Ticket #4424: fix4424.dpatch

File fix4424.dpatch, 161.0 KB (added by nominolo, 4 years ago)

Patch to fix.

Line 
11 patch for repository http://darcs.haskell.org/ghc:
2
3Thu Oct 21 18:04:32 BST 2010  Thomas Schilling <[email protected]>
4  * Fix #4424.  Allow `-hide-all-packages` to be set dynamically.
5
6New patches:
7
8[Fix #4424.  Allow `-hide-all-packages` to be set dynamically.
9Thomas Schilling <[email protected]>**20101021170432
10 Ignore-this: 59fa27d6c74252a89a14290fa57fa093
11] hunk ./compiler/main/Packages.lhs 173
12 initPackages dflags = do
13   pkg_db <- case pkgDatabase dflags of
14                 Nothing -> readPackageConfigs dflags
15-                Just db -> return db
16+                Just db -> return $ maybeHidePackages dflags db
17   (pkg_state, preload, this_pkg)       
18         <- mkPackageState dflags pkg_db [] (thisPackage dflags)
19   return (dflags{ pkgDatabase = Just pkg_db,
20
21Context:
22
23[Fix warnings
24[email protected]**20101013040335
25 Ignore-this: 4e5875827d2840de863000cdb35afb0e
26]
27[RegAlloc: Track slot liveness over jumps in spill cleaner
28[email protected]**20101013015414
29 Ignore-this: ccd4a148908b7fbdc6ea76acf527c16b
30]
31[Bump Cabal dep
32Ian Lynagh <[email protected]>**20101012154528
33 Ignore-this: da8539b957b74c7da91efff8ac25872
34]
35[Remove __HASKELL1__, __HASKELL98__, __CONCURRENT_HASKELL__
36Ian Lynagh <[email protected]>**20101012134700
37 Ignore-this: d0db531be58537c52802de1c62694eb1
38 We used to define these CPP symbols, but nothing on hackage uses them
39 and the first 2 are no longer correct (as we support multiple Haskell
40 versions).
41]
42[Follow Cabal changes: Cabal no longer has a docbook userguide
43Ian Lynagh <[email protected]>**20101012130538
44 Ignore-this: bf0b30e465c13f82725d72aec145e939
45 For now we don't build the Cabal userguide, but we should add markdown
46 support so that we can do so.
47]
48[Fix build on Windows: ghc-pkg/Main.hs needs ForeignFunctionInterface
49Ian Lynagh <[email protected]>**20101012112111]
50[Remove unnecessary import
51Ian Lynagh <[email protected]>**20101010222231
52 Ignore-this: 6c7d5cee72837e2af43c2807e10ad602
53]
54[Make "./validate --slow" run the full testsuite
55Ian Lynagh <[email protected]>**20101007004327
56 Ignore-this: 2dee8262c19fd5d5034a186203e20d0f
57]
58[Fix build following haskell98 and -fglasgow-exts changes
59Ian Lynagh <[email protected]>**20101006160656
60 Ignore-this: c56bdb0f01da897b3de75bcab1e2887c
61]
62[Don't automatically link the haskell98 package
63Ian Lynagh <[email protected]>**20101006130235
64 The default language is now Haskell2010, so this was a little odd.
65 Also, --make is now on by default, so this was largely irrelevant.
66]
67[Deprecate -fglasgow-exts
68Ian Lynagh <[email protected]>**20101006124413
69 Ignore-this: 70be67898a633ab31d76f6fca557d55
70]
71[Remove Opt_GADTs and Opt_TypeFamilies from -fglasgow-exts
72Ian Lynagh <[email protected]>**20101006122000
73 Ignore-this: 7153ec880b2b5f9d332ae2f57b97afcc
74 This means most code doesn't get caught by monomorphic local bindings.
75]
76[Fix Trac #4360: omitted case in combineCtLoc
77[email protected]**20101008135747
78 Ignore-this: 834636a97af6469862a822809253db41
79]
80[Beautiful new approach to the skolem-escape check and untouchable
81[email protected]**20101008133751
82 Ignore-this: 33517a772cfdfbf4aa4678609f3dcd71
83 
84 Instead of keeping a *set* of untouchable variables in each
85 implication contraints, we keep a *range* of uniques for the
86 *touchable* variables of an implication.  This are precisely
87 the ones we would call the "existentials" if we were French.
88 
89 It turns out that the code is more efficient, and vastly easier
90 to get right, than the set-based approach.
91 
92 Fixes Trac #4355 among others
93]
94[Do less simplification when doing let-generalisation
95[email protected]**20101008133542
96 Ignore-this: 71366d0de37f10ffba2edc9f3927ddbe
97 
98 This fixes Trac #4361.  In a rather delicate way, but
99 no more delicate than before.  A more remoseless typechecker
100 would reject #4361 altogether.
101 
102 See Note [Avoid unecessary constraint simplification]
103]
104[Suppress ambiguity errors if any other errors occur
105[email protected]**20101008111318
106 Ignore-this: 40f014265c1ab15fe172baaf76c23c87
107]
108[Fix Trac #4361: be more discerning when inferring types
109[email protected]**20101008111227
110 Ignore-this: 33656f9a151f494b26b6318b5fcffef
111 
112 Note [Avoid unecessary constraint simplification] in TcSimplify
113]
114[Float out partial applications
115Simon Marlow <[email protected]>**20101008092709
116 Ignore-this: 2dc9d10597c19d0598ef2fc3cf74156d
117 
118 This fixes at least one case of performance regression in 7.0, and
119 is nice win on nofib:
120 
121         Program           Size    Allocs   Runtime   Elapsed
122             Min          +0.3%    -63.0%    -38.5%    -38.7%
123             Max          +1.2%     +0.2%     +0.9%     +0.9%
124  Geometric Mean          +0.6%     -3.0%     -6.4%     -6.6%
125]
126[Suppress knock-on typechecker errors
127[email protected]**20101008094348
128 Ignore-this: 8d125926286a7614fa1ce998e3b26d04
129 
130 The error cascade caused puzzling errors in T4093b, and
131 suppressing some seems like a good plan.  Very few test
132 outputs change.
133]
134[Some refactoring and simplification in TcInteract.occurCheck
135[email protected]**20101007163500
136 Ignore-this: d43d09370ab27b8796062e2e98ce7e9
137]
138[Comments only
139[email protected]**20101007130301
140 Ignore-this: ab46592edd3d24786bbce42c50feb4fd
141]
142[Implement auto-specialisation of imported Ids
143[email protected]**20101007111051
144 Ignore-this: 45257ff6e9597e4fa4de10b0657e27d6
145 
146 This big-ish patch arranges that if an Id 'f' is
147   * Type-class overloaded
148        f :: Ord a => [a] -> [a]
149   * Defined with an INLINABLE pragma
150        {-# INLINABLE f #-}
151   * Exported from its defining module 'D'
152 
153 then in any module 'U' that imports D
154 
155 1. Any call of 'f' at a fixed type will generate
156    (a) a specialised version of f in U
157    (b) a RULE that rewrites unspecialised calls to the
158        specialised on
159 
160   e.g. if the call is (f Int dOrdInt xs) then the
161   specialiser will generate
162      $sfInt :: [Int] -> [Int]
163      $sfInt = <code for f, imported from D, specialised>
164      {-# RULE forall d.  f Int d = $sfInt #-}
165 
166 2. In addition, you can give an explicit {-# SPECIALISE -#}
167    pragma for the imported Id
168      {-# SPECIALISE f :: [Bool] -> [Bool] #-}
169    This too generates a local specialised definition,
170    and the corresponding RULE
171 
172 The new RULES are exported from module 'U', so that any module
173 importing U will see the specialised versions of 'f', and will
174 not re-specialise them.
175 
176 There's a flag -fwarn-auto-orphan that warns you if the auto-generated
177 RULES are orphan rules. It's not in -Wall, mainly to avoid lots of
178 error messages with existing packages.
179 
180 Main implementation changes
181 
182  - A new flag on a CoreRule to say if it was auto-generated.
183    This is persisted across interface files, so there's a small
184    change in interface file format.
185 
186  - Quite a bit of fiddling with plumbing, to get the
187    {-# SPECIALISE #-} pragmas for imported Ids.  In particular, a
188    new field tgc_imp_specs in TcGblEnv, to keep the specialise
189    pragmas for imported Ids between the typechecker and the desugarer.
190 
191  - Some new code (although surprisingly little) in Specialise,
192    to deal with calls of imported Ids
193]
194[Make NameEnv back into type NameEnv a = UniqFM a
195[email protected]**20101007104638
196 Ignore-this: 24857c013461788be354520e84f4c286
197 
198 I don't think the type distinction of declaring NameEnv with a newtype
199 (as it was) is really useful to us. Moreover, VarEnv is a UniqFM, and
200 I do sometimes want to build an envt with Ids and look up with Names.
201 
202 This may not be the last word on the subject.
203]
204[Improve the rule-matcher
205[email protected]**20101007103700
206 Ignore-this: 9de96237dc4b73a43326bd568e34b53b
207 
208 Previously it was rejecting the match
209 
210   Template: forall s t. map s t
211   Actual:   map Int t
212 
213 which should obviously be fine.  It turns out that this kind of match
214 comes up when specialising.  By freshening that t we could avoid the
215 difficulty, but morally the (forall t) binds t and the rule should
216 be alpha-equivalent regardless of the forall'd variables.
217 
218 This patch makes it so, and incidentally makes matching a little
219 more efficient.  See Note [Eta expansion] in VarEnv.
220]
221[Fix Trac #4345: simplifier bug
222[email protected]**20101007102720
223 Ignore-this: 261c1c9f094df344ce34de814f8b60c5
224 
225 This is another long-standing bug, in which there was a possibility
226 that a loop-breaker could lose its loop-breaker-hood OccInfo,
227 and then the simplifer re-simplified the expression. Result, either
228 non-termination or, in the case of #4345, an unbound identifier.
229 
230 The fix is very simple, in Id.transferPolyIdInfo.
231 See Note [transferPolyIdInfo].
232]
233[Avoid redundant simplification
234[email protected]**20101007095935
235 Ignore-this: 61bd1a2c508260f558866e6a88c29fa3
236 
237 When adding specialisation for imported Ids, I noticed that the
238 Glorious Simplifier was repeatedly (and fruitlessly) simplifying the
239 same term.  It turned out to be easy to fix this, because I already
240 had a flag in the ApplyTo and Select constructors of SimplUtils.SimplCont.
241 
242 See Note [Avoid redundant simplification]
243]
244[Make the occurrence analyser deal correctly with RULES for imported Ids
245[email protected]**20101007094100
246 Ignore-this: 335b1cad013524e42b31e88c0a7a00f6
247 
248 This patch fixes a long-standing lurking bug, but it surfaced when I
249 was adding specialisation for imported Ids.
250 
251 See Note [ImpRuleUsage], which explains the issue.   The solution
252 seems more complicated than the problem really deserves, but I
253 could not think of a simpler way, so I just bit the bullet and
254 wrote the code.  Improvements welcome.
255]
256[Make warning-free
257[email protected]**20101007092007
258 Ignore-this: 4bae0c470a8a1f96d21990d1f3cf1f93
259]
260[This is just white-space and layout
261[email protected]**20101007091618
262 Ignore-this: 759c0335df70fce32558e967f140803a
263 
264 (At least, I don't think there is anything else.)
265]
266[Fix an ASSERT failure in FamInstEnv
267[email protected]**20101007091327
268 Ignore-this: a8c08ccb7ec2bc65864a674b5441539
269 
270 I added a lot of comments too, to explain the preconditions;
271 esp Note [FamInstEnv]
272]
273[Fix a looping bug in the new occur-check code
274[email protected]**20101007084104
275 Ignore-this: a02a2deafb9ec986ef1565f4596049ed
276]
277[Fix test T4235 with -O
278[email protected]**20101006155223
279 Ignore-this: f0fa0fe2f0c493e362d528d71f7a64e1
280 
281 The tag2Enum rule wasn't doing the right thing for
282 enumerations with a phantom type parameter, like
283    data T a = A | B
284]
285[Make warning-free
286[email protected]**20101006155033
287 Ignore-this: 221a3c95a6079c6ecc0468996a38b048
288]
289[Major bugfixing pass through the type checker
290[email protected]**20101006152854]
291[Typechecker performance fixes and flatten skolem bugfixing
292[email protected]**20101004130200
293 Ignore-this: 86721ba3f09479c146a0710796b43459
294]
295[Performance bug fixes
296[email protected]**20100923143918]
297[Fix Trac #4371: matching of view patterns
298[email protected]**20101006115316
299 Ignore-this: 494b28b91f1e6392b2f1521cda0e83b1
300]
301[Remove unused NoMatchContext construtor
302[email protected]**20101006115251
303 Ignore-this: 8985ff1dac51fb652bd65657a630a792
304]
305[Refactoring: mainly rename ic_env_tvs to ic_untch
306[email protected]**20101006102830
307 Ignore-this: 32999403a3f447e14b59cec7896027ff
308 
309 Plus remember to zonk the free_tvs in TcUnify.newImplication
310]
311[remove unnecessary/broken definition of mask_
312Simon Marlow <[email protected]>**20101002195118
313 Ignore-this: 4cdc9c95d40e01cdfe2d0ac411476603
314]
315[-fwarn-tabs: add "Warning" to the message
316Simon Marlow <[email protected]>**20101002195100
317 Ignore-this: 589a36daa3426ab51f2fb140e38df6c
318]
319[give a better error message in the non-threaded RTS for out-of-range FDs
320Simon Marlow <[email protected]>**20100929212916
321 Ignore-this: e94c9f390b8f79d24895a80f9d16c8d9
322 
323 # ./aw
324 aw: file descriptor 1027 out of range for select (0--1024).
325 Recompile with -threaded to work around this.
326]
327[Fix a very rare crash in GHCi
328Simon Marlow <[email protected]>**20101005144735
329 Ignore-this: dad1cd08934bae2ba47e72c0c000acfa
330 When a BCO with a zero-length bitmap was right at the edge of
331 allocated memory, we were reading a word of non-existent memory.
332 
333 This showed up as a segfault in T789(ghci) for me, but the crash was
334 extremely sensitive and went away with most changes.
335 
336 Also, optimised scavenge_large_bitmap a bit while I was in there.
337]
338[Using 'stdcall' when it is not supported is only a warning now (#3336)
339Simon Marlow <[email protected]>**20100924152445
340 Ignore-this: 66c5903a600a47485a7583535bb38455
341]
342[remove unnecessary stg_noForceIO (#3508)
343Simon Marlow <[email protected]>**20100924150202
344 Ignore-this: dec52de9cd9da7dcedae12b20691aba9
345]
346[Replace an outputStr with putStrLn calls; fixes #4332
347Ian Lynagh <[email protected]>**20101003125707
348 Ignore-this: eb4b5f60d9d9d3f7dc203869927b28ba
349]
350[make test and fulltest targets in the main Makefile; fixes #4297
351Ian Lynagh <[email protected]>**20100930224741
352 You can now run "make test" in the root, and the fast testsuite will be
353 run with cleaning enabled. It will also put the summary in
354 testsuite_summary.txt.
355]
356[Don't show the loaded packages in ":show packages"; fixes #4300
357Ian Lynagh <[email protected]>**20100930210128
358 It's never worked properly, and the information is in ":show linker".
359]
360[Handle EXTRA_LIBRARIES when building programs
361Ian Lynagh <[email protected]>**20100930192552
362 Ignore-this: 401a26e18d25dcaee010b13eaed8f011
363 And set hp2ps's EXTRA_LIBRARIES. Based on a patch from Sergei Trofimovich.
364]
365[Fix the doc directory on Windows
366Ian Lynagh <[email protected]>**20100929133328]
367[Remove an unused import on Windows
368Ian Lynagh <[email protected]>**20100929000024
369 Ignore-this: 2899e0e5a47122e637fb5c8aa0df52ab
370]
371[Use showCommandForUser when showing tracing commands
372Ian Lynagh <[email protected]>**20100928235844
373 Ignore-this: 8a4a9c9f8a8029e708c4297b096b6ef1
374]
375[Fix hsc2hs docs: 'gcc' is now the default compiler, not 'ghc'; fixes #4341
376Ian Lynagh <[email protected]>**20100928201938]
377[Use an empty signal handler for SIGPIPE instead of SIG_IGN
378Simon Marlow <[email protected]>**20100925193548
379 Ignore-this: b4dc5de32740a7c5fd8fe4b3bfb1300f
380 
381 This is so that the SIGPIPE handler gets reset to the default
382 automatically on exec().
383]
384[Fix the TH deps
385Ian Lynagh <[email protected]>**20100925210029
386 Ignore-this: 32b832301a3625d4ba70f84c5c4f94d2
387]
388[Check inplace doesn't exist before we try to create it
389Ian Lynagh <[email protected]>**20100924191858
390 This fixes rerunning configure in a tree which already has an inplace
391 directory. Edward Z Yang ran into this; I guess whether it actually
392 fails depends on details of your installation, or we'd have run into
393 it sooner.
394]
395[Fix an egregious bug: INLINE pragmas on monomorphic Ids were being ignored
396[email protected]**20100924155815
397 Ignore-this: 38c6eec6710a92df7662a55fc5132c15
398 
399 I had do to some refactoring to make this work nicely
400 but now it does. I can't think how this escaped our
401 attention for so long!
402]
403[Eta expand only lambdas that bind a non-dictionary Id
404[email protected]**20100924155707
405 Ignore-this: 7cc265eaf6c0bb3fa12eb311d92594ac
406 
407 See Note [When to eta expand]. The idea is that dictionary
408 lambdas are invisible to the user, so we shouldn't eta
409 expand them.
410]
411[Add a comment
412[email protected]**20100924155620
413 Ignore-this: de210a1afdd40328824803e1d77b4d7f
414]
415[Add a debug print
416[email protected]**20100924155614
417 Ignore-this: 1a58b6d297fc77d6ded8eec7ea9f895d
418]
419[Just moving comments around
420[email protected]**20100924155600
421 Ignore-this: 96635b8e8c9d88b50d82938568152ef8
422]
423[use putStrLn instead of Haskeline's outputStrLn
424Simon Marlow <[email protected]>**20100924133154
425 Ignore-this: 7581ae11714a9a52e78ba098c3c216b3
426 use of the latter caused problems for Claus Reinke's macros that
427 redirect stdout.
428]
429[Change "OPTIONS" to "OPTIONS_GHC" in error messages; fixes #4327
430Ian Lynagh <[email protected]>**20100924120423
431 Ignore-this: 1697c83a5c346db640c0a2e22c69ff55
432]
433[Add deps for TH uses in vector
434Ian Lynagh <[email protected]>**20100923220244
435 Ignore-this: 54c3386b1c268821fcdd34b84bc8c6a4
436]
437[Bump Cabal dep
438Ian Lynagh <[email protected]>**20100923143241]
439[Update Cabal's version number
440Ian Lynagh <[email protected]>**20100923141719]
441[Build primitive with stage2
442Ian Lynagh <[email protected]>**20100923140525
443 Ignore-this: 110a819b78a57629a7edf1d4facdc191
444]
445[Fix the Windows __chkstk build error (missing Linker symbol)
446Simon Marlow <[email protected]>**20100924113837
447 Ignore-this: 48f0907bb1bd5eaa0730b94a6bd94ea
448]
449[emit a helpful error message for missing DPH packages
450Simon Marlow <[email protected]>**20100923141957
451 Ignore-this: 55ff2ee90c94524e023e014243bfe5df
452]
453[Fix computation of installed packages
454[email protected]**20100924084737
455 Ignore-this: a597d2fa8be5135ba8ead6d2624b3d71
456 
457 This is a follow-on to Simon's patch yesterday, developed
458 with him.  It cleans up the computation of how packages
459 are installed, and installs the right ones.
460]
461[Fix braino in WwLib/Literal patch
462[email protected]**20100924070914
463 Ignore-this: f6eb3a42e10f8aa7920de541cdfe76d8
464]
465[For now, switch off incomplete-pattern warnings in containers
466[email protected]**20100923130117
467 Ignore-this: 7ffa58567f7a33aafe256492999da325
468 
469 Put it back on when my patch is applied to the containers repo.
470 (the one that removes two refuable lambdas)
471]
472[Make -funfolding-dict-threshold work properly
473[email protected]**20100923130032
474 Ignore-this: 417788f5b09d1d624f6b6371852c80c7
475 
476 and increase its default value. This makes overloaded functions
477 a bit keener to inline.  Which fixes Trac #4321
478]
479[Impredicative types is no longer deprecated
480[email protected]**20100923125910
481 Ignore-this: 2bbaeb38b5e8424551677c0add627683
482]
483[Do not make FunctionalDependencies force MonoLocalBinds
484[email protected]**20100923125900
485 Ignore-this: f4ae1fd07c87ec14f60bdfe3863ba7a9
486]
487[move CHECKED settings to the right place
488Simon Marlow <[email protected]>**20100923123558
489 Ignore-this: e00a0eb5855463cc9b953670b3bbf211
490]
491[turn off -Werror for primitive and vector
492Simon Marlow <[email protected]>**20100923122055
493 Ignore-this: 54d7b80f3f893385e1c3ef431e2a8a7b
494]
495[Add primitive and vector packages for DPH support
496Simon Marlow <[email protected]>**20100923104542
497 Ignore-this: c070d015385b0a0797394132dcbb7670
498 DPH is now using the public vector package instead of its internal
499 version.
500 
501 vector and primitive are not "boot" packages; they aren't required to
502 build GHC, but they are required to validate (because we include DPH
503 when validating).
504 
505 If you say './darcs-all get --no-dph' then you don't get DPH, vector,
506 or primitive.
507]
508[Refactoring and tidy up in the build system
509Simon Marlow <[email protected]>**20100923095642
510 Ignore-this: f7bf3a1fd160149d89b26f464b064fb1
511 
512 Instead of the ghc-stage and ghc-stage2-package files in a package, we
513 now have a list of these in ghc.mk.  There are other similar lists (of
514 boot-packages and non-installable packages), so this is not too bad,
515 and is simpler.
516 
517 While poking around in the top-level ghc.mk file I spotted various
518 opportunities to clean up and re-order some of the cruft that has
519 accumulated over time.
520]
521[Allow absent State# RealWorld arguments
522[email protected]**20100923111356
523 Ignore-this: c2d57633dec0293ebe6723ea3a4bb5df
524]
525[Add notSCCNote, and use it
526[email protected]**20100923105949
527 Ignore-this: c8cc758656558a7f366bf784d75f0304
528 
529 The point here is that SCCs get in the way of eta
530 expansion and we must treat them uniformly.
531]
532[Remove use of lambda with a refutable pattern
533[email protected]**20100923105901
534 Ignore-this: d7d48b94e5744717a838591a1cc79cf0
535]
536[Avoid ASSERT black hole
537[email protected]**20100923105820
538 Ignore-this: 5419d450871be22c8781ac3f0f40d76a
539 
540 When this ASSERT tripped in CoreToStg it tried to print out
541 too much, which tripped the asssertion again.  Result: an
542 infinite loop with no output at all.  Hard to debug!
543]
544[Rejig the absent-arg stuff for unlifted types
545[email protected]**20100923105732
546 Ignore-this: 69daa35816b948b0c4d259c73a5e928e
547 
548 This is what was giving the "absent entered" messages
549 See Note [Absent errors] in WwLib.  We now return a
550 suitable literal for absent values of unlifted type.
551]
552[Remove -fwarn-simple-patterns, and make -fwarn-incomplete-patterns include lambdas
553[email protected]**20100922133934
554 Ignore-this: e851a2fb0377e10c28c506f0bf14cc85
555 
556 This makes
557      \(x:xs) -> e
558 want when you have -fwarn-incomplete-patterns, which is consistent.
559]
560[Get rid of non-exhaustive lambda
561[email protected]**20100922133801
562 Ignore-this: 748b2d5b43b02b6591b81abe7c105cd6
563]
564[Fix an ASSERT failure with profiling
565[email protected]**20100922133741
566 Ignore-this: 170b2e94d6ee8cc7444cc4bb515328a0
567 
568 The problem arose with this kind of thing
569 
570    x = (,) (scc "blah" Nothing)
571 
572 Then 'x' is marked NoCafRefs by CoreTidy, becuase it has
573 arity 1, and doesn't mention any caffy things.
574 
575 That in turns means that CorePrep must not float out the
576 sat binding to give
577 
578   sat = scc "blah" Nothing
579   x = (,) sat
580 
581 Rather we must generate
582 
583   x = \eta. let sat = scc "blah" Nothing
584             in (,) sat eta
585 
586 URGH! This Caf stuff is such a mess.
587]
588[Remove an out of date paragraph from the user guide; fixes #4331
589Ian Lynagh <[email protected]>**20100922225239]
590[Fix bindisttest when GhcProfiled = YES
591Ian Lynagh <[email protected]>**20100921222634
592 Ignore-this: 47c620fd6bec745e3eb699d9f53441d8
593]
594[Fixes for when HADDOCK_DOCS=NO
595Ian Lynagh <[email protected]>**20100921213916
596 Ignore-this: e0e069555c6db9d01a8ac70ba4dde591
597]
598[Bump version to 7.1
599Ian Lynagh <[email protected]>**20100921195935
600 Ignore-this: 4563987e6885d5ef55995ec0fa0d5ae8
601]
602[Don't use -march=i686 on powerpc-apple-darwin
603Ian Lynagh <[email protected]>**20100921193721
604 Thorikil ran into this when doing a PPC OS X build. We now also don't
605 use -m32 on PPC/OSX, but I don't think it should be necessary. We can
606 add it back if it does turn out to be.
607]
608[add a simple trace facility to the build system
609Simon Marlow <[email protected]>**20100921134729
610 Ignore-this: d23ea2d62a648d0711b4b07d98e1b79f
611 
612 saying
613 
614   make TRACE=1
615 
616 prints most of the macro calls and their arguments.  It's easy to
617 trace new macros; see rules/trace.mk.
