Ticket #3516: missing-dependency-on-ffitarget-h.patch

File missing-dependency-on-ffitarget-h.patch, 167.1 KB (added by PHO, 4 years ago)
Line 
11 patch for repository http://darcs.haskell.org/ghc:
2
3Sun Oct 17 20:33:21 JST 2010  pho@cielonegro.org
4  * libffi: missing dependency on ffitarget.h
5  dist-install/build/ffi.h should have a dependency on ffitarget.h as *_stub.c requires it during the stage-2 build.
6-----BEGIN PGP SIGNED MESSAGE-----
7Hash: SHA1
8
9
10New patches:
11
12[libffi: missing dependency on ffitarget.h
13pho@cielonegro.org**20101017113321
14 Ignore-this: 3e7f0c802e39184596219e3dcfd06fe9
15 dist-install/build/ffi.h should have a dependency on ffitarget.h as *_stub.c requires it during the stage-2 build.
16] hunk ./libffi/ghc.mk 142
17 
18        touch $@
19 
20- -libffi/dist-install/build/ffi.h: $(libffi_STAMP_CONFIGURE) | $$(dir $$@)/.
21+libffi/dist-install/build/ffi.h: $(libffi_STAMP_CONFIGURE) | libffi/dist-install/build/ffitarget.h $$(dir $$@)/.
22        "$(CP)" libffi/build/include/ffi.h $@
23 
24 libffi/dist-install/build/ffitarget.h: $(libffi_STAMP_CONFIGURE) | $$(dir $$@)/.
25
26Context:
27
28[Only put the boot packages in the haddock contents/index
29Ian Lynagh <igloo@earth.li>**20101016180031
30 Ignore-this: 3dc276dc734fec207def24a1d10eb369
31 We don't install dph etc, so don't put them in the doc index.
32]
33[Correct the regexp used to search for extra packages
34Ian Lynagh <igloo@earth.li>**20101016123421
35 Ignore-this: 98e3f8fda1c5678730a00ec1ac25fde3
36 We weren't ignoring comment lines
37]
38[Add a -fghci-sandbox flag so that we can en/disable the ghci sandbox
39Ian Lynagh <igloo@earth.li>**20101015172746
40 Ignore-this: 52bc5729437a93a925d3586d9803623c
41 It's on by default (which matches the previous behaviour).
42 
43 Motivation:
44 GLUT on OS X needs to run on the main thread. If you
45 try to use it from another thread then you just get a
46 white rectangle rendered. For this, or anything else
47 with such restrictions, you can turn the GHCi sandbox off
48 and things will be run in the main thread.
49]
50[Fix boot; it was failing if darcs-all or validate were missing
51Ian Lynagh <igloo@earth.li>**20101015164549
52 Ignore-this: 14b3e98836647aff345e35bca2102f93
53 (which is the case in sdists)
54]
55[Comments and layout
56simonpj@microsoft.com**20101015131924
57 Ignore-this: 126fbdb629a08c1380c7a1f5cd967d97
58]
59[Make (Located a) an instance of Eq, Ord
60simonpj@microsoft.com**20101015131857
61 Ignore-this: 5da50f8dab06fcbc23ce149cd5080062
62 
63 Fulfils Trac #4369
64]
65[Give user-defined rules precedence over built-in rules
66simonpj@microsoft.com**20101015131814
67 Ignore-this: 76fbf36cb1a09e31c10c68c06b98937e
68 
69 This fixes Trac #4397.  See comments with 'isMoreSpecific'.
70]
71[Fix Trac #4401: meta-tyvars allocated by the constraint solver are always touchable
72simonpj@microsoft.com**20101015130818
73 Ignore-this: 7fcdaee825d9cadcf69c8c9b8967a446
74 
75   See Note [Touchable meta type variables] in TcSMonad
76]
77[Remove GHC.extendGlobalRdrScope, GHC.extendGlobalTypeScope
78simonpj@microsoft.com**20101013091107
79 Ignore-this: df65a43d261353ad53811b25507e913e
80 
81 These functions were added by
82 
83    Tue Apr 18 03:36:06 BST 2006  Lemmih <lemmih@gmail.com>
84    * Make the initial rdr and type scope available in the ghc-api
85 
86 The are extremely dubious, because they extend the Rdr and Type
87 env for every compilation.  The right thing to do is to use
88 the InteractiveContext for temporary extensions.
89 
90 So far as we know, no one uses them.  And if they are being used
91 it's probably a mistake.  So we're backing them out.
92]
93[InlinePrag needs an arity only for INLINE, not INLINABLE
94Simon Marlow <marlowsd@gmail.com>**20101015094925
95 Ignore-this: 3b338f0b2b49fa714b195067e0026ef7
96 This doesn't fix anything (we think), but it's morally correct.
97]
98[Fix #4346 (INLINABLE pragma not behaving consistently)
99Simon Marlow <marlowsd@gmail.com>**20101015094836
100 Ignore-this: 9a9d8ad42cfdf7f619adfac284bae021
101 Debugged thanks to lots of help from Simon PJ: we weren't updating the
102 UnfoldingGuidance when the unfolding changed.
103 Also, a bit of refactoring and additinoal comments.
104]
105[Have boot check that we have the dph packages when validating
106Ian Lynagh <igloo@earth.li>**20101014140556
107 Ignore-this: 61e365127933f4dbfb62be66115455bd
108]
109[Add more documentation for interruptible foreign calls
110Simon Marlow <marlowsd@gmail.com>**20101014084253
111 Ignore-this: 28ec0ddd958926a08ab816b6975da344
112]
113[minor refactoring
114Simon Marlow <marlowsd@gmail.com>**20100926105819
115 Ignore-this: 70d38e0a31a096c94dad5f346b0d91f6
116]
117[Fix for interruptible FFI handling
118Simon Marlow <marlowsd@gmail.com>**20100925193442
119 Ignore-this: a63ba3c60f577002a1d32b30bb45090c
120 Set tso->why_blocked before calling maybePerformBlockedException(), so
121 that throwToSingleThreaded() doesn't try to unblock the current thread
122 (it is already unblocked).
123]
124[interruptible FFI: more robust handling of the exception case in the interpreter
125Simon Marlow <marlowsd@gmail.com>**20100925193317
126 Ignore-this: 7f9e67835a3bd096154db2dcf533ec66
127]
128[Don't interrupt when task blocks exceptions, don't immediately start exception.
129Edward Z. Yang <ezyang@mit.edu>**20100925033026
130 Ignore-this: 6c0669995a2b66abf29d68b3711cb78e
131]
132[Interruptible FFI calls with pthread_kill and CancelSynchronousIO. v4
133Edward Z. Yang <ezyang@mit.edu>**20100919002905
134 Ignore-this: 43c260f90eeb9c03413f6e749e23101d
135 
136 This is patch that adds support for interruptible FFI calls in the form
137 of a new foreign import keyword 'interruptible', which can be used
138 instead of 'safe' or 'unsafe'.  Interruptible FFI calls act like safe
139 FFI calls, except that the worker thread they run on may be interrupted.
140 
141 Internally, it replaces BlockedOnCCall_NoUnblockEx with
142 BlockedOnCCall_Interruptible, and changes the behavior of the RTS
143 to not modify the TSO_ flags on the event of an FFI call from
144 a thread that was interruptible.  It also modifies the bytecode
145 format for foreign call, adding an extra Word16 to indicate
146 interruptibility.
147 
148 The semantics of interruption vary from platform to platform, but the
149 intent is that any blocking system calls are aborted with an error code.
150 This is most useful for making function calls to system library
151 functions that support interrupting.  There is no support for pre-Vista
152 Windows.
153 
154 There is a partner testsuite patch which adds several tests for this
155 functionality.
156]
157[Remove ghc-pkg's dependency on haskell98
158Ian Lynagh <igloo@earth.li>**20101013194356
159 Ignore-this: 273fdede3f21e682faa4dc6e61f4e7aa
160]
161[Build haskell98 and haskell2010 with stage2
162Ian Lynagh <igloo@earth.li>**20101013182759
163 Stops us accidentally depending on them
164]
165[Fix warning: Remove unused import
166Ian Lynagh <igloo@earth.li>**20101013141224
167 Ignore-this: ae8891440811f4676d9bb8e721ad22ca
168]
169[Fix warnings
170benl@ouroborus.net**20101013040335
171 Ignore-this: 4e5875827d2840de863000cdb35afb0e
172]
173[RegAlloc: Track slot liveness over jumps in spill cleaner
174benl@ouroborus.net**20101013015414
175 Ignore-this: ccd4a148908b7fbdc6ea76acf527c16b
176]
177[Bump Cabal dep
178Ian Lynagh <igloo@earth.li>**20101012154528
179 Ignore-this: da8539b957b74c7da91efff8ac25872
180]
181[Remove __HASKELL1__, __HASKELL98__, __CONCURRENT_HASKELL__
182Ian Lynagh <igloo@earth.li>**20101012134700
183 Ignore-this: d0db531be58537c52802de1c62694eb1
184 We used to define these CPP symbols, but nothing on hackage uses them
185 and the first 2 are no longer correct (as we support multiple Haskell
186 versions).
187]
188[Follow Cabal changes: Cabal no longer has a docbook userguide
189Ian Lynagh <igloo@earth.li>**20101012130538
190 Ignore-this: bf0b30e465c13f82725d72aec145e939
191 For now we don't build the Cabal userguide, but we should add markdown
192 support so that we can do so.
193]
194[Fix build on Windows: ghc-pkg/Main.hs needs ForeignFunctionInterface
195Ian Lynagh <igloo@earth.li>**20101012112111]
196[Remove unnecessary import
197Ian Lynagh <igloo@earth.li>**20101010222231
198 Ignore-this: 6c7d5cee72837e2af43c2807e10ad602
199]
200[Make "./validate --slow" run the full testsuite
201Ian Lynagh <igloo@earth.li>**20101007004327
202 Ignore-this: 2dee8262c19fd5d5034a186203e20d0f
203]
204[Fix build following haskell98 and -fglasgow-exts changes
205Ian Lynagh <igloo@earth.li>**20101006160656
206 Ignore-this: c56bdb0f01da897b3de75bcab1e2887c
207]
208[Don't automatically link the haskell98 package
209Ian Lynagh <igloo@earth.li>**20101006130235
210 The default language is now Haskell2010, so this was a little odd.
211 Also, --make is now on by default, so this was largely irrelevant.
212]
213[Deprecate -fglasgow-exts
214Ian Lynagh <igloo@earth.li>**20101006124413
215 Ignore-this: 70be67898a633ab31d76f6fca557d55
216]
217[Remove Opt_GADTs and Opt_TypeFamilies from -fglasgow-exts
218Ian Lynagh <igloo@earth.li>**20101006122000
219 Ignore-this: 7153ec880b2b5f9d332ae2f57b97afcc
220 This means most code doesn't get caught by monomorphic local bindings.
221]
222[Fix Trac #4360: omitted case in combineCtLoc
223simonpj@microsoft.com**20101008135747
224 Ignore-this: 834636a97af6469862a822809253db41
225]
226[Beautiful new approach to the skolem-escape check and untouchable
227simonpj@microsoft.com**20101008133751
228 Ignore-this: 33517a772cfdfbf4aa4678609f3dcd71
229 
230 Instead of keeping a *set* of untouchable variables in each
231 implication contraints, we keep a *range* of uniques for the
232 *touchable* variables of an implication.  This are precisely
233 the ones we would call the "existentials" if we were French.
234 
235 It turns out that the code is more efficient, and vastly easier
236 to get right, than the set-based approach.
237 
238 Fixes Trac #4355 among others
239]
240[Do less simplification when doing let-generalisation
241simonpj@microsoft.com**20101008133542
242 Ignore-this: 71366d0de37f10ffba2edc9f3927ddbe
243 
244 This fixes Trac #4361.  In a rather delicate way, but
245 no more delicate than before.  A more remoseless typechecker
246 would reject #4361 altogether.
