias Blume
458 :     Date: 2000/03/07 04:01:04
459 :     Tag: blume_main_v110_26_2
460 :     Description:
461 :     - size info in BOOTLIST
462 :     * no fixed upper limits for number of bootfiles or length of
463 :     bootfile names in runtime
464 :     * falling back to old behavior if no BOOTLIST size info found
465 :     - allocation size heuristics in .run-sml
466 :     * tries to read cache size from /proc/cpuinfo (this is important for
467 :     small-cache Celeron systems!)
468 :     - install.sh robustified
469 :     - CM manual updates
470 :     - paranoid mode
471 :     * no more CMB.deliver() (i.e., all done by CMB.make())
472 :     * can re-use existing sml.boot.* files
473 :     * init.cmi now treated as library
474 :     * library stamps for consistency checks
475 :     - sml.boot.<arch>-<os>/PIDMAP file
476 :     * This file is read by the CM startup code. This is used to minimize
477 :     the amount of dynamic state that needs to be stowed away for the
478 :     purpose of sharing between interactive system and user code.
479 :     - CM.Anchor.anchor instead of CM.Anchor.{set,cancel}
480 :     * Upon request by Elsa. Anchors now controlled by get-set-pair
481 :     like most other CM state variables.
482 :     - Compiler.CMSA eliminated
483 :     * No longer supported by CM anyway.
484 :     - fixed bugs in pickler that kept biting Stefan
485 :     * past refs to past refs (was caused by the possibility that
486 :     ad-hoc sharing is more discriminating than hash-cons sharing)
487 :     * integer overflow on LargeInt.minInt
488 :     - ml-{lex,yacc} build scripts now use new mechanism
489 :     for building standalone programs
490 :     - fixed several gcc -Wall warnings that were caused by missing header
491 :     files, missing initializations, etc., in runtime (not all warnings
492 :     eliminated, though)
493 :    
494 :     ----------------------------------------------------------------------
495 :     Name: Matthias Blume
496 :     Date: 2000/03/09 10:23:53
497 :     Tag: blume_main_v110p26p1_0
498 :     Description:
499 :    
500 :     * Complete re-organization of library names. Many libraries have been
501 :     consolidated so that they share the same path anchor. For example,
502 :     all MLRISC-related libraries are anchored at MLRISC, most libraries that
503 :     are SML/NJ-specific are under "smlnj". Notice that names like
504 :     host-cmb.cm or host-compiler.cm no longer exist. See system/README
505 :     for a complete description of the new naming scheme. Quick reference:
506 :    
507 :     host-cmb.cm -> smlnj/cmb.cm
508 :     host-compiler.cm -> smlnj/compiler.cm
509 :     full-cm.cm -> smlnj/cm.cm
510 :     <arch>-<os>.cm -> smlnj/cmb/<arch>-<os>.cm
511 :     <arch>-compiler.cm -> smlnj/compiler/<arch>.cm
512 :    
513 :     * Bug fixes in CM.
514 :     - exceptions in user code are being passed through (i.e., reach top level)
515 :     - more bugs in paranoia mode fixed
516 :     - bug related to checking group owners fixed
517 :    
518 :     * New install.sh script that automagically fetches archive files:
519 :     The new file config/srcarchiveurl must contain the URL of the
520 :     (remote) directory that contains bin files (or other source archives).
521 :     If install.sh does not find the archive locally, it tries to get
522 :     it from that remote directory.
523 :     This should simplify installation further: For machines that have
524 :     access to the internet, just fetch <version>-config.tgz, unpack it,
525 :     edit config/targets, and go (run config/install.sh). The scipt will
526 :     fetch everything else that it might need all by itself.
527 :    
528 :     For CVS users, this mechanism is not relevant for source archives, but
529 :     it is convenient for getting new sets of binfiles.
530 :    
531 :     Archives should be tar files compressed with either gzip, compress, or
532 :     bzip2. The script recognizes .tgz, .tar, tar.gz, tz, .tar.Z, and .tar.bz2.
533 :    
534 :     ----------------------------------------------------------------------
535 :     Name: Matthias Blume
536 :     Date: 2000/03/10 07:48:34
537 :     Tag: blume_main_v110p26p1_1
538 :     Description:
539 :    
540 :     I added a re-written version of Dave's fixpt script to src/system.