618]
619[fix building with extra packages (packages were added to BUILD_DIRS twice)
620Simon Marlow <[email protected]>**20100921100153
621 Ignore-this: 4b425dff9777871ad5ba3e05e1d14483
622 Also add some comments about what extra-packages is doing
623]
624[add extra packages to $(EXTRA_PACKAGES), so we avoid installing them by default
625Simon Marlow <[email protected]>**20100920144307
626 Ignore-this: 3395825d911a8bf7ba8385518d8b517b
627]
628[Fix indexing error in archive loader
629Ian Lynagh <[email protected]>**20100921121642]
630[Add some -Dl belches
631Ian Lynagh <[email protected]>**20100921121624]
632[Add casts to fix warnings
633Ian Lynagh <[email protected]>**20100921121714]
634[Add support for BSD-variant large filenames in .a archives
635Ian Lynagh <igloo@earth.li>**20100921000451]
636[Tell Cabal that we're not building GHCi libs if UseArchivesForGhci=YES
637Ian Lynagh <igloo@earth.li>**20100920230449]
638["UseArchivesForGhci = YES" on darwin
639Ian Lynagh <igloo@earth.li>**20100920211538]
640[Add a dependency that my OS X build has recently started tripping up over
641Ian Lynagh <igloo@earth.li>**20100920210239]
642[Add "Use archives for ghci" to --info output
643Ian Lynagh <igloo@earth.li>**20100920210523]
644[Implement archive loading for ghci
645Ian Lynagh <igloo@earth.li>**20100920201620]
646[Tweak gen_contents_index now dph may not be there
647Ian Lynagh <igloo@earth.li>**20100920201513]
648[Filter out the FFI library when loading package in ghci
649Ian Lynagh <igloo@earth.li>**20100920181032
650 The FFI GHCi import lib isn't needed as
651 compiler/ghci/Linker.lhs + rts/Linker.c link the
652 interpreted references to FFI to the compiled FFI.
653 We therefore filter it out so that we don't get
654 duplicate symbol errors.
655]
656[Loosen the conditions for -XUndecidableInstances; fixes Trac #4200
657simonpj@microsoft.com**20100919162623
658 Ignore-this: 2f4323e278b1ce9250549727ffd0aa1b
659]
660[Further improvements in error messages
661simonpj@microsoft.com**20100919153355
662 Ignore-this: b6fa0b11ae893df1a3ca68f78e427fa
663]
664[Add a flag -fwarn-missing-local-sigs, and improve -fwarn-mising-signatures
665simonpj@microsoft.com**20100919153327
666 Ignore-this: fda8dfca450054ea692be0ee30b01885
667 
668 The new flag prints out a warning if you have a local,
669 polymorphic binding that lacks a type signature. It's meant
670 to help with the transition to the new typechecker, which
671 discourages local let-generalisation.
672 
673 At the same time I moved the missing-signature code to TcHsSyn,
674 where it takes place as part of zonking.  That way the
675 types are reported after all typechecking is complete,
676 thereby fixing Trac #3696.  (It's even more important for
677 local bindings, which is why I made the change.)
678]
679[Include the "stupid theta" in the type of $con2tag
680simonpj@microsoft.com**20100919152201
681 Ignore-this: d95fae78a0e66f48bbd5862573a11f4d
682]
683[Add a release note about the typechecker
684Ian Lynagh <igloo@earth.li>**20100919132927]
685[Enable shared libs on OpenBSD
686Matthias Kilian <kili@outback.escape.de>**20100918205040
687 Ignore-this: 729dd7ac0bba5d758f43bc31b541896
688]
689[Add separate functions for querying DynFlag and ExtensionFlag options
690Ian Lynagh <igloo@earth.li>**20100918163815
691 and remove the temporary DOpt class workaround.
692]
693[Fix mkUserGuidePart deps
694Ian Lynagh <igloo@earth.li>**20100918145904
695 We need to directly depend on the stage1 libs. The stage1 compiler lib
696 doesn't depend on them.
697]
698[Fix build on cygwin: Normalise slashes in .depend files to be /
699Ian Lynagh <igloo@earth.li>**20100918132328
700 Ignore-this: 664f5ef4a41a4461eb34fe2ca7f2729a
701]
702[extra packages info is now read from packages file
703Ian Lynagh <igloo@earth.li>**20100917224409
704 rather than being repeated in the build system
705]
706[Tweak darcs-all
707Ian Lynagh <igloo@earth.li>**20100917194435]
708[Bump dependencies
709Ian Lynagh <igloo@earth.li>**20100917183609]
710[Library release notes for 7.0.1
711Ian Lynagh <igloo@earth.li>**20100917174850]
712[Fix overriding of implicit parameters in the solver
713simonpj@microsoft.com**20100917140403
714 Ignore-this: af76732309c7e2ca6b04f49327e9c14b
715]
716[Minor type printing amomaly
717simonpj@microsoft.com**20100917140204
718 Ignore-this: c90cb2e51421b4543a827e096051772e
719]
720[Spaces only
721simonpj@microsoft.com**20100917140156
722 Ignore-this: 7e34479502f7cb87d762355e40cbd012
723]
724[Minor refactoring
725simonpj@microsoft.com**20100917140150
726 Ignore-this: 6c0648b949b91b7e2f23c136b124faf2
727]
728[Add types of implicit parameters as untouchable
729simonpj@microsoft.com**20100917140138
730 Ignore-this: ba80740a557a9ba062dc7756e2320d17
731 
732 This is a tricky point:
733    see Note [Implicit parameter untouchables]
734]
735[Better pretty printing of implicit parameters
736simonpj@microsoft.com**20100917140054
737 Ignore-this: 867dd67818a5bd687b2b6a1b59e15775
738]
739[Yet more error message improvement
740simonpj@microsoft.com**20100917121206
741 Ignore-this: 647fe8129d1d39d81e8249debd8df94e
742]
743[More error message wibbles
744simonpj@microsoft.com**20100917094721
745 Ignore-this: 8ec2f150b96b26af2e9ab7ac2b371fc7
746]
747[More error refactoring
748simonpj@microsoft.com**20100917092834
749 Ignore-this: 2d570ac0b9cc11305ddd33d093d11324
750]
751[Refactor type errors a bit
752simonpj@microsoft.com**20100917080726
753 Ignore-this: 33da4549373f585064e2ee22b50ad6ac
754 
755 Improves kind error messages in paticular
756]
757[Fix a very subtle shadowing bug in optCoercion
758simonpj@microsoft.com**20100916170452
759 Ignore-this: 9041cfb3c93e27a5e644e57815313aae
760 
761 See Note [Subtle shadowing in coercions]
762 
763 This is what was going wrong in Trac 4160.
764]
765[Fix bad error in tyVarsOfType
766simonpj@microsoft.com**20100916170348
767 Ignore-this: 67c8ce96a668cf6e3a38b82c893bcd81
768 
769 We weren't gathering the type variables free in the kind
770 of a coercion binder!
771]
772[More assertions
773simonpj@microsoft.com**20100916170310
774 Ignore-this: 7fdcb53c99d791621a3d7e01ef454404
775]
776[Add more location info in CoreLint
777simonpj@microsoft.com**20100916170229
778 Ignore-this: 6558bab544b4f30189e0430668db87c3
779]
780[Print coercion variables as such (debugging change only)
781simonpj@microsoft.com**20100916165944
782 Ignore-this: c6d2001c1d8279a2288cb63bc339577d
783]
784[Remove pprTrace
785simonpj@microsoft.com**20100915225935
786 Ignore-this: 28185bbfa9732386f3c0f3eb4781a637
787]
788[Remove dead code dealing with type refinement
789simonpj@microsoft.com**20100915223230
790 Ignore-this: 62824b5c2ec1077c7642163352559621
791]
792[Use mkAppTy
793simonpj@microsoft.com**20100915223205
794 Ignore-this: e79e087b6a49219e9088846a1253a153
795 
796 Using AppTy in CoreLint was giving a bogus Lint failure
797]
798[Comments only
799simonpj@microsoft.com**20100915221253
800 Ignore-this: 3a45ea614188ccbb4a462de5cac96eda
801]
802[Extend eta reduction to work with casted arguments
803simonpj@microsoft.com**20100915221229
804 Ignore-this: 24b103dcdf70331211507af929789f86
805 
806 See Trac #4201, and
807 Note [Eta reduction with casted arguments]
808 
809 Thanks to Louis Wasserman for suggesting this, and
810 implementing an early version of the patch
811]
812[Allow "INLINEABLE" as a synonym
813simonpj@microsoft.com**20100915154249
814 Ignore-this: f41f80cb769e9acd5b463b170df698d0
815]
816[Documentation for INLINABLE
817simonpj@microsoft.com**20100915154235
818 Ignore-this: f942c02bcadc0d2d2f05b9369f93e280
819]
820[Implement TH reification of instances (Trac #1835)
821simonpj@microsoft.com**20100915151242
822 Ignore-this: 97dfa83db7da8f6cbd1b96801a57f8c5
823 
824 Accompanying patch for template-haskell package is reqd
825]
826[errno corresponding to ERROR_NO_DATA should be EPIPE (non-threaded RTS)
827Simon Marlow <marlowsd@gmail.com>**20100915141809
828 Ignore-this: 709c7280fbaa762e7071fb8796e8c01e
829]
830[Windows: use a thread-local variable for myTask()
831Simon Marlow <marlowsd@gmail.com>**20100915120627
832 Ignore-this: 13ffa4f19ebd319fe672af53af8d0b9a
833 Which entailed fixing an incorrect #ifdef in Task.c
834]
835[Fix typo
836Ian Lynagh <igloo@earth.li>**20100915140814]
837[Add quotes in error message
838simonpj@microsoft.com**20100915144724
839 Ignore-this: c5158047c0aa41947a79e4c8edbe54f4
840]
841[Fix isDefaultInlinePragma
842simonpj@microsoft.com**20100915144710
843 Ignore-this: c9addf6bf811b23dc12603cf8521aa6c
844]
845[Implement INLINABLE pragma
846simonpj@microsoft.com**20100915124442
847 Ignore-this: 80a4ab2c2d65b27868dc9b2e954d6c6f
848 
849 Implements Trac #4299.  Documentation to come.
850]
851[Less voluminous error when derived code doesn't typecheck
852simonpj@microsoft.com**20100915072301
853 Ignore-this: eca7871dcc50c1070a0b530711adea27
854]
855[Improve pretty-printing of family instances
856simonpj@microsoft.com**20100915123219
857 Ignore-this: 25ec6bcc7e8a7f7c303b38ca201db90e
858 
859 Fixed Trac #4246
860]
861[Fix Trac #4240: -ddump-minimal-imports
862simonpj@microsoft.com**20100915121937
863 Ignore-this: ab85057cb829a42ea44a92f7b4af24a3
864 
865 See Note [Partial export] for the details.
866 I also fixed one egregious bug that was just
867 waiting to bite: we were using loadSysInterface
868 instead of loadSrcInterface.
869]
870[Comments only
871simonpj@microsoft.com**20100915105707
872 Ignore-this: ab3a5f16f8260b7b8570e748bf97998a
873]
874[implement setThreadAffinity on Windows (#1741)
875Simon Marlow <marlowsd@gmail.com>**20100914155844
876 Ignore-this: a14c7b4ef812007042342d0a25478f0b
877]
878[COFF: cope with new debug sections in gcc 4.x (fixes ghciprog004)
879Simon Marlow <marlowsd@gmail.com>**20100914153026
880 Ignore-this: f340e40a2b0390836bc61bba144a04ed
881 Also updated the object file parser to properly handle the overflow
882 case for section names longer than 8 chars.
883]
884[eliminate clutter from make output
885Simon Marlow <marlowsd@gmail.com>**20100915105712
886 Ignore-this: bfa4480dd239dda2a02ac391b6a9219c
887]
888[rts_isProfiled should be a visible API (fixes T2615(dyn))
889Simon Marlow <marlowsd@gmail.com>**20100915083941
890 Ignore-this: b8ac09bb9d1a929bf45c6122f8485561
891]
892[Fix the "lost due to fragmentation" calculation
893Simon Marlow <marlowsd@gmail.com>**20100914145945
894 Ignore-this: cdffcc9f3061c3a33da5171be111fc43
895 It was counting the space used by block descriptors as "lost"
896]
897[fix +RTS -S output: use peak_mblocks_allocated, now that mblocks can be freed
898Simon Marlow <marlowsd@gmail.com>**20100914135030
899 Ignore-this: 65d21e5f86d3ab6ab4d6c255180b6968
900]
901[Fix egregious bug in deeplyInstantiate
902simonpj@microsoft.com**20100915070325
903 Ignore-this: 22ede973038877af2673339aaf5de6cf
904 
905 This resulted in an infinite loop in applyTypeToArgs, in syb
906]
907[Improve HsSyn pretty printing
908simonpj@microsoft.com**20100915070255
909 Ignore-this: 7c8e2d86a482453c7e69e22bc31cb03f
910]
911[Remove (most of) the FiniteMap wrapper
912Ian Lynagh <igloo@earth.li>**20100914201703
913 We still have
914     insertList, insertListWith, deleteList
915 which aren't in Data.Map, and
916     foldRightWithKey
917 which works around the fold(r)WithKey addition and deprecation.
918]
919[Improve ASSERT
920simonpj@microsoft.com**20100914113900
921 Ignore-this: dbc0363be5924f543316e77f7d18dd77
922]
923[Comment on what an "enumeration" type is
924simonpj@microsoft.com**20100914113850
925 Ignore-this: c09c8591e3140f305d55fbf945adbf95
926]
927[Make absent-arg wrappers work for unlifted types (fix Trac #4306)
928simonpj@microsoft.com**20100914113827
929 Ignore-this: 1945e56779329e8b79780403710aba98
930 
931 Previously we were simply passing arguments of unlifted
932 type to a wrapper, even if they were absent, which was
933 stupid.
934 
935 See Note [Absent error Id] in WwLib.
936]
937[Comments only
938simonpj@microsoft.com**20100914113641
939 Ignore-this: 3191ce856c9b5d9700cedc9b149b8097
940]
941[Move error-ids to MkCore (from PrelRules)
942simonpj@microsoft.com**20100914113635
943 Ignore-this: c3d820db62ba6139dd7c96bf97e51bb5
944 
945 and adjust imports accordingly
946]
947[More wibbles to deriving error messages
948simonpj@microsoft.com**20100914113523
949 Ignore-this: bd2df662644961138fa209aec843a2aa
950]
951[Fix getThreadCPUTime()
952Simon Marlow <marlowsd@gmail.com>**20100913153838
953 Ignore-this: 950e048a5724086534b74c609c7d5ed
954 ever since the patch "Check with sysconf _POSIX_THREAD_CPUTIME", it
955 has been returning incorrect results, because the sysconf variable to
956 check should have been _SC_THREAD_CPUTIME, not _POSIX_THREAD_CPUTIME.
957]
958[filter out the gcc-lib directory from the rts package's library-dirs
959Simon Marlow <marlowsd@gmail.com>**20100913101259
960 Ignore-this: 46dc1dccbfee8a65f9243e125eee117f
961 fixes problems when building with GHC 6.10 on Windows
962]
963[Don't include GC time in heap profiles (#4225)
964Simon Marlow <marlowsd@gmail.com>**20100913133852
965 Ignore-this: 68ac48b004b311384b5996c6b33ba5cc
966]
967[Use clock_gettime (if available) to measure the process CPU time
968Simon Marlow <marlowsd@gmail.com>**20100913133818
969 Ignore-this: 8c9300df9b929bfc1db4713c9b6065b3
970 This is much more accurate than getrusage, which was giving misleading
971 results when trying to time very quick operations like a minor GC.
972]
973[make stg_arg_bitmaps public, and available via the GHCi linker (#3672)
974Simon Marlow <marlowsd@gmail.com>**20100913105235
975 Ignore-this: e18efd0bd77c521e5530fb59e93b5a42
976]
977[fix typo
978Simon Marlow <marlowsd@gmail.com>**20100913105100
979 Ignore-this: 6049eea21208864203b2d79db2edd143
980]
981[Update release notes and docs with LLVM info.
982David Terei <davidterei@gmail.com>**20100914072135
983 Ignore-this: 5b3d0e5c9d5da98ed6ae9c2e8e1f6f30
984]
985[Remove defaultExtensionFlags
986Ian Lynagh <igloo@earth.li>**20100913165949
987 The default should do into languageExtensions instead
988]
989[Improve crash message
990simonpj@microsoft.com**20100913170407
991 Ignore-this: 5c26a9979f18be8cd12cea823c9f4b5a
992]
993[Fix Trac #4302, plus a little refactoring
994simonpj@microsoft.com**20100913170355
995 Ignore-this: cf6886b587aa0e8d723362183625d946
996]
997[Fix build with 6.10
998Ian Lynagh <igloo@earth.li>**20100913160048]
999[Haddock fixes
1000simonpj@microsoft.com**20100913120510
1001 Ignore-this: f3157d6969f10d4cbd593000a477138b
1002]
1003[Remove two old junk files
1004simonpj@microsoft.com**20100913103426
1005 Ignore-this: ed7af5ef1b9592178909a8d876345302
1006]
1007[Super-monster patch implementing the new typechecker -- at last
1008simonpj@microsoft.com**20100913095048
1009 Ignore-this: 14d14a1e4d7a414f5ae8d9d89d1c6a4b
1010 
1011 This major patch implements the new OutsideIn constraint solving
1012 algorithm in the typecheker, following our JFP paper "Modular type
1013 inference with local assumptions". 
1014 
1015 Done with major help from Dimitrios Vytiniotis and Brent Yorgey.
1016 
1017]
1018[Fix simplifier statistics
1019simonpj@microsoft.com**20100909085441
1020 Ignore-this: 48e383655aafc912dea15c4d94382863
1021]
1022[Trace output
1023simonpj@microsoft.com**20100908170056
1024 Ignore-this: 4b67fa4b310fbf0a16b852686d2d3294
1025]
1026[Better debug output
1027simonpj@microsoft.com**20100908170047
1028 Ignore-this: 410cef00616dda7c0c162f65216e8ca3
1029]
1030[Add Outputable instance for OccEncl
1031simonpj@microsoft.com**20100908150510
1032 Ignore-this: 6362ef9028287d84f070eaf8963c1bfc
1033]
1034[Better simplifier counting
1035simonpj@microsoft.com**20100907214840
1036 Ignore-this: 9d4722703f8f47447e86a28c8c50e0ea
1037]
1038[Put liftStringName into the known-key names
1039simonpj@microsoft.com**20100906112415
1040 Ignore-this: 287064d14ff484da1a6dea6924bc9235
1041]
1042[Deprecate NoRelaxedPolyRec
1043simonpj@microsoft.com**20100903234519
1044 Ignore-this: 607217e77f6bc1b91bf57dfd8dd2b967
1045]
1046[Buglet in Core Lint
1047simonpj@microsoft.com**20100903234457
1048 Ignore-this: 277535d51b396d3b4b0265a0939c2d4
1049]
1050[Give seqId the right type
1051simonpj@microsoft.com**20100903093556
1052 Ignore-this: d1fc7a73dea160614a8d4ddc930f99cd
1053]
1054[Remove dead code
1055simonpj@microsoft.com**20100903093548
1056 Ignore-this: 92cc3f7651445aa349ee7f114d3ec758
1057]
1058[Comments and layout
1059simonpj@microsoft.com**20100903093502
1060 Ignore-this: 9987d1409e654992c1cb1be35cb87728
1061]
1062[Remove checkFreeness; no longer needed
1063simonpj@microsoft.com**20100902233227
1064 Ignore-this: c96a12ac9794290aa30402317d88c095
1065]
1066[Assert
1067simonpj@microsoft.com**20100902073642
1068 Ignore-this: 4be1ab2f6096665ae5ec7fdd1f025a67
1069]
1070[Add aserts
1071simonpj@microsoft.com**20100902073211
1072 Ignore-this: e1409441217fd070c5a7f9ee4cca99ab
1073]
1074[Wibbles
1075simonpj@microsoft.com**20100831113540
1076 Ignore-this: 903811ab493a7b560a62eb86fcf3ee25
1077]
1078[Wibble to allow phantom types in Enum
1079simonpj@microsoft.com**20100825112711
1080 Ignore-this: fdef1c50d92b4a3d46bbe4cbfd8a83ea
1081]
1082[Add HsCoreTy to HsType
1083simonpj@microsoft.com**20100824141845
1084 Ignore-this: 4ca742b099f9cc90af3167f1012dbba6
1085 
1086 The main thing here is to allow us to provide type
1087 signatures for 'deriving' bindings without pain.
1088]
1089[Comments
1090simonpj@microsoft.com**20100823223654
1091 Ignore-this: dd412a55839430c436902d8699d6900b
1092]
1093[Wibbles to error message
1094simonpj@microsoft.com**20100823163308
1095 Ignore-this: 4d6cd8e613762dca8135c2e3b09264ec
1096]
1097[Correct type signatures
1098simonpj@microsoft.com**20100823153045
1099 Ignore-this: 42942309221a443258246098f9c0a13b
1100]
1101[Add missing signatures
1102simonpj@microsoft.com**20100823112413
1103 Ignore-this: 8ee1ce40456306de469938c02df4fed5
1104]
1105[Add type signatures in "deriving" bindings
1106simonpj@microsoft.com**20100820234230
1107 Ignore-this: 4726b28968cf65ec16cb65b7e0e7303e
1108]
1109[Minor
1110dimitris@microsoft.com**20100820131021]
1111[Be a bit less aggressive in mark-many inside a cast
1112simonpj@microsoft.com**20100819104804
1113 Ignore-this: 3fd48fe7647ec7a58c2032cd86ca4d4f
1114]
1115[Wibble
1116simonpj@microsoft.com**20100818185738
1117 Ignore-this: d5c939311377c0d0c7244aa339193315
1118]
1119[Pretty printing change
1120simonpj@microsoft.com**20100818065436
1121 Ignore-this: 4f7e70976dbe52f95effb3e634dfef5d
1122]
1123[Remember to zonk FlatSkols!
1124simonpj@microsoft.com**20100811143555
1125 Ignore-this: 84f7f9dbda97f561a918c69308ddef9a
1126]
1127[De-polymorphise
1128simonpj@microsoft.com**20100730151217
1129 Ignore-this: a9304487b983e517a9083fd697f77576
1130]
1131[Work around missing type signature in Happy
1132simonpj@microsoft.com**20100730122405
1133 Ignore-this: 7f241a655d93c5ad7763a7ffe8db0c7a
1134 
1135 Happy generates
1136 
1137       notHappyAtAll = error "Blah"
1138 
1139 without a type signature, and currently the new
1140 typechecker doesn't generalise it.  This patch
1141 says "no monomorphism restriction" which makes it
1142 generalise again.
1143 
1144 Better would be to add a type sig to Happy's template
1145]
1146[Add two local type signatures
1147simonpj@microsoft.com**20100729152611
1148 Ignore-this: afa99bcc515469aa0990d44d8c18a9e6
1149]
1150[Second test from Simon's laptop
1151simonpj@microsoft.com**20100729091703
1152 Ignore-this: 4dc64cadae314a5a1b05cc5326918a83
1153]
1154[Test commit from Simon's laptop
1155simonpj@microsoft.com**20100729091344
1156 Ignore-this: 109eff835cc19e9f93799d12f09b0ba7
1157]
1158[Add OutsideIn flag
1159simonpj@microsoft.com**20100728075525
1160 Ignore-this: 69c2f5c3a15fa653f6da80598aa8d74d
1161]
1162[Layout only
1163simonpj@microsoft.com**20100727141539
1164 Ignore-this: 1a58a8fe80ba8bced18ae81a2efb9495
1165]
1166[Improvement to SimplUtils.mkLam
1167simonpj@microsoft.com**20100727131659
1168 Ignore-this: 739beaefa79baa7e0ebeb5b2b6d1ea91
1169]
1170[Give the correct kind to unsafeCoerce#
1171simonpj@microsoft.com**20100727131538
1172 Ignore-this: 6b787de3b398c6d7a61fa04fccd15fd6
1173]
1174[Suppress warnings about recursive INLINE in output of desugarer
1175simonpj@microsoft.com**20100727094549
1176 Ignore-this: a361f7238c0fcba526d46326722c42e
1177]
1178[Rename CorePrep.tryEtaReduce to tryEtaReducePrep
1179simonpj@microsoft.com**20100726231253
1180 Ignore-this: 4375ddace205745244ba224ae012252
1181 
1182 This avoids the name clash with the similar but
1183 not identical CoreUtils.tryEtaReduce
1184]
1185[Add a trace message
1186simonpj@microsoft.com**20100719211111
1187 Ignore-this: b5daebe46e50c8cf28cc693f84bbf099
1188]
1189[Don't use RelaxedPolyRec in the compiler; it's built in now
1190simonpj@microsoft.com**20100719170441
1191 Ignore-this: a2e4489cdf63478e46282a421ee7aec3
1192]
1193[Remove duplicated #defines for FreeBSD
1194Matthias Kilian <kili@outback.escape.de>**20100912181518
1195 Ignore-this: d16214fef8635c7c9ef4edec4e8e7896
1196]
1197[Don't fail with absolute silence
1198Matthias Kilian <kili@outback.escape.de>**20100912150506
1199 Ignore-this: 479e2321f39b263fa2d9f80491e5e9f7
1200]
1201[Add a release note: "-dynload wrapper" removed
1202Ian Lynagh <igloo@earth.li>**20100911195809]
1203[put back the conversion of warn-lazy-unlifted-bindings into an error until 7.2
1204Ian Lynagh <igloo@earth.li>**20100911193434
1205 I think we'll currently still have too many people with old versions of
1206 alex/happy to want to make this an error now.