247 
248 See Note [Avoid unecessary constraint simplification]
249]
250[Suppress ambiguity errors if any other errors occur
251simonpj@microsoft.com**20101008111318
252 Ignore-this: 40f014265c1ab15fe172baaf76c23c87
253]
254[Fix Trac #4361: be more discerning when inferring types
255simonpj@microsoft.com**20101008111227
256 Ignore-this: 33656f9a151f494b26b6318b5fcffef
257 
258 Note [Avoid unecessary constraint simplification] in TcSimplify
259]
260[Float out partial applications
261Simon Marlow <marlowsd@gmail.com>**20101008092709
262 Ignore-this: 2dc9d10597c19d0598ef2fc3cf74156d
263 
264 This fixes at least one case of performance regression in 7.0, and
265 is nice win on nofib:
266 
267         Program           Size    Allocs   Runtime   Elapsed
268             Min          +0.3%    -63.0%    -38.5%    -38.7%
269             Max          +1.2%     +0.2%     +0.9%     +0.9%
270  Geometric Mean          +0.6%     -3.0%     -6.4%     -6.6%
271]
272[Suppress knock-on typechecker errors
273simonpj@microsoft.com**20101008094348
274 Ignore-this: 8d125926286a7614fa1ce998e3b26d04
275 
276 The error cascade caused puzzling errors in T4093b, and
277 suppressing some seems like a good plan.  Very few test
278 outputs change.
279]
280[Some refactoring and simplification in TcInteract.occurCheck
281simonpj@microsoft.com**20101007163500
282 Ignore-this: d43d09370ab27b8796062e2e98ce7e9
283]
284[Comments only
285simonpj@microsoft.com**20101007130301
286 Ignore-this: ab46592edd3d24786bbce42c50feb4fd
287]
288[Implement auto-specialisation of imported Ids
289simonpj@microsoft.com**20101007111051
290 Ignore-this: 45257ff6e9597e4fa4de10b0657e27d6
291 
292 This big-ish patch arranges that if an Id 'f' is
293   * Type-class overloaded
294        f :: Ord a => [a] -> [a]
295   * Defined with an INLINABLE pragma
296        {-# INLINABLE f #-}
297   * Exported from its defining module 'D'
298 
299 then in any module 'U' that imports D
300 
301 1. Any call of 'f' at a fixed type will generate
302    (a) a specialised version of f in U
303    (b) a RULE that rewrites unspecialised calls to the
304        specialised on
305 
306   e.g. if the call is (f Int dOrdInt xs) then the
307   specialiser will generate
308      $sfInt :: [Int] -> [Int]
309      $sfInt = <code for f, imported from D, specialised>
310      {-# RULE forall d.  f Int d = $sfInt #-}
311 
312 2. In addition, you can give an explicit {-# SPECIALISE -#}
313    pragma for the imported Id
314      {-# SPECIALISE f :: [Bool] -> [Bool] #-}
315    This too generates a local specialised definition,
316    and the corresponding RULE
317 
318 The new RULES are exported from module 'U', so that any module
319 importing U will see the specialised versions of 'f', and will
320 not re-specialise them.
321 
322 There's a flag -fwarn-auto-orphan that warns you if the auto-generated
323 RULES are orphan rules. It's not in -Wall, mainly to avoid lots of
324 error messages with existing packages.
325 
326 Main implementation changes
327 
328  - A new flag on a CoreRule to say if it was auto-generated.
329    This is persisted across interface files, so there's a small
330    change in interface file format.
331 
332  - Quite a bit of fiddling with plumbing, to get the
333    {-# SPECIALISE #-} pragmas for imported Ids.  In particular, a
334    new field tgc_imp_specs in TcGblEnv, to keep the specialise
335    pragmas for imported Ids between the typechecker and the desugarer.
336 
337  - Some new code (although surprisingly little) in Specialise,
338    to deal with calls of imported Ids
339]
340[Make NameEnv back into type NameEnv a = UniqFM a
341simonpj@microsoft.com**20101007104638
342 Ignore-this: 24857c013461788be354520e84f4c286
343 
344 I don't think the type distinction of declaring NameEnv with a newtype
345 (as it was) is really useful to us. Moreover, VarEnv is a UniqFM, and
346 I do sometimes want to build an envt with Ids and look up with Names.
347 
348 This may not be the last word on the subject.
349]
350[Improve the rule-matcher
351simonpj@microsoft.com**20101007103700
352 Ignore-this: 9de96237dc4b73a43326bd568e34b53b
353 
354 Previously it was rejecting the match
355 
356   Template: forall s t. map s t
357   Actual:   map Int t
358 
359 which should obviously be fine.  It turns out that this kind of match
360 comes up when specialising.  By freshening that t we could avoid the
361 difficulty, but morally the (forall t) binds t and the rule should
362 be alpha-equivalent regardless of the forall'd variables.
363 
364 This patch makes it so, and incidentally makes matching a little
365 more efficient.  See Note [Eta expansion] in VarEnv.
366]
367[Fix Trac #4345: simplifier bug
368simonpj@microsoft.com**20101007102720
369 Ignore-this: 261c1c9f094df344ce34de814f8b60c5
370 
371 This is another long-standing bug, in which there was a possibility
372 that a loop-breaker could lose its loop-breaker-hood OccInfo,
373 and then the simplifer re-simplified the expression. Result, either
374 non-termination or, in the case of #4345, an unbound identifier.
375 
376 The fix is very simple, in Id.transferPolyIdInfo.
377 See Note [transferPolyIdInfo].
378]
379[Avoid redundant simplification
380simonpj@microsoft.com**20101007095935
381 Ignore-this: 61bd1a2c508260f558866e6a88c29fa3
382 
383 When adding specialisation for imported Ids, I noticed that the
384 Glorious Simplifier was repeatedly (and fruitlessly) simplifying the
385 same term.  It turned out to be easy to fix this, because I already
386 had a flag in the ApplyTo and Select constructors of SimplUtils.SimplCont.
387 
388 See Note [Avoid redundant simplification]
389]
390[Make the occurrence analyser deal correctly with RULES for imported Ids
391simonpj@microsoft.com**20101007094100
392 Ignore-this: 335b1cad013524e42b31e88c0a7a00f6
393 
394 This patch fixes a long-standing lurking bug, but it surfaced when I
395 was adding specialisation for imported Ids.
396 
397 See Note [ImpRuleUsage], which explains the issue.   The solution
398 seems more complicated than the problem really deserves, but I
399 could not think of a simpler way, so I just bit the bullet and
400 wrote the code.  Improvements welcome.
401]
402[Make warning-free
403simonpj@microsoft.com**20101007092007
404 Ignore-this: 4bae0c470a8a1f96d21990d1f3cf1f93
405]
406[This is just white-space and layout
407simonpj@microsoft.com**20101007091618
408 Ignore-this: 759c0335df70fce32558e967f140803a
409 
410 (At least, I don't think there is anything else.)
411]
412[Fix an ASSERT failure in FamInstEnv
413simonpj@microsoft.com**20101007091327
414 Ignore-this: a8c08ccb7ec2bc65864a674b5441539
415 
416 I added a lot of comments too, to explain the preconditions;
417 esp Note [FamInstEnv]
418]
419[Fix a looping bug in the new occur-check code
420simonpj@microsoft.com**20101007084104
421 Ignore-this: a02a2deafb9ec986ef1565f4596049ed
422]
423[Fix test T4235 with -O
424simonpj@microsoft.com**20101006155223
425 Ignore-this: f0fa0fe2f0c493e362d528d71f7a64e1
426 
427 The tag2Enum rule wasn't doing the right thing for
428 enumerations with a phantom type parameter, like
429    data T a = A | B
430]
431[Make warning-free
432simonpj@microsoft.com**20101006155033
433 Ignore-this: 221a3c95a6079c6ecc0468996a38b048
434]
435[Major bugfixing pass through the type checker
436dimitris@microsoft.com**20101006152854]
437[Typechecker performance fixes and flatten skolem bugfixing
438dimitris@microsoft.com**20101004130200
439 Ignore-this: 86721ba3f09479c146a0710796b43459
440]
441[Performance bug fixes
442dimitris@microsoft.com**20100923143918]
443[Fix Trac #4371: matching of view patterns
444simonpj@microsoft.com**20101006115316
445 Ignore-this: 494b28b91f1e6392b2f1521cda0e83b1
446]
447[Remove unused NoMatchContext construtor
448simonpj@microsoft.com**20101006115251
449 Ignore-this: 8985ff1dac51fb652bd65657a630a792
450]
451[Refactoring: mainly rename ic_env_tvs to ic_untch
452simonpj@microsoft.com**20101006102830
453 Ignore-this: 32999403a3f447e14b59cec7896027ff
454 
455 Plus remember to zonk the free_tvs in TcUnify.newImplication
456]
457[remove unnecessary/broken definition of mask_
458Simon Marlow <marlowsd@gmail.com>**20101002195118
459 Ignore-this: 4cdc9c95d40e01cdfe2d0ac411476603
460]
461[-fwarn-tabs: add "Warning" to the message
462Simon Marlow <marlowsd@gmail.com>**20101002195100
463 Ignore-this: 589a36daa3426ab51f2fb140e38df6c
464]
465[give a better error message in the non-threaded RTS for out-of-range FDs
466Simon Marlow <marlowsd@gmail.com>**20100929212916
467 Ignore-this: e94c9f390b8f79d24895a80f9d16c8d9
468 
469 # ./aw
470 aw: file descriptor 1027 out of range for select (0--1024).
471 Recompile with -threaded to work around this.
472]
473[Fix a very rare crash in GHCi
474Simon Marlow <marlowsd@gmail.com>**20101005144735
475 Ignore-this: dad1cd08934bae2ba47e72c0c000acfa
476 When a BCO with a zero-length bitmap was right at the edge of
477 allocated memory, we were reading a word of non-existent memory.
478 
479 This showed up as a segfault in T789(ghci) for me, but the crash was
480 extremely sensitive and went away with most changes.
481 
482 Also, optimised scavenge_large_bitmap a bit while I was in there.
483]
484[Using 'stdcall' when it is not supported is only a warning now (#3336)
485Simon Marlow <marlowsd@gmail.com>**20100924152445
486 Ignore-this: 66c5903a600a47485a7583535bb38455
487]
488[remove unnecessary stg_noForceIO (#3508)
489Simon Marlow <marlowsd@gmail.com>**20100924150202
490 Ignore-this: dec52de9cd9da7dcedae12b20691aba9
491]
492[Replace an outputStr with putStrLn calls; fixes #4332
493Ian Lynagh <igloo@earth.li>**20101003125707
494 Ignore-this: eb4b5f60d9d9d3f7dc203869927b28ba
495]
496[make test and fulltest targets in the main Makefile; fixes #4297
497Ian Lynagh <igloo@earth.li>**20100930224741
498 You can now run "make test" in the root, and the fast testsuite will be
499 run with cleaning enabled. It will also put the summary in
500 testsuite_summary.txt.
501]
502[Don't show the loaded packages in ":show packages"; fixes #4300
503Ian Lynagh <igloo@earth.li>**20100930210128
504 It's never worked properly, and the information is in ":show linker".
505]
506[Handle EXTRA_LIBRARIES when building programs
507Ian Lynagh <igloo@earth.li>**20100930192552
508 Ignore-this: 401a26e18d25dcaee010b13eaed8f011
509 And set hp2ps's EXTRA_LIBRARIES. Based on a patch from Sergei Trofimovich.
510]
511[Fix the doc directory on Windows
512Ian Lynagh <igloo@earth.li>**20100929133328]
513[Remove an unused import on Windows
514Ian Lynagh <igloo@earth.li>**20100929000024
515 Ignore-this: 2899e0e5a47122e637fb5c8aa0df52ab
516]
517[Use showCommandForUser when showing tracing commands
518Ian Lynagh <igloo@earth.li>**20100928235844
519 Ignore-this: 8a4a9c9f8a8029e708c4297b096b6ef1
520]
521[Fix hsc2hs docs: 'gcc' is now the default compiler, not 'ghc'; fixes #4341
522Ian Lynagh <igloo@earth.li>**20100928201938]
523[New member "archiveMemberName" for struct _ObjectCode
524pho@cielonegro.org**20100927224145
525 Ignore-this: 628bc605e7dc4f0c4856c6f7ad23d9ee
526 
527 struct _ObjectCode should be able to retain the name of archive members.
528 Though currently the only use of those names are for debugging outputs.
529]
530[Use an empty signal handler for SIGPIPE instead of SIG_IGN
531Simon Marlow <marlowsd@gmail.com>**20100925193548
532 Ignore-this: b4dc5de32740a7c5fd8fe4b3bfb1300f
533 
534 This is so that the SIGPIPE handler gets reset to the default
535 automatically on exec().