541 :     Changes relative to the original version:
542 :     - sh-ified (not everybody has ksh)
543 :     - automatically figures out which architecture it runs on
544 :     - uses ./makeml a bit more cleverly
545 :     - never invokes ./installml (and, thus, does not clobber your
546 :     good and working installation of sml in case something goes wrong)
547 :     - accepts max iteration count using option "-iter <n>"
548 :     - accepts a "base" name using option "-base <base>"
549 :    
550 :     It does not build any extraneous heap images but directly rebuilds
551 :     bin- and boot-hierarchies using makeml's "-rebuild" switch. Finally,
552 :     it can incorporate existing bin- and boot- hierarchies. For example,
553 :     suppose the base is set to "sml" (which is the default). Then it
554 :     successively builds
555 :    
556 :     sml.bin.<arch>-unix and sml.boot.<arch>-unix
557 :     then sml1.bin.<arch>-unix and sml1.boot.<arch>-unix
558 :     then sml2.bin.<arch>-unix and sml2.boot.<arch>-unix
559 :     ...
560 :     then sml<n>.bin.<arch>-unix and sml<n>.boot.<arch>-unix
561 :    
562 :     and so on. If any of these already exist, it will just use what's
563 :     there. In particular, many people will have the initial set of bin
564 :     and boot files around, so this saves time for at least one full
565 :     rebuild. Having sets of the form <base><k>.{bin,boot}.<arch>-unix for
566 :     <k>=1,2,... is normally not a good idea when invoking fixpt. However,
567 :     they might be the result of an earlier partial run of fixpt (which
568 :     perhaps got accidentially killed). In this case, fixpt will quickly
569 :     move through what exists before continuing where it left off earlier,
570 :     and, thus, saves a lot of time.
571 :    
572 :     ----------------------------------------------------------------------
573 :     Name: Matthias Blume
574 :     Date: 2000/03/14 14:15:32
575 :     Tag: blume_main_v110p26p1_2
576 :     Description:
577 :    
578 :     1. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an
579 :     additional argument called "template" which is an optional string that
580 :     specifiel the layout of the tool command line. See the CM manual for
581 :     explanation.
582 :    
583 :     2. A special-purpose tool can be "regisitered" by simply dropping the
584 :     corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same
585 :     directory where the .cm file lives that uses this tool. (The
586 :     behavior/misfeature until now was to look for the tool description
587 :     files in the current working directory.) As before, tool description
588 :     files could also be anchored -- in which case they can live anywhere
589 :     they like. Following the recent e-mail discussion, this change should
590 :     make it easier to have special-purpose tools that are shipped together
591 :     with the sources of the program that uses them.
592 :    
593 :     ----------------------------------------------------------------------
594 :     Name: Matthias Blume
595 :     Date: 2000/03/30 18:00:00 JST
596 :     Tag: blume_main_v110p26p2_0
597 :     Description:
598 :    
599 :     !!!!! WARNING !!!!!!
600 :     !! New binfiles !!
601 :     !!!!!!!!!!!!!!!!!!!!
602 :    
603 :     This update contains:
604 :    
605 :     1. Moderate changes to CM:
606 :    
607 :     - Changes to CM's tools mechanism. In particular, it is now possible
608 :     to have tools that accept additional "command line" parameters
609 :     (specified in the .cm file at each instance where the tool's class is
610 :     used).
611 :    
612 :     This was done to accomodate the new "make" and "shell" tools which
613 :     facilitate fairly seemless hookup to portions of code managed using
614 :     Makefiles or Shell scripts.
615 :    
616 :     There are no classes "shared" or "private" anymore. Instead, the
617 :     sharing annotation is now a parameter to the "sml" class.
618 :    
619 :     There is a bit of generic machinery for implementing one's own
620 :     tools that accept command-line parameters. However, I am not yet fully
621 :     satisfied with that part, so expect changes here in the future.
622 :    
623 :     All existing tools are described in the CM manual.
624 :    
625 :     - Slightly better error handling. (CM now surpresses many followup
626 :     error messages that tended to be more annoying than helpful.)
627 :    
628 :     2. Major changes to the compiler's static environment data structures.
629 :    
630 :     - no CMStaticEnv anymore.
631 :     - no CMEnv, no "BareEnvironment" (actually, _only_ BareEnvironment,
632 :     but it is called Environment), no conversions between different
633 :     kinds of static environments
634 :    
635 :     - There is still a notion of a "modmap", but such modmaps are generated
636 :     on demand at the time when they are needed. This sounds slow, but I
637 :     sped up the code that generates modmaps enough for this not to lead to
638 :     a slowdown of the compiler (at least I didn't detect any).