1207]
1208[6.14 -> 7.0
1209Ian Lynagh <igloo@earth.li>**20100911192837]
1210[Add a couple more release notes
1211Ian Lynagh <igloo@earth.li>**20100911162059]
1212[Document -dsuppress-module-prefixes
1213Ian Lynagh <igloo@earth.li>**20100911162005]
1214[Enable -fregs-graph with -O2; fixes #2790
1215Ian Lynagh <igloo@earth.li>**20100910191301]
1216[Remove unused code
1217Ian Lynagh <igloo@earth.li>**20100909170207]
1218[Fix warnings
1219Ian Lynagh <igloo@earth.li>**20100909154348]
1220[Fix warnings
1221Ian Lynagh <igloo@earth.li>**20100909150957]
1222[Remove context completion
1223lykahb@gmail.com**20100901160153
1224 Ignore-this: dc61b259dcb7063f0c76f56788b5d2af
1225 Now completion suggests to remove only modules added to context before.
1226]
1227[avoid Foreign.unsafePerformIO
1228Ross Paterson <ross@soi.city.ac.uk>**20100909125531
1229 Ignore-this: 5cabeae4cffec8fc17ef7c0cabbea22a
1230]
1231[updates to the release notes
1232Simon Marlow <marlowsd@gmail.com>**20100909111450
1233 Ignore-this: a4d25ad8815c305b7e0f21fd4f6ee37b
1234]
1235[newAlignedPinnedByteArray#: avoid allocating an extra word sometimes
1236Simon Marlow <marlowsd@gmail.com>**20100909110805
1237 Ignore-this: 996a3c0460068ab2835b4920905b3e75
1238]
1239[Finish breaking up vectoriser utils
1240benl@ouroborus.net**20100909061311
1241 Ignore-this: 217fe1d58a3e8bb13200bcb81353a416
1242]
1243[Move VectType module to Vectorise tree
1244benl@ouroborus.net**20100909042451
1245 Ignore-this: 5af8cf394d4835911259ca3ffb6774c5
1246]
1247[Sort all the PADict/PData/PRDict/PRepr stuff into their own modules
1248benl@ouroborus.net**20100909035147
1249 Ignore-this: 53436329773347cad793adbd83e90a9e
1250]
1251[Break out Repr and PADict stuff for vectorisation of ADTs to their own modules
1252benl@ouroborus.net**20100909025759
1253 Ignore-this: d2b7d2f79332eda13416449742f7cf1c
1254]
1255[Break out conversion functions to own module
1256benl@ouroborus.net**20100909023332
1257 Ignore-this: 613f2666b6ca7f2f8876fcc1e4a59593
1258]
1259[Comments and formatting only
1260benl@ouroborus.net**20100909022117
1261 Ignore-this: c8e30139d730669e5db44f0ef491a588
1262]
1263[Remove "-dynload wrapper"; fixes trac #4275
1264Ian Lynagh <igloo@earth.li>**20100908213251]
1265[Don't set visibility on Windows
1266Ian Lynagh <igloo@earth.li>**20100905122442
1267 With gcc 4.5.0-1, using visibility hidden gives:
1268     error: visibility attribute not supported in this configuration; ignored
1269]
1270[Fix warnings on Windows
1271Ian Lynagh <igloo@earth.li>**20100905111201
1272 Ignore-this: c5cce63bb1e0c7a27271bed78d25fbc5
1273]
1274[Fix gcc wrapper for new mingw binaries
1275Ian Lynagh <igloo@earth.li>**20100905001807
1276 Ignore-this: f6acc8c911055ffce632bac138ccc939
1277]
1278[Don't pass our gcc options to stage0 ghc's gcc; they may not be suitable
1279Ian Lynagh <igloo@earth.li>**20100905103129]
1280[Update intree-mingw creation
1281Ian Lynagh <igloo@earth.li>**20100904225559]
1282[Update commands to build in-tree mingw
1283Ian Lynagh <igloo@earth.li>**20100904215112]
1284[Break out hoisting utils into their own module
1285benl@ouroborus.net**20100908074102
1286 Ignore-this: e3ba4ed0252a2def1ed88a9e14c58fea
1287]
1288[Break out closure utils into own module
1289benl@ouroborus.net**20100908072040
1290 Ignore-this: 216172b046ff101cf31a1753667a5383
1291]
1292[Move VectVar module to Vectorise tree
1293benl@ouroborus.net**20100908065904
1294 Ignore-this: 1fba5333d29927dba4275381e1a7f315
1295]
1296[Break out vectorisation of expressions into own module
1297benl@ouroborus.net**20100908065128
1298 Ignore-this: 6a952b80fb024b5291f166477eb1976
1299]
1300[Break out TyCon classifier into own module
1301benl@ouroborus.net**20100908063111
1302 Ignore-this: da754c4ef6960b4e152ea1bf8c04ab6f
1303]
1304[Break out vectorisation of TyConDecls into own module
1305benl@ouroborus.net**20100908052004
1306 Ignore-this: c0ab4fb2a05ca396efe348b384db1ebf
1307]
1308[Break out type vectorisation into own module
1309benl@ouroborus.net**20100907110311
1310 Ignore-this: 67bd70a21d16468daf68dd3ec1ff7d62
1311]
1312[Tidy up the ArchHasAdjustorSupport definition
1313Ian Lynagh <igloo@earth.li>**20100904144234]
1314[ppc: switch handling of 'foreign import wrapper' (FIW) to libffi
1315Sergei Trofimovich <slyfox@community.haskell.org>**20100829192859
1316 Ignore-this: 662ea926681ebea0759e2a04a38e82b7
1317 
1318 Joseph Jezak reported darcs-2.4.4 SIGSEGV in interactive mode in ghc-6.12.3.
1319 So I've concluded ppc also has rotten native adjustor. I don't have hardware
1320 to verify the patch (ticket #3516 should help to test it), but I think it will
1321 help (as similar patch helped for ia64 and ppc64).
1322]
1323[Binary no longer has unusable UNPACK pragmas, so no need to turn of -Werror
1324Ian Lynagh <igloo@earth.li>**20100904133339]
1325[Don't haddock packages that we aren't going to install
1326Ian Lynagh <igloo@earth.li>**20100903231921]
1327[Give haddock per-package source entity paths; fixes #3810
1328Ian Lynagh <igloo@earth.li>**20100903221335]
1329[update for containers-0.4
1330Simon Marlow <marlowsd@gmail.com>**20100903105131
1331 Ignore-this: 556eac0e4926c9b8af6b66d7b069302c
1332]
1333[Fix for nursery resizing: the first block's back pointer should be NULL
1334Simon Marlow <marlowsd@gmail.com>**20100827102818
1335 Ignore-this: fb68938e3f1e291e3c9e5e8047f9dcd2
1336 I'm not sure if this could lead to a crash or not, but it upsets +RTS -DS
1337 Might be related to #4265
1338]
1339[Add some -no-user-package-conf flags
1340Ian Lynagh <igloo@earth.li>**20100902224726
1341 Stops user-installed packages breaking the build
1342]
1343[Fix warnings: Remove unused imports
1344Ian Lynagh <igloo@earth.li>**20100902204342]
1345[Finish breaking up VectBuiltIn and VectMonad, and add comments
1346benl@ouroborus.net**20100831100724
1347 Ignore-this: 65604f3d22d03433abc12f10be40050d
1348]
1349[Fix warnings
1350benl@ouroborus.net**20100830083746
1351 Ignore-this: 2a0e000985f694582a6f9a9261ff2739
1352]
1353[Break up vectoriser builtins module
1354benl@ouroborus.net**20100830070900
1355 Ignore-this: b86bd36a7875abdcf16763902ba2e637
1356]
1357[Move VectCore to Vectorise tree
1358benl@ouroborus.net**20100830053415
1359 Ignore-this: d5763ca6424285b39a58c7792f4a84a1
1360]
1361[Split out vectoriser environments into own module
1362benl@ouroborus.net**20100830050252
1363 Ignore-this: 5319111c74831394d2c29b9aedf5a766
1364]
1365[Comments and formatting to vectoriser, and split out varish stuff into own module
1366benl@ouroborus.net**20100830042722
1367 Ignore-this: d3f0c98ed8124dd0fca9a2ccea3e15fd
1368]
1369[Fix warnings
1370benl@ouroborus.net**20100830040340
1371 Ignore-this: d6cfad803ad4617e7fdaa62e4a895282
1372]
1373[Fix warning about multiply exported name
1374benl@ouroborus.net**20100830035243
1375 Ignore-this: 27ce2c1d22d9f99929d16a426343044e
1376]
1377[Vectorisation of method types
1378benl@ouroborus.net**20100830032941
1379 Ignore-this: 75614571d5c246a4906edb3b39ab1e0b
1380]
1381[Comments and formatting to vectoriser
1382benl@ouroborus.net**20100830032516
1383 Ignore-this: fe665b77108501c7960d858be3290761
1384]
1385[Implement -dsuppress-module-prefixes
1386benl@ouroborus.net**20100830032428
1387 Ignore-this: 2bb8bad9c60ef9044132bba118010687
1388]
1389[Whitespace only
1390benl@ouroborus.net**20100527045629
1391 Ignore-this: 4c160dfa77727e659817b6af9c84684a
1392]
1393[Disambiguate a function name
1394Ian Lynagh <igloo@earth.li>**20100828225827]
1395[users_guide.xml is now generated
1396Ian Lynagh <igloo@earth.li>**20100828225751]
1397[Pass more -pgm flags in the ghc wrapper; fixes #3863
1398Ian Lynagh <igloo@earth.li>**20100827204537]
1399[Add a new-IO manager release note
1400Ian Lynagh <igloo@earth.li>**20100827171616]
1401[Merge a duplicate release note
1402Ian Lynagh <igloo@earth.li>**20100827171511]
1403[Typo, spotted by Johan Tibell
1404Ian Lynagh <igloo@earth.li>**20100827153914]
1405[First pass at 6.14.1 release notes
1406Ian Lynagh <igloo@earth.li>**20100826220811]
1407[Fix typo
1408Ian Lynagh <igloo@earth.li>**20100824201330]
1409[FIX BUILD: add rts_isProfiled to the symbol table
1410Simon Marlow <marlowsd@gmail.com>**20100826094319
1411 Ignore-this: 9536ddb0a94721c8dec03a2a981cfa83
1412]
1413[Fix the DPH package cleaning/profiled mess even more (the build was broken)
1414Simon Marlow <marlowsd@gmail.com>**20100826084436
1415 Ignore-this: 49d7e4db2fb53b856c213c74c8969d82
1416]
1417[Remove the debugging memory allocator - valgrind does a better job
1418Simon Marlow <marlowsd@gmail.com>**20100824113537
1419 Ignore-this: a3731a83dc18b0fd0de49452e695a7ca
1420 
1421 I got fed up with the constant bogus output from the debugging memory
1422 allocator in RtsUtils.c.  One problem is that we allocate memory in
1423 constructors that then isn't tracked, because the debugging allocator
1424 hasn't been initialised yet.
1425 
1426 The bigger problem is that for a given piece of leaking memory it's
1427 impossible to find out where it was allocated; however Valgrind gives
1428 output like this:
1429 
1430 ==6967== 8 bytes in 1 blocks are still reachable in loss record 1 of 7
1431 ==6967==    at 0x4C284A8: malloc (vg_replace_malloc.c:236)
1432 ==6967==    by 0x4C28522: realloc (vg_replace_malloc.c:525)
1433 ==6967==    by 0x6745E9: stgReallocBytes (RtsUtils.c:213)
1434 ==6967==    by 0x68D812: setHeapAlloced (MBlock.c:91)
1435 ==6967==    by 0x68D8E2: markHeapAlloced (MBlock.c:116)
1436 ==6967==    by 0x68DB56: getMBlocks (MBlock.c:240)
1437 ==6967==    by 0x684F55: alloc_mega_group (BlockAlloc.c:305)
1438 ==6967==    by 0x6850C8: allocGroup (BlockAlloc.c:358)
1439 ==6967==    by 0x69484F: allocNursery (Storage.c:390)
1440 ==6967==    by 0x694ABD: allocNurseries (Storage.c:436)
1441 ==6967==    by 0x6944F2: initStorage (Storage.c:217)
1442 ==6967==    by 0x673E3C: hs_init (RtsStartup.c:160)
1443 
1444 which tells us exactly what the leaking bit of memory is.  So I don't
1445 think we need our own debugging allocator.
1446]
1447[free the entries in the thread label table on exit
1448Simon Marlow <marlowsd@gmail.com>**20100824112606
1449 Ignore-this: c9d577c06548cda80791e590e40d35b3
1450]
1451[Panic in the right way
1452simonpj@microsoft.com**20100825091614
1453 Ignore-this: e6ea4f6dfd2aea088828ea7a945ddd5f
1454]
1455[Fix the DPH/profiled make thing (again)
1456simonpj@microsoft.com**20100825091602
1457 Ignore-this: bc58fa48034ac40cf7be4170958ea29e
1458]
1459[Don't test for gcc flags before we've found gcc
1460Ian Lynagh <igloo@earth.li>**20100824131401]
1461[Change how the dblatex/lndir problem is worked around
1462Ian Lynagh <igloo@earth.li>**20100824130938
1463 Hack: dblatex normalises the name of the main input file using
1464 os.path.realpath, which means that if we're in a linked build tree,
1465 it find the real source files rather than the symlinks in our link
1466 tree. This is fine for the static sources, but it means it can't
1467 find the generated sources.
1468 
1469 We therefore also generate the main input file, so that it really
1470 is in the link tree, and thus dblatex can find everything.
1471]
1472[Clean the generated userguide sources
1473Ian Lynagh <igloo@earth.li>**20100824105827
1474 Ignore-this: 39b4f9702c688c053ed3273b20969597
1475]
1476[DPH should not even be built if GhcProfiled
1477simonpj@microsoft.com**20100823133439
1478 Ignore-this: 62acbf83de5b70ff6d27ab38ce9218ae
1479 
1480 It's not just when cleaning!
1481]
1482[The templateHaskellOk check should only run in stage2
1483simonpj@microsoft.com**20100823133353
1484 Ignore-this: f6dc9292923a1ca201953c5f58c0af3c
1485 
1486 Because rtsIsProfiled is only available in stage2
1487]
1488[Add a couple of missing tests for EAGER_BLACKHOLE
1489Simon Marlow <marlowsd@gmail.com>**20100823104654
1490 Ignore-this: 70c981b86370b0c7564b29b057650897
1491 This was leading to looping and excessive allocation, when the
1492 computation should have just blocked on the black hole. 
1493 
1494 Reported by Christian Höner zu Siederdissen <choener@tbi.univie.ac.at>
1495 on glasgow-haskell-users.
1496]
1497[Don't check for swept blocks in -DS.
1498Marco Túlio Gontijo e Silva <marcot@marcot.eti.br>**20100718225526
1499 Ignore-this: ad5dcf3c247bc19fbef5122c1142f3b2
1500 
1501 The checkHeap function assumed the allocated part of the block contained only
1502 alive objects and slops.  This was not true for blocks that are collected using
1503 mark sweep.  The code in this patch skip the test for this kind of blocks.
1504]
1505[Fix "darcs get"
1506Ian Lynagh <igloo@earth.li>**20100822183542]
1507[Add "darcs-all upstreampull"
1508Ian Lynagh <igloo@earth.li>**20100822163419
1509 This pulls from the upstream repos, for those packages which have
1510 upstreams
1511]
1512[Generate the bit in the user guide where we say what -fglasgow-exts does
1513Ian Lynagh <igloo@earth.li>**20100822155514
1514 Stops the docs going out of sync with the code.
1515]
1516[Factor out the packages file parsing in darcs-all
1517Ian Lynagh <igloo@earth.li>**20100822154813]
1518[Document --supported-extensions
1519Ian Lynagh <igloo@earth.li>**20100822134530]
1520[fix extraction of command stack of arguments of arrow "forms" (fixes #4236)
1521Ross Paterson <ross@soi.city.ac.uk>**20100822090022
1522 Ignore-this: a93db04ec4f20540642a19cdc67d1666
1523 
1524 The command stack was being extracted (by unscramble) with the outermost
1525 type first, contrary to the comment on the function.
1526]
1527[minor fix to comment
1528Ross Paterson <ross@soi.city.ac.uk>**20100822085838
1529 Ignore-this: 8d203ba2600eaf4cf21b043dcfa96cdc
1530]
1531[Add the RTS library path to the library search path
1532Ian Lynagh <igloo@earth.li>**20100820155523
1533 In case the RTS is being explicitly linked in. For #3807.
1534]
1535[Remove some duplication of C flags
1536Ian Lynagh <igloo@earth.li>**20100819233743
1537 We now use the CONF_CC_OPTS_STAGEn C flags in machdepCCOpts, rather than
1538 repeating them there.
1539]
1540[Set -fno-stack-protector in CONF_CC_OPTS_STAGE* rathre than extra-gcc-opts
1541Ian Lynagh <igloo@earth.li>**20100819233031
1542 The latter is only used when compiling .hc files, whereas we need it for
1543 .c files too.
1544]
1545[Give clearer errors for bad input in the packages file; suggested by pejo
1546Ian Lynagh <igloo@earth.li>**20100819232420]
1547[Set -march=i686 on OS X x86 in the configure variables
1548Ian Lynagh <igloo@earth.li>**20100819230939
1549 We used to set it only in machdepCCOpts, so this is more consistent
1550]
1551[Give each stage its own Config.hs
1552Ian Lynagh <igloo@earth.li>**20100819224709
1553 This also means the file is generated in a dist directory, not a
1554 source directory.
1555]
1556[Fix cleaning when GhcProfiled = YES
1557Ian Lynagh <igloo@earth.li>**20100819131346
1558 We need to include the DPH cleaning rules, even though we don't build DPH
1559 when GhcProfiled = YES.
1560]
1561[stgReallocBytes(DEBUG): don't fail when the ptr passed in is NULL
1562Simon Marlow <marlowsd@gmail.com>**20100817150836
1563 Ignore-this: 4b5063e65e01399f64a33f0d0555ff38
1564]
1565[Use make-command in rules/bindist.mk
1566Ian Lynagh <igloo@earth.li>**20100818191243
1567 Rather than it having its own specialised version
1568]
1569[Use make-command when installing packages
1570Ian Lynagh <igloo@earth.li>**20100818190600]
1571[Add _DATA_FILES to package-data.mk files
1572Ian Lynagh <igloo@earth.li>**20100818185801]
1573[Add a "make-command" utility Makefile function
1574Ian Lynagh <igloo@earth.li>**20100818183055]
1575[LLVM: Nicer format for lack of shared lib warning
1576David Terei <davidterei@gmail.com>**20100817145207
1577 Ignore-this: 753d45762601d87761614937a1bb6716
1578]
1579[fix FP_CHECK_ALIGNMENT for autoconf 2.66 (fixes #4252)
1580Ross Paterson <ross@soi.city.ac.uk>**20100816142442
1581 Ignore-this: cd784b8888d32b3b2cc2cc0969ec40f
1582 
1583 Recent versions of AS_LITERAL_IF don't like *'s.  Fix from
1584 
1585 http://blog.gmane.org/gmane.comp.sysutils.autoconf.general/month=20100701
1586]
1587[Refactor the command-line argument parsing (again)
1588simonpj@microsoft.com**20100816074453
1589 Ignore-this: 26dc9e37a88660a887a2e316ed7a9803
1590 
1591 This change allows the client of CmdLineParser a bit more flexibility,
1592 by giving him an arbitrary computation (not just a deprecation
1593 message) for each flag. 
1594 
1595 There are several clients, so there are lots of boilerplate changes.
1596 
1597 Immediate motivation: if RTS is not profiled, we want to make
1598 Template Haskell illegal.  That wasn't with the old setup.
1599]
1600[Add upstream repo to the packages file
1601Ian Lynagh <igloo@earth.li>**20100815154741]
1602[Make the "tag" column of the packages file always present
1603Ian Lynagh <igloo@earth.li>**20100815151657
1604 It makes the parsing simpler if we always have the same number of columns
1605]
1606[Disable object splitting on OSX; works around #4013
1607Ian Lynagh <igloo@earth.li>**20100815134759]
1608[Return memory to the OS; trac #698
1609Ian Lynagh <igloo@earth.li>**20100813170402]
1610[Reduce the xargs -s value we use on Windows
1611Ian Lynagh <igloo@earth.li>**20100812223721
1612 With 30000 I was getting:
1613     xargs: value for -s option should be < 28153
1614]
1615[LLVM: Enable shared lib support on Linux x64
1616David Terei <davidterei@gmail.com>**20100813191534
1617 Ignore-this: 642ed37af38e5f17d419bf4f09332671
1618]
1619[Re-do the arity calculation mechanism again (fix Trac #3959)
1620simonpj@microsoft.com**20100813161151
1621 Ignore-this: d4a2aa48150b503b20c25351a79decfb
1622 
1623 After rumination, yet again, on the subject of arity calculation,
1624 I have redone what an ArityType is (it's purely internal to the
1625 CoreArity module), and documented it better.  The result should
1626 fix #3959, and I hope the related #3961, #3983.
1627 
1628 There is lots of new documentation: in particular
1629  * Note [ArityType] 
1630    describes the new datatype for arity info
1631 
1632  * Note [State hack and bottoming functions]
1633    says how bottoming functions are dealt with, particularly
1634    covering catch# and Trac #3959
1635 
1636 I also found I had to be careful not to eta-expand single-method
1637 class constructors; see Note [Newtype classes and eta expansion].
1638 I think this part could be done better, but it works ok.
1639]
1640[Comments only
1641simonpj@microsoft.com**20100813161019
1642 Ignore-this: baf68300d8bc630bf0b7ab27647b33a0
1643]
1644[Modify FloatOut to fix Trac #4237
1645simonpj@microsoft.com**20100813163120
1646 Ignore-this: ffc8d00d4b7f0a8a785fcef312900413
1647 
1648 The problem was that a strict binding was getting floated
1649 out into a letrec. This only happened when profiling was
1650 on.  It exposed a fragility in the floating strategy.  This
1651 patch makes it more robust.  See
1652       Note [Avoiding unnecessary floating]
1653]
1654[Fix egregious bug in SetLevels.notWorthFloating
1655simonpj@microsoft.com**20100813161429
1656 Ignore-this: d22865f48d417e6a6b732de3dfba378f
1657 
1658 This bug just led to stupid code, which would
1659 later be optimised away, but better not to generate
1660 stupid code in the first place.
1661]
1662[Delete GhcLibProfiled
1663simonpj@microsoft.com**20100813140152
1664 Ignore-this: 2e1a3f677308be726bd022f45e2fd856
1665 
1666 Simon M and I looked at this, and we think GhcLibProfiled is
1667 (a) not needed (b) confusing.
1668 
1669 Ian should review.
1670 
1671 Really, if GhcProfiled is on we should also
1672 check that 'p' is in the GhcLibWays
1673]
1674[Do not build DPH when GhcProfiled (fixes #4172)
1675simonpj@microsoft.com**20100813140021
1676 Ignore-this: 9e20181643b456e841f845ae0cab0a9a
1677 
1678 Reason: DPH uses Template Haskell and TH doesn't work
1679 in a profiled compiler
1680]
1681[Fix Trac #4220
1682simonpj@microsoft.com**20100812131319
1683 Ignore-this: 33141cfd81627592150a9e5973411ff8
1684 
1685 For deriving Functor, Foldable, Traversable with empty
1686 data cons I just generate a null equation
1687    f _ = error "urk"
1688 
1689 There are probably more lurking (eg Enum) but this will do for now.
1690]
1691[Improve the Specialiser, fixing Trac #4203
1692simonpj@microsoft.com**20100812131133
1693 Ignore-this: 482afbf75165e24a80527a6e52080c07
1694 
1695 Simply fixing #4203 is a tiny fix: in case alterantives we should
1696 do dumpUDs *including* the case binder. 
1697 
1698 But I realised that we can do better and wasted far too much time
1699 implementing the idea.  It's described in
1700     Note [Floating dictionaries out of cases]
1701]
1702[Comments
1703simonpj@microsoft.com**20100812101456
1704 Ignore-this: 6362fe887d25688c12ef2c3cf5554ce4
1705]
1706[Comments only
1707simonpj@microsoft.com**20100812101439
1708 Ignore-this: 7ed2f5fc08811cbe9958c2309a9ed1fa
1709]
1710[Fix bug in linting of shadowed case-alternative binders
1711simonpj@microsoft.com**20100812101413
1712 Ignore-this: 9212a5e2c03421749f5935b3944ecf53
1713]
1714[Comments and spacing only
1715simonpj@microsoft.com**20100812101347
1716 Ignore-this: ed59a7dae7decb24470709dc1c118dbb
1717]
1718[Add more info to more parse error messages (#3811)
1719Ian Lynagh <igloo@earth.li>**20100809233108]
1720[Run finalizers *after* updating the stable pointer table (#4221)
1721Simon Marlow <marlowsd@gmail.com>**20100810133739
1722 Ignore-this: b0462f80dd64eac71e599d8a9f6dd665
1723 Silly bug really, we were running the C finalizers while the StablePtr
1724 table was still in a partially-updated state during GC, but finalizers
1725 are allowed to call freeStablePtr() (via hs_free_fun_ptr(), for
1726 example), and chaos ensues.
1727]
1728[Do the dependency-omitting for 'make 1' in a slightly different way
1729Simon Marlow <marlowsd@gmail.com>**20100810093446
1730 Ignore-this: af15edd3a1492cbd93111316b57e02e4
1731 
1732 I encountered a couple of things that broke after Ian's previous
1733 patch: one was my nightly build scripts that use 'make stage=2' at the
1734 top level, and the other is 'make fast' in libraries/base, which uses
1735 'stage=0' to avoid building any compilers.
1736 
1737 So my version of this patch is more direct: it just turns off the
1738 appropriate dependencies using a variable set by 'make 1', 'make 2',
1739 etc.