536]
537[Fix the TH deps
538Ian Lynagh <igloo@earth.li>**20100925210029
539 Ignore-this: 32b832301a3625d4ba70f84c5c4f94d2
540]
541[Check inplace doesn't exist before we try to create it
542Ian Lynagh <igloo@earth.li>**20100924191858
543 This fixes rerunning configure in a tree which already has an inplace
544 directory. Edward Z Yang ran into this; I guess whether it actually
545 fails depends on details of your installation, or we'd have run into
546 it sooner.
547]
548[Fix an egregious bug: INLINE pragmas on monomorphic Ids were being ignored
549simonpj@microsoft.com**20100924155815
550 Ignore-this: 38c6eec6710a92df7662a55fc5132c15
551 
552 I had do to some refactoring to make this work nicely
553 but now it does. I can't think how this escaped our
554 attention for so long!
555]
556[Eta expand only lambdas that bind a non-dictionary Id
557simonpj@microsoft.com**20100924155707
558 Ignore-this: 7cc265eaf6c0bb3fa12eb311d92594ac
559 
560 See Note [When to eta expand]. The idea is that dictionary
561 lambdas are invisible to the user, so we shouldn't eta
562 expand them.
563]
564[Add a comment
565simonpj@microsoft.com**20100924155620
566 Ignore-this: de210a1afdd40328824803e1d77b4d7f
567]
568[Add a debug print
569simonpj@microsoft.com**20100924155614
570 Ignore-this: 1a58b6d297fc77d6ded8eec7ea9f895d
571]
572[Just moving comments around
573simonpj@microsoft.com**20100924155600
574 Ignore-this: 96635b8e8c9d88b50d82938568152ef8
575]
576[use putStrLn instead of Haskeline's outputStrLn
577Simon Marlow <marlowsd@gmail.com>**20100924133154
578 Ignore-this: 7581ae11714a9a52e78ba098c3c216b3
579 use of the latter caused problems for Claus Reinke's macros that
580 redirect stdout.
581]
582[Change "OPTIONS" to "OPTIONS_GHC" in error messages; fixes #4327
583Ian Lynagh <igloo@earth.li>**20100924120423
584 Ignore-this: 1697c83a5c346db640c0a2e22c69ff55
585]
586[Add deps for TH uses in vector
587Ian Lynagh <igloo@earth.li>**20100923220244
588 Ignore-this: 54c3386b1c268821fcdd34b84bc8c6a4
589]
590[Bump Cabal dep
591Ian Lynagh <igloo@earth.li>**20100923143241]
592[Update Cabal's version number
593Ian Lynagh <igloo@earth.li>**20100923141719]
594[Build primitive with stage2
595Ian Lynagh <igloo@earth.li>**20100923140525
596 Ignore-this: 110a819b78a57629a7edf1d4facdc191
597]
598[Fix the Windows __chkstk build error (missing Linker symbol)
599Simon Marlow <marlowsd@gmail.com>**20100924113837
600 Ignore-this: 48f0907bb1bd5eaa0730b94a6bd94ea
601]
602[emit a helpful error message for missing DPH packages
603Simon Marlow <marlowsd@gmail.com>**20100923141957
604 Ignore-this: 55ff2ee90c94524e023e014243bfe5df
605]
606[Fix computation of installed packages
607simonpj@microsoft.com**20100924084737
608 Ignore-this: a597d2fa8be5135ba8ead6d2624b3d71
609 
610 This is a follow-on to Simon's patch yesterday, developed
611 with him.  It cleans up the computation of how packages
612 are installed, and installs the right ones.
613]
614[Fix braino in WwLib/Literal patch
615simonpj@microsoft.com**20100924070914
616 Ignore-this: f6eb3a42e10f8aa7920de541cdfe76d8
617]
618[For now, switch off incomplete-pattern warnings in containers
619simonpj@microsoft.com**20100923130117
620 Ignore-this: 7ffa58567f7a33aafe256492999da325
621 
622 Put it back on when my patch is applied to the containers repo.
623 (the one that removes two refuable lambdas)
624]
625[Make -funfolding-dict-threshold work properly
626simonpj@microsoft.com**20100923130032
627 Ignore-this: 417788f5b09d1d624f6b6371852c80c7
628 
629 and increase its default value. This makes overloaded functions
630 a bit keener to inline.  Which fixes Trac #4321
631]
632[Impredicative types is no longer deprecated
633simonpj@microsoft.com**20100923125910
634 Ignore-this: 2bbaeb38b5e8424551677c0add627683
635]
636[Do not make FunctionalDependencies force MonoLocalBinds
637simonpj@microsoft.com**20100923125900
638 Ignore-this: f4ae1fd07c87ec14f60bdfe3863ba7a9
639]
640[move CHECKED settings to the right place
641Simon Marlow <marlowsd@gmail.com>**20100923123558
642 Ignore-this: e00a0eb5855463cc9b953670b3bbf211
643]
644[turn off -Werror for primitive and vector
645Simon Marlow <marlowsd@gmail.com>**20100923122055
646 Ignore-this: 54d7b80f3f893385e1c3ef431e2a8a7b
647]
648[Add primitive and vector packages for DPH support
649Simon Marlow <marlowsd@gmail.com>**20100923104542
650 Ignore-this: c070d015385b0a0797394132dcbb7670
651 DPH is now using the public vector package instead of its internal
652 version.
653 
654 vector and primitive are not "boot" packages; they aren't required to
655 build GHC, but they are required to validate (because we include DPH
656 when validating).
657 
658 If you say './darcs-all get --no-dph' then you don't get DPH, vector,
659 or primitive.
660]
661[Refactoring and tidy up in the build system
662Simon Marlow <marlowsd@gmail.com>**20100923095642
663 Ignore-this: f7bf3a1fd160149d89b26f464b064fb1
664 
665 Instead of the ghc-stage and ghc-stage2-package files in a package, we
666 now have a list of these in ghc.mk.  There are other similar lists (of
667 boot-packages and non-installable packages), so this is not too bad,
668 and is simpler.
669 
670 While poking around in the top-level ghc.mk file I spotted various
671 opportunities to clean up and re-order some of the cruft that has
672 accumulated over time.
673]
674[Allow absent State# RealWorld arguments
675simonpj@microsoft.com**20100923111356
676 Ignore-this: c2d57633dec0293ebe6723ea3a4bb5df
677]
678[Add notSCCNote, and use it
679simonpj@microsoft.com**20100923105949
680 Ignore-this: c8cc758656558a7f366bf784d75f0304
681 
682 The point here is that SCCs get in the way of eta
683 expansion and we must treat them uniformly.
684]
685[Remove use of lambda with a refutable pattern
686simonpj@microsoft.com**20100923105901
687 Ignore-this: d7d48b94e5744717a838591a1cc79cf0
688]
689[Avoid ASSERT black hole
690simonpj@microsoft.com**20100923105820
691 Ignore-this: 5419d450871be22c8781ac3f0f40d76a
692 
693 When this ASSERT tripped in CoreToStg it tried to print out
694 too much, which tripped the asssertion again.  Result: an
695 infinite loop with no output at all.  Hard to debug!
696]
697[Rejig the absent-arg stuff for unlifted types
698simonpj@microsoft.com**20100923105732
699 Ignore-this: 69daa35816b948b0c4d259c73a5e928e
700 
701 This is what was giving the "absent entered" messages
702 See Note [Absent errors] in WwLib.  We now return a
703 suitable literal for absent values of unlifted type.
704]
705[Remove -fwarn-simple-patterns, and make -fwarn-incomplete-patterns include lambdas
706simonpj@microsoft.com**20100922133934
707 Ignore-this: e851a2fb0377e10c28c506f0bf14cc85
708 
709 This makes
710      \(x:xs) -> e
711 want when you have -fwarn-incomplete-patterns, which is consistent.
712]
713[Get rid of non-exhaustive lambda
714simonpj@microsoft.com**20100922133801
715 Ignore-this: 748b2d5b43b02b6591b81abe7c105cd6
716]
717[Fix an ASSERT failure with profiling
718simonpj@microsoft.com**20100922133741
719 Ignore-this: 170b2e94d6ee8cc7444cc4bb515328a0
720 
721 The problem arose with this kind of thing
722 
723    x = (,) (scc "blah" Nothing)
724 
725 Then 'x' is marked NoCafRefs by CoreTidy, becuase it has
726 arity 1, and doesn't mention any caffy things.
727 
728 That in turns means that CorePrep must not float out the
729 sat binding to give
730 
731   sat = scc "blah" Nothing
732   x = (,) sat
733 
734 Rather we must generate
735 
736   x = \eta. let sat = scc "blah" Nothing
737             in (,) sat eta
738 
739 URGH! This Caf stuff is such a mess.
740]
741[Remove an out of date paragraph from the user guide; fixes #4331
742Ian Lynagh <igloo@earth.li>**20100922225239]
743[Fix bindisttest when GhcProfiled = YES
744Ian Lynagh <igloo@earth.li>**20100921222634
745 Ignore-this: 47c620fd6bec745e3eb699d9f53441d8
746]
747[Fixes for when HADDOCK_DOCS=NO
748Ian Lynagh <igloo@earth.li>**20100921213916
749 Ignore-this: e0e069555c6db9d01a8ac70ba4dde591
750]
751[Bump version to 7.1
752Ian Lynagh <igloo@earth.li>**20100921195935
753 Ignore-this: 4563987e6885d5ef55995ec0fa0d5ae8
754]
755[Don't use -march=i686 on powerpc-apple-darwin
756Ian Lynagh <igloo@earth.li>**20100921193721
757 Thorikil ran into this when doing a PPC OS X build. We now also don't
758 use -m32 on PPC/OSX, but I don't think it should be necessary. We can
759 add it back if it does turn out to be.
760]
761[add a simple trace facility to the build system
762Simon Marlow <marlowsd@gmail.com>**20100921134729
763 Ignore-this: d23ea2d62a648d0711b4b07d98e1b79f
764 
765 saying
766 
767   make TRACE=1
768 
769 prints most of the macro calls and their arguments.  It's easy to
770 trace new macros; see rules/trace.mk.
771]
772[fix building with extra packages (packages were added to BUILD_DIRS twice)
773Simon Marlow <marlowsd@gmail.com>**20100921100153
774 Ignore-this: 4b425dff9777871ad5ba3e05e1d14483
775 Also add some comments about what extra-packages is doing
776]
777[add extra packages to $(EXTRA_PACKAGES), so we avoid installing them by default
778Simon Marlow <marlowsd@gmail.com>**20100920144307
779 Ignore-this: 3395825d911a8bf7ba8385518d8b517b
780]
781[Fix indexing error in archive loader
782Ian Lynagh <igloo@earth.li>**20100921121642]
783[Add some -Dl belches
784Ian Lynagh <igloo@earth.li>**20100921121624]
785[Add casts to fix warnings
786Ian Lynagh <igloo@earth.li>**20100921121714]
787[Add support for BSD-variant large filenames in .a archives
788Ian Lynagh <igloo@earth.li>**20100921000451]
789[Tell Cabal that we're not building GHCi libs if UseArchivesForGhci=YES
790Ian Lynagh <igloo@earth.li>**20100920230449]
791["UseArchivesForGhci = YES" on darwin
792Ian Lynagh <igloo@earth.li>**20100920211538]
793[Add a dependency that my OS X build has recently started tripping up over
794Ian Lynagh <igloo@earth.li>**20100920210239]
795[Add "Use archives for ghci" to --info output
796Ian Lynagh <igloo@earth.li>**20100920210523]
797[Implement archive loading for ghci
798Ian Lynagh <igloo@earth.li>**20100920201620]
799[Tweak gen_contents_index now dph may not be there
800Ian Lynagh <igloo@earth.li>**20100920201513]
801[Filter out the FFI library when loading package in ghci
802Ian Lynagh <igloo@earth.li>**20100920181032
803 The FFI GHCi import lib isn't needed as
804 compiler/ghci/Linker.lhs + rts/Linker.c link the
805 interpreted references to FFI to the compiled FFI.
806 We therefore filter it out so that we don't get
807 duplicate symbol errors.