639 :    
640 :     - To facilitate rapid modmap generation, static environments now
641 :     contain an (optional) "modtree" structure. Modtree annotations are
642 :     constructed by the unpickler during unpickling. (This means that
643 :     the elaborator does not have to worry about modtrees at all.)
644 :     Modtrees have the advantage that they are compositional in the same
645 :     way as the environment data structure itself is compositional.
646 :     As a result, modtrees never hang on to parts of an environment that
647 :     has already been rendered "stale" by filtering or rebinding.
648 :    
649 :     - I went through many, many trials and errors before arriving at the
650 :     current solution. (The initial idea of "linkpaths" did not work.)
651 :     But the result of all this is that I have touched a lot of files that
652 :     depend on the "modules" and "types" data structures (most of the
653 :     elaborator). There were a lot of changes during my "linkpath" trials
654 :     that could have been reverted to their original state but weren't.
655 :     Please, don't be too harsh on me for messing with this code a bit more
656 :     than what was strictly necessary... (I _did_ resist the tempation
657 :     of doing any "global reformatting" to avoid an untimely death at
658 :     Dave's hands. :)
659 :    
660 :     - One positive aspect of the previous point: At least I made sure that
661 :     all files that I touched now compile without warnings (other than
662 :     "polyEqual").
663 :    
664 :     - compiler now tends to run "leaner" (i.e., ties up less memory in
665 :     redundant modmaps)
666 :    
667 :     ----------------------------------------------------------------------
668 :     Name: Matthias Blume
669 :     Date: 2000/03/31 18:00:00 JST
670 :     Tag: blume_main_v110p26p2_1
671 :     Description:
672 :    
673 :     This update contains:
674 :    
675 :     1. A small change to CM's handling of stable libraries:
676 :     CM now maintains one "global" modmap that is used for all stable
677 :     libraries. The use of such a global modmap maximizes sharing and
678 :     minimizes the need for re-traversing parts of environments during
679 :     modmap construction. (However, this has minor impact since modmap
680 :     construction seems to account for just one percent or less of total
681 :     compile time.)
682 :    
683 :     2. I added a "genmap" phase to the statistics. This is where I got the
684 :     "one percent" number (see above).
685 :    
686 :     3. CM's new tool parameter mechanism just became _even_ better. :)
687 :     - The parser understands named parameters and recursive options.
688 :     - The "make" and "shell" tools use these new features.
689 :     (This makes it a lot easier to cascade these tools.)
690 :     - There is a small syntax change: named parameters use a
691 :    
692 :     <name> : ( <option> ... ) or
693 :     <name> : <string>
694 :    
695 :     syntax. Previously, named parameters were implemented in an
696 :     ad-hoc fashion by each tool individually (by parsing strings)
697 :     and had the form
698 :    
699 :     <name>=<string>
700 :    
701 :     See the CM manual for a full description of these issues.
702 :    
703 :     ----------------------------------------------------------------------
704 :     Name: Matthias Blume
705 :     Date: 2000/04/03 16:05:00 JST
706 :     Tag: blume_main_v110p26p2_2
707 :     Description:
708 :    
709 :     I eliminated coreEnv from compInfo. Access to the "Core" structure is
710 :     now done via the ordinary static environment that is context to each
711 :     compilation unit.
712 :    
713 :     To this end, I arranged that instead of "structure Core" as "structure
714 :     _Core" is bound in the pervasive environment. Core access is done via
715 :     _Core (which can never be accidentially rebound because _Core is not a
716 :     legal surface-syntax symbol).
717 :    
718 :     The current solution is much cleaner because the core environment is
719 :     now simply part of the pervasive environment which is part of every
720 :     compilation unit's context anyway. In particular, this eliminates all
721 :     special-case handling that was necessary until now in order to deal
722 :     with dynamic and symbolic parts of the core environment.
723 :    
724 :     Remaining hackery (to bind the "magic" symbol _Core) is localized in the
725 :     compilation mananger's bootstrap compiler (actually: in the "init group"
726 :     handling). See the comments in src/system/smlnj/init/init.cmi for
727 :     more details.
728 :    
729 :     I also tried to track down all mentions of "Core" (as string argument
730 :     to Symbol.strSymbol) in the compiler and replaced them with a
731 :     reference to the new CoreSym.coreSym. Seems cleaner since the actual
732 :     name appears in one place only.
733 :    
734 :     Binfile and bootfile format have not changed, but the switchover from
735 :     the old "init.cmi" to the new one is a bit tricky, so I supplied new
736 :     bootfiles anyway.
737 :    

root@smlnj-gforge.cs.uchicago.edu
ViewVC Help
Powered by ViewVC 1.0.0