1740]
1741[Integrate new I/O manager, with signal support
1742Johan Tibell <johan.tibell@gmail.com>**20100724102355
1743 Ignore-this: eb092857a2a1b0ca966649caffe7ac2b
1744]
1745[Add DoAndIfThenElse support
1746Ian Lynagh <igloo@earth.li>**20100808194625]
1747[Make another parse error more informative
1748Ian Lynagh <igloo@earth.li>**20100808193340]
1749[Make a parse error say what it is failing to parse; part of #3811
1750Ian Lynagh <igloo@earth.li>**20100808155732]
1751[Send ghc progress output to stdout; fixes #3636
1752Ian Lynagh <igloo@earth.li>**20100808142542]
1753[Fix the HsColour test in the build system
1754Ian Lynagh <igloo@earth.li>**20100805155319
1755 Ignore-this: ba2752b04801a253e891b31e1914485d
1756]
1757[Fix the -lm configure test; fixes #4155
1758Ian Lynagh <igloo@earth.li>**20100805142508
1759 Ignore-this: 358b8b1074d2d22fb8d362ea6d8b80d6
1760]
1761[Don't restrict filenames in line pragmas to printable characters; fixes #4207
1762Ian Lynagh <igloo@earth.li>**20100805135011
1763 Ignore-this: e3d32312127165e40e6eaa919193d60b
1764 "printable" is ASCII-only, whereas in other locales we can get things like
1765 # 1 "<línea-de-orden>"
1766]
1767[Ensure extension flags are flattened in the Cmm phase
1768Ian Lynagh <igloo@earth.li>**20100805133614
1769 If we start with a .cmmcpp file then they don't get flattened in
1770 the CmmCpp phase, as we don't run that phase.
1771]
1772[Add "cmmcpp" as a Haskellish source suffix
1773Ian Lynagh <igloo@earth.li>**20100805132555]
1774[On amd64/OSX we don't need to be given memory in the first 31bits
1775Ian Lynagh <igloo@earth.li>**20100805120600
1776 Ignore-this: 42eb64e25ad4b66ae022884305e0297b
1777 as PIC is always on
1778]
1779[NCG: Don't worry about trying to re-freeze missing coalescences
1780benl@ouroborus.net**20100702053319
1781 Ignore-this: ea05cbee19b6c5c410db41292cbb64b0
1782]
1783[Make -rtsopts more flexible
1784Ian Lynagh <igloo@earth.li>**20100805011137
1785 The default is a new "some" state, which allows only known-safe flags
1786 that we want on by default. Currently this is only "--info".
1787]
1788[Test for (fd < 0) before trying to FD_SET it
1789Ian Lynagh <igloo@earth.li>**20100804173636]
1790[Remove "On by default" comments in DynFlags
1791Ian Lynagh <igloo@earth.li>**20100802110803
1792 Ignore-this: 2a51055277b5ce9f0e98e1438b212027
1793 These make less sense now we support multiple languges. The
1794 "languageExtensions" function gives the defaults.
1795]
1796[Fix build: Add newline to end of ghc-pkg/Main.hs
1797Ian Lynagh <igloo@earth.li>**20100801183206]
1798[Add a versions haddock binary for Windows
1799Ian Lynagh <igloo@earth.li>**20100801180917]
1800[ghc-pkg: don't fail, if a file is already removed
1801ich@christoph-bauer.net**20100725162606
1802 Ignore-this: 5501d6812c31f4da525c7fb24f6dcaed
1803]
1804[Remove push-all from file list in boot script (push-all no longer exists)
1805Ian Lynagh <igloo@earth.li>**20100801121841
1806 Ignore-this: eec130f06610d8728a57626682860a1a
1807]
1808[Add error checking to boot-pkgs script
1809Ian Lynagh <igloo@earth.li>**20100801121432
1810 Ignore-this: 8afd6663db443c774bad45d75bbfe950
1811]
1812[Add more error checking to the boot script
1813Ian Lynagh <igloo@earth.li>**20100801113628]
1814[Remove libHSrtsmain.a before creating it
1815Ian Lynagh <igloo@earth.li>**20100801005432
1816 Otherwise it isn't updated properly if rts/Main.c changes
1817]
1818[Expose the functions haddock needs even when haddock is disabled; #3558
1819Ian Lynagh <igloo@earth.li>**20100731115506]
1820[Always haddock by default
1821Ian Lynagh <igloo@earth.li>**20100730235001
1822 Revert this patch:
1823     Matthias Kilian <kili@outback.escape.de>**20090920181319
1824     Don't build haddock if HADDOC_DOCS = NO, and disable HADDOC_DOCS
1825         if GhcWithInterpreter = NO
1826     Haddock uses TcRnDriver.tcRnGetInfo, which is only available if
1827     GHCI is built. Set HADDOC_DOCS to NO if GhcWithInterpreter is NO,
1828     and disable the haddock build if HADDOC_DOCS = NO.
1829]
1830[Add a debugTrace for the phases that we run
1831Ian Lynagh <igloo@earth.li>**20100729201503]
1832[* Add StringPrimL as a constructor for Template Haskell (Trac #4168)
1833simonpj@microsoft.com**20100730131922
1834 Ignore-this: 520d0a0a14b499b299e8b2be8d148ff0
1835   
1836 There are already constructors for IntPrim, FloatPrim etc,
1837 so this makes it more uniform.
1838   
1839 There's a corresponding patch for the TH library
1840]
1841[Add thread affinity support for FreeBSD
1842Gabor Pali <pgj@FreeBSD.org>**20100720001409
1843 Ignore-this: 6c117b8219bfb45445089e82ee470410
1844 - Implement missing functions for setting thread affinity and getting real
1845   number of processors.
1846 - It is available starting from 7.1-RELEASE, which includes a native support
1847   for managing CPU sets.
1848 - Add __BSD_VISIBLE, since it is required for certain types to be visible in
1849   addition to POSIX & C99.
1850]
1851[Disable symbol visibility pragmas for FreeBSD
1852Ian Lynagh <igloo@earth.li>**20100729012507
1853 Do not use GCC pragmas for controlling visibility, because it causes
1854 "undefined reference" errors at link-time.  The true reasons are
1855 unknown, however FreeBSD 8.x includes GCC 4.2.1 in the base system,
1856 which might be buggy.
1857]
1858[Fix numeric escape sequences parsing
1859Anton Nikishaev <anton.nik@gmail.com>**20100721194208
1860 Ignore-this: dd71935b1866b5624f7975c45ad519a1
1861 This fixes trac bug #1344
1862]
1863[Explicitly give the right path to perl when making the OS X installer; #4183
1864Ian Lynagh <igloo@earth.li>**20100728163030]
1865[Set -fno-stack-protector in extra-gcc-opts; fixes #4206
1866Ian Lynagh <igloo@earth.li>**20100728161957
1867 We were using it only when building the RTS, and only on certain
1868 platforms. However, some versions of OS X need the flag, while others
1869 don't support it, so we now test for it properly.
1870]
1871[Make PersistentLinkerState fields strict; fixes #4208
1872Ian Lynagh <igloo@earth.li>**20100727201911
1873 Ignore-this: fc5cfba48cd16624f6bb15a7a03a3b4
1874 We modify fields a lot, so we retain the old value if they aren't forced.
1875]
1876[Don't rebuild dependency files unnecessarily when doing "make 1" etc
1877Ian Lynagh <igloo@earth.li>**20100726211512
1878 Ignore-this: d91a729e5113aa964cc67768e92e57ef
1879]
1880[LLVM: If user specifies optlo, don't use '-O' levels
1881David Terei <davidterei@gmail.com>**20100726105650
1882 Ignore-this: e05e103b09d1de937540ffad7983f88e
1883]
1884[Flatten flags for ghci's :show
1885Ian Lynagh <igloo@earth.li>**20100725135320]
1886[Add support for Haskell98 and Haskell2010 "languages"
1887Ian Lynagh <igloo@earth.li>**20100724230121]
1888[Rename "language" varibles etc to "extension", and add --supported-extensions
1889Ian Lynagh <igloo@earth.li>**20100724223624]
1890[Separate language option handling into 2 phases
1891Ian Lynagh <igloo@earth.li>**20100724212013
1892 We now first collect the option instructions (from the commandline,
1893 from pragmas in source files, etc), and then later flatten them into
1894 the list of enabled options. This will enable us to use different
1895 standards (H98, H2010, etc) as a base upon which to apply the
1896 instructions, when we don't know what the base will be when we start
1897 collecting instructions.
1898]
1899[Separate the language flags from the other DynFlag's
1900Ian Lynagh <igloo@earth.li>**20100724133103
1901 Ignore-this: 47bb8d42e621e47016b66c7472bd6cb5
1902]
1903[Set stage-specific CC/LD opts in the bindist configure.ac
1904Ian Lynagh <igloo@earth.li>**20100724113748
1905 Ignore-this: f06926d185a35ddd05490ca4a257e992
1906]
1907[Use different CC/LD options for different stages
1908Ian Lynagh <igloo@earth.li>**20100723223059]
1909[Add some error belchs to the linker, when we find bad magic numbers
1910Ian Lynagh <igloo@earth.li>**20100723200822]
1911[Add some more linker debugging prints
1912Ian Lynagh <igloo@earth.li>**20100723180237]
1913[When (un)loading an object fails, say which object in teh panic
1914Ian Lynagh <igloo@earth.li>**20100723162649]
1915[Add a release note: GHCi import syntax
1916Ian Lynagh <igloo@earth.li>**20100721193647]
1917[Deprecate NewQualifiedOperators extension (rejected by H')
1918Ian Lynagh <igloo@earth.li>**20100719150909
1919 Ignore-this: 6e7e3bedc5360c5975f73497b3e6cba5
1920]
1921[LLVM: Allow optlc and optlo to override default params for these systools
1922David Terei <davidterei@gmail.com>**20100722181631
1923 Ignore-this: e60af7941996f7170fb3bfb02a002082
1924]
1925[LLVM: Code and speed improvement to dominateAllocs pass.
1926David Terei <davidterei@gmail.com>**20100721143654
1927 Ignore-this: 9fb7058c8a2afc005521298c7b8d0036
1928]
1929[Comments only
1930simonpj@microsoft.com**20100721144257
1931 Ignore-this: b3091ddcd1df271eb85fe90978ab7adc
1932]
1933[Fix inlining for default methods
1934simonpj@microsoft.com**20100721144248
1935 Ignore-this: 61a11a8f741f775000c6318aae4b3191
1936 
1937 This was discombobulated by a patch a week ago;
1938 now fixed, quite straightforwardly.  See
1939 Note [Default methods and instances]
1940]
1941[Allow reification of existentials and GADTs
1942simonpj@microsoft.com**20100721090437
1943 Ignore-this: 20f1ccd336cc25aff4d4d67a9ac2211a
1944 
1945 It turns out that TH.Syntax is rich enough to express even GADTs,
1946 provided we express them in equality-predicate form.  So for
1947 example
1948 
1949   data T a where
1950      MkT1 :: a -> T [a]
1951      MkT2 :: T Int
1952 
1953 will appear in TH syntax like this
1954 
1955   data T a = forall b. (a ~ [b]) => MkT1 b
1956            | (a ~ Int) => MkT2
1957 
1958 While I was at it I also improved the reification of types,
1959 so that we use TH.TupleT and TH.ListT when we can.
1960]
1961[add numSparks# primop (#4167)
1962Simon Marlow <marlowsd@gmail.com>**20100720153746
1963 Ignore-this: f3f925e7de28f3f895213aefbdbe0b0f
1964]
1965[LLVM: Decrease max opt level used under OSX to avoid bug
1966David Terei <davidterei@gmail.com>**20100720160938
1967 Ignore-this: 34b0b3550f00b27b00ad92f8232745e5
1968 
1969 Currently, many programs compiled with GHC at -O2 and LLVM
1970 set to -O3 will segfault (only under OSX). Until this issue
1971 is fixed I have simply 'solved' the segfault by lowering
1972 the max opt level for LLVM used to -O2 under OSX.
1973 
1974 All these recent changes to OSX should mean its finally as
1975 stable as Linux and Windows.
1976]
1977[LLVM: Fix OSX to work again with TNTC disabled.
1978David Terei <davidterei@gmail.com>**20100720160845
1979 Ignore-this: 8dc98139cfa536b2a64aa364d040b581
1980]
1981[LLVM: Fix printing of local vars so LLVM works with -fnew-codegen
1982David Terei <davidterei@gmail.com>**20100720160302
1983 Ignore-this: d883c433dfaed67921a8c5360e1f9f6a
1984]
1985[Use a separate mutex to protect all_tasks, avoiding a lock-order-reversal
1986Simon Marlow <marlowsd@gmail.com>**20100716150832
1987 Ignore-this: ffbdb4ee502e0f724d57acb9bfbe9d92
1988 In GHC 6.12.x I found a rare deadlock caused by this
1989 lock-order-reversal:
1990 
1991 AQ cap->lock
1992   startWorkerTask
1993     newTask
1994       AQ sched_mutex
1995 
1996 scheduleCheckBlackHoles
1997   AQ sched_mutex
1998    unblockOne_
1999     wakeupThreadOnCapabilty
2000       AQ cap->lock
2001 
2002 so sched_mutex and cap->lock are taken in a different order in two
2003 places.
2004 
2005 This doesn't happen in the HEAD because we don't have
2006 scheduleCheckBlackHoles, but I thought it would be prudent to make
2007 this less likely to happen in the future by using a different mutex in
2008 newTask.  We can clearly see that the all_tasks mutex cannot be
2009 involved in a deadlock, becasue we never call anything else while
2010 holding it.
2011]
2012['make fast' in a package does not build any compilers
2013Simon Marlow <marlowsd@gmail.com>**20100715125904
2014 Ignore-this: f27e70faf3944831dad16e89a4e273da
2015]
2016[LLVM: Fix up botched last commit
2017David Terei <davidterei@gmail.com>**20100719104823
2018 Ignore-this: a32e0f6a38cb9e02527eb8ca69b3eb59
2019]
2020[LLVM: Fix warning introduce in last commit.
2021David Terei <davidterei@gmail.com>**20100719103411
2022 Ignore-this: e9c92a9402aff50d60ab26e6ad441bfc
2023]
2024[LLVM: Use mangler to fix up stack alignment issues on OSX
2025David Terei <davidterei@gmail.com>**20100718231000
2026 Ignore-this: 9f6e8cb855269cb3a5ac1a23480d0e71
2027]
2028[Fix #4195 (isGadtSyntaxTyCon returns opposite result)
2029illissius@gmail.com**20100715134134
2030 Ignore-this: a90403f893030432b5c15d743647f350
2031]
2032[Update to time 1.2.0.3
2033Ian Lynagh <igloo@earth.li>**20100717181810
2034 Ignore-this: 1ccb4801a73f399e6718ce556543ede1
2035]
2036[Reorder RTS --info output
2037Ian Lynagh <igloo@earth.li>**20100717162356]
2038[Fix unreg prof build: Define CCS_SYSTEM in stg/MiscClosures.h
2039Ian Lynagh <igloo@earth.li>**20100717142832
2040 Ignore-this: 9675f3f51b6dac40483155344e7f45b6
2041]
2042[Make mkDerivedConstants as a stage 1 program
2043Ian Lynagh <igloo@earth.li>**20100717000827
2044 Ignore-this: 5357403461b209b8606f1d33defb51cf
2045 This way it gets the defines for the right platform when cross-compiling
2046]
2047[Don't generate Haskell dependencies if we don't have any Haskell sources
2048Ian Lynagh <igloo@earth.li>**20100717000800
2049 Ignore-this: 454abd0358f535b7e789327125c9206c
2050]
2051[Link programs that have no Haskell objects with gcc rather than ghc
2052Ian Lynagh <igloo@earth.li>**20100716235303
2053 Ignore-this: f65588b69675edea616cc434e769b0a4
2054]
2055[Use gcc to build C programs for stages >= 1
2056Ian Lynagh <igloo@earth.li>**20100716223703
2057 Ignore-this: 9f843a4e17285cda582117504707f9e7
2058]
2059[Add platform info to "ghc --info" output
2060Ian Lynagh <igloo@earth.li>**20100716141953]
2061[Tidy up Config.hs generation
2062Ian Lynagh <igloo@earth.li>**20100716140630]
2063[Fix HC porting test in makefiles
2064Ian Lynagh <igloo@earth.li>**20100716010808
2065 Ignore-this: 6052c1dd022a6108ab2236a299ee1d84
2066 Now that we are trying to support cross compilation, we can't use
2067     "$(TARGETPLATFORM)" != "$(HOSTPLATFORM)"
2068 as a test for HC-porting.
2069]
2070[Change a BUILD var to a HOST var
2071Ian Lynagh <igloo@earth.li>**20100716002558]
2072[Remove an unnecessary #include
2073Ian Lynagh <igloo@earth.li>**20100715233930
2074 Ignore-this: dcede249de6be7e3c9305c9279c2ca07
2075]
2076[Split up some make commands, so that errors aren't overlooked
2077Ian Lynagh <igloo@earth.li>**20100715152237
2078 Ignore-this: fb69b0a25d9ca71dae5e75d38db675cd
2079 When we ask make to run "a | b", if a fails then the pipeline might
2080 still exit successfuly.
2081]
2082[Remove an unnecessary #include
2083Ian Lynagh <igloo@earth.li>**20100715143000
2084 Ignore-this: 4e098cac5dda2dd595ca0a0f5121853c
2085]
2086[Simplify some more CPP __GLASGOW_HASKELL__ tests
2087Ian Lynagh <igloo@earth.li>**20100715142500]
2088[Remove some code only used with GHC 6.11.*
2089Ian Lynagh <igloo@earth.li>**20100715141720]
2090[__GLASGOW_HASKELL__ >= 609 is now always true
2091Ian Lynagh <igloo@earth.li>**20100715141544]
2092[Correct the values in ghc_boot_platform.h
2093Ian Lynagh <igloo@earth.li>**20100714223717
2094 Ignore-this: 4c99116f7ac73fadbd6d16807f57a693
2095]
2096[Change some TARGET checks to HOST checks
2097Ian Lynagh <igloo@earth.li>**20100714184715]
2098[LLVM: Add inline assembly to binding.
2099David Terei <davidterei@gmail.com>**20100714152530
2100 Ignore-this: 72a7b5460c128ed511e8901e5889fe2b
2101]
2102[LLVM: Fix mistype in last commit which broke TNTC under win/linux.
2103David Terei <davidterei@gmail.com>**20100714153339
2104 Ignore-this: 302d7957e3dded80368ebade5312ab35
2105]
2106[Remove unnecessary #include
2107Ian Lynagh <igloo@earth.li>**20100713153704
2108 Ignore-this: c37d3127b1dc68f59270c07173994c28
2109]
2110[Change some TARGET tests to HOST tests in the RTS
2111Ian Lynagh <igloo@earth.li>**20100713141034
2112 Which was being used seemed to be random
2113]
2114[LLVM: Add in new LLVM mangler for implementing TNTC on OSX
2115David Terei <davidterei@gmail.com>**20100713183243
2116 Ignore-this: 394fb74d7f9657d8b454bd0148d24bf7
2117]
2118[Refactor where an error message is generated
2119simonpj@microsoft.com**20100713115733
2120 Ignore-this: f94467856238586fcbbe48537141cf78
2121]
2122[Comments only
2123simonpj@microsoft.com**20100713115703
2124 Ignore-this: 5815442c4e69b9ec331b34242a596253
2125]
2126[Comments on data type families
2127simonpj@microsoft.com**20100713115640
2128 Ignore-this: 90a333bb7f7d64a49fb7dd180d893f6b
2129]
2130[Fix Trac #T4136: take care with nullary symbol constructors
2131simonpj@microsoft.com**20100707135945
2132 Ignore-this: 2a717a24fefcd593ea41c23dad351db0
2133 
2134 When a nullary constructor is a symbol eg (:=:) we need
2135 to take care.  Annoying.
2136]
2137[Fix Trac #4127 (and hence #4173)
2138simonpj@microsoft.com**20100707123125
2139 Ignore-this: 98bb6d0f7182b59f8c93596c61f9785d
2140 
2141 The change involves a little refactoring, so that the default
2142 method Ids are brought into scope earlier, before the value
2143 declarations are compiled.  (Since a value decl may contain
2144 an instance decl in a quote.)
2145 
2146 See Note [Default method Ids and Template Haskell] in
2147 TcTyClsDcls.
2148]
2149[Fix second bug in Trac #4127
2150simonpj@microsoft.com**20100701140124
2151 Ignore-this: c8d1cc27364fe9ee5a52acb1ecb5cdd9
2152 
2153 This bug concerned the awkward shadowing we do for
2154 Template Haskell declaration brackets.  Lots of
2155 comments in
2156 
2157   Note [Top-level Names in Template Haskell decl quotes]
2158]
2159[ia64: switch handling of 'foreign import wrapper' (FIW) to libffi
2160Sergei Trofimovich <slyfox@community.haskell.org>**20100709213922
2161 Ignore-this: fd07687e0089aebabf62de85d2be693
2162 
2163 I tried to build darcs-2.4.4 with ghc-6.12.3 and got coredumps when darcs is used
2164 in interactive mode. I tried test from ticket #3516 and found out FIW code is broken.
2165 Instead of fixing it I just switched to libffi. Result built successfully, passed
2166 'foreign import wrapper' test from ticket #3516 and builds working darcs.
2167]
2168[* storage manager: preserve upper address bits on 64bit machines (thanks to zygoloid)
2169Sergei Trofimovich <slyfox@community.haskell.org>**20100709115917
2170 Ignore-this: 9f1958a19992091ddc2761c389ade940
2171 
2172 Patch does not touch amd64 as it's address lengts is 48 bits at most, so amd64 is unaffected.
2173 
2174 the issue: during ia64 ghc bootstrap (both 6.10.4 and 6.12.3) I
2175 got the failure on stage2 phase:
2176     "inplace/bin/ghc-stage2"   -H32m -O -H64m -O0 -w ...
2177     ghc-stage2: internal error: evacuate: strange closure type 15
2178         (GHC version 6.12.3 for ia64_unknown_linux)
2179         Please report this as a GHC bug:  http://www.haskell.org/ghc/reportabug
2180     make[1]: *** [libraries/dph/dph-base/dist-install/build/Data/Array/Parallel/Base/Hyperstrict.o] Aborted
2181 
2182 gdb backtrace (break on 'barf'):
2183 Breakpoint 1 at 0x400000000469ec31: file rts/RtsMessages.c, line 39.
2184 (gdb) run -B/var/tmp/portage/dev-lang/ghc-6.12.3/work/ghc-6.12.3/inplace/bin --info
2185 Starting program: /var/tmp/portage/dev-lang/ghc-6.12.3/work/ghc-6.12.3/inplace/lib/ghc-stage2 -B/var/tmp/portage/dev-lang/ghc-6.12.3/work/ghc-6.12.3/inplace/bin --info
2186 [Thread debugging using libthread_db enabled]
2187 
2188 Breakpoint 1, barf (s=0x40000000047915b0 "evacuate: strange closure type %d") at rts/RtsMessages.c:39
2189 39        va_start(ap,s);
2190 (gdb) bt
2191 #0  barf (s=0x40000000047915b0 "evacuate: strange closure type %d") at rts/RtsMessages.c:39
2192 #1  0x400000000474a1e0 in evacuate (p=0x6000000000147958) at rts/sm/Evac.c:756
2193 #2  0x40000000046d68c0 in scavenge_srt (srt=0x6000000000147958, srt_bitmap=7) at rts/sm/Scav.c:348
2194 ...
2195 
2196 > 16:52:53 < zygoloid> slyfox: i'm no ghc expert but it looks like HEAP_ALLOCED_GC(q)
2197 >                      is returning true for a FUN_STATIC closure
2198 > 17:18:43 < zygoloid> try: p HEAP_ALLOCED_miss((unsigned long)(*p) >> 20, *p)
2199 > 17:19:12 < slyfox> (gdb) p HEAP_ALLOCED_miss((unsigned long)(*p) >> 20, *p)
2200 > 17:19:12 < slyfox> $1 = 0
2201 > 17:19:40 < zygoloid> i /think/ that means the mblock_cache is broken
2202 > 17:22:45 < zygoloid> i can't help further. however i am suspicious that you seem to have pointers with similar-looking low 33
2203 >                      bits and different high 4 bits, and it looks like such pointers get put into the same bucket in
2204 >                      mblock_cache.
2205 ...
2206 > 17:36:16 < zygoloid> slyfox: try changing the definition of MbcCacheLine to StgWord64, see if that helps
2207 > 17:36:31 < zygoloid> that's in includes/rts/storage/MBlock.h
2208 And it helped!
2209]
2210[Fixing link failure of compiler on ia64 ('-Wl,' prefixed value passed directly to ld)
2211Sergei Trofimovich <slyfox@community.haskell.org>**20100708180943
2212 Ignore-this: ced99785e1f870ee97e5bec658e2504f
2213 
2214     /usr/bin/ld -Wl,--relax -r -o dist-stage1/build/HSghc-6.10.4.o \
2215                                   dist-stage1/build/BasicTypes.o dist-stage1/build/DataCon.o ...
2216     /usr/bin/ld: unrecognized option '-Wl,--relax'
2217 
2218 If we just drop '-Wl,' part it will not help as '-r' and '--relax' are incompatible.
2219 
2220 Looks like '-Wl,--relax' was skipped by earlier binutils' ld as unknown option.
2221 Removing ia64 specific path.
2222]
2223[LLVM: Allow getelementptr to use LlvmVar for indexes.
2224David Terei <davidterei@gmail.com>**20100712152529
2225 Ignore-this: 9e158d9b89a86bca8abf11d082328278
2226]
2227[Move all the warning workarounds to one place
2228Ian Lynagh <igloo@earth.li>**20100710161723]
2229[xhtml is now warning-free
2230Ian Lynagh <igloo@earth.li>**20100710144635]
2231[Move a bit of build system code
2232Ian Lynagh <igloo@earth.li>**20100709224534]
2233[adapt to the new async exceptions API
2234Simon Marlow <marlowsd@gmail.com>**20100709125238
2235 Ignore-this: 55d845e40b9daed3575c1479d8dda1d5
2236]
2237[quiet some new spewage
2238Simon Marlow <marlowsd@gmail.com>**20100709091521
2239 Ignore-this: de7f91976bbc9789e6fd7091f05c25c0
2240]
2241[New asynchronous exception control API (ghc parts)
2242Simon Marlow <marlowsd@gmail.com>**20100708144851
2243 Ignore-this: 56320c5fc61ae3602d586609387aae22
2244 
2245 As discussed on the libraries/haskell-cafe mailing lists
2246   http://www.haskell.org/pipermail/libraries/2010-April/013420.html
2247 
2248 This is a replacement for block/unblock in the asychronous exceptions
2249 API to fix a problem whereby a function could unblock asynchronous
2250 exceptions even if called within a blocked context.