808]
809[Loosen the conditions for -XUndecidableInstances; fixes Trac #4200
810simonpj@microsoft.com**20100919162623
811 Ignore-this: 2f4323e278b1ce9250549727ffd0aa1b
812]
813[Further improvements in error messages
814simonpj@microsoft.com**20100919153355
815 Ignore-this: b6fa0b11ae893df1a3ca68f78e427fa
816]
817[Add a flag -fwarn-missing-local-sigs, and improve -fwarn-mising-signatures
818simonpj@microsoft.com**20100919153327
819 Ignore-this: fda8dfca450054ea692be0ee30b01885
820 
821 The new flag prints out a warning if you have a local,
822 polymorphic binding that lacks a type signature. It's meant
823 to help with the transition to the new typechecker, which
824 discourages local let-generalisation.
825 
826 At the same time I moved the missing-signature code to TcHsSyn,
827 where it takes place as part of zonking.  That way the
828 types are reported after all typechecking is complete,
829 thereby fixing Trac #3696.  (It's even more important for
830 local bindings, which is why I made the change.)
831]
832[Include the "stupid theta" in the type of $con2tag
833simonpj@microsoft.com**20100919152201
834 Ignore-this: d95fae78a0e66f48bbd5862573a11f4d
835]
836[Add a release note about the typechecker
837Ian Lynagh <igloo@earth.li>**20100919132927]
838[Enable shared libs on OpenBSD
839Matthias Kilian <kili@outback.escape.de>**20100918205040
840 Ignore-this: 729dd7ac0bba5d758f43bc31b541896
841]
842[Add separate functions for querying DynFlag and ExtensionFlag options
843Ian Lynagh <igloo@earth.li>**20100918163815
844 and remove the temporary DOpt class workaround.
845]
846[Fix mkUserGuidePart deps
847Ian Lynagh <igloo@earth.li>**20100918145904
848 We need to directly depend on the stage1 libs. The stage1 compiler lib
849 doesn't depend on them.
850]
851[Fix build on cygwin: Normalise slashes in .depend files to be /
852Ian Lynagh <igloo@earth.li>**20100918132328
853 Ignore-this: 664f5ef4a41a4461eb34fe2ca7f2729a
854]
855[extra packages info is now read from packages file
856Ian Lynagh <igloo@earth.li>**20100917224409
857 rather than being repeated in the build system
858]
859[Tweak darcs-all
860Ian Lynagh <igloo@earth.li>**20100917194435]
861[Bump dependencies
862Ian Lynagh <igloo@earth.li>**20100917183609]
863[Library release notes for 7.0.1
864Ian Lynagh <igloo@earth.li>**20100917174850]
865[Fix overriding of implicit parameters in the solver
866simonpj@microsoft.com**20100917140403
867 Ignore-this: af76732309c7e2ca6b04f49327e9c14b
868]
869[Minor type printing amomaly
870simonpj@microsoft.com**20100917140204
871 Ignore-this: c90cb2e51421b4543a827e096051772e
872]
873[Spaces only
874simonpj@microsoft.com**20100917140156
875 Ignore-this: 7e34479502f7cb87d762355e40cbd012
876]
877[Minor refactoring
878simonpj@microsoft.com**20100917140150
879 Ignore-this: 6c0648b949b91b7e2f23c136b124faf2
880]
881[Add types of implicit parameters as untouchable
882simonpj@microsoft.com**20100917140138
883 Ignore-this: ba80740a557a9ba062dc7756e2320d17
884 
885 This is a tricky point:
886    see Note [Implicit parameter untouchables]
887]
888[Better pretty printing of implicit parameters
889simonpj@microsoft.com**20100917140054
890 Ignore-this: 867dd67818a5bd687b2b6a1b59e15775
891]
892[Yet more error message improvement
893simonpj@microsoft.com**20100917121206
894 Ignore-this: 647fe8129d1d39d81e8249debd8df94e
895]
896[More error message wibbles
897simonpj@microsoft.com**20100917094721
898 Ignore-this: 8ec2f150b96b26af2e9ab7ac2b371fc7
899]
900[More error refactoring
901simonpj@microsoft.com**20100917092834
902 Ignore-this: 2d570ac0b9cc11305ddd33d093d11324
903]
904[Refactor type errors a bit
905simonpj@microsoft.com**20100917080726
906 Ignore-this: 33da4549373f585064e2ee22b50ad6ac
907 
908 Improves kind error messages in paticular
909]
910[Fix a very subtle shadowing bug in optCoercion
911simonpj@microsoft.com**20100916170452
912 Ignore-this: 9041cfb3c93e27a5e644e57815313aae
913 
914 See Note [Subtle shadowing in coercions]
915 
916 This is what was going wrong in Trac 4160.
917]
918[Fix bad error in tyVarsOfType
919simonpj@microsoft.com**20100916170348
920 Ignore-this: 67c8ce96a668cf6e3a38b82c893bcd81
921 
922 We weren't gathering the type variables free in the kind
923 of a coercion binder!
924]
925[More assertions
926simonpj@microsoft.com**20100916170310
927 Ignore-this: 7fdcb53c99d791621a3d7e01ef454404
928]
929[Add more location info in CoreLint
930simonpj@microsoft.com**20100916170229
931 Ignore-this: 6558bab544b4f30189e0430668db87c3
932]
933[Print coercion variables as such (debugging change only)
934simonpj@microsoft.com**20100916165944
935 Ignore-this: c6d2001c1d8279a2288cb63bc339577d
936]
937[Remove pprTrace
938simonpj@microsoft.com**20100915225935
939 Ignore-this: 28185bbfa9732386f3c0f3eb4781a637
940]
941[Remove dead code dealing with type refinement
942simonpj@microsoft.com**20100915223230
943 Ignore-this: 62824b5c2ec1077c7642163352559621
944]
945[Use mkAppTy
946simonpj@microsoft.com**20100915223205
947 Ignore-this: e79e087b6a49219e9088846a1253a153
948 
949 Using AppTy in CoreLint was giving a bogus Lint failure
950]
951[Comments only
952simonpj@microsoft.com**20100915221253
953 Ignore-this: 3a45ea614188ccbb4a462de5cac96eda
954]
955[Extend eta reduction to work with casted arguments
956simonpj@microsoft.com**20100915221229
957 Ignore-this: 24b103dcdf70331211507af929789f86
958 
959 See Trac #4201, and
960 Note [Eta reduction with casted arguments]
961 
962 Thanks to Louis Wasserman for suggesting this, and
963 implementing an early version of the patch
964]
965[Allow "INLINEABLE" as a synonym
966simonpj@microsoft.com**20100915154249
967 Ignore-this: f41f80cb769e9acd5b463b170df698d0
968]
969[Documentation for INLINABLE
970simonpj@microsoft.com**20100915154235
971 Ignore-this: f942c02bcadc0d2d2f05b9369f93e280
972]
973[Implement TH reification of instances (Trac #1835)
974simonpj@microsoft.com**20100915151242
975 Ignore-this: 97dfa83db7da8f6cbd1b96801a57f8c5
976 
977 Accompanying patch for template-haskell package is reqd
978]
979[errno corresponding to ERROR_NO_DATA should be EPIPE (non-threaded RTS)
980Simon Marlow <marlowsd@gmail.com>**20100915141809
981 Ignore-this: 709c7280fbaa762e7071fb8796e8c01e
982]
983[Windows: use a thread-local variable for myTask()
984Simon Marlow <marlowsd@gmail.com>**20100915120627
985 Ignore-this: 13ffa4f19ebd319fe672af53af8d0b9a
986 Which entailed fixing an incorrect #ifdef in Task.c
987]
988[Fix typo
989Ian Lynagh <igloo@earth.li>**20100915140814]
990[Add quotes in error message
991simonpj@microsoft.com**20100915144724
992 Ignore-this: c5158047c0aa41947a79e4c8edbe54f4
993]
994[Fix isDefaultInlinePragma
995simonpj@microsoft.com**20100915144710
996 Ignore-this: c9addf6bf811b23dc12603cf8521aa6c
997]
998[Implement INLINABLE pragma
999simonpj@microsoft.com**20100915124442
1000 Ignore-this: 80a4ab2c2d65b27868dc9b2e954d6c6f
1001 
1002 Implements Trac #4299.  Documentation to come.
1003]
1004[Less voluminous error when derived code doesn't typecheck
1005simonpj@microsoft.com**20100915072301
1006 Ignore-this: eca7871dcc50c1070a0b530711adea27
1007]
1008[Improve pretty-printing of family instances
1009simonpj@microsoft.com**20100915123219
1010 Ignore-this: 25ec6bcc7e8a7f7c303b38ca201db90e
1011 
1012 Fixed Trac #4246
1013]
1014[Fix Trac #4240: -ddump-minimal-imports
1015simonpj@microsoft.com**20100915121937
1016 Ignore-this: ab85057cb829a42ea44a92f7b4af24a3
1017 
1018 See Note [Partial export] for the details.
1019 I also fixed one egregious bug that was just
1020 waiting to bite: we were using loadSysInterface
1021 instead of loadSrcInterface.
1022]
1023[Comments only
1024simonpj@microsoft.com**20100915105707
1025 Ignore-this: ab3a5f16f8260b7b8570e748bf97998a
1026]
1027[implement setThreadAffinity on Windows (#1741)
1028Simon Marlow <marlowsd@gmail.com>**20100914155844
1029 Ignore-this: a14c7b4ef812007042342d0a25478f0b
1030]
1031[COFF: cope with new debug sections in gcc 4.x (fixes ghciprog004)
1032Simon Marlow <marlowsd@gmail.com>**20100914153026
1033 Ignore-this: f340e40a2b0390836bc61bba144a04ed
1034 Also updated the object file parser to properly handle the overflow
1035 case for section names longer than 8 chars.
1036]
1037[eliminate clutter from make output
1038Simon Marlow <marlowsd@gmail.com>**20100915105712
1039 Ignore-this: bfa4480dd239dda2a02ac391b6a9219c
1040]
1041[rts_isProfiled should be a visible API (fixes T2615(dyn))
1042Simon Marlow <marlowsd@gmail.com>**20100915083941
1043 Ignore-this: b8ac09bb9d1a929bf45c6122f8485561
1044]
1045[Fix the "lost due to fragmentation" calculation
1046Simon Marlow <marlowsd@gmail.com>**20100914145945
1047 Ignore-this: cdffcc9f3061c3a33da5171be111fc43
1048 It was counting the space used by block descriptors as "lost"
1049]
1050[fix +RTS -S output: use peak_mblocks_allocated, now that mblocks can be freed
1051Simon Marlow <marlowsd@gmail.com>**20100914135030
1052 Ignore-this: 65d21e5f86d3ab6ab4d6c255180b6968
1053]
1054[Fix egregious bug in deeplyInstantiate
1055simonpj@microsoft.com**20100915070325
1056 Ignore-this: 22ede973038877af2673339aaf5de6cf
1057 
1058 This resulted in an infinite loop in applyTypeToArgs, in syb
1059]
1060[Improve HsSyn pretty printing
1061simonpj@microsoft.com**20100915070255
1062 Ignore-this: 7c8e2d86a482453c7e69e22bc31cb03f
1063]
1064[Remove (most of) the FiniteMap wrapper
1065Ian Lynagh <igloo@earth.li>**20100914201703
1066 We still have
1067     insertList, insertListWith, deleteList
1068 which aren't in Data.Map, and
1069     foldRightWithKey
1070 which works around the fold(r)WithKey addition and deprecation.
1071]
1072[Improve ASSERT
1073simonpj@microsoft.com**20100914113900
1074 Ignore-this: dbc0363be5924f543316e77f7d18dd77
1075]
1076[Comment on what an "enumeration" type is
1077simonpj@microsoft.com**20100914113850
1078 Ignore-this: c09c8591e3140f305d55fbf945adbf95
1079]
1080[Make absent-arg wrappers work for unlifted types (fix Trac #4306)
1081simonpj@microsoft.com**20100914113827
1082 Ignore-this: 1945e56779329e8b79780403710aba98
1083 
1084 Previously we were simply passing arguments of unlifted
1085 type to a wrapper, even if they were absent, which was
1086 stupid.
1087 
1088 See Note [Absent error Id] in WwLib.
1089]
1090[Comments only
1091simonpj@microsoft.com**20100914113641
1092 Ignore-this: 3191ce856c9b5d9700cedc9b149b8097
1093]
1094[Move error-ids to MkCore (from PrelRules)
1095simonpj@microsoft.com**20100914113635
1096 Ignore-this: c3d820db62ba6139dd7c96bf97e51bb5
1097 
1098 and adjust imports accordingly
1099]
1100[More wibbles to deriving error messages
1101simonpj@microsoft.com**20100914113523
1102 Ignore-this: bd2df662644961138fa209aec843a2aa
1103]
1104[Fix getThreadCPUTime()
1105Simon Marlow <marlowsd@gmail.com>**20100913153838
1106 Ignore-this: 950e048a5724086534b74c609c7d5ed
1107 ever since the patch "Check with sysconf _POSIX_THREAD_CPUTIME", it
1108 has been returning incorrect results, because the sysconf variable to
1109 check should have been _SC_THREAD_CPUTIME, not _POSIX_THREAD_CPUTIME.
1110]
1111[filter out the gcc-lib directory from the rts package's library-dirs
1112Simon Marlow <marlowsd@gmail.com>**20100913101259
1113 Ignore-this: 46dc1dccbfee8a65f9243e125eee117f
1114 fixes problems when building with GHC 6.10 on Windows
1115]
1116[Don't include GC time in heap profiles (#4225)
1117Simon Marlow <marlowsd@gmail.com>**20100913133852
1118 Ignore-this: 68ac48b004b311384b5996c6b33ba5cc
1119]
1120[Use clock_gettime (if available) to measure the process CPU time
1121Simon Marlow <marlowsd@gmail.com>**20100913133818
1122 Ignore-this: 8c9300df9b929bfc1db4713c9b6065b3
1123 This is much more accurate than getrusage, which was giving misleading
1124 results when trying to time very quick operations like a minor GC.
1125]
1126[make stg_arg_bitmaps public, and available via the GHCi linker (#3672)
1127Simon Marlow <marlowsd@gmail.com>**20100913105235
1128 Ignore-this: e18efd0bd77c521e5530fb59e93b5a42
1129]
1130[fix typo
1131Simon Marlow <marlowsd@gmail.com>**20100913105100
1132 Ignore-this: 6049eea21208864203b2d79db2edd143
1133]
1134[Update release notes and docs with LLVM info.
1135David Terei <davidterei@gmail.com>**20100914072135
1136 Ignore-this: 5b3d0e5c9d5da98ed6ae9c2e8e1f6f30
1137]
1138[Remove defaultExtensionFlags
1139Ian Lynagh <igloo@earth.li>**20100913165949
1140 The default should do into languageExtensions instead
1141]
1142[Improve crash message
1143simonpj@microsoft.com**20100913170407
1144 Ignore-this: 5c26a9979f18be8cd12cea823c9f4b5a
1145]
1146[Fix Trac #4302, plus a little refactoring
1147simonpj@microsoft.com**20100913170355
1148 Ignore-this: cf6886b587aa0e8d723362183625d946
1149]
1150[Fix build with 6.10
1151Ian Lynagh <igloo@earth.li>**20100913160048]
1152[Haddock fixes
1153simonpj@microsoft.com**20100913120510
1154 Ignore-this: f3157d6969f10d4cbd593000a477138b
1155]
1156[Remove two old junk files
1157simonpj@microsoft.com**20100913103426
1158 Ignore-this: ed7af5ef1b9592178909a8d876345302
1159]
1160[Super-monster patch implementing the new typechecker -- at last
1161simonpj@microsoft.com**20100913095048
1162 Ignore-this: 14d14a1e4d7a414f5ae8d9d89d1c6a4b
1163 
1164 This major patch implements the new OutsideIn constraint solving
1165 algorithm in the typecheker, following our JFP paper "Modular type
1166 inference with local assumptions". 
1167 
1168 Done with major help from Dimitrios Vytiniotis and Brent Yorgey.
1169 
1170]
1171[Fix simplifier statistics
1172simonpj@microsoft.com**20100909085441
1173 Ignore-this: 48e383655aafc912dea15c4d94382863
1174]
1175[Trace output
1176simonpj@microsoft.com**20100908170056
1177 Ignore-this: 4b67fa4b310fbf0a16b852686d2d3294
1178]
1179[Better debug output
1180simonpj@microsoft.com**20100908170047
1181 Ignore-this: 410cef00616dda7c0c162f65216e8ca3
1182]
1183[Add Outputable instance for OccEncl
1184simonpj@microsoft.com**20100908150510
1185 Ignore-this: 6362ef9028287d84f070eaf8963c1bfc
1186]
1187[Better simplifier counting
1188simonpj@microsoft.com**20100907214840
1189 Ignore-this: 9d4722703f8f47447e86a28c8c50e0ea
1190]
1191[Put liftStringName into the known-key names
1192simonpj@microsoft.com**20100906112415
1193 Ignore-this: 287064d14ff484da1a6dea6924bc9235
1194]
1195[Deprecate NoRelaxedPolyRec
1196simonpj@microsoft.com**20100903234519
1197 Ignore-this: 607217e77f6bc1b91bf57dfd8dd2b967
1198]
1199[Buglet in Core Lint
1200simonpj@microsoft.com**20100903234457
1201 Ignore-this: 277535d51b396d3b4b0265a0939c2d4
1202]
1203[Give seqId the right type
1204simonpj@microsoft.com**20100903093556
1205 Ignore-this: d1fc7a73dea160614a8d4ddc930f99cd
1206]
1207[Remove dead code
1208simonpj@microsoft.com**20100903093548
1209 Ignore-this: 92cc3f7651445aa349ee7f114d3ec758
1210]
1211[Comments and layout
1212simonpj@microsoft.com**20100903093502
1213 Ignore-this: 9987d1409e654992c1cb1be35cb87728
1214]
1215[Remove checkFreeness; no longer needed
1216simonpj@microsoft.com**20100902233227
1217 Ignore-this: c96a12ac9794290aa30402317d88c095
1218]
1219[Assert
1220simonpj@microsoft.com**20100902073642
1221 Ignore-this: 4be1ab2f6096665ae5ec7fdd1f025a67
1222]
1223[Add aserts
1224simonpj@microsoft.com**20100902073211
1225 Ignore-this: e1409441217fd070c5a7f9ee4cca99ab
1226]
1227[Wibbles
1228simonpj@microsoft.com**20100831113540
1229 Ignore-this: 903811ab493a7b560a62eb86fcf3ee25
1230]
1231[Wibble to allow phantom types in Enum
1232simonpj@microsoft.com**20100825112711
1233 Ignore-this: fdef1c50d92b4a3d46bbe4cbfd8a83ea
1234]
1235[Add HsCoreTy to HsType
1236simonpj@microsoft.com**20100824141845
1237 Ignore-this: 4ca742b099f9cc90af3167f1012dbba6
1238 
1239 The main thing here is to allow us to provide type
1240 signatures for 'deriving' bindings without pain.
1241]
1242[Comments
1243simonpj@microsoft.com**20100823223654
1244 Ignore-this: dd412a55839430c436902d8699d6900b
1245]
1246[Wibbles to error message
1247simonpj@microsoft.com**20100823163308
1248 Ignore-this: 4d6cd8e613762dca8135c2e3b09264ec
1249]
1250[Correct type signatures
1251simonpj@microsoft.com**20100823153045
1252 Ignore-this: 42942309221a443258246098f9c0a13b
1253]
1254[Add missing signatures
1255simonpj@microsoft.com**20100823112413
1256 Ignore-this: 8ee1ce40456306de469938c02df4fed5
1257]
1258[Add type signatures in "deriving" bindings
1259simonpj@microsoft.com**20100820234230
1260 Ignore-this: 4726b28968cf65ec16cb65b7e0e7303e
1261]
1262[Minor
1263dimitris@microsoft.com**20100820131021]
1264[Be a bit less aggressive in mark-many inside a cast
1265simonpj@microsoft.com**20100819104804
1266 Ignore-this: 3fd48fe7647ec7a58c2032cd86ca4d4f
1267]
1268[Wibble
1269simonpj@microsoft.com**20100818185738
1270 Ignore-this: d5c939311377c0d0c7244aa339193315
1271]
1272[Pretty printing change
1273simonpj@microsoft.com**20100818065436
1274 Ignore-this: 4f7e70976dbe52f95effb3e634dfef5d
1275]
1276[Remember to zonk FlatSkols!
1277simonpj@microsoft.com**20100811143555
1278 Ignore-this: 84f7f9dbda97f561a918c69308ddef9a
1279]
1280[De-polymorphise
1281simonpj@microsoft.com**20100730151217
1282 Ignore-this: a9304487b983e517a9083fd697f77576
1283]
1284[Work around missing type signature in Happy
1285simonpj@microsoft.com**20100730122405
1286 Ignore-this: 7f241a655d93c5ad7763a7ffe8db0c7a
1287 
1288 Happy generates
1289 
1290       notHappyAtAll = error "Blah"
1291 
1292 without a type signature, and currently the new
1293 typechecker doesn't generalise it.  This patch
1294 says "no monomorphism restriction" which makes it
1295 generalise again.
1296 
1297 Better would be to add a type sig to Happy's template
1298]
1299[Add two local type signatures
1300simonpj@microsoft.com**20100729152611
1301 Ignore-this: afa99bcc515469aa0990d44d8c18a9e6
1302]
1303[Second test from Simon's laptop
1304simonpj@microsoft.com**20100729091703
1305 Ignore-this: 4dc64cadae314a5a1b05cc5326918a83
1306]
1307[Test commit from Simon's laptop
1308simonpj@microsoft.com**20100729091344
1309 Ignore-this: 109eff835cc19e9f93799d12f09b0ba7
1310]
1311[Add OutsideIn flag
1312simonpj@microsoft.com**20100728075525
1313 Ignore-this: 69c2f5c3a15fa653f6da80598aa8d74d
1314]
1315[Layout only
1316simonpj@microsoft.com**20100727141539
1317 Ignore-this: 1a58a8fe80ba8bced18ae81a2efb9495
1318]
1319[Improvement to SimplUtils.mkLam
1320simonpj@microsoft.com**20100727131659
1321 Ignore-this: 739beaefa79baa7e0ebeb5b2b6d1ea91
1322]
1323[Give the correct kind to unsafeCoerce#
1324simonpj@microsoft.com**20100727131538
1325 Ignore-this: 6b787de3b398c6d7a61fa04fccd15fd6
1326]
1327[Suppress warnings about recursive INLINE in output of desugarer
1328simonpj@microsoft.com**20100727094549
1329 Ignore-this: a361f7238c0fcba526d46326722c42e
1330]
1331[Rename CorePrep.tryEtaReduce to tryEtaReducePrep
1332simonpj@microsoft.com**20100726231253
1333 Ignore-this: 4375ddace205745244ba224ae012252
1334 
1335 This avoids the name clash with the similar but
1336 not identical CoreUtils.tryEtaReduce
1337]
1338[Add a trace message
1339simonpj@microsoft.com**20100719211111
1340 Ignore-this: b5daebe46e50c8cf28cc693f84bbf099
1341]
1342[Don't use RelaxedPolyRec in the compiler; it's built in now
1343simonpj@microsoft.com**20100719170441
1344 Ignore-this: a2e4489cdf63478e46282a421ee7aec3
1345]
1346[Remove duplicated #defines for FreeBSD
1347Matthias Kilian <kili@outback.escape.de>**20100912181518
1348 Ignore-this: d16214fef8635c7c9ef4edec4e8e7896
1349]
1350[Don't fail with absolute silence
1351Matthias Kilian <kili@outback.escape.de>**20100912150506
1352 Ignore-this: 479e2321f39b263fa2d9f80491e5e9f7
1353]
1354[Add a release note: "-dynload wrapper" removed
1355Ian Lynagh <igloo@earth.li>**20100911195809]
1356[put back the conversion of warn-lazy-unlifted-bindings into an error until 7.2
1357Ian Lynagh <igloo@earth.li>**20100911193434
1358 I think we'll currently still have too many people with old versions of
1359 alex/happy to want to make this an error now.