2251 
2252 The new terminology is "mask" rather than "block" (to avoid confusion
2253 due to overloaded meanings of the latter).
2254 
2255 In GHC, we changed the names of some primops:
2256 
2257   blockAsyncExceptions#   -> maskAsyncExceptions#
2258   unblockAsyncExceptions# -> unmaskAsyncExceptions#
2259   asyncExceptionsBlocked# -> getMaskingState#
2260 
2261 and added one new primop:
2262 
2263   maskUninterruptible#
2264 
2265 See the accompanying patch to libraries/base for the API changes.
2266]
2267[remove outdated comment
2268Simon Marlow <marlowsd@gmail.com>**20100708100840
2269 Ignore-this: afb2e9f6fe1f1acda51b0cbdf2637176
2270]
2271[remove 'mode: xml' emacs settings (#2208)
2272Simon Marlow <marlowsd@gmail.com>**20100708100817
2273 Ignore-this: 3a8d997fb90e01ca88dc47fb95feeba0
2274]
2275[typo in comment
2276Simon Marlow <marlowsd@gmail.com>**20100616111359
2277 Ignore-this: d3ef9288d6d6b9ab3bacbe09e0d9801c
2278]
2279[Win32 getProcessElapsedTime: use a higher-resolution time source
2280Simon Marlow <marlowsd@gmail.com>**20100708093223
2281 Ignore-this: 821989d4ff7ff2bff40cee71a881521c
2282 QueryPerformanceCounter() on Windows gives much better resolution than
2283 GetSystemTimeAsFileTime().
2284]
2285[alpha: switch handling of 'foreign import wrapper' (FIW) to libffi
2286Sergei Trofimovich <slyfox@community.haskell.org>**20100708065318
2287 Ignore-this: ddee15876737a6aa7f6dabc8ff79ce0d
2288 
2289 I tried to build ghc-6.12.3 and found out FIW part of code
2290 does not compile anymore. It uses absent functions under #ifdef.
2291 Instead of fixing it I just switched to libffi. Result built successfully
2292 and passed 'foreign import wrapper' test I wrote for trac ticket #3516.
2293 
2294 I didn't try to build -HEAD yet, but this patch only removes code, so
2295 it should not make -HEAD worse.
2296]
2297[Reorder the CPP flags so -optP can override the platform defines
2298Ian Lynagh <igloo@earth.li>**20100708203523]
2299[Add docs for DatatypeContexts extension
2300Ian Lynagh <igloo@earth.li>**20100707230907
2301 Ignore-this: 8158f03b35a2d7442a75fe85d6f1b1c7
2302]
2303[Make datatype contexts an extension (on by default) (DatatypeContexts)
2304Ian Lynagh <igloo@earth.li>**20100707212529
2305 Ignore-this: 6885ff510a0060610eeeba65122caef5
2306]
2307[LLVM: Fix various typos in comments
2308David Terei <davidterei@gmail.com>**20100707220448
2309 Ignore-this: 1ba3e722f150492da2f9d485c5795e80
2310]
2311[Handle haddock headers when looking for LANGUAGE/OPTIONS_GHC pragmas
2312Ian Lynagh <igloo@earth.li>**20100707120423
2313 Ignore-this: a75aa67690284a6cee3e62c943d4fd01
2314]
2315[Make pragState call mkPState, rather than duplicating everything
2316Ian Lynagh <igloo@earth.li>**20100706173007
2317 Ignore-this: 61fe24b99dbe7a42efff1a9dd703a75c
2318 This also means that extsBitmap gets set, whereas is was just being set
2319 to 0 before.
2320]
2321[LLVM: Add alias type defenitions to LlvmModule.
2322David Terei <davidterei@gmail.com>**20100707142053
2323 Ignore-this: eee6ad5385563ccf08e664d2634a03f2
2324]
2325[LLVM: Use packed structure type instead of structure type
2326David Terei <davidterei@gmail.com>**20100707120320
2327 Ignore-this: a06e0359d182291b81cae56993ca385e
2328 
2329 The regular structure type adds padding to conform to the platform ABI,
2330 which causes problems with structures storing doubles under windows since
2331 we don't conform to the platform ABI there. So we use packed structures
2332 instead now that don't do any padding.
2333]
2334[Make mkPState and pragState take their arguments in the same order
2335Ian Lynagh <igloo@earth.li>**20100706172611]
2336[Remove an out-of-date comment
2337Ian Lynagh <igloo@earth.li>**20100706172217
2338 Ignore-this: 710ebd7d2dc01c1b0f1e58a5b6f85701
2339]
2340[LLVM: Stop llvm saving stg caller-save regs across C calls
2341David Terei <davidterei@gmail.com>**20100705162629
2342 Ignore-this: 28b4877b31b9358e682e38fc54b88658
2343 
2344 This is already handled by the Cmm code generator so LLVM is simply
2345 duplicating work. LLVM also doesn't know which ones are actually live
2346 so saves them all which causes a fair performance overhead for C calls
2347 on x64. We stop llvm saving them across the call by storing undef to
2348 them just before the call.
2349]
2350[LLVM: Add in literal undefined value to binding
2351David Terei <davidterei@gmail.com>**20100705161544
2352 Ignore-this: 95d8361b11584aaeec44c30e76916470
2353]
2354[LLVM: Add a literal NULL value to binding
2355David Terei <davidterei@gmail.com>**20100705161308
2356 Ignore-this: 9507b4b12c1157498704a9d1e5860f12
2357 
2358 Patch from Erik de Castro Lopo <erikd@mega-nerd.com>.
2359]
2360[refactor import declaration support (#2362)
2361Simon Marlow <marlowsd@gmail.com>**20100705104557
2362 Ignore-this: ee034ac377078a7e92bfada1907c86a0
2363]
2364[Disable dynamic linking optimisations on OS X
2365Simon Marlow <marlowsd@gmail.com>**20100705103014
2366 Ignore-this: b04420d3705c51112797758d17b2e40c
2367 To improve performance of the RTS when dynamically linked on x86, I
2368 previously disabled -fPIC for certain critical modules (the GC, and a
2369 few others).  However, build reports suggest that the dynamic linker
2370 on OS X doesn't like this, so I'm disabling this optimsation on that
2371 platform.
2372]
2373[trac #2362 (full import syntax in ghci)
2374amsay@amsay.net**20100625032632
2375 Ignore-this: a9d0859d84956beb74e27b797431bf9c
2376 'import' syntax is seperate from ':module' syntax
2377]
2378[Simplify ghc-pkg's Cabal dependencies
2379Ian Lynagh <igloo@earth.li>**20100704184155
2380 We no longer support building with a compiler that doesn't come with
2381 base 4.
2382]
2383[Use Cabal to configure the dist-install ghc-pkg; fixes trac #4156
2384Ian Lynagh <igloo@earth.li>**20100704132612]
2385[Remove dead code (standalone deriving flag no longer used in parser)
2386Ian Lynagh <igloo@earth.li>**20100701162058]
2387[LLVM: Use the inbounds keyword for getelementptr instructions.
2388David Terei <davidterei@gmail.com>**20100702160511
2389 Ignore-this: 3708e658a4c82b78b1402393f4405541
2390]
2391[threadPaused: fix pointer arithmetic
2392Simon Marlow <marlowsd@gmail.com>**20100701085046
2393 Ignore-this: b78210e5d978f18ffd235f1c78a55a23
2394 Noticed by Henrique Ferreiro <hferreiro@udc.es>, thanks!
2395]
2396[LLVM: Change more operations to use getelementptr
2397David Terei <davidterei@gmail.com>**20100701161856
2398 Ignore-this: fb24eb124e203f50680c6fec3ff9fe7d
2399]
2400[Add the haskell2010 package
2401Simon Marlow <marlowsd@gmail.com>**20100630125532
2402 Ignore-this: e9b011313f283a8ff2fcda7d029a01f
2403]
2404[LLVM: Use getelementptr instruction for a lot of situations
2405David Terei <davidterei@gmail.com>**20100630181157
2406 Ignore-this: 34d314dd8dffad9bdcffdc525261a49d
2407 
2408 LLVM supports creating pointers in two ways, firstly through
2409 pointer arithmetic (by casting between pointers and ints)
2410 and secondly using the getelementptr instruction. The second way
2411 is preferable as it gives LLVM more information to work with.
2412 
2413 This patch changes a lot of pointer related code from the first
2414 method to the getelementptr method.
2415]
2416[remove out of date comments; point to the wiki
2417Simon Marlow <marlowsd@gmail.com>**20100625100313
2418 Ignore-this: 95f363a373534b9471b1818102ec592d
2419]
2420[NCG: allocatableRegs is only giving us 8 SSE regs to allocate to
2421benl@ouroborus.net**20100629054321
2422 Ignore-this: b3e0fa0b4ce988a0258dc12261989ee0
2423]
2424[LLVM: Use intrinsic functions for pow, sqrt, sin, cos
2425David Terei <davidterei@gmail.com>**20100628182949
2426 Ignore-this: 98a0befaca3fe2b36d710d8ff9f062c4
2427 
2428 Instead of calling the C library for these Cmm functions
2429 we use intrinsic functions provided by llvm. LLVM will
2430 then either create a compile time constant if possible, or
2431 use a cpu instruction or as a last resort call the C
2432 library.
2433]
2434[LLVM: Fix test '2047' under linux-x64
2435David Terei <davidterei@gmail.com>**20100628165256
2436 Ignore-this: 41735d4f431a430db636621650ccd71e
2437]
2438[LLVM: Fix test 'ffi005' under linux-x64
2439David Terei <davidterei@gmail.com>**20100628155355
2440 Ignore-this: 841f3142c63cc898ac4c3f89698a837e
2441]
2442[LLVM: Update to use new fp ops introduced in 2.7
2443David Terei <davidterei@gmail.com>**20100628144037
2444 Ignore-this: 5dd2e5964e3c039d297ed586841e706b
2445]
2446[Add noalias and nocapture attributes to pointer stg registers
2447David Terei <davidterei@gmail.com>**20100628115120
2448 Ignore-this: 492a1e723cb3a62498d240d7de92dd7
2449 
2450 At the moment this gives a very slight performance boost of around 1 - 2%.
2451 Future changes to the generated code though so that pointers are kept as
2452 pointers more often instead of being cast to integer types straight away
2453 should hopefully improve the benefit this brings.
2454 
2455]
2456[during shutdown, only free the heap if we waited for foreign calls to exit
2457Simon Marlow <marlowsd@gmail.com>**20100628090536
2458 Ignore-this: d545384a4f641d701455d08ef1217479
2459]
2460[Fix typo in -ddump-pass's document.
2461shelarcy <shelarcy@gmail.com>**20100620070759
2462 Ignore-this: f4f1ddb53f147949e948147d89190c37
2463]
2464[Add #undefs for posix source symbols when including papi.h
2465dmp@rice.edu**20100624163514
2466 Ignore-this: 8a1cba21b880d12a75a75f7e96882053
2467 
2468 Validation fails when validating with PAPI support (i.e. GhcRtsWithPapi  = YES
2469 in validate.mk).  The problem is that the posix symbols are defined by a header
2470 included from papi.h. Compilation then fails because these symbols are
2471 redefined in PosixSource.h.
2472 
2473 This patch adds an undefine for the posix symbols after including papi.h and
2474 before including PosixSource.h. The #undefines are localized to Papi.c since
2475 that is the only case where they are getting defined twice.
2476]
2477[Use machdepCCOpts in runPhase_MoveBinary; fixes trac #3952
2478Ian Lynagh <igloo@earth.li>**20100625220953]
2479[LLVM: Fix bug with calling tail with empty list
2480David Terei <davidterei@gmail.com>**20100625115729
2481 Ignore-this: 46b4b32c8d92372a2d49794a96fe1613
2482]
2483[Fix warnings
2484benl@ouroborus.net**20100624091339
2485 Ignore-this: 5ba4bbd6abb9c9d1fb8c5d21ab73f218
2486]
2487[NCG: Comments and formatting only
2488benl@ouroborus.net**20100624083121
2489 Ignore-this: 86002e72c30d06bcc876d8c49f4caa5a
2490]
2491[NCG: Do the actual reversing of SCCs
2492benl@ouroborus.net**20100624082717
2493 Ignore-this: 12d2027ea118e751fbb48b27126150ef
2494]
2495[NCG: Fix dumping of graphs in regalloc stats for graph allocator
2496benl@ouroborus.net**20100624082625
2497 Ignore-this: 2b971bc9e0318099a9afb0e0db135730
2498]
2499[NCG: Reverse SCCs after each round in the graph allocator
2500benl@ouroborus.net**20100624082437
2501 Ignore-this: f0152e4039d6f16f7b5a99b286538116
2502]
2503[NCG: Don't actually complain on unreachable code blocks
2504benl@ouroborus.net**20100624081445
2505 Ignore-this: e7335ae6120917cb858c38c7c6da8e24
2506]
2507[NCG: Do explicit check for precondition of computeLiveness
2508benl@ouroborus.net**20100624080747
2509 Ignore-this: e7053c4e5e4c3c746b5ebf016913424a
2510 
2511  computeLiveness requires the SCCs of blocks to be in reverse dependent
2512  order, and if they're not it was silently giving bad liveness info,
2513  yielding a bad allocation.
2514 
2515  Now it complains, loudly.
2516]
2517[NCG: Fix off-by-one error in realRegSqueeze
2518benl@ouroborus.net**20100623095813
2519 Ignore-this: ab0698686d4c250da8e207f734f8252d
2520]
2521[NCG: Handle stripping of liveness info from procs with no blocks (like stg_split_marker)
2522benl@ouroborus.net**20100623091209
2523 Ignore-this: c0319b6cc62ec713afe4eb03790406e3
2524]
2525[NCG: Emit a warning on unreachable code block instead of panicing
2526benl@ouroborus.net**20100623085002
2527 Ignore-this: d20314b79e3c31e764ed4cd97290c696
2528]
2529[NCG: Remember to keep the entry block first when erasing liveness info
2530Ben.Lippmeier@anu.edu.au**20090917104429
2531 Ignore-this: 1b0c1df19d622858d50ffb6a01f2cef0
2532]
2533[NCG: Refactor representation of code with liveness info
2534Ben.Lippmeier@anu.edu.au**20090917090730
2535 Ignore-this: 2aebb3b02ebd92e547c5abad9feb0f0d
2536 
2537  * I've pushed the SPILL and RELOAD instrs down into the
2538    LiveInstr type to make them easier to work with.
2539 
2540  * When the graph allocator does a spill cycle it now just
2541    re-annotates the LiveCmmTops instead of converting them
2542    to NatCmmTops and back.
2543 
2544  * This saves working out the SCCS again, and avoids rewriting
2545    the SPILL and RELOAD meta instructions into real machine
2546    instructions.
2547]
2548[NCG: Add sanity checking to linear allocator
2549Ben.Lippmeier@anu.edu.au**20090917090335
2550 Ignore-this: 5a442be8b5087d04bc8b58dffa9ea080
2551 If there are are unreachable basic blocks in the native code then the
2552 linear allocator might loop. Detect this case and panic instead.
2553]
2554[NCG: Refactor LiveCmmTop to hold a list of SCCs instead of abusing ListGraph
2555Ben.Lippmeier@anu.edu.au**20090917060332
2556 Ignore-this: 3fec8d69ed0f760e53a202f873d5d9cb
2557]
2558[NCG: Allow the liveness map in a LiveInfo to be Nothing
2559Ben.Lippmeier@anu.edu.au**20090917043937
2560 Ignore-this: 5f82422d54d1b0ffc0589eb7e82fb7a4
2561]
2562[NCG: Also show the result of applying coalesings with -ddump-asm-regalloc-stages
2563Ben.Lippmeier.anu.edu.au**20090917034427
2564 Ignore-this: 76bd6d5ca43adb2167cb25832cbaa80b
2565]
2566[Fix panic when running "ghc -H"; trac #3364
2567Ian Lynagh <igloo@earth.li>**20100624234011
2568 The problem is that showing SDoc's looks at the static flags global
2569 variables, but those are panics while we are parsing the static flags.
2570 We work around this by explicitly using a fixed prettyprinter style.
2571]
2572[Allow for stg registers to have pointer type in llvm BE.
2573David Terei <davidterei@gmail.com>**20100621175839
2574 Ignore-this: fc09b1a8314aef0bde945c77af1124fb
2575 
2576 Before all the stg registers were simply a bit type or
2577 floating point type but now they can be declared to have
2578 a pointer type to one of these. This will allow various
2579 optimisations in the future in llvm since the type is
2580 more accurate.
2581]
2582[Add support for parameter attributes to the llvm BE binding
2583David Terei <davidterei@gmail.com>**20100624111744
2584 Ignore-this: 77f3c0c7bf8f81c4a154dc835ae7bcba
2585 
2586 These allow annotations of the code produced by the backend
2587 which should bring some perforamnce gains. At the moment
2588 the attributes aren't being used though.
2589]
2590[Cast some more nats to StgWord to be on the safe side
2591Simon Marlow <marlowsd@gmail.com>**20100624105700
2592 Ignore-this: e6176683856f9872fdeb2358bb065bb8
2593 And add a comment about the dangers of int overflow
2594]
2595[comments only
2596Simon Marlow <marlowsd@gmail.com>**20100624105105
2597 Ignore-this: fc8f762f4c3a5ffca2f8da2bc63ac2a4
2598]
2599[Fix an arithmetic overflow bug causing crashes with multi-GB heaps
2600Simon Marlow <marlowsd@gmail.com>**20100624104654
2601 Ignore-this: 67210755aa098740ff5230347be0fd5d
2602]
2603[Add support for collecting PAPI native events
2604dmp@rice.edu**20100622195953
2605 Ignore-this: 7269f9c4dfb2912a024eb632200fcd1
2606 
2607 This patch extends the PAPI support in the RTS to allow collection of native
2608 events. PAPI can collect data for native events that are exposed by the
2609 hardware beyond the PAPI present events. The native events supported on your
2610 hardware can found by using the papi_native_avail tool.
2611 
2612 The RTS already allows users to specify PAPI preset events from the command
2613 line. This patch extends that support to allow users to specify native events.
2614 The changes needed are:
2615 
2616 1) New option (#) for the RTS PAPI flag for native events. For example, to
2617    collect the native event 0x40000000, use ./a.out +RTS -a#0x40000000 -sstderr
2618 
2619 2) Update the PAPI_FLAGS struct to store whether the user specified event is a
2620    papi preset or a native event
2621 
2622 3) Update init_countable_events function to add the native events after parsing
2623    the event code and decoding the name using PAPI_event_code_to_name
2624 
2625]
2626[Don't warn about unused bindings with parents in .hs-boot files; trac #3449
2627Ian Lynagh <igloo@earth.li>**20100624110351]
2628[fix the home_imps filter to allow for 'import "this" <module>'
2629Simon Marlow <marlowsd@gmail.com>**20100621125535
2630 Ignore-this: da4e605b0513afc32a4e7caa921a2c76
2631 In the PackageImports extension, import "this" means "import from the
2632 current package".
2633]
2634[Use the standard C wrapper code for the ghc-$version.exe wrapper
2635Ian Lynagh <igloo@earth.li>**20100622202859
2636 Ignore-this: 60cd3e6db3afb63e6ba9e2db3b033580
2637]
2638[Don't rely on "-packagefoo" working; use "-package foo" instead
2639Ian Lynagh <igloo@earth.li>**20100622202547]
2640[Remove unnecessary C #includes
2641Ian Lynagh <igloo@earth.li>**20100622172919]
2642[Make the ghci.exe wrapper call the right ghc.exe
2643Ian Lynagh <igloo@earth.li>**20100622172247]
2644[More updates to datalayout description in llvm BE
2645David Terei <davidterei@gmail.com>**20100622165339
2646 Ignore-this: b0c604fe7673b0aa7c7064694d574437
2647]
2648[Remove LlvmAs phase as the llvm opt tool now handles this phase
2649David Terei <davidterei@gmail.com>**20100622144044
2650 Ignore-this: b9fd8f959702b6af014e2fa654bede3
2651 
2652 This phase originally invoked the llvm-as tool that turns a textual
2653 llvm assembly file into a bit code file for the rest of llvm to deal
2654 with. Now the llvm opt tool can do this itself, so we don't need to
2655 use llvm-as anymore.
2656]
2657[Update datalayout info in llvm BE
2658David Terei <davidterei@gmail.com>**20100622123457
2659 Ignore-this: 89b043d211225dcd819f30549afe1840
2660]
2661[Fix handling of float literals in llvm BE
2662David Terei <davidterei@gmail.com>**20100622121642
2663 Ignore-this: a3b5f382ad4b5a426ad4b581664506fa
2664]
2665[Declare some top level globals to be constant when appropriate
2666David Terei <davidterei@gmail.com>**20100621174954
2667 Ignore-this: 44832f65550d4f995d11c01cc1affef5
2668 
2669 This involved removing the old constant handling mechanism
2670 which was fairly hard to use. Now being constant or not is
2671 simply a property of a global variable instead of a separate
2672 type.
2673]
2674[Reduce the number of passes over the cmm in llvm BE
2675David Terei <davidterei@gmail.com>**20100621125220
2676 Ignore-this: cb2f4e54e8d0f982d5087fbeee35c73c
2677]
2678[Fix negate op not working for -0 in llvm backend
2679David Terei <davidterei@gmail.com>**20100621123606
2680 Ignore-this: c5d76e5cffa781fed074137851b1347f
2681]
2682[ROLLBACK: picCCOpts: -dynamic should not entail -optc-fPIC
2683Simon Marlow <marlowsd@gmail.com>**20100621100409
2684 Ignore-this: f2fac7df33d3919199befc59bd455414
2685 and add a comment to explain why it was wrong.  This fixes the dyn
2686 test failures that sprang up recently.
2687]
2688[Check files are really created in libffi
2689Ian Lynagh <igloo@earth.li>**20100620163724
2690 when we think that the libffi build creates them, so they just depend
2691 on the libffi build stamp.
2692]
2693[Improve the missing-import-list warning
2694Ian Lynagh <igloo@earth.li>**20100620124320
2695 Ignore-this: 551e5fdf2dfb56b49d249e0cebaa6115
2696]
2697[Tweak missing-import-list warning
2698Ian Lynagh <igloo@earth.li>**20100620122622
2699 Ignore-this: 360cdf59ae13d66ded181129325506c4
2700]
2701[trac #1789 (warnings for missing import lists)
2702amsay@amsay.net**20100618150649
2703 Ignore-this: b0b0b1e048fbca0817c1e6fade1153fa
2704]
2705[Refix docs for sizeofByteArray#/sizeofMutableByteArray# (#3800)
2706Ian Lynagh <igloo@earth.li>**20100620103749]
2707[Remove some old commented out code
2708Ian Lynagh <igloo@earth.li>**20100620000459]
2709[SET_ARR_HDR's last argument is now a number of bytes, rather than words
2710Ian Lynagh <igloo@earth.li>**20100619235214
2711 This avoids unnecessary work and potential loss of information
2712]
2713[Replace an (incorrect) bytes-to-words calculation with ROUNDUP_BYTES_TO_WDS
2714Ian Lynagh <igloo@earth.li>**20100619234310]
2715[FIX #38000 Store StgArrWords payload size in bytes
2716Antoine Latter <aslatter@gmail.com>**20100101183346
2717 Ignore-this: 7bf3ab4fc080c46311fc10b179361bb6
2718]
2719[Add win32 datalayout support to llvm backend
2720David Terei <davidterei@gmail.com>**20100618131733
2721 Ignore-this: 4b7bffaa8ef38c628ab852c1a6c1c009
2722]
2723[Remove unused 'ddump-opt-llvm' flag
2724David Terei <davidterei@gmail.com>**20100618101237
2725 Ignore-this: f78467496d986897e49d82646ee2907e
2726]
2727[generate "movl lbl(%reg1), %reg2" instructions, better codegen for -fPIC
2728Simon Marlow <marlowsd@gmail.com>**20100618082258
2729 Ignore-this: a25567ebff9f575303ddc8f2deafebbf
2730]
2731[joinToTargets: fix a case of panic "handleComponent cyclic"
2732Simon Marlow <marlowsd@gmail.com>**20100618082147
2733 Ignore-this: 765baeefbb5a41724004acd92405cecc
2734]
2735[comment typo
2736Simon Marlow <marlowsd@gmail.com>**20100618082102
2737 Ignore-this: e495610b7dd5ec30b02938638b56cb7
2738]
2739[Add support of TNTC to llvm backend
2740David Terei <davidterei@gmail.com>**20100618093205
2741 Ignore-this: 2c27d21668374a5b0d5e844882c69439
2742 
2743 We do this through a gnu as feature called subsections,
2744 where you can put data/code into a numbered subsection
2745 and those subsections will be joined together in descending
2746 order by gas at compile time.
2747]
2748[Don't automatically insert a -fvia-C flag in an unregisterised compiler
2749Ian Lynagh <igloo@earth.li>**20100617190901
2750 Ignore-this: eb25a9a338fade9e17c153da7c5f27e9
2751 The default object mode is already HscC, so it's unnecessary, and
2752 -fvia-C generates a deprecated flag warning now.
2753]
2754[In PosixSource.h, conditionally define things based on platform
2755Ian Lynagh <igloo@earth.li>**20100617174912
2756 This may not be ideal, but it should get GHC building on all platforms
2757 again.