1360]
1361[6.14 -> 7.0
1362Ian Lynagh <igloo@earth.li>**20100911192837]
1363[Add a couple more release notes
1364Ian Lynagh <igloo@earth.li>**20100911162059]
1365[Document -dsuppress-module-prefixes
1366Ian Lynagh <igloo@earth.li>**20100911162005]
1367[Enable -fregs-graph with -O2; fixes #2790
1368Ian Lynagh <igloo@earth.li>**20100910191301]
1369[Remove unused code
1370Ian Lynagh <igloo@earth.li>**20100909170207]
1371[Fix warnings
1372Ian Lynagh <igloo@earth.li>**20100909154348]
1373[Fix warnings
1374Ian Lynagh <igloo@earth.li>**20100909150957]
1375[Remove context completion
1376lykahb@gmail.com**20100901160153
1377 Ignore-this: dc61b259dcb7063f0c76f56788b5d2af
1378 Now completion suggests to remove only modules added to context before.
1379]
1380[avoid Foreign.unsafePerformIO
1381Ross Paterson <ross@soi.city.ac.uk>**20100909125531
1382 Ignore-this: 5cabeae4cffec8fc17ef7c0cabbea22a
1383]
1384[updates to the release notes
1385Simon Marlow <marlowsd@gmail.com>**20100909111450
1386 Ignore-this: a4d25ad8815c305b7e0f21fd4f6ee37b
1387]
1388[newAlignedPinnedByteArray#: avoid allocating an extra word sometimes
1389Simon Marlow <marlowsd@gmail.com>**20100909110805
1390 Ignore-this: 996a3c0460068ab2835b4920905b3e75
1391]
1392[Finish breaking up vectoriser utils
1393benl@ouroborus.net**20100909061311
1394 Ignore-this: 217fe1d58a3e8bb13200bcb81353a416
1395]
1396[Move VectType module to Vectorise tree
1397benl@ouroborus.net**20100909042451
1398 Ignore-this: 5af8cf394d4835911259ca3ffb6774c5
1399]
1400[Sort all the PADict/PData/PRDict/PRepr stuff into their own modules
1401benl@ouroborus.net**20100909035147
1402 Ignore-this: 53436329773347cad793adbd83e90a9e
1403]
1404[Break out Repr and PADict stuff for vectorisation of ADTs to their own modules
1405benl@ouroborus.net**20100909025759
1406 Ignore-this: d2b7d2f79332eda13416449742f7cf1c
1407]
1408[Break out conversion functions to own module
1409benl@ouroborus.net**20100909023332
1410 Ignore-this: 613f2666b6ca7f2f8876fcc1e4a59593
1411]
1412[Comments and formatting only
1413benl@ouroborus.net**20100909022117
1414 Ignore-this: c8e30139d730669e5db44f0ef491a588
1415]
1416[Remove "-dynload wrapper"; fixes trac #4275
1417Ian Lynagh <igloo@earth.li>**20100908213251]
1418[Don't set visibility on Windows
1419Ian Lynagh <igloo@earth.li>**20100905122442
1420 With gcc 4.5.0-1, using visibility hidden gives:
1421     error: visibility attribute not supported in this configuration; ignored
1422]
1423[Fix warnings on Windows
1424Ian Lynagh <igloo@earth.li>**20100905111201
1425 Ignore-this: c5cce63bb1e0c7a27271bed78d25fbc5
1426]
1427[Fix gcc wrapper for new mingw binaries
1428Ian Lynagh <igloo@earth.li>**20100905001807
1429 Ignore-this: f6acc8c911055ffce632bac138ccc939
1430]
1431[Don't pass our gcc options to stage0 ghc's gcc; they may not be suitable
1432Ian Lynagh <igloo@earth.li>**20100905103129]
1433[Update intree-mingw creation
1434Ian Lynagh <igloo@earth.li>**20100904225559]
1435[Update commands to build in-tree mingw
1436Ian Lynagh <igloo@earth.li>**20100904215112]
1437[Break out hoisting utils into their own module
1438benl@ouroborus.net**20100908074102
1439 Ignore-this: e3ba4ed0252a2def1ed88a9e14c58fea
1440]
1441[Break out closure utils into own module
1442benl@ouroborus.net**20100908072040
1443 Ignore-this: 216172b046ff101cf31a1753667a5383
1444]
1445[Move VectVar module to Vectorise tree
1446benl@ouroborus.net**20100908065904
1447 Ignore-this: 1fba5333d29927dba4275381e1a7f315
1448]
1449[Break out vectorisation of expressions into own module
1450benl@ouroborus.net**20100908065128
1451 Ignore-this: 6a952b80fb024b5291f166477eb1976
1452]
1453[Break out TyCon classifier into own module
1454benl@ouroborus.net**20100908063111
1455 Ignore-this: da754c4ef6960b4e152ea1bf8c04ab6f
1456]
1457[Break out vectorisation of TyConDecls into own module
1458benl@ouroborus.net**20100908052004
1459 Ignore-this: c0ab4fb2a05ca396efe348b384db1ebf
1460]
1461[Break out type vectorisation into own module
1462benl@ouroborus.net**20100907110311
1463 Ignore-this: 67bd70a21d16468daf68dd3ec1ff7d62
1464]
1465[Tidy up the ArchHasAdjustorSupport definition
1466Ian Lynagh <igloo@earth.li>**20100904144234]
1467[ppc: switch handling of 'foreign import wrapper' (FIW) to libffi
1468Sergei Trofimovich <slyfox@community.haskell.org>**20100829192859
1469 Ignore-this: 662ea926681ebea0759e2a04a38e82b7
1470 
1471 Joseph Jezak reported darcs-2.4.4 SIGSEGV in interactive mode in ghc-6.12.3.
1472 So I've concluded ppc also has rotten native adjustor. I don't have hardware
1473 to verify the patch (ticket #3516 should help to test it), but I think it will
1474 help (as similar patch helped for ia64 and ppc64).
1475]
1476[Binary no longer has unusable UNPACK pragmas, so no need to turn of -Werror
1477Ian Lynagh <igloo@earth.li>**20100904133339]
1478[Don't haddock packages that we aren't going to install
1479Ian Lynagh <igloo@earth.li>**20100903231921]
1480[Give haddock per-package source entity paths; fixes #3810
1481Ian Lynagh <igloo@earth.li>**20100903221335]
1482[update for containers-0.4
1483Simon Marlow <marlowsd@gmail.com>**20100903105131
1484 Ignore-this: 556eac0e4926c9b8af6b66d7b069302c
1485]
1486[Fix for nursery resizing: the first block's back pointer should be NULL
1487Simon Marlow <marlowsd@gmail.com>**20100827102818
1488 Ignore-this: fb68938e3f1e291e3c9e5e8047f9dcd2
1489 I'm not sure if this could lead to a crash or not, but it upsets +RTS -DS
1490 Might be related to #4265
1491]
1492[Add some -no-user-package-conf flags
1493Ian Lynagh <igloo@earth.li>**20100902224726
1494 Stops user-installed packages breaking the build
1495]
1496[Fix warnings: Remove unused imports
1497Ian Lynagh <igloo@earth.li>**20100902204342]
1498[Finish breaking up VectBuiltIn and VectMonad, and add comments
1499benl@ouroborus.net**20100831100724
1500 Ignore-this: 65604f3d22d03433abc12f10be40050d
1501]
1502[Fix warnings
1503benl@ouroborus.net**20100830083746
1504 Ignore-this: 2a0e000985f694582a6f9a9261ff2739
1505]
1506[Break up vectoriser builtins module
1507benl@ouroborus.net**20100830070900
1508 Ignore-this: b86bd36a7875abdcf16763902ba2e637
1509]
1510[Move VectCore to Vectorise tree
1511benl@ouroborus.net**20100830053415
1512 Ignore-this: d5763ca6424285b39a58c7792f4a84a1
1513]
1514[Split out vectoriser environments into own module
1515benl@ouroborus.net**20100830050252
1516 Ignore-this: 5319111c74831394d2c29b9aedf5a766
1517]
1518[Comments and formatting to vectoriser, and split out varish stuff into own module
1519benl@ouroborus.net**20100830042722
1520 Ignore-this: d3f0c98ed8124dd0fca9a2ccea3e15fd
1521]
1522[Fix warnings
1523benl@ouroborus.net**20100830040340
1524 Ignore-this: d6cfad803ad4617e7fdaa62e4a895282
1525]
1526[Fix warning about multiply exported name
1527benl@ouroborus.net**20100830035243
1528 Ignore-this: 27ce2c1d22d9f99929d16a426343044e
1529]
1530[Vectorisation of method types
1531benl@ouroborus.net**20100830032941
1532 Ignore-this: 75614571d5c246a4906edb3b39ab1e0b
1533]
1534[Comments and formatting to vectoriser
1535benl@ouroborus.net**20100830032516
1536 Ignore-this: fe665b77108501c7960d858be3290761
1537]
1538[Implement -dsuppress-module-prefixes
1539benl@ouroborus.net**20100830032428
1540 Ignore-this: 2bb8bad9c60ef9044132bba118010687
1541]
1542[Whitespace only
1543benl@ouroborus.net**20100527045629
1544 Ignore-this: 4c160dfa77727e659817b6af9c84684a
1545]
1546[Disambiguate a function name
1547Ian Lynagh <igloo@earth.li>**20100828225827]
1548[users_guide.xml is now generated
1549Ian Lynagh <igloo@earth.li>**20100828225751]
1550[Pass more -pgm flags in the ghc wrapper; fixes #3863
1551Ian Lynagh <igloo@earth.li>**20100827204537]
1552[Add a new-IO manager release note
1553Ian Lynagh <igloo@earth.li>**20100827171616]
1554[Merge a duplicate release note
1555Ian Lynagh <igloo@earth.li>**20100827171511]
1556[Typo, spotted by Johan Tibell
1557Ian Lynagh <igloo@earth.li>**20100827153914]
1558[First pass at 6.14.1 release notes
1559Ian Lynagh <igloo@earth.li>**20100826220811]
1560[Fix typo
1561Ian Lynagh <igloo@earth.li>**20100824201330]
1562[FIX BUILD: add rts_isProfiled to the symbol table
1563Simon Marlow <marlowsd@gmail.com>**20100826094319
1564 Ignore-this: 9536ddb0a94721c8dec03a2a981cfa83
1565]
1566[Fix the DPH package cleaning/profiled mess even more (the build was broken)
1567Simon Marlow <marlowsd@gmail.com>**20100826084436
1568 Ignore-this: 49d7e4db2fb53b856c213c74c8969d82
1569]
1570[Remove the debugging memory allocator - valgrind does a better job
1571Simon Marlow <marlowsd@gmail.com>**20100824113537
1572 Ignore-this: a3731a83dc18b0fd0de49452e695a7ca
1573 
1574 I got fed up with the constant bogus output from the debugging memory
1575 allocator in RtsUtils.c.  One problem is that we allocate memory in
1576 constructors that then isn't tracked, because the debugging allocator
1577 hasn't been initialised yet.
1578 
1579 The bigger problem is that for a given piece of leaking memory it's
1580 impossible to find out where it was allocated; however Valgrind gives
1581 output like this:
1582 
1583 ==6967== 8 bytes in 1 blocks are still reachable in loss record 1 of 7
1584 ==6967==    at 0x4C284A8: malloc (vg_replace_malloc.c:236)
1585 ==6967==    by 0x4C28522: realloc (vg_replace_malloc.c:525)
1586 ==6967==    by 0x6745E9: stgReallocBytes (RtsUtils.c:213)
1587 ==6967==    by 0x68D812: setHeapAlloced (MBlock.c:91)
1588 ==6967==    by 0x68D8E2: markHeapAlloced (MBlock.c:116)
1589 ==6967==    by 0x68DB56: getMBlocks (MBlock.c:240)
1590 ==6967==    by 0x684F55: alloc_mega_group (BlockAlloc.c:305)
1591 ==6967==    by 0x6850C8: allocGroup (BlockAlloc.c:358)
1592 ==6967==    by 0x69484F: allocNursery (Storage.c:390)
1593 ==6967==    by 0x694ABD: allocNurseries (Storage.c:436)
1594 ==6967==    by 0x6944F2: initStorage (Storage.c:217)
1595 ==6967==    by 0x673E3C: hs_init (RtsStartup.c:160)
1596 
1597 which tells us exactly what the leaking bit of memory is.  So I don't
1598 think we need our own debugging allocator.
1599]
1600[free the entries in the thread label table on exit
1601Simon Marlow <marlowsd@gmail.com>**20100824112606
1602 Ignore-this: c9d577c06548cda80791e590e40d35b3
1603]
1604[Panic in the right way
1605simonpj@microsoft.com**20100825091614
1606 Ignore-this: e6ea4f6dfd2aea088828ea7a945ddd5f
1607]
1608[Fix the DPH/profiled make thing (again)
1609simonpj@microsoft.com**20100825091602
1610 Ignore-this: bc58fa48034ac40cf7be4170958ea29e
1611]
1612[Don't test for gcc flags before we've found gcc
1613Ian Lynagh <igloo@earth.li>**20100824131401]
1614[Change how the dblatex/lndir problem is worked around
1615Ian Lynagh <igloo@earth.li>**20100824130938
1616 Hack: dblatex normalises the name of the main input file using
1617 os.path.realpath, which means that if we're in a linked build tree,
1618 it find the real source files rather than the symlinks in our link
1619 tree. This is fine for the static sources, but it means it can't
1620 find the generated sources.
1621 
1622 We therefore also generate the main input file, so that it really
1623 is in the link tree, and thus dblatex can find everything.
1624]
1625[Clean the generated userguide sources
1626Ian Lynagh <igloo@earth.li>**20100824105827
1627 Ignore-this: 39b4f9702c688c053ed3273b20969597
1628]
1629[DPH should not even be built if GhcProfiled
1630simonpj@microsoft.com**20100823133439
1631 Ignore-this: 62acbf83de5b70ff6d27ab38ce9218ae
1632 
1633 It's not just when cleaning!
1634]
1635[The templateHaskellOk check should only run in stage2
1636simonpj@microsoft.com**20100823133353
1637 Ignore-this: f6dc9292923a1ca201953c5f58c0af3c
1638 
1639 Because rtsIsProfiled is only available in stage2
1640]
1641[Add a couple of missing tests for EAGER_BLACKHOLE
1642Simon Marlow <marlowsd@gmail.com>**20100823104654
1643 Ignore-this: 70c981b86370b0c7564b29b057650897
1644 This was leading to looping and excessive allocation, when the
1645 computation should have just blocked on the black hole. 
1646 
1647 Reported by Christian Höner zu Siederdissen <choener@tbi.univie.ac.at>
1648 on glasgow-haskell-users.
1649]
1650[Don't check for swept blocks in -DS.
1651Marco Túlio Gontijo e Silva <marcot@marcot.eti.br>**20100718225526
1652 Ignore-this: ad5dcf3c247bc19fbef5122c1142f3b2
1653 
1654 The checkHeap function assumed the allocated part of the block contained only
1655 alive objects and slops.  This was not true for blocks that are collected using
1656 mark sweep.  The code in this patch skip the test for this kind of blocks.
1657]
1658[Fix "darcs get"
1659Ian Lynagh <igloo@earth.li>**20100822183542]
1660[Add "darcs-all upstreampull"
1661Ian Lynagh <igloo@earth.li>**20100822163419
1662 This pulls from the upstream repos, for those packages which have
1663 upstreams
1664]
1665[Generate the bit in the user guide where we say what -fglasgow-exts does
1666Ian Lynagh <igloo@earth.li>**20100822155514
1667 Stops the docs going out of sync with the code.
1668]
1669[Factor out the packages file parsing in darcs-all
1670Ian Lynagh <igloo@earth.li>**20100822154813]
1671[Document --supported-extensions
1672Ian Lynagh <igloo@earth.li>**20100822134530]
1673[fix extraction of command stack of arguments of arrow "forms" (fixes #4236)
1674Ross Paterson <ross@soi.city.ac.uk>**20100822090022
1675 Ignore-this: a93db04ec4f20540642a19cdc67d1666
1676 
1677 The command stack was being extracted (by unscramble) with the outermost
1678 type first, contrary to the comment on the function.
1679]
1680[minor fix to comment
1681Ross Paterson <ross@soi.city.ac.uk>**20100822085838
1682 Ignore-this: 8d203ba2600eaf4cf21b043dcfa96cdc
1683]
1684[Add the RTS library path to the library search path
1685Ian Lynagh <igloo@earth.li>**20100820155523
1686 In case the RTS is being explicitly linked in. For #3807.
1687]
1688[Remove some duplication of C flags
1689Ian Lynagh <igloo@earth.li>**20100819233743
1690 We now use the CONF_CC_OPTS_STAGEn C flags in machdepCCOpts, rather than
1691 repeating them there.
1692]
1693[Set -fno-stack-protector in CONF_CC_OPTS_STAGE* rathre than extra-gcc-opts
1694Ian Lynagh <igloo@earth.li>**20100819233031
1695 The latter is only used when compiling .hc files, whereas we need it for
1696 .c files too.
1697]
1698[Give clearer errors for bad input in the packages file; suggested by pejo
1699Ian Lynagh <igloo@earth.li>**20100819232420]
1700[Set -march=i686 on OS X x86 in the configure variables
1701Ian Lynagh <igloo@earth.li>**20100819230939
1702 We used to set it only in machdepCCOpts, so this is more consistent
1703]
1704[Give each stage its own Config.hs
1705Ian Lynagh <igloo@earth.li>**20100819224709
1706 This also means the file is generated in a dist directory, not a
1707 source directory.
1708]
1709[Fix cleaning when GhcProfiled = YES
1710Ian Lynagh <igloo@earth.li>**20100819131346
1711 We need to include the DPH cleaning rules, even though we don't build DPH
1712 when GhcProfiled = YES.
1713]
1714[stgReallocBytes(DEBUG): don't fail when the ptr passed in is NULL
1715Simon Marlow <marlowsd@gmail.com>**20100817150836
1716 Ignore-this: 4b5063e65e01399f64a33f0d0555ff38
1717]
1718[Use make-command in rules/bindist.mk
1719Ian Lynagh <igloo@earth.li>**20100818191243
1720 Rather than it having its own specialised version
1721]
1722[Use make-command when installing packages
1723Ian Lynagh <igloo@earth.li>**20100818190600]
1724[Add _DATA_FILES to package-data.mk files
1725Ian Lynagh <igloo@earth.li>**20100818185801]
1726[Add a "make-command" utility Makefile function
1727Ian Lynagh <igloo@earth.li>**20100818183055]
1728[LLVM: Nicer format for lack of shared lib warning
1729David Terei <davidterei@gmail.com>**20100817145207
1730 Ignore-this: 753d45762601d87761614937a1bb6716
1731]
1732[fix FP_CHECK_ALIGNMENT for autoconf 2.66 (fixes #4252)
1733Ross Paterson <ross@soi.city.ac.uk>**20100816142442
1734 Ignore-this: cd784b8888d32b3b2cc2cc0969ec40f
1735 
1736 Recent versions of AS_LITERAL_IF don't like *'s.  Fix from
1737 
1738 http://blog.gmane.org/gmane.comp.sysutils.autoconf.general/month=20100701
1739]
1740[Refactor the command-line argument parsing (again)
1741simonpj@microsoft.com**20100816074453
1742 Ignore-this: 26dc9e37a88660a887a2e316ed7a9803
1743 
1744 This change allows the client of CmdLineParser a bit more flexibility,
1745 by giving him an arbitrary computation (not just a deprecation
1746 message) for each flag. 
1747 
1748 There are several clients, so there are lots of boilerplate changes.
1749 
1750 Immediate motivation: if RTS is not profiled, we want to make
1751 Template Haskell illegal.  That wasn't with the old setup.
1752]
1753[Add upstream repo to the packages file
1754Ian Lynagh <igloo@earth.li>**20100815154741]
1755[Make the "tag" column of the packages file always present
1756Ian Lynagh <igloo@earth.li>**20100815151657
1757 It makes the parsing simpler if we always have the same number of columns
1758]
1759[Disable object splitting on OSX; works around #4013
1760Ian Lynagh <igloo@earth.li>**20100815134759]
1761[Return memory to the OS; trac #698
1762Ian Lynagh <igloo@earth.li>**20100813170402]
1763[Reduce the xargs -s value we use on Windows
1764Ian Lynagh <igloo@earth.li>**20100812223721
1765 With 30000 I was getting:
1766     xargs: value for -s option should be < 28153
1767]
1768[LLVM: Enable shared lib support on Linux x64
1769David Terei <davidterei@gmail.com>**20100813191534
1770 Ignore-this: 642ed37af38e5f17d419bf4f09332671
1771]
1772[Re-do the arity calculation mechanism again (fix Trac #3959)
1773simonpj@microsoft.com**20100813161151
1774 Ignore-this: d4a2aa48150b503b20c25351a79decfb
1775 
1776 After rumination, yet again, on the subject of arity calculation,
1777 I have redone what an ArityType is (it's purely internal to the
1778 CoreArity module), and documented it better.  The result should
1779 fix #3959, and I hope the related #3961, #3983.
1780 
1781 There is lots of new documentation: in particular
1782  * Note [ArityType] 
1783    describes the new datatype for arity info
1784 
1785  * Note [State hack and bottoming functions]
1786    says how bottoming functions are dealt with, particularly
1787    covering catch# and Trac #3959
1788 
1789 I also found I had to be careful not to eta-expand single-method
1790 class constructors; see Note [Newtype classes and eta expansion].
1791 I think this part could be done better, but it works ok.
1792]
1793[Comments only
1794simonpj@microsoft.com**20100813161019
1795 Ignore-this: baf68300d8bc630bf0b7ab27647b33a0
1796]
1797[Modify FloatOut to fix Trac #4237
1798simonpj@microsoft.com**20100813163120
1799 Ignore-this: ffc8d00d4b7f0a8a785fcef312900413
1800 
1801 The problem was that a strict binding was getting floated
1802 out into a letrec. This only happened when profiling was
1803 on.  It exposed a fragility in the floating strategy.  This
1804 patch makes it more robust.  See
1805       Note [Avoiding unnecessary floating]
1806]
1807[Fix egregious bug in SetLevels.notWorthFloating
1808simonpj@microsoft.com**20100813161429
1809 Ignore-this: d22865f48d417e6a6b732de3dfba378f
1810 
1811 This bug just led to stupid code, which would
1812 later be optimised away, but better not to generate
1813 stupid code in the first place.
1814]
1815[Delete GhcLibProfiled
1816simonpj@microsoft.com**20100813140152
1817 Ignore-this: 2e1a3f677308be726bd022f45e2fd856
1818 
1819 Simon M and I looked at this, and we think GhcLibProfiled is
1820 (a) not needed (b) confusing.
1821 
1822 Ian should review.
1823 
1824 Really, if GhcProfiled is on we should also
1825 check that 'p' is in the GhcLibWays
1826]
1827[Do not build DPH when GhcProfiled (fixes #4172)
1828simonpj@microsoft.com**20100813140021
1829 Ignore-this: 9e20181643b456e841f845ae0cab0a9a
1830 
1831 Reason: DPH uses Template Haskell and TH doesn't work
1832 in a profiled compiler
1833]
1834[Fix Trac #4220
1835simonpj@microsoft.com**20100812131319
1836 Ignore-this: 33141cfd81627592150a9e5973411ff8
1837 
1838 For deriving Functor, Foldable, Traversable with empty
1839 data cons I just generate a null equation
1840    f _ = error "urk"
1841 
1842 There are probably more lurking (eg Enum) but this will do for now.
1843]
1844[Improve the Specialiser, fixing Trac #4203
1845simonpj@microsoft.com**20100812131133
1846 Ignore-this: 482afbf75165e24a80527a6e52080c07
1847 
1848 Simply fixing #4203 is a tiny fix: in case alterantives we should
1849 do dumpUDs *including* the case binder. 