2758]
2759[disable -fPIC for the GC for performance reasons
2760Simon Marlow <marlowsd@gmail.com>**20100617140025
2761 Ignore-this: c7c152bbff71ef7891eaee8ff39fc281
2762 see comment for details
2763]
2764[picCCOpts: -dynamic should not entail -optc-fPIC
2765Simon Marlow <marlowsd@gmail.com>**20100617115259
2766 Ignore-this: d71e42bd56e4bd107d2c431b801855e5
2767]
2768[Make getAllocations() visible
2769Simon Marlow <marlowsd@gmail.com>**20100617113259
2770 Ignore-this: 1b7fb38a01358c0acbe8987df07d23f2
2771]
2772[Fix the symbol visibility pragmas
2773Simon Marlow <marlowsd@gmail.com>**20100617105758
2774 Ignore-this: 76552500865473a1dbebbc1cc2def9f0
2775]
2776[pick up changes to $(GhcStage1HcOpts) without re-configuring the ghc package
2777Simon Marlow <marlowsd@gmail.com>**20100616124718
2778 Ignore-this: afb56d5560c813051285607fefb15493
2779]
2780[Fix bindisttest Makefile
2781Ian Lynagh <igloo@earth.li>**20100616205611
2782 Ignore-this: 39cd352152422f378572fc3859c5a377
2783]
2784[Remove some more unused make variables
2785Ian Lynagh <igloo@earth.li>**20100616180519]
2786[Convert some more variable names to FOO_CMD, for consistency
2787Ian Lynagh <igloo@earth.li>**20100616175916]
2788[Rename some variables from FOO to FOO_CMD
2789Ian Lynagh <igloo@earth.li>**20100616161108
2790 This fixes a problem with commands like gzip, where if $GZIP is exported
2791 in the environment, then when make runs a command it'll put the Makefile
2792 variable's value in the environment. But gzip treats $GZIP as arguments
2793 for itself, so when we run gzip it thinks we're giving it "gzip" as an
2794 argument.
2795]
2796[Make the "show" target work anywhere in the build tree
2797Ian Lynagh <igloo@earth.li>**20100616122910
2798 Ignore-this: 299d40cbe16112accd9f14e56fa12158
2799]
2800[Change ghc-pwd's license to a string Cabal recognises
2801Ian Lynagh <igloo@earth.li>**20100615204015
2802 Ignore-this: c935b6ad7f605aab0168997a90b40fc6
2803]
2804[fix warning
2805Simon Marlow <marlowsd@gmail.com>**20100604205933
2806 Ignore-this: 2aaa4ed6a8b9ae1e39adc4696aaf14a3
2807]
2808[--install-signal-handles=no does not affect the timer signal (#1908)
2809Simon Marlow <marlowsd@gmail.com>**20100527214627
2810 Ignore-this: b0c51f1abdb159dc360662485095a11a
2811]
2812[Small optimisation: allocate nursery blocks contiguously
2813Simon Marlow <marlowsd@gmail.com>**20100509194928
2814 Ignore-this: e650e99e9ea9493d2efb245d565beef4
2815 This lets automatic prefetching work better, for a tiny performance boost
2816]
2817[fix -fforce-recomp setting: module is PrimOp, not PrimOps
2818Simon Marlow <marlowsd@gmail.com>**20100507084507
2819 Ignore-this: f76e0d9b643682ec0e8fb7d91afdea68
2820]
2821[it should be an error to use relative directories (#4134)
2822Simon Marlow <marlowsd@gmail.com>**20100615151740
2823 Ignore-this: 2068021701832e018ca41b22877921d5
2824]
2825[missing include-dirs or library-dirs is only a warning now (#4104)
2826Simon Marlow <marlowsd@gmail.com>**20100615151702
2827 Ignore-this: e3114123cef147bbd28ccb64581a1afb
2828]
2829[fix #3822: desugaring case command in arrow notation
2830Ross Paterson <ross@soi.city.ac.uk>**20100615225110
2831 Ignore-this: 477d6c460b4174b94b4cd113fa5b9d19
2832 
2833 Get the set of free variables from the generated case expression:
2834 includes variables in the guards and decls that were missed before,
2835 and is also a bit simpler.
2836]
2837[Deprecate the -fvia-C flag; trac #3232
2838Ian Lynagh <igloo@earth.li>**20100615151836
2839 Ignore-this: c2452b2648bf7e44546465c1b964fce
2840]
2841[Avoid using the new ~~ perl operator in the mangler
2842Ian Lynagh <igloo@earth.li>**20100615151236
2843 Ignore-this: 709a7ba4e514b1596841b3ba7e5c6cc
2844]
2845[stmAddInvariantToCheck: add missing init of invariant->lock (#4057)
2846Simon Marlow <marlowsd@gmail.com>**20100615123643
2847 Ignore-this: 3b132547fa934cecf71a846db2a5f70e
2848]
2849[Add new LLVM code generator to GHC. (Version 2)
2850David Terei <davidterei@gmail.com>**20100615094714
2851 Ignore-this: 4dd2fe5854b64a3f0339d484fd5c238
2852 
2853 This was done as part of an honours thesis at UNSW, the paper describing the
2854 work and results can be found at:
2855 
2856 http://www.cse.unsw.edu.au/~pls/thesis/davidt-thesis.pdf
2857 
2858 A Homepage for the backend can be found at:
2859 
2860 http://hackage.haskell.org/trac/ghc/wiki/Commentary/Compiler/Backends/LLVM
2861 
2862 Quick summary of performance is that for the 'nofib' benchmark suite, runtimes
2863 are within 5% slower than the NCG and generally better than the C code
2864 generator.  For some code though, such as the DPH projects benchmark, the LLVM
2865 code generator outperforms the NCG and C code generator by about a 25%
2866 reduction in run times.
2867 
2868]
2869[Fix Trac #4127: build GlobalRdrEnv in GHCi correctly
2870simonpj@microsoft.com**20100615070626
2871 Ignore-this: d907e3bfa7882878cea0af172aaf6e84
2872 
2873 GHCi was building its GlobalRdrEnv wrongly, so that the
2874 gre_par field was bogus.  That in turn fooled the renamer.
2875 The fix is easy: use the right function!  Namely, call
2876 RnNames.gresFromAvail rather than availsToNameSet.
2877]
2878[Comments, and improvement to pretty-printing of HsGroup
2879simonpj@microsoft.com**20100615070409
2880 Ignore-this: ec8358f2485370b20226a97ec84e9024
2881]
2882[Don't reverse bindings in rnMethodBinds (fix Trac #4126)
2883simonpj@microsoft.com**20100614163935
2884 Ignore-this: a6ffbb5af6f51b142ed0aeae8ee5e3a9
2885]
2886[Fix Trac #4120: generate a proper coercion when unifying forall types
2887simonpj@microsoft.com**20100614134311
2888 Ignore-this: 601592bb505305f1954cbe730f168da4
2889 
2890 This was just a blatant omission, which hasn't come up before.
2891 Easily fixed, happily.
2892]
2893[Use mkFunTy to ensure that invariants are respected
2894simonpj@microsoft.com**20100614134159
2895 Ignore-this: 67dcada7a4e8d9927581cd77af71b6f
2896]
2897[Remove redundant debug code
2898simonpj@microsoft.com**20100601154151
2899 Ignore-this: e6ff11c04c631cf6aac73788cbcf02b5
2900]
2901[Fix Trac #4099: better error message for type functions
2902simonpj@microsoft.com**20100531140413
2903 Ignore-this: 3f53ca98cf770577818b9c0937482577
2904 
2905 Now we only want about "T is a type function and might not be
2906 injective" when matchin (T x) against (T y), which is the case
2907 that is really confusing.
2908]
2909[Gruesome fix in CorePrep to fix embarassing Trac #4121
2910simonpj@microsoft.com**20100614132726
2911 Ignore-this: fe82d15474afaac3e6133adfd7a7e055
2912 
2913 This is a long-lurking bug that has been flushed into
2914 the open by other arity-related changes.  There's a
2915 long comment
2916 
2917      Note [CafInfo and floating]
2918 
2919 to explain. 
2920 
2921 I really hate the contortions we have to do through to keep correct
2922 CafRef information on top-level binders.  The Right Thing, I believe,
2923 is to compute CAF and arity information later, and merge it into the
2924 interface-file information when the latter is generated.
2925 
2926 But for now, this hackily fixes the problem.
2927]
2928[Fix a bug in CorePrep that meant output invariants not satisfied
2929simonpj@microsoft.com**20100531150013
2930 Ignore-this: d34eb36d8877d3caf1cf2b20de426abd
2931 
2932 In cpePair I did things in the wrong order so that something that
2933 should have been a CprRhs wasn't.  Result: a crash in CoreToStg.
2934 Fix is easy, and I added more informative type signatures too.
2935]
2936[Robustify the treatement of DFunUnfolding
2937simonpj@microsoft.com**20100531145332
2938 Ignore-this: 8f5506ada4d89f6ab8ad1e8c3ffb09ba
2939 
2940 See Note [DFun unfoldings] in CoreSyn.  The issue here is that
2941 you can't tell how many dictionary arguments a DFun needs just
2942 from looking at the Arity of the DFun Id: if the dictionary is
2943 represented by a newtype the arity might include the dictionary
2944 and value arguments of the (single) method.
2945 
2946 So we need to record the number of arguments need by the DFun
2947 in the DFunUnfolding itself.  Details in
2948    Note [DFun unfoldings] in CoreSyn
2949]
2950[Fix spelling in comment
2951simonpj@microsoft.com**20100614132259
2952 Ignore-this: bbf0d55f2e5f10ef9c74592c12f9201c
2953]
2954[Update docs on view patterns
2955simonpj@microsoft.com**20100614074801
2956 Ignore-this: 8617b9078800d4942d71f142a5b6c831
2957]
2958[Fix printing of splices; part of #4124
2959Ian Lynagh <igloo@earth.li>**20100613154838
2960 Just putting parens around non-atomic expressions isn't sufficient
2961 for splices, as only the $x and $(e) forms are valid input.
2962]
2963[In ghci, catch IO exceptions when calling canonicalizePath
2964Ian Lynagh <igloo@earth.li>**20100613134627
2965 We now get an exception if the path doesn't exist
2966]
2967[Whitespace only
2968Ian Lynagh <igloo@earth.li>**20100612213119]
2969[Whitespace only
2970Ian Lynagh <igloo@earth.li>**20100612165450]
2971[Update ghci example output in user guide; patch from YitzGale in #4111
2972Ian Lynagh <igloo@earth.li>**20100612162250]
2973[Fix #4131 missing UNTAG_CLOSURE in messageBlackHole()
2974benl@ouroborus.net**20100611044614]
2975[messageBlackHole: fix deadlock bug caused by a missing 'volatile'
2976Simon Marlow <marlowsd@gmail.com>**20100610080636
2977 Ignore-this: 3cda3054bb45408aa9bd2d794b69c938
2978]
2979[Pass --no-tmp-comp-dir to Haddock (see comment)
2980Simon Marlow <marlowsd@gmail.com>**20100604083214
2981 Ignore-this: bfa4d74038637bd149f4d878b4eb8a87
2982]
2983[Track changes to DPH libs
2984Roman Leshchinskiy <rl@cse.unsw.edu.au>**20100607052903
2985 Ignore-this: 4dbc3f8418af3e74b3fc4f9a9dfe7764
2986]
2987[Track changes to DPH libs
2988Roman Leshchinskiy <rl@cse.unsw.edu.au>**20100607012642
2989 Ignore-this: 5d4e498171a3c57ab02621bfaea82cff
2990]
2991[In ghc-pkg, send warnings to stderr
2992Ian Lynagh <igloo@earth.li>**20100606161726
2993 Ignore-this: 56927d13b5e1c1ce2752734f0f9b665b
2994]
2995[Re-add newlines to enable layout for multi-line input.
2996Ian Lynagh <igloo@earth.li>**20100602180737
2997 Patch from Adam Vogt <vogt.adam@gmail.com>
2998 Partial fix for #3984
2999]
3000[Don't use unnecessary parens when printing types (Fix Trac 4107)
3001simonpj@microsoft.com**20100604110143
3002 Ignore-this: a833714ab13013c4345b222f4e87db1d
3003 
3004    f :: Eq a => a -> a
3005 rather than
3006    f :: (Eq a) => a -> a
3007]
3008[Track DPH library changes
3009Roman Leshchinskiy <rl@cse.unsw.edu.au>**20100604005728
3010 Ignore-this: 32bc2fbea6ad975e89545d4c42fd7c30
3011]
3012[fix --source-entity option passed to Haddock: we needed to escape a #
3013Simon Marlow <marlowsd@gmail.com>**20100603125459
3014 Ignore-this: d52ae6188b510c482bcebb23f0e553ae
3015]
3016[__stg_EAGER_BLACKHOLE_INFO -> __stg_EAGER_BLACKHOLE_info (#4106)
3017Simon Marlow <marlowsd@gmail.com>**20100602091419
3018 Ignore-this: 293315ac8f86fd366b8d61992ecc7961
3019]
3020[Add xhtml package (a new dependency of Haddock; not installed/shipped)
3021Simon Marlow <marlowsd@gmail.com>**20100602090101
3022 Ignore-this: af0ac8b91abe98f7fdb624ea0a4dee20
3023]
3024[Use UserInterrupt rather than our own Interrupted exception (#4100)
3025Simon Marlow <marlowsd@gmail.com>**20100602082345
3026 Ignore-this: 1909acf2f452593138b9f85024711714
3027]
3028[Add the global package DB to ghc --info (#4103)
3029Simon Marlow <marlowsd@gmail.com>**20100602082233
3030 Ignore-this: fd5c0e207e70eb0f62606c45dc5b8124
3031]
3032[rts/sm/GC.c: resize_generations(): Remove unneeded check of number of generations.
3033Marco Túlio Gontijo e Silva <marcot@debian.org>**20100528115612
3034 Ignore-this: 6f1bea62917c01c7adac636146132c97
3035 
3036 This "if" is inside another "if" which checks for RtsFlags.GcFlags.generations
3037 > 1, so testing this again is redundant, assuming the number of generations
3038 won't change during program execution.
3039]
3040[rts/sm/BlockAlloc.c: Small comment correction.
3041Marco Túlio Gontijo e Silva <marcot@debian.org>**20100526205839
3042 Ignore-this: bd2fcd4597cc872d80b0e2eeb1c3998a
3043]
3044[rts/sm/GC.c: Annotate constants.
3045Marco Túlio Gontijo e Silva <marcot@debian.org>**20100526205707
3046 Ignore-this: f232edb89383564d759ed890a18f602f
3047]
3048[includes/rts/storage/GC.h: generation_: n_words: Improve comment.
3049Marco Túlio Gontijo e Silva <marcot@debian.org>**20100526204615
3050 Ignore-this: f5d5feefa8f7b552303978f1804fea23
3051]
3052[Add PPC_RELOC_LOCAL_SECTDIFF support; patch from PHO in #3654
3053Ian Lynagh <igloo@earth.li>**20100601204211
3054 Ignore-this: 51293b7041cdce3ce7619ef11cf7ceb
3055]
3056[powerpc-apple-darwin now supports shared libs
3057Ian Lynagh <igloo@earth.li>**20100601173325]
3058[PIC support for PowerPC
3059pho@cielonegro.org**20100508143900
3060 Ignore-this: 3673859a305398c4acae3f4d7c997615
3061 
3062 PPC.CodeGen.getRegister was not properly handling PicBaseReg.
3063 It seems working with this patch, but I'm not sure this change is correct.
3064]
3065[Vectoriser: only treat a function as scalar if it actually computes something
3066Roman Leshchinskiy <rl@cse.unsw.edu.au>**20100601045630
3067 Ignore-this: e5d99a6ddb62052e3520094a5af47552
3068]
3069[Add a release notes file for 6.14.1
3070Ian Lynagh <igloo@earth.li>**20100530171117
3071 Ignore-this: 1941e6d3d1f4051b69ca2f17a1cf84d6
3072]
3073[Check dblatex actually creates the files we tell it to
3074Ian Lynagh <igloo@earth.li>**20100530171043
3075 Ignore-this: ccc72caea2313be05cbac59bb54c0603
3076 If it fails, it still exits successfully.
3077]
3078[Add darwin to the list of OSes for which we use mmap
3079Ian Lynagh <igloo@earth.li>**20100529145016
3080 Ignore-this: a86d12a3334aaaafc86f7af9dbb0a7ae
3081 Patch from Barney Stratford
3082]
3083[Simplify the CPP logic in rts/Linker.c
3084Ian Lynagh <igloo@earth.li>**20100529144929
3085 Ignore-this: 1288f5b752cc1ab8b1c90cfd0ecfdf68
3086]
3087[Fix validate on OS X
3088Ian Lynagh <igloo@earth.li>**20100529154726]
3089[OS X x86_64 fix from Barney Stratford
3090Ian Lynagh <igloo@earth.li>**20100529122440]
3091[OS X 64 installer fixes from Barney Stratford
3092Ian Lynagh <igloo@earth.li>**20100528234935]
3093[fix warning
3094Simon Marlow <marlowsd@gmail.com>**20100525155812
3095 Ignore-this: f34eee3fe3d89579fd8d381c91ced750
3096]
3097[Fix doc bugs (#4071)
3098Simon Marlow <marlowsd@gmail.com>**20100525155728
3099 Ignore-this: aa25be196de567de360075022a1942f7
3100]
3101[Make sparks into weak pointers (#2185)
3102Simon Marlow <marlowsd@gmail.com>**20100525150435
3103 Ignore-this: feea0bb5006007b82c932bc3006124d7
3104 The new strategies library (parallel-2.0+, preferably 2.2+) is now
3105 required for parallel programming, otherwise parallelism will be lost.
3106]
3107[If you say 'make' or 'make stage=2' here, pretend we're in the ghc dir
3108Simon Marlow <marlowsd@gmail.com>**20100525085301
3109 Ignore-this: 78b740337aa460915c812cbbcdae5321
3110]
3111[Another attempt to get these #defines right
3112Simon Marlow <marlowsd@gmail.com>**20100525154313
3113 Ignore-this: 460ca0c47d81cd25eae6542114f67899
3114 Apparently on Solaris it is an error to omit _ISOC99_SOURCE when using
3115 _POSIX_C_SOURCE==200112L.
3116]
3117[Add configure flags for the location of GMP includes/library; fixes #4022
3118Ian Lynagh <igloo@earth.li>**20100525221616
3119 Ignore-this: fc3060caf995d07274ec975eeefbdf3e
3120]
3121[Refactor pretty printing of TyThings to fix Trac #4015
3122simonpj@microsoft.com**20100525153126
3123 Ignore-this: 8f15053b7554f62caa84201d2e4976d2
3124]
3125[When haddocking, we need the dependencies to have been built
3126Ian Lynagh <igloo@earth.li>**20100525145830
3127 as haddock loads the .hi files with the GHC API.
3128]
3129[Fix profiling output; spotted by jlouis
3130Ian Lynagh <igloo@earth.li>**20100525111217
3131 We were outputing the number of words allocated in a column titled "bytes".
3132]
3133[Improve printing of TyThings; fixes Trac #4087
3134simonpj@microsoft.com**20100525114045
3135 Ignore-this: da2a757a533454bba80b9b77cc5a771
3136]
3137[Spelling in comments
3138simonpj@microsoft.com**20100525114001
3139 Ignore-this: 270f3da655e526cf04e27db7a01e29c0
3140]
3141[Refactor (again) the handling of default methods
3142simonpj@microsoft.com**20100525113910
3143 Ignore-this: 6686f6cdb878d57abf6b49fec64fcbb1
3144 
3145 This patch fixes Trac #4056, by
3146 
3147  a) tidying up the treatment of default method names
3148  b) removing the 'module' argument to newTopSrcBinder
3149 
3150 The details aren't that interesting, but the result
3151 is much tidier. The original bug was a 'nameModule' panic,
3152 caused by trying to find the module of a top-level name.
3153 But TH quotes generate Internal top-level names that don't
3154 have a module, and that is generally a good thing. 
3155 
3156 Fixing that in turn led to the default-method refactoring,
3157 which also makes the Name for a default method be handled
3158 in the same way as other derived names, generated in BuildTyCl
3159 via a call newImplicitBinder.  Hurrah.
3160]
3161[Don't do SpecConstr on NOINLINE things (Trac #4064)
3162simonpj@microsoft.com**20100525112807
3163 Ignore-this: 452be0a2cef0042fb67275c2827b5f72
3164 
3165 Since the RULE from specialising gets the same Activation as
3166 the inlining for the Id itself there's no point in specialising
3167 a NOINLINE thing, because the rule will be permanently switched
3168 off.
3169 
3170 See Note [Transfer activation] in SpecConstr
3171 and Note [Auto-specialisation and RULES] in Specialise.
3172]
3173[Change our #defines to work on FreeBSD too
3174Simon Marlow <marlowsd@gmail.com>**20100524105828
3175 Ignore-this: b23ede46211e67859206c0ec57d6a86f
3176 With glibc, things like _POSIX_C_SOURCE and _ISOC99_SOURCE are
3177 additive, but on FreeBSD they are mutually exclusive.  However, it
3178 turns out we only need to define _POSIX_C_SOURCE and _XOPEN_SOURCE to
3179 get all the C99 stuff we need too, so there's no need for any #ifdefs.
3180 
3181 Submitted by: Gabor PALI <pgj@FreeBSD.org>
3182]
3183[Add a missing UNTAG_CLOSURE, causing bus errors on Sparc
3184Simon Marlow <marlowsd@gmail.com>**20100524105547
3185 Ignore-this: a590b5391d6f05d50c8c088456c3c166
3186 We just about got away with this on x86 which isn't
3187 alignment-sensitive.  The result of the memory load is compared
3188 against a few different values, but there is a fallback case that
3189 happened to be the right thing when the pointer was tagged.  A good
3190 bug to find, nonetheless.
3191]
3192[Add wiki links
3193Simon Marlow <marlowsd@gmail.com>**20100520095953
3194 Ignore-this: c22f126cde166e6207922b2eb51d29e3
3195]
3196[the 'stage=0' trick to disable all compiler builds stopped working; fix it
3197Simon Marlow <marlowsd@gmail.com>**20100520104455
3198 Ignore-this: bb6fae9056471612c8dbf06916188c33
3199]
3200[Comments and formatting only
3201benl@ouroborus.net**20100524014021
3202 Ignore-this: 64579c38154728b632e358bec751cc0b
3203]
3204[Core prettyprinter fixes. Patch from Tim Chevalier. Fixes #4085
3205Ian Lynagh <igloo@earth.li>**20100522225048]
3206[Correct install-name for dynamic Darwin rts
3207pho@cielonegro.org**20100508151155
3208 Ignore-this: 6d31716c8c113dcb46e9cb925c4201df
3209]
3210[Fix the RTS debug_p build
3211Ian Lynagh <igloo@earth.li>**20100522163127]
3212[Unset $CFLAGS for "GNU non-executable stack" configure test; fixes #3889
3213Ian Lynagh <igloo@earth.li>**20100521165005
3214 With gcc 4.4 we get
3215     Error: can't resolve `.note.GNU-stack' {.note.GNU-stack section} - `.Ltext0' {.text section}
3216 when running gcc with the -g flag. To work around this we unset
3217 CFLAGS when running the test.
3218]
3219[Don't run "set -o igncr" before configuring libffi
3220Ian Lynagh <igloo@earth.li>**20100520162918
3221 Ignore-this: 489fa94df23f2adf4ff63c8ede2c0794
3222 It used to make the build work on cygwin, but now it breaks it instead:
3223     config.status: creating include/Makefile
3224     gawk: ./confLqjohp/subs.awk:1: BEGIN {\r
3225     gawk: ./confLqjohp/subs.awk:1: ^ backslash not last character on line
3226     config.status: error: could not create include/Makefile
3227     make[2]: *** [libffi/stamp.ffi.configure-shared] Error 1
3228     make[1]: *** [all] Error 2
3229]
3230[Stop passing -Wl,-macosx_version_min to gcc
3231Ian Lynagh <igloo@earth.li>**20100520154003
3232 Fixes a build failure on OS X 10.6. When linking
3233     rts/dist/build/libHSrts-ghc6.13.20100519.dylib
3234 we got
3235     ld: symbol dyld_stub_binding_helper not defined (usually in crt1.o/dylib1.o/bundle1.o)
3236     collect2: ld returned 1 exit status
3237]
3238[Fix build on FreeBSD; patch from Gabor PALI
3239Ian Lynagh <igloo@earth.li>**20100519140552]
3240[Fix package shadowing order (#4072)
3241Simon Marlow <marlowsd@gmail.com>**20100519104617
3242 Ignore-this: 26ea5e4bb5dff18618b807a54c7d6ebb
3243 
3244 Later packages are supposed to shadow earlier ones in the stack,
3245 unless the ordering is overriden with -package-id flags.
3246 Unfortunately an earlier fix for something else had sorted the list of
3247 packages so that it was in lexicographic order by installedPackageId,
3248 and sadly our test (cabal/shadow) didn't pick this up because the
3249 lexicographic ordering happened to work for the test.  I've now fixed
3250 the test so it tries both orderings.
3251]
3252[Set more env variables when configuring libffi
3253Ian Lynagh <igloo@earth.li>**20100518185014
3254 We now tell it where to find ld, nm and ar
3255]
3256[Set the location of ar to be the in-tree ar on Windows
3257Ian Lynagh <igloo@earth.li>**20100518181556]
3258[Change another / to </> to avoid building paths containing \/
3259Ian Lynagh <igloo@earth.li>**20100518172015
3260 This will hopefully fix #2889.
3261]
3262[Fix #4074 (I hope).
3263Simon Marlow <marlowsd@gmail.com>**20100518113214
3264 Ignore-this: 73cd70f5bc6f5add5247b61985c03fc1
3265 
3266 1. allow multiple threads to call startTimer()/stopTimer() pairs
3267 2. disable the timer around fork() in forkProcess()
3268 
3269 A corresponding change to the process package is required.
3270]
3271[we don't have a gcc-lib in LIB_DIR any more
3272Simon Marlow <marlowsd@gmail.com>**20100401102351
3273 Ignore-this: f41acd2d8f8e6763aa8bd57a0b44a7e4
3274]
3275[In validate, use gmake if available; based on a patch from Gabor PALI
3276Ian Lynagh <igloo@earth.li>**20100517200654]
3277[Remove duplicate "./configure --help" output; fixes #4075
3278Ian Lynagh <igloo@earth.li>**20100516141206]
3279[Update various 'sh boot's to 'perl boot'
3280Ian Lynagh <igloo@earth.li>**20100516122609
3281 Spotted by Marco Túlio Gontijo e Silva
3282]
3283[add missing initialisation for eventBufMutex
3284Simon Marlow <marlowsd@gmail.com>**20100514094943
3285 Ignore-this: 7f75594a8cb54fbec5aebd46bb959f45
3286]
3287[Undo part of #4003 patch
3288Simon Marlow <marlowsd@gmail.com>**20100513142017
3289 Ignore-this: cb65db86a38a7e5ccee9f779e489d104
3290 We still need the workaround for when compiling HEAD with 6.12.2
3291 
3292]
3293[Fix makefile loop (#4050)
3294pho@cielonegro.org**20100507140707
3295 Ignore-this: 3a1cb13d0600977e74d17ac26cbef83d
3296 
3297 The libtool creates "libffi.dylib" and "libffi.5.dylib" but not "libffi.5.0.9.dylib". Having it in libffi_DYNAMIC_LIBS causes an infinite makefile loop.
3298]
3299[fix !TABLES_NEXT_TO_CODE
3300Simon Marlow <marlowsd@gmail.com>**20100510151934
3301 Ignore-this: fccb859b114bef1c3122c98e60af51
3302]
3303[looksLikeModuleName: allow apostrophe in module names (#4051)
3304Simon Marlow <marlowsd@gmail.com>**20100510094741
3305 Ignore-this: df9348f3ba90608bec57257b47672985
3306]
3307[add the proper library dependencies for GhcProfiled=YES
3308Simon Marlow <marlowsd@gmail.com>**20100506122118
3309 Ignore-this: 6236993aa308ab5b5e1e5ea5f65982a
3310]
3311[Fix Trac #4003: fix the knot-tying in checkHiBootIface
3312simonpj@microsoft.com**20100511075026
3313 Ignore-this: a9ce2a318386fdc8782848df84592002
3314 
3315 I had incorrectly "optimised" checkHiBootIface so that it forgot
3316 to update the "knot-tied" type environment.
3317 
3318 This patch fixes the HEAD
3319]
3320[Re-engineer the derived Ord instance generation code (fix Trac #4019)
3321simonpj@microsoft.com**20100510133333
3322 Ignore-this: 8fe46e4dad27fbee211a7928acf372c2
3323   
3324 As well as fixing #4019, I rejigged the way that Ord instances are
3325 generated, which should make them faster in general.  See the
3326 Note [Generating Ord instances].
3327 
3328 I tried to measure the performance difference from this change, but
3329 the #4019 fix only removes one conditional branch per iteration, and
3330 I couldn't measure a consistent improvement.  But still, tihs is
3331 better than before.
3332]
3333[Make arity of INLINE things consistent
3334simonpj@microsoft.com**20100510133005
3335 Ignore-this: 15e7abf803d1dcb3f4ca760d2d939d0d
3336 
3337 We eta-expand things with INLINE pragmas;
3338 see Note [Eta-expanding INLINE things].
3339 
3340 But I eta-expanded it the wrong amount when the function
3341 was overloaded.  Ooops.
3342]
3343[Compacting GC fix, we forgot to thread the new bq field of StgTSO.
3344Simon Marlow <marlowsd@gmail.com>**20100510082325
3345 Ignore-this: a079c8446e2ad53efff6fd95d0f3ac80
3346]
3347[Add version constraints for the boot packages; fixes trac #3852
3348Ian Lynagh <igloo@earth.li>**20100509175051
3349 When using the bootstrapping compiler, we now explicitly constrain
3350 the version of boot packages (Cabal, extensible-exceptions, etc) to the
3351 in-tree version, so that the build system is less fragile should the
3352 user have a newer version installed for the bootstrapping compiler.
3353]
3354[Don't include inter-package dependencies when compiling with stage 0; #4031
3355Ian Lynagh <igloo@earth.li>**20100509130511
3356 This fixes a problem when building with GHC 6.12 on Windows, where
3357 dependencies on stage 0 (bootstrapping compiler) packages have absolute
3358 paths c:/ghc/..., and make gets confused by the colon.
3359]
3360[Add a ghc.mk for bindisttest/
3361Ian Lynagh <igloo@earth.li>**20100508223911]
3362[Move some make variables around so they are available when cleaning
3363Ian Lynagh <igloo@earth.li>**20100508212405]
3364[Optimise checkremove a bit
3365Ian Lynagh <igloo@earth.li>**20100508202006]
3366[Improve the bindisttest Makefile
3367Ian Lynagh <igloo@earth.li>**20100508195450]
3368[Add tools to test that cleaning works properly
3369Ian Lynagh <igloo@earth.li>**20100508194105]
3370[Tweak the ghc-pkg finding code
3371Ian Lynagh <igloo@earth.li>**20100508125815
3372 It now understand the ghc-stage[123] names we use in-tree, and it won't
3373 go looking for any old ghc-pkg if it can't find the one that matches
3374 ghc.
3375]
3376[Add a way to show what cleaning would be done, without actually doing it
3377Ian Lynagh <igloo@earth.li>**20100508122438]
3378[Tidy up the "rm" flags in the build system
3379Ian Lynagh <igloo@earth.li>**20100508115745]
3380[Fix crash in nested callbacks (#4038)
3381Simon Marlow <marlowsd@gmail.com>**20100507093222
3382 Ignore-this: cade85e361534ce711865a4820276388
3383 Broken by "Split part of the Task struct into a separate struct
3384 InCall".
3385]
3386[Add $(GhcDynamic) knob, set to YES to get stage2 linked with -dynamic
3387Simon Marlow <marlowsd@gmail.com>**20100428205241
3388 Ignore-this: 1db8bccf92099785ecac39aebd27c92d
3389 Default currently NO.
3390 
3391 Validate passed with GhcDynamic=YES on x86/Linux here.
3392 
3393 The compiler is currently slower on x86 when linked -dynamic,
3394 because the GC inner loop has been adversely affected by -fPIC, I'm
3395 looking into how to fix it.
3396]
3397[omit "dyn" from the way appended to the __stginit label
3398Simon Marlow <marlowsd@gmail.com>**20100428204914
3399 Ignore-this: 14183f3defa9f2bde68fda6729b740bc
3400 When GHCi is linked dynamically, we still want to be able to load
3401 non-dynamic object files.
3402]
3403[improvements to findPtr(), a neat hack for browsing the heap in gdb
3404Simon Marlow <marlowsd@gmail.com>**20100506115427
3405 Ignore-this: ac57785bb3e13b97a5945f753f068738
3406]
3407[Fix +RTS -G1
3408Simon Marlow <marlowsd@gmail.com>**20100506110739
3409 Ignore-this: 86a5de39a94d3331a4ee1213f82be497
3410]
3411[Enable the "redundant specialise pragmas" warning; fixes trac #3855
3412Ian Lynagh <igloo@earth.li>**20100506175351]
3413[Find the correct external ids when there's a wrapper
3414simonpj@microsoft.com**20100506164135
3415 Ignore-this: 636266407b174b05b2b8646cc73062c0
3416 
3417 We were failing to externalise the wrapper id for a function
3418 that had one.
3419]
3420[Add a comment about pattern coercions
3421simonpj@microsoft.com**20100506164027
3422 Ignore-this: 17428089f3df439f65d892e23e8ed61a
3423]
3424[Comments only
3425simonpj@microsoft.com**20100506163829
3426 Ignore-this: 169167b6463873ab173cc5750c5be469
3427]
3428[Make a missing name in mkUsageInfo into a panic
3429simonpj@microsoft.com**20100506163813
3430 Ignore-this: b82ff1b8bf89f74f146db7cb5cc4c4d7
3431 
3432 We really want to know about this!
3433]
3434[Refactoring of hsXxxBinders
3435simonpj@microsoft.com**20100506163737
3436 Ignore-this: 97c6667625262b160f9746f7bea1c980
3437 
3438 This patch moves various functions that extract the binders
3439 from a HsTyClDecl, HsForeignDecl etc into HsUtils, and gives
3440 them consistent names.
3441]
3442[Fix Trac #3966: warn about useless UNPACK pragmas
3443simonpj@microsoft.com**20100506163337
3444 Ignore-this: 5beb24b686eda6113b614dfac8490df1
3445 
3446 Warning about useless UNPACK pragmas wasn't as easy as I thought.
3447 I did quite a bit of refactoring, which improved the code by refining
3448 the types somewhat.  In particular notice that in DataCon, we have
3449 
3450     dcStrictMarks   :: [HsBang]
3451     dcRepStrictness :: [StrictnessMarks]
3452 
3453 The former relates to the *source-code* annotation, the latter to
3454 GHC's representation choice.
3455]
3456[Make tcg_dus behave more sanely; fixes a mkUsageInfo panic
3457simonpj@microsoft.com**20100506162719
3458 Ignore-this: d000bca15b0e127e297378ded1bfb81b
3459 
3460 The tcg_dus field used to contain *uses* of type and class decls,
3461 but not *defs*.  That was inconsistent, and it really went wrong
3462 for Template Haskell bracket.  What happened was that
3463  foo = [d| data A = A
3464                   f :: A -> A
3465                   f x = x |]
3466 would find a "use" of A when processing the top level of the module,
3467 which in turn led to a mkUsageInfo panic in MkIface.  The cause was
3468 the fact that the tcg_dus for the nested quote didn't have defs for
3469 A.
3470]
3471[Add a HsExplicitFlag to SpliceDecl, to improve Trac #4042
3472simonpj@microsoft.com**20100506161523
3473 Ignore-this: e4e563bac2fd831cc9e94612f5b4fa9d
3474 
3475 The issue here is that
3476 
3477     g :: A -> A
3478     f
3479     data A = A
3480 
3481 is treated as if you'd written $(f); that is the call of
3482 f is a top-level Template Haskell splice.  This patch
3483 makes sure that we *first* check the -XTemplateHaskellFlag
3484 and bleat about a parse error if it's off.  Othewise we
3485 get strange seeing "A is out of scope" errors.
3486]
3487[Change an assert to a warn
3488simonpj@microsoft.com**20100506161111
3489 Ignore-this: 739a4fb4c7940376b0f2c8ad52a1966c
3490 
3491 This is in the constraint simplifier which I'm about
3492 to rewrite, so I'm hoping the assert isn't fatal!
3493]
3494[Tidy up debug print a little
3495simonpj@microsoft.com**20100506161027
3496 Ignore-this: bd5492878e06bee1cddcbb3fc4df66d8
3497]
3498[Remove useless UNPACK pragmas
3499simonpj@microsoft.com**20100506161012
3500 Ignore-this: 3e5ab1a7cf58107034412a798bc214e5
3501]
3502[Add WARNM2 macro, plus some refactoring
3503simonpj@microsoft.com**20100506160808
3504 Ignore-this: 2ab4f1f0b5d94be683036e77aec09255
3505]
3506[Use -Wwarn for the binary package, becuase it has redundant UNPACK pragmas
3507simonpj@microsoft.com**20100506160750
3508 Ignore-this: cf0d3a11473e28bfce9602e716e69a5f
3509]
3510[Fix Trac #3966: warn about unused UNPACK pragmas
3511simonpj@microsoft.com**20100409201812
3512 Ignore-this: c96412596b39c918b5fb9b3c39ce2119
3513]
3514[Fix Trac #3953: fail earlier when using a bogus quasiquoter
3515simonpj@microsoft.com**20100409201748
3516 Ignore-this: ef48e39aa932caed538643985234f043
3517]
3518[Fix Trac #3965: tighten conditions when deriving Data
3519simonpj@microsoft.com**20100409184420
3520 Ignore-this: 96f7d7d2da11565d26b465d7d0497ac9
3521 
3522 It's tricky to set up the context for a Data instance.  I got it wrong
3523 once, and fixed it -- hence the "extra_constraints" in
3524 TcDeriv.inferConstraints. 
3525 
3526 But it still wasn't right!  The tricky bit is that dataCast1 is only
3527 generated when T :: *->*, and dataCast2 when T :: *->*->*. (See
3528 the code in TcGenDeriv for dataCastX.
3529]
3530[Fix Trac #3964: view patterns in DsArrows
3531simonpj@microsoft.com**20100409165557
3532 Ignore-this: d823c182831d5e2e592e995b16180e2f
3533 
3534 Just a missing case; I've eliminated the catch-all so
3535 that we get a warning next time we extend HsPat
3536]
3537[Fix Trac #3955: renamer and type variables
3538simonpj@microsoft.com**20100409163710
3539 Ignore-this: bd5ec64d76c0f583bf5f224792bf294c
3540 
3541 The renamer wasn't computing the free variables of a type declaration
3542 properly.  This patch refactors a bit, and makes it more robust,
3543 fixing #3955 and several other closely-related bugs.  (We were
3544 omitting some free variables and that could just possibly lead to a
3545 usage-version tracking error.
3546]
3547[Layout only
3548simonpj@microsoft.com**20100409163506
3549 Ignore-this: 1f14990b5aa0b9821b84452fb34e9f41
3550]
3551[Give a better deprecated message for INCLUDE pragmas; fixes #3933
3552Ian Lynagh <igloo@earth.li>**20100506130910
3553 We now have a DeprecatedFullText constructor, so we can override the
3554 "-#include is deprecated: " part of the warning.
3555]
3556[De-haddock a comment that confuses haddock
3557Ian Lynagh <igloo@earth.li>**20100506123607]
3558[Fix comment to not confuse haddock
3559Ian Lynagh <igloo@earth.li>**20100506113642]
3560[Detect EOF when trying to parse a string in hp2ps
3561Ian Lynagh <igloo@earth.li>**20100506000830]
3562[Make the demand analyser sdd demands for strict constructors
3563simonpj@microsoft.com**20100505200936
3564 Ignore-this: eb32632adbc354eb7a5cf884c263e0d3
3565 
3566 This opportunity was spotted by Roman, and is documented in
3567 Note [Add demands for strict constructors] in DmdAnal.
3568]
3569[Fix interaction of exprIsCheap and the lone-variable inlining check
3570simonpj@microsoft.com**20100505200723
3571 Ignore-this: f3cb65085c5673a99153d5d7b6559ab1
3572 
3573 See Note [Interaction of exprIsCheap and lone variables] in CoreUnfold
3574 
3575 This buglet meant that a nullary definition with an INLINE pragma
3576 counter-intuitively didn't get inlined at all.  Roman identified
3577 the bug.
3578]
3579[Matching cases in SpecConstr and Rules
3580simonpj@microsoft.com**20100505200543
3581 Ignore-this: f5c28c780fbf8badce84c6fdc9aa1779
3582 
3583 This patch has zero effect.  It includes comments,
3584 a bit of refactoring, and a tiny bit of commment-out
3585 code go implement the "matching cases" idea below.
3586 
3587 In the end I've left it disabled because while I think
3588 it does no harm I don't think it'll do any good either.
3589 But I didn't want to lose the idea totally. There's
3590 a thread called "Storable and constant memory" on
3591 the libraries@haskell.org list (Apr 2010) about it.
3592 
3593 Note [Matching cases]
3594 ~~~~~~~~~~~~~~~~~~~~~
3595 {- NOTE: This idea is currently disabled.  It really only works if
3596          the primops involved are OkForSpeculation, and, since
3597         they have side effects readIntOfAddr and touch are not.
3598         Maybe we'll get back to this later .  -}
3599   
3600 Consider
3601    f (case readIntOffAddr# p# i# realWorld# of { (# s#, n# #) ->
3602       case touch# fp s# of { _ ->
3603       I# n# } } )
3604 This happened in a tight loop generated by stream fusion that
3605 Roman encountered.  We'd like to treat this just like the let
3606 case, because the primops concerned are ok-for-speculation.
3607 That is, we'd like to behave as if it had been
3608    case readIntOffAddr# p# i# realWorld# of { (# s#, n# #) ->
3609    case touch# fp s# of { _ ->
3610    f (I# n# } } )
3611]
3612[Comments only
3613simonpj@microsoft.com**20100504163629
3614 Ignore-this: 3be12df04714aa820bce706b5dc8a9cb
3615]
3616[Comments only
3617simonpj@microsoft.com**20100504163529
3618 Ignore-this: 791e2fd39c7d880ce1dc80ebdf3a5398
3619]
3620[Comments only
3621simonpj@microsoft.com**20100504163457
3622 Ignore-this: f19e9ffeb3d65770b1595bca5f97a59d
3623]
3624[Comments only (about type families)
3625simonpj@microsoft.com**20100417145032
3626 Ignore-this: dd39425ef2155d52dbf55a4d5fd97cb8
3627]
3628[Fix hp2ps when the .hp file has large string literals
3629Ian Lynagh <igloo@earth.li>**20100505191921]
3630[In build system, call package-config after including package data
3631Ian Lynagh <igloo@earth.li>**20100504225035
3632 Otherwise the $1_$2_HC_OPTS variable gets clobbered.
3633]
3634[runghc: flush stdout/stderr on an exception (#3890)
3635Simon Marlow <marlowsd@gmail.com>**20100505133848
3636 Ignore-this: 224c1898cec64cb1c94e0d7033e7590e
3637]
3638[Remove the Unicode alternative for ".." (#3894)
3639Simon Marlow <marlowsd@gmail.com>**20100505121202
3640 Ignore-this: 2452cd67281667106f9169747b6d784f
3641]
3642[tidyup; no functional changes
3643Simon Marlow <marlowsd@gmail.com>**20100505115015
3644 Ignore-this: d0787e5cdeef1dee628682fa0a46019
3645]
3646[Make the running_finalizers flag task-local
3647Simon Marlow <marlowsd@gmail.com>**20100505114947
3648 Ignore-this: 345925d00f1dca203941b3c5d84c90e1
3649 Fixes a bug reported by Lennart Augustsson, whereby we could get an
3650 incorrect error from the RTS about re-entry from a finalizer,
3651]
3652[add a MAYBE_GC() in killThread#, fixes throwto003(threaded2) looping
3653Simon Marlow <marlowsd@gmail.com>**20100505114746
3654 Ignore-this: efea04991d6feed04683a42232fc85da
3655]
3656[Allow filepath-1.2.*
3657Simon Marlow <marlowsd@gmail.com>**20100505101139
3658 Ignore-this: 1b5580cd9cd041ec48f40cd37603326a
3659]
3660[BlockedOnMsgThrowTo is possible in resurrectThreads (#4030)
3661Simon Marlow <marlowsd@gmail.com>**20100505094534
3662 Ignore-this: ac24a22f95ffeaf480187a1620fdddb2
3663]
3664[Don't raise a throwTo when the target is masking and BlockedOnBlackHole
3665Simon Marlow <marlowsd@gmail.com>**20100505094506
3666 Ignore-this: 302616931f61667030d77ddfbb02374e
3667]
3668[Fix build with GHC 6.10
3669Ian Lynagh <igloo@earth.li>**20100504180302
3670 In GHC 6.10, intersectionWith is (a -> b -> a) instead of (a -> b -> c),
3671 so we need to jump through some hoops to get the more general type.
3672]
3673[The libffi patches are no longer needed
3674Ian Lynagh <igloo@earth.li>**20100504171603]
3675[Use the in-tree windres; fixes trac #4032
3676Ian Lynagh <igloo@earth.li>**20100504170941]
3677[Print unfoldings on lambda-bound variables
3678Simon PJ <simonpj@microsoft.com>**20100503181822
3679 Ignore-this: 2fd5a7502cc6273d96258e0914f0f8cd
3680 
3681 ...in the unusual case where they have one;
3682 see Note [Case binders and join points] in Simplify.lhs
3683]
3684[Replace FiniteMap and UniqFM with counterparts from containers.
3685Milan Straka <fox@ucw.cz>**20100503171315
3686 Ignore-this: a021972239163dbf728284b19928cebb
3687 
3688 The original interfaces are kept. There is small performance improvement:
3689 - when compiling for five nofib, we get following speedups:
3690     Average                -----           -2.5%
3691     Average                -----           -0.6%
3692     Average                -----           -0.5%
3693     Average                -----           -5.5%
3694     Average                -----          -10.3%
3695 - when compiling HPC ten times, we get:
3696     switches                          oldmaps   newmaps
3697     -O -fasm                          117.402s  116.081s (98.87%)
3698     -O -fasm -fregs-graph             119.993s  118.735s (98.95%)
3699     -O -fasm -fregs-iterative         120.191s  118.607s (98.68%)
3700]
3701[Make the demand analyser take account of lambda-bound unfoldings
3702Simon PJ <simonpj@microsoft.com>**20100503151630
3703 Ignore-this: 2ee8e27d4df2debfc79e6b8a17c32bc1
3704 
3705 This is a long-standing lurking bug. See Note [Lamba-bound unfoldings]
3706 in DmdAnal.
3707 
3708 I'm still not really happy with this lambda-bound-unfolding stuff.
3709]
3710[Fix dynamic libs on OS X, and enable them by default
3711Ian Lynagh <igloo@earth.li>**20100503150302]
3712[Switch back to using bytestring from the darcs repo; partially fixes #3855
3713Ian Lynagh <igloo@earth.li>**20100502113458]
3714[Fix some cpp warnings when building on FreeBSD; patch from Gabor PALI
3715Ian Lynagh <igloo@earth.li>**20100428150700]
3716[Fix "make 2"
3717Ian Lynagh <igloo@earth.li>**20100427162212
3718 The new Makefile logic was enabling the stage 1 rules when stage=2,
3719 so "make 2" was rebuilding stage 1.
3720]
3721[Inplace programs depend on their shell wrappers
3722Ian Lynagh <igloo@earth.li>**20100427160038]
3723[--make is now the default (#3515), and -fno-code works with --make (#3783)
3724Simon Marlow <marlowsd@gmail.com>**20100427122851
3725 Ignore-this: 33330474fa4703f32bf9997462b4bf3c
3726 If the command line contains any Haskell source files, then we behave
3727 as if --make had been given.
3728 
3729 The meaning of the -c flag has changed (back): -c now selects one-shot
3730 compilation, but stops before linking.  However, to retain backwards
3731 compatibility, -c is still allowed with --make, and means the same as
3732 --make -no-link.  The -no-link flag has been un-deprecated.
3733 
3734 -fno-code is now allowed with --make (#3783); the fact that it was
3735 disabled before was largely accidental, it seems.  We also had some
3736 regressions in this area: it seems that -fno-code was causing a .hc
3737 file to be emitted in certain cases.  I've tidied up the code, there
3738 was no need for -fno-code to be a "mode" flag, as far as I can tell.
3739 
3740 -fno-code does not emit interface files, nor does it do recompilation
3741 checking, as suggested in #3783.  This would make Haddock emit
3742 interface files, for example, and I'm fairly sure we don't want to do
3743 that.  Compiling with -fno-code is pretty quick anyway, perhaps we can
3744 get away without recompilation checking.
3745]
3746[remove duplicate docs for -e in --help output (#4010)
3747Simon Marlow <marlowsd@gmail.com>**20100426140642
3748 Ignore-this: 187ff893ba8ffa0ec127867a7590e38d
3749]
3750[workaround for #4003, fixes HEAD build with 6.12.2
3751Simon Marlow <marlowsd@gmail.com>**20100426103428
3752 Ignore-this: c4bc445dc8052d4e6efef3f1daf63562
3753]
3754[Make sure all the clean rules are always included
3755Ian Lynagh <igloo@earth.li>**20100424181823
3756 In particular, this fixes a problem where stage3 bits weren't being cleaned
3757]
3758[Correct the name of the amd64/FreeBSD platform in PlatformSupportsSharedLibs
3759Ian Lynagh <igloo@earth.li>**20100424132830
3760 We weren't getting sharedlibs on amd64/FreeBSD because of this
3761]
3762[Include DPH docs in bindists
3763Ian Lynagh <igloo@earth.li>**20100424123101]
3764[reinstate eta-expansion during SimplGently, to fix inlining of sequence_
3765Simon Marlow <marlowsd@gmail.com>**20100423124853
3766 Ignore-this: 4fa0fd5bafe0d6b58fc81076f50d5f8d
3767]
3768[fix 64-bit value for W_SHIFT, which thankfully appears to be not used
3769Simon Marlow <marlowsd@gmail.com>**20100422213605
3770 Ignore-this: 525c062d2456c224ec8d0e083edd3b55
3771]
3772[Add missing constant folding and optimisation for unsigned division
3773Simon Marlow <marlowsd@gmail.com>**20100422213443
3774 Ignore-this: fb10d1cda0852fab0cbcb47247498fb3
3775 Noticed by Denys Rtveliashvili <rtvd@mac.com>, see #4004
3776]
3777[Fix the GHC API link in the main doc index.html
3778Ian Lynagh <igloo@earth.li>**20100422213226]
3779[Give the right exit code in darcs-all
3780Ian Lynagh <igloo@earth.li>**20100421171339
3781 Our END block was calling system, which alters $?. So now we save and
3782 restore it.
3783]
3784[Use StgWord64 instead of ullong
3785Ian Lynagh <igloo@earth.li>**20100421162336
3786 This patch also fixes ullong_format_string (renamed to showStgWord64)
3787 so that it works with values outside the 32bit range (trac #3979), and
3788 simplifies the without-commas case.
3789]
3790[Implement try10Times in Makefile
3791Ian Lynagh <igloo@earth.li>**20100420165909
3792 Avoid using seq, as FreeBSD has jot instead.
3793]
3794[Fix crash in non-threaded RTS on Windows
3795Simon Marlow <marlowsd@gmail.com>**20100420122125
3796 Ignore-this: 28b0255a914a8955dce02d89a7dfaca
3797 The tso->block_info field is now overwritten by pushOnRunQueue(), but
3798 stg_block_async_info was assuming that it still held a pointer to the
3799 StgAsyncIOResult.  We must therefore save this value somewhere safe
3800 before putting the TSO on the run queue.
3801]
3802[Expand the scope of the event_buf_mutex to cover io_manager_event
3803Simon Marlow <marlowsd@gmail.com>**20100420122026
3804 Ignore-this: 185a6d84f7d4a35997f10803f6dacef1
3805 I once saw a failure that I think was due to a race on
3806 io_manager_event, this should fix it.
3807]
3808[Flags -auto and -auto-all operate only on functions not marked INLINE.
3809Milan Straka <fox@ucw.cz>**20100331191050
3810 Ignore-this: 3b63580cfcb3c33d62ad697c36d94d05
3811]
3812[Spelling correction for LANGUAGE pragmas
3813Max Bolingbroke <batterseapower@hotmail.com>**20100413192825
3814 Ignore-this: 311b51ba8d43f6c7fd32f48db9a88dee
3815]
3816[Update the user guide so it talks about the newer "do rec" notation everywhere
3817Ian Lynagh <igloo@earth.li>**20100416205416
3818 Some of the problems highlighted in trac #3968.
3819]
3820[Fix typo
3821Ian Lynagh <igloo@earth.li>**20100416205412]
3822[Fix Trac #3950: unifying types of different kinds
3823simonpj@microsoft.com**20100412151845
3824 Ignore-this: d145b9de5ced136ef2c39f3ea4a04f4a
3825 
3826 I was assuming that the unifer only unified types of the
3827 same kind, but now we can "defer" unsolved constraints that
3828 invariant no longer holds.  Or at least is's more complicated
3829 to ensure. 
3830 
3831 This patch takes the path of not assuming the invariant, which
3832 is simpler and more robust.  See
3833 Note [Mismatched type lists and application decomposition]
3834]
3835[Fix Trac #3943: incorrect unused-variable warning
3836simonpj@microsoft.com**20100412151630
3837 Ignore-this: 52459f2b8b02c3cb120abe674dc9a060
3838 
3839 In fixing this I did the usual little bit of refactoring
3840]
3841[Convert boot and boot-pkgs to perl
3842Ian Lynagh <igloo@earth.li>**20100415143919
3843 This stops us having to worry about sh/sed/... portability.
3844]
3845[Use $(MAKE), not make, when recursively calling make
3846Ian Lynagh <igloo@earth.li>**20100415121453]
3847[Remove the ghc_ge_609 makefile variables
3848Ian Lynagh <igloo@earth.li>**20100412235658
3849 They are now guaranteed to be YES
3850]
3851[Increase the minimum version number required to 6.10 in configure.ac
3852Ian Lynagh <igloo@earth.li>**20100412235313]
3853[The bootstrapping compiler is now required to be > 609
3854Ian Lynagh <igloo@earth.li>**20100409161046]
3855[Handle IND_STATIC in isRetainer
3856Ian Lynagh <igloo@earth.li>**20100409104207
3857 IND_STATIC used to be an error, but at the moment it can happen
3858 as isAlive doesn't look through IND_STATIC as it ignores static
3859 closures. See trac #3956 for a program that hit this error.
3860]
3861[Add Data and Typeable instances to HsSyn
3862David Waern <david.waern@gmail.com>**20100330011020
3863 Ignore-this: c3f2717207b15539fea267c36b686e6a
3864 
3865 The instances (and deriving declarations) have been taken from the ghc-syb
3866 package.
3867]
3868[Fix for derefing ThreadRelocated TSOs in MVar operations
3869Simon Marlow <marlowsd@gmail.com>**20100407092824
3870 Ignore-this: 94dd7c68a6094eda667e2375921a8b78
3871]
3872[sanity check fix
3873Simon Marlow <marlowsd@gmail.com>**20100407092746
3874 Ignore-this: 9c18cd5f5393e5049015ca52e62a1269
3875]
3876[get the reg liveness right in the putMVar# heap check
3877Simon Marlow <marlowsd@gmail.com>**20100407092724
3878 Ignore-this: b1ba07a59ecfae00e9a1f8391741abc
3879]
3880[initialise the headers of MSG_BLACKHOLE objects properly
3881Simon Marlow <marlowsd@gmail.com>**20100407081712
3882 Ignore-this: 183dcd0ca6a395d08db2be12b02bdd79
3883]
3884[initialise the headers of MVAR_TSO_QUEUE objects properly
3885Simon Marlow <marlowsd@gmail.com>**20100407081514
3886 Ignore-this: 4b4a2f30cf2fb69ca4128c41744687bb
3887]
3888[undo debugging code
3889Simon Marlow <marlowsd@gmail.com>**20100406142740
3890 Ignore-this: 323c2248f817b6717c19180482fc4b00
3891]
3892[putMVar#: fix reg liveness in the heap check
3893Simon Marlow <marlowsd@gmail.com>**20100406135832
3894 Ignore-this: cddd2c7807ac7612c9b2c4c0d384d284
3895]
3896[account for the new BLACKHOLEs in the GHCi debugger
3897Simon Marlow <marlowsd@gmail.com>**20100406133406
3898 Ignore-this: 4d4aeb4bbada3f50dc1fb0123f565e8f
3899]
3900[don't forget to deRefTSO() in tryWakeupThread()
3901Simon Marlow <marlowsd@gmail.com>**20100406130411
3902 Ignore-this: 171d57c4f8653835dec0b69f9be9881c
3903]
3904[Fix bug in popRunQueue
3905Simon Marlow <marlowsd@gmail.com>**20100406091453
3906 Ignore-this: 9d3cec8f18f5c5cbd51751797386eb6f
3907]
3908[fix bug in migrateThread()
3909Simon Marlow <marlowsd@gmail.com>**20100401105840
3910 Ignore-this: 299bcf0d1ea0f8865f3e845eb93d2ad3
3911]
3912[Remove the IND_OLDGEN and IND_OLDGEN_PERM closure types
3913Simon Marlow <marlowsd@gmail.com>**20100401093519
3914 Ignore-this: 95f2480c8a45139835eaf5610217780b
3915 These are no longer used: once upon a time they used to have different
3916 layout from IND and IND_PERM respectively, but that is no longer the
3917 case since we changed the remembered set to be an array of addresses
3918 instead of a linked list of closures.
3919]
3920[Change the representation of the MVar blocked queue
3921Simon Marlow <marlowsd@gmail.com>**20100401091605
3922 Ignore-this: 20a35bfabacef2674df362905d7834fa
3923 
3924 The list of threads blocked on an MVar is now represented as a list of
3925 separately allocated objects rather than being linked through the TSOs
3926 themselves.  This lets us remove a TSO from the list in O(1) time
3927 rather than O(n) time, by marking the list object.  Removing this
3928 linear component fixes some pathalogical performance cases where many
3929 threads were blocked on an MVar and became unreachable simultaneously
3930 (nofib/smp/threads007), or when sending an asynchronous exception to a
3931 TSO in a long list of thread blocked on an MVar.
3932 
3933 MVar performance has actually improved by a few percent as a result of
3934 this change, slightly to my surprise.
3935 
3936 This is the final cleanup in the sequence, which let me remove the old
3937 way of waking up threads (unblockOne(), MSG_WAKEUP) in favour of the
3938 new way (tryWakeupThread and MSG_TRY_WAKEUP, which is idempotent).  It
3939 is now the case that only the Capability that owns a TSO may modify
3940 its state (well, almost), and this simplifies various things.  More of
3941 the RTS is based on message-passing between Capabilities now.
3942]
3943[eliminate some duplication with a bit of CPP
3944Simon Marlow <marlowsd@gmail.com>**20100330154355
3945 Ignore-this: 838f7d341f096ca14c86ab9c81193e36
3946]
3947[Make ioManagerDie() idempotent
3948Simon Marlow <marlowsd@gmail.com>**20100401100705
3949 Ignore-this: a5996b43cdb2e2d72e6e971d7ea925fb
3950 Avoids screeds of "event buffer overflowed; event dropped" in
3951 conc059(threaded1).
3952]
3953[Move a thread to the front of the run queue when another thread blocks on it
3954Simon Marlow <marlowsd@gmail.com>**20100329144521
3955 Ignore-this: c518ff0d41154680edc811d891826a29
3956 This fixes #3838, and was made possible by the new BLACKHOLE
3957 infrastructure.  To allow reording of the run queue I had to make it
3958 doubly-linked, which entails some extra trickiness with regard to
3959 GC write barriers and suchlike.
3960]
3961[remove non-existent MUT_CONS symbols
3962Simon Marlow <marlowsd@gmail.com>**20100330152600
3963 Ignore-this: 885628257a9d03f2ece2a754d993014a
3964]
3965[change throwTo to use tryWakeupThread rather than unblockOne
3966Simon Marlow <marlowsd@gmail.com>**20100329144613
3967 Ignore-this: 10ad4965e6c940db71253f1c72218bbb
3968]
3969[tiny GC optimisation
3970Simon Marlow <marlowsd@gmail.com>**20100329144551
3971 Ignore-this: 9e095b9b73fff0aae726f9937846ba92
3972]
3973[New implementation of BLACKHOLEs
3974Simon Marlow <marlowsd@gmail.com>**20100329144456
3975 Ignore-this: 96cd26793b4e6ab9ddd0d59aae5c2f1d
3976 
3977 This replaces the global blackhole_queue with a clever scheme that
3978 enables us to queue up blocked threads on the closure that they are
3979 blocked on, while still avoiding atomic instructions in the common
3980 case.
3981 
3982 Advantages:
3983 
3984  - gets rid of a locked global data structure and some tricky GC code
3985    (replacing it with some per-thread data structures and different
3986    tricky GC code :)
3987 
3988  - wakeups are more prompt: parallel/concurrent performance should
3989    benefit.  I haven't seen anything dramatic in the parallel
3990    benchmarks so far, but a couple of threading benchmarks do improve
3991    a bit.
3992 
3993  - waking up a thread blocked on a blackhole is now O(1) (e.g. if
3994    it is the target of throwTo).
3995 
3996  - less sharing and better separation of Capabilities: communication
3997    is done with messages, the data structures are strictly owned by a
3998    Capability and cannot be modified except by sending messages.
3999 
4000  - this change will utlimately enable us to do more intelligent
4001    scheduling when threads block on each other.  This is what started
4002    off the whole thing, but it isn't done yet (#3838).
4003 
4004 I'll be documenting all this on the wiki in due course.
4005 
4006]
4007[Fix warnings (allow pushOnRunQueue() to not be inlined)
4008Simon Marlow <marlowsd@gmail.com>**20100401114559
4009 Ignore-this: f40bfbfad70a5165a946d11371605b7d
4010]
4011[remove out of date comment
4012Simon Marlow <marlowsd@gmail.com>**20100401105853
4013 Ignore-this: 26af88dd418ee0bcda7223b3b7e4e8d2
4014]
4015[tidy up spacing in stderr traces
4016Simon Marlow <marlowsd@gmail.com>**20100326163122
4017 Ignore-this: 16558b0433a274be217d4bf39aa4946
4018]
4019[Fix an assertion that was not safe when running in parallel
4020Simon Marlow <marlowsd@gmail.com>**20100325143656
4021 Ignore-this: cad08fb8900eb3a475547af0189fcc47
4022]
4023[Never jump directly to a thunk's entry code, even if it is single-entry
4024Simon Marlow <marlowsd@gmail.com>**20100325114847
4025 Ignore-this: 938da172c06a97762ef605c8fccfedf1
4026 I don't think this fixes any bugs as we don't have single-entry thunks
4027 at the moment, but it could cause problems for parallel execution if
4028 we ever did re-introduce update avoidance.
4029]
4030[Rename forgotten -dverbose-simpl to -dverbose-core2core in the docs.
4031Milan Straka <fox@ucw.cz>**20100331153626
4032 Ignore-this: 2da58477fb96e1cfb80f37dddd7c422c
4033]
4034[Add -pa and -V to the documentation of time profiling options.
4035Milan Straka <fox@ucw.cz>**20100329191121
4036 Ignore-this: be74d216481ec5a19e5f40f85e6e3d65
4037]
4038[Keep gcc 4.5 happy
4039Simon Marlow <marlowsd@gmail.com>**20100330120425
4040 Ignore-this: 7811878cc2bd1ce9cfbb5bf102fe3454
4041]
4042[Fix warning compiling Linker.c for PPC Mac
4043naur@post11.tele.dk**20100403182355
4044 Ignore-this: e2d2448770c9714ce17dd6cf3e297063
4045 The warning message eliminated is:
4046 > rts/Linker.c:4756:0:
4047 >      warning: nested extern declaration of 'symbolsWithoutUnderscore'
4048]
4049[Fix error compiling AsmCodeGen.lhs for PPC Mac (mkRtsCodeLabel)
4050naur@post11.tele.dk**20100403181656
4051 Ignore-this: deb7524ea7852a15a2ac0849c8c82f74
4052 The error messages eliminated are:
4053 > compiler/nativeGen/AsmCodeGen.lhs:875:31:
4054 >     Not in scope: `mkRtsCodeLabel'
4055 > compiler/nativeGen/AsmCodeGen.lhs:879:31:
4056 >     Not in scope: `mkRtsCodeLabel'
4057 > compiler/nativeGen/AsmCodeGen.lhs:883:31:
4058 >     Not in scope: `mkRtsCodeLabel'
4059]
4060[Fix error compiling AsmCodeGen.lhs for PPC Mac (DestBlockId)
4061naur@post11.tele.dk**20100403180643
4062 Ignore-this: 71e833e94ed8371b2ffabc2cf80bf585
4063 The error message eliminated is:
4064 > compiler/nativeGen/AsmCodeGen.lhs:637:16:
4065 >     Not in scope: data constructor `DestBlockId'
4066]
4067[Fix boot-pkgs's sed usage to work with Solaris's sed
4068Ian Lynagh <igloo@earth.li>**20100401153441]
4069[Pass "-i org.haskell.GHC" to packagemaker when building the OS X installer
4070Ian Lynagh <igloo@earth.li>**20100331144707
4071 This seems to fix this failure:
4072 [...]
4073 ** BUILD SUCCEEDED **
4074 rm -f -f GHC-system.pmdoc/*-contents.xml
4075 /Developer/usr/bin/packagemaker -v --doc GHC-system.pmdoc\
4076              -o /Users/ian/to_release/ghc-6.12.1.20100330/GHC-6.12.1.20100330-i386.pkg
4077 2010-03-31 15:08:15.695 packagemaker[13909:807] Setting to : 0 (null)
4078 2010-03-31 15:08:15.709 packagemaker[13909:807] Setting to : 0 org.haskell.glasgowHaskellCompiler.ghc.pkg
4079 2010-03-31 15:08:15.739 packagemaker[13909:807] relocate: (null) 0
4080 2010-03-31 15:08:15.740 packagemaker[13909:807] *** Terminating app due to uncaught exception 'NSInvalidArgumentException', reason: '*** -[NSXMLDocument initWithXMLString:options:error:]: nil argument'
4081 2010-03-31 15:08:15.741 packagemaker[13909:807] Stack: (
4082     2511962091,
4083     2447007291,
4084     2511961547,
4085     2511961610,
4086     2432803204,
4087     453371,
4088     447720,
4089     436209,
4090     435510,
4091     9986,
4092     9918
4093 )
4094 make[1]: *** [framework-pkg] Trace/BPT trap
4095 make: *** [framework-pkg] Error 2
4096]
4097[Use machdepCCOpts when compiling the file to toggle -(no-)rtsopts
4098Ian Lynagh <igloo@earth.li>**20100331161302
4099 Should fix toggling on OS X "Snow Leopard". Diagnosed by Roman Leshchinskiy.
4100]
4101[Avoid a non-portable use of tar reported by Roman Leshchinskiy
4102Ian Lynagh <igloo@earth.li>**20100330145802]
4103[Don't install EXTRA_PACKAGES by default
4104Simon Marlow <marlowsd@gmail.com>**20100330142714
4105 Ignore-this: d4cc8f87a6de8d9d1d6dc9b77130b3
4106]
4107[fix a non-portable printf format
4108Simon Marlow <marlowsd@gmail.com>**20100330134437
4109 Ignore-this: d41c23c54ec29654cb2049de1e588570
4110]
4111[avoid single quote in #error
4112Simon Marlow <marlowsd@gmail.com>**20100330120346
4113 Ignore-this: 663f39e7a27fead2f648fbf22d345bb4
4114]
4115[use FMT_Word64 instead of locally-defined version
4116Simon Marlow <marlowsd@gmail.com>**20100330114650
4117 Ignore-this: 82697b8095dffb3a8e196c687006ece0
4118]
4119[remove old/unused DotnetSupport and GhcLibsWithUnix
4120Simon Marlow <marlowsd@gmail.com>**20100330123732
4121 Ignore-this: c68814868b3671abdc369105bbeafe6c
4122]
4123[fix return type cast in f.i.wrapper when using libffi (#3516)
4124Simon Marlow <marlowsd@gmail.com>**20100329154220
4125 Ignore-this: f898eb8c9ae2ca2009e539735b92c438
4126 
4127 Original fix submitted by
4128   Sergei Trofimovich <slyfox@community.haskell.org>
4129 modified by me:
4130  - exclude 64-bit types
4131  - compare uniques, not strings
4132  - #include "ffi.h" is conditional
4133]
4134[libffi: install 'ffitarget.h' header as sole 'ffi.h' is unusable
4135Simon Marlow <marlowsd@gmail.com>**20100329135734
4136 Ignore-this: f9b555ea289d8df1aa22cb6faa219a39
4137 Submitted by: Sergei Trofimovich <slyfox@community.haskell.org>
4138 Re-recorded against HEAD.
4139]
4140[avoid a fork deadlock (see comments)
4141Simon Marlow <marlowsd@gmail.com>**20100329132329
4142 Ignore-this: 3377f88b83bb3b21e42d7fc5f0d866f
4143]
4144[tidy up the end of the all_tasks list after forking
4145Simon Marlow <marlowsd@gmail.com>**20100329132253
4146 Ignore-this: 819d679875be5f344e816210274d1c29
4147]
4148[Add a 'setKeepCAFs' external function (#3900)
4149Simon Marlow <marlowsd@gmail.com>**20100329110036
4150 Ignore-this: ec532a18cad4259a09847b0b9ae2e1d2
4151]
4152[Explicitly check whether ar supports the @file syntax
4153Ian Lynagh <igloo@earth.li>**20100329123325
4154 rather than assuming that all GNU ar's do.
4155 Apparently OpenBSD's older version doesn't.
4156]
4157[Fix the format specifier for Int64/Word64 on Windows
4158Ian Lynagh <igloo@earth.li>**20100327182126
4159 mingw doesn't understand %llu/%lld - it treats them as 32-bit rather
4160 than 64-bit. We use %I64u/%I64d instead.
4161]
4162[Fix the ghci startmenu item
4163Ian Lynagh <igloo@earth.li>**20100326235934
4164 I'm not sure what changed, but it now doesn't work for me without
4165 the "Start in" field being set.
4166]
4167[Fix paths to docs in "Start Menu" entries in Windows installer; fixes #3847
4168Ian Lynagh <igloo@earth.li>**20100326155917]
4169[Add a licence file for the Windows installer to use
4170Ian Lynagh <igloo@earth.li>**20100326155130]
4171[Add gcc-g++ to the inplace mingw installation; fixes #3893
4172Ian Lynagh <igloo@earth.li>**20100326154714]
4173[Add the licence file to the Windows installer. Fixes #3934
4174Ian Lynagh <igloo@earth.li>**20100326152449]
4175[Quote the paths to alex and happy in configure
4176Ian Lynagh <igloo@earth.li>**20100325143449
4177 Ignore-this: d6d6e1a250f88985bbeea760e63a79db
4178]
4179[Use </> rather than ++ "/"
4180Ian Lynagh <igloo@earth.li>**20100325133237
4181 This stops us generating paths like
4182     c:\foo\/ghc460_0/ghc460_0.o
4183 which windres doesn't understand.
4184]
4185[Append $(exeext) to utils/ghc-pkg_dist_PROG
4186Ian Lynagh <igloo@earth.li>**20100324233447
4187 Fixes bindist creation
4188]
4189[A sanity check
4190Simon Marlow <marlowsd@gmail.com>**20100325110500
4191 Ignore-this: 3b3b76d898c822456857e506b7531e65
4192]
4193[do_checks: do not set HpAlloc if the stack check fails
4194Simon Marlow <marlowsd@gmail.com>**20100325110328
4195 Ignore-this: 899ac8c29ca975d03952dbf4608d758
4196 
4197 This fixes a very rare heap corruption bug, whereby
4198 
4199  - a context switch is requested, which sets HpLim to zero
4200    (contextSwitchCapability(), called by the timer signal or
4201    another Capability).
4202 
4203  - simultaneously a stack check fails, in a code fragment that has
4204    both a stack and a heap check.
4205 
4206 The RTS then assumes that a heap-check failure has occurred and
4207 subtracts HpAlloc from Hp, although in fact it was a stack-check
4208 failure and retreating Hp will overwrite valid heap objects.  The bug
4209 is that HpAlloc should only be set when Hp has been incremented by the
4210 heap check.  See comments in rts/HeapStackCheck.cmm for more details.
4211 
4212 This bug is probably incredibly rare in practice, but I happened to be
4213 working on a test that triggers it reliably:
4214 concurrent/should_run/throwto001, compiled with -O -threaded, args 30
4215 300 +RTS -N2, run repeatedly in a loop.
4216]
4217[comments and formatting only
4218Simon Marlow <marlowsd@gmail.com>**20100325104617
4219 Ignore-this: c0a211e15b5953bb4a84771bcddd1d06
4220]
4221[Change how perl scripts get installed; partially fixes #3863
4222Ian Lynagh <igloo@earth.li>**20100324171422
4223 We now regenerate them when installing, which means the path for perl
4224 doesn't get baked in
4225]
4226[Pass the location of gcc in the ghc wrapper script; partially fixes #3863
4227Ian Lynagh <igloo@earth.li>**20100324171408
4228 This means we don't rely on baking a path to gcc into the executable
4229]
4230[Quote the ar path in configure
4231Ian Lynagh <igloo@earth.li>**20100324162043]
4232[Remove unused cUSER_WAY_NAMES cUSER_WAY_OPTS
4233Ian Lynagh <igloo@earth.li>**20100324145048]
4234[Remove unused cCONTEXT_DIFF
4235Ian Lynagh <igloo@earth.li>**20100324145013]
4236[Remove unused cEnableWin32DLLs
4237Ian Lynagh <igloo@earth.li>**20100324144841]
4238[Remove unused cGHC_CP
4239Ian Lynagh <igloo@earth.li>**20100324144656]
4240[Fix the build for non-GNU-ar
4241Ian Lynagh <igloo@earth.li>**20100324132907]
4242[Tweak the Makefile code for making .a libs; fixes trac #3642
4243Ian Lynagh <igloo@earth.li>**20100323221325
4244 The main change is that, rather than using "xargs ar" we now put
4245 all the filenames into a file, and do "ar @file". This means that
4246 ar adds all the files at once, which works around a problem where
4247 files with the same basename in a later invocation were overwriting
4248 the existing file in the .a archive.
4249]
4250[Enable shared libraries on Windows; fixes trac #3879
4251Ian Lynagh <igloo@earth.li>**20100320231414
4252 Ignore-this: c93b35ec5b7a7fa6ddb286d17a616216
4253]
4254[Add the external core PDF to the new build system
4255Ian Lynagh <igloo@earth.li>**20100321161909]
4256[Allow specifying $threads directly when validating
4257Ian Lynagh <igloo@earth.li>**20100321112835]
4258[Remove LazyUniqFM; fixes trac #3880
4259Ian Lynagh <igloo@earth.li>**20100320213837]
4260[UNDO: slight improvement to scavenging ...
4261Simon Marlow <marlowsd@gmail.com>**20100319153413
4262 Ignore-this: f0ab581c07361f7b57eae02dd6ec893c
4263 
4264 Accidnetally pushed this patch which, while it validates, isn't
4265 correct.
4266 
4267 rolling back:
4268 
4269 Fri Mar 19 11:21:27 GMT 2010  Simon Marlow <marlowsd@gmail.com>
4270   * slight improvement to scavenging of update frames when a collision has occurred
4271 
4272     M ./rts/sm/Scav.c -19 +15
4273]
4274[slight improvement to scavenging of update frames when a collision has occurred
4275Simon Marlow <marlowsd@gmail.com>**20100319112127
4276 Ignore-this: 6de2bb9614978975f17764a0f259d9bf
4277]
4278[Don't install the utf8-string package
4279Ian Lynagh <igloo@earth.li>**20100317212709]
4280[Don't use -Bsymbolic when linking the RTS
4281Ian Lynagh <igloo@earth.li>**20100316233357
4282 This makes the RTS hooks work when doing dynamic linking
4283]
4284[Fix Trac #3920: Template Haskell kinds
4285simonpj@microsoft.com**20100317123519
4286 Ignore-this: 426cac7920446e04f3cc30bd1d9f76e2
4287 
4288 Fix two places where we were doing foldl instead of foldr
4289 after decomposing a Kind.  Strange that the same bug appears
4290 in two quite different places!
4291]
4292[copy_tag_nolock(): fix write ordering and add a write_barrier()
4293Simon Marlow <marlowsd@gmail.com>**20100316143103
4294 Ignore-this: ab7ca42904f59a0381ca24f3eb38d314
4295 
4296 Fixes a rare crash in the parallel GC.
4297 
4298 If we copy a closure non-atomically during GC, as we do for all
4299 immutable values, then before writing the forwarding pointer we better
4300 make sure that the closure itself is visible to other threads that
4301 might follow the forwarding pointer.  I imagine this doesn't happen
4302 very often, but I just found one case of it: in scavenge_stack, the
4303 RET_FUN case, after evacuating ret_fun->fun we then follow it and look
4304 up the info pointer.
4305]
4306[Add sliceP mapping to vectoriser builtins
4307benl@ouroborus.net**20100316060517
4308 Ignore-this: 54c3cafff584006b6fbfd98124330aa3
4309]
4310[Comments only
4311benl@ouroborus.net**20100311064518
4312 Ignore-this: d7dc718cc437d62aa5b1b673059a9b22
4313]
4314[TAG 2010-03-16
4315Ian Lynagh <igloo@earth.li>**20100316005137
4316 Ignore-this: 234e3bc29e2f26cc59d7b03d780cc352
4317]
4318Patch bundle hash:
43193a9c2edcce685d8e88a3fccb6f2432c909556bd9