1850 
1851 But I realised that we can do better and wasted far too much time
1852 implementing the idea.  It's described in
1853     Note [Floating dictionaries out of cases]
1854]
1855[Comments
1856simonpj@microsoft.com**20100812101456
1857 Ignore-this: 6362fe887d25688c12ef2c3cf5554ce4
1858]
1859[Comments only
1860simonpj@microsoft.com**20100812101439
1861 Ignore-this: 7ed2f5fc08811cbe9958c2309a9ed1fa
1862]
1863[Fix bug in linting of shadowed case-alternative binders
1864simonpj@microsoft.com**20100812101413
1865 Ignore-this: 9212a5e2c03421749f5935b3944ecf53
1866]
1867[Comments and spacing only
1868simonpj@microsoft.com**20100812101347
1869 Ignore-this: ed59a7dae7decb24470709dc1c118dbb
1870]
1871[Add more info to more parse error messages (#3811)
1872Ian Lynagh <igloo@earth.li>**20100809233108]
1873[Run finalizers *after* updating the stable pointer table (#4221)
1874Simon Marlow <marlowsd@gmail.com>**20100810133739
1875 Ignore-this: b0462f80dd64eac71e599d8a9f6dd665
1876 Silly bug really, we were running the C finalizers while the StablePtr
1877 table was still in a partially-updated state during GC, but finalizers
1878 are allowed to call freeStablePtr() (via hs_free_fun_ptr(), for
1879 example), and chaos ensues.
1880]
1881[Do the dependency-omitting for 'make 1' in a slightly different way
1882Simon Marlow <marlowsd@gmail.com>**20100810093446
1883 Ignore-this: af15edd3a1492cbd93111316b57e02e4
1884 
1885 I encountered a couple of things that broke after Ian's previous
1886 patch: one was my nightly build scripts that use 'make stage=2' at the
1887 top level, and the other is 'make fast' in libraries/base, which uses
1888 'stage=0' to avoid building any compilers.
1889 
1890 So my version of this patch is more direct: it just turns off the
1891 appropriate dependencies using a variable set by 'make 1', 'make 2',
1892 etc.
1893]
1894[Integrate new I/O manager, with signal support
1895Johan Tibell <johan.tibell@gmail.com>**20100724102355
1896 Ignore-this: eb092857a2a1b0ca966649caffe7ac2b
1897]
1898[Add DoAndIfThenElse support
1899Ian Lynagh <igloo@earth.li>**20100808194625]
1900[Make another parse error more informative
1901Ian Lynagh <igloo@earth.li>**20100808193340]
1902[Make a parse error say what it is failing to parse; part of #3811
1903Ian Lynagh <igloo@earth.li>**20100808155732]
1904[Send ghc progress output to stdout; fixes #3636
1905Ian Lynagh <igloo@earth.li>**20100808142542]
1906[Fix the HsColour test in the build system
1907Ian Lynagh <igloo@earth.li>**20100805155319
1908 Ignore-this: ba2752b04801a253e891b31e1914485d
1909]
1910[Fix the -lm configure test; fixes #4155
1911Ian Lynagh <igloo@earth.li>**20100805142508
1912 Ignore-this: 358b8b1074d2d22fb8d362ea6d8b80d6
1913]
1914[Don't restrict filenames in line pragmas to printable characters; fixes #4207
1915Ian Lynagh <igloo@earth.li>**20100805135011
1916 Ignore-this: e3d32312127165e40e6eaa919193d60b
1917 "printable" is ASCII-only, whereas in other locales we can get things like
1918 # 1 "<línea-de-orden>"
1919]
1920[Ensure extension flags are flattened in the Cmm phase
1921Ian Lynagh <igloo@earth.li>**20100805133614
1922 If we start with a .cmmcpp file then they don't get flattened in
1923 the CmmCpp phase, as we don't run that phase.
1924]
1925[Add "cmmcpp" as a Haskellish source suffix
1926Ian Lynagh <igloo@earth.li>**20100805132555]
1927[On amd64/OSX we don't need to be given memory in the first 31bits
1928Ian Lynagh <igloo@earth.li>**20100805120600
1929 Ignore-this: 42eb64e25ad4b66ae022884305e0297b
1930 as PIC is always on
1931]
1932[NCG: Don't worry about trying to re-freeze missing coalescences
1933benl@ouroborus.net**20100702053319
1934 Ignore-this: ea05cbee19b6c5c410db41292cbb64b0
1935]
1936[Make -rtsopts more flexible
1937Ian Lynagh <igloo@earth.li>**20100805011137
1938 The default is a new "some" state, which allows only known-safe flags
1939 that we want on by default. Currently this is only "--info".
1940]
1941[Test for (fd < 0) before trying to FD_SET it
1942Ian Lynagh <igloo@earth.li>**20100804173636]
1943[Remove "On by default" comments in DynFlags
1944Ian Lynagh <igloo@earth.li>**20100802110803
1945 Ignore-this: 2a51055277b5ce9f0e98e1438b212027
1946 These make less sense now we support multiple languges. The
1947 "languageExtensions" function gives the defaults.
1948]
1949[Fix build: Add newline to end of ghc-pkg/Main.hs
1950Ian Lynagh <igloo@earth.li>**20100801183206]
1951[Add a versions haddock binary for Windows
1952Ian Lynagh <igloo@earth.li>**20100801180917]
1953[ghc-pkg: don't fail, if a file is already removed
1954ich@christoph-bauer.net**20100725162606
1955 Ignore-this: 5501d6812c31f4da525c7fb24f6dcaed
1956]
1957[Remove push-all from file list in boot script (push-all no longer exists)
1958Ian Lynagh <igloo@earth.li>**20100801121841
1959 Ignore-this: eec130f06610d8728a57626682860a1a
1960]
1961[Add error checking to boot-pkgs script
1962Ian Lynagh <igloo@earth.li>**20100801121432
1963 Ignore-this: 8afd6663db443c774bad45d75bbfe950
1964]
1965[Add more error checking to the boot script
1966Ian Lynagh <igloo@earth.li>**20100801113628]
1967[Remove libHSrtsmain.a before creating it
1968Ian Lynagh <igloo@earth.li>**20100801005432
1969 Otherwise it isn't updated properly if rts/Main.c changes
1970]
1971[Expose the functions haddock needs even when haddock is disabled; #3558
1972Ian Lynagh <igloo@earth.li>**20100731115506]
1973[Always haddock by default
1974Ian Lynagh <igloo@earth.li>**20100730235001
1975 Revert this patch:
1976     Matthias Kilian <kili@outback.escape.de>**20090920181319
1977     Don't build haddock if HADDOC_DOCS = NO, and disable HADDOC_DOCS
1978         if GhcWithInterpreter = NO
1979     Haddock uses TcRnDriver.tcRnGetInfo, which is only available if
1980     GHCI is built. Set HADDOC_DOCS to NO if GhcWithInterpreter is NO,
1981     and disable the haddock build if HADDOC_DOCS = NO.
1982]
1983[Add a debugTrace for the phases that we run
1984Ian Lynagh <igloo@earth.li>**20100729201503]
1985[* Add StringPrimL as a constructor for Template Haskell (Trac #4168)
1986simonpj@microsoft.com**20100730131922
1987 Ignore-this: 520d0a0a14b499b299e8b2be8d148ff0
1988   
1989 There are already constructors for IntPrim, FloatPrim etc,
1990 so this makes it more uniform.
1991   
1992 There's a corresponding patch for the TH library
1993]
1994[Add thread affinity support for FreeBSD
1995Gabor Pali <pgj@FreeBSD.org>**20100720001409
1996 Ignore-this: 6c117b8219bfb45445089e82ee470410
1997 - Implement missing functions for setting thread affinity and getting real
1998   number of processors.
1999 - It is available starting from 7.1-RELEASE, which includes a native support
2000   for managing CPU sets.
2001 - Add __BSD_VISIBLE, since it is required for certain types to be visible in
2002   addition to POSIX & C99.
2003]
2004[Disable symbol visibility pragmas for FreeBSD
2005Ian Lynagh <igloo@earth.li>**20100729012507
2006 Do not use GCC pragmas for controlling visibility, because it causes
2007 "undefined reference" errors at link-time.  The true reasons are
2008 unknown, however FreeBSD 8.x includes GCC 4.2.1 in the base system,
2009 which might be buggy.
2010]
2011[Fix numeric escape sequences parsing
2012Anton Nikishaev <anton.nik@gmail.com>**20100721194208
2013 Ignore-this: dd71935b1866b5624f7975c45ad519a1
2014 This fixes trac bug #1344
2015]
2016[Explicitly give the right path to perl when making the OS X installer; #4183
2017Ian Lynagh <igloo@earth.li>**20100728163030]
2018[Set -fno-stack-protector in extra-gcc-opts; fixes #4206
2019Ian Lynagh <igloo@earth.li>**20100728161957
2020 We were using it only when building the RTS, and only on certain
2021 platforms. However, some versions of OS X need the flag, while others
2022 don't support it, so we now test for it properly.
2023]
2024[Make PersistentLinkerState fields strict; fixes #4208
2025Ian Lynagh <igloo@earth.li>**20100727201911
2026 Ignore-this: fc5cfba48cd16624f6bb15a7a03a3b4
2027 We modify fields a lot, so we retain the old value if they aren't forced.
2028]
2029[Don't rebuild dependency files unnecessarily when doing "make 1" etc
2030Ian Lynagh <igloo@earth.li>**20100726211512
2031 Ignore-this: d91a729e5113aa964cc67768e92e57ef
2032]
2033[LLVM: If user specifies optlo, don't use '-O' levels
2034David Terei <davidterei@gmail.com>**20100726105650
2035 Ignore-this: e05e103b09d1de937540ffad7983f88e
2036]
2037[Flatten flags for ghci's :show
2038Ian Lynagh <igloo@earth.li>**20100725135320]
2039[Add support for Haskell98 and Haskell2010 "languages"
2040Ian Lynagh <igloo@earth.li>**20100724230121]
2041[Rename "language" varibles etc to "extension", and add --supported-extensions
2042Ian Lynagh <igloo@earth.li>**20100724223624]
2043[Separate language option handling into 2 phases
2044Ian Lynagh <igloo@earth.li>**20100724212013
2045 We now first collect the option instructions (from the commandline,
2046 from pragmas in source files, etc), and then later flatten them into
2047 the list of enabled options. This will enable us to use different
2048 standards (H98, H2010, etc) as a base upon which to apply the
2049 instructions, when we don't know what the base will be when we start
2050 collecting instructions.
2051]
2052[Separate the language flags from the other DynFlag's
2053Ian Lynagh <igloo@earth.li>**20100724133103
2054 Ignore-this: 47bb8d42e621e47016b66c7472bd6cb5
2055]
2056[Set stage-specific CC/LD opts in the bindist configure.ac
2057Ian Lynagh <igloo@earth.li>**20100724113748
2058 Ignore-this: f06926d185a35ddd05490ca4a257e992
2059]
2060[Use different CC/LD options for different stages
2061Ian Lynagh <igloo@earth.li>**20100723223059]
2062[Add some error belchs to the linker, when we find bad magic numbers
2063Ian Lynagh <igloo@earth.li>**20100723200822]
2064[Add some more linker debugging prints
2065Ian Lynagh <igloo@earth.li>**20100723180237]
2066[When (un)loading an object fails, say which object in teh panic
2067Ian Lynagh <igloo@earth.li>**20100723162649]
2068[Add a release note: GHCi import syntax
2069Ian Lynagh <igloo@earth.li>**20100721193647]
2070[Deprecate NewQualifiedOperators extension (rejected by H')
2071Ian Lynagh <igloo@earth.li>**20100719150909
2072 Ignore-this: 6e7e3bedc5360c5975f73497b3e6cba5
2073]
2074[LLVM: Allow optlc and optlo to override default params for these systools
2075David Terei <davidterei@gmail.com>**20100722181631
2076 Ignore-this: e60af7941996f7170fb3bfb02a002082
2077]
2078[LLVM: Code and speed improvement to dominateAllocs pass.
2079David Terei <davidterei@gmail.com>**20100721143654
2080 Ignore-this: 9fb7058c8a2afc005521298c7b8d0036
2081]
2082[Comments only
2083simonpj@microsoft.com**20100721144257
2084 Ignore-this: b3091ddcd1df271eb85fe90978ab7adc
2085]
2086[Fix inlining for default methods
2087simonpj@microsoft.com**20100721144248
2088 Ignore-this: 61a11a8f741f775000c6318aae4b3191
2089 
2090 This was discombobulated by a patch a week ago;
2091 now fixed, quite straightforwardly.  See
2092 Note [Default methods and instances]
2093]
2094[Allow reification of existentials and GADTs
2095simonpj@microsoft.com**20100721090437
2096 Ignore-this: 20f1ccd336cc25aff4d4d67a9ac2211a
2097 
2098 It turns out that TH.Syntax is rich enough to express even GADTs,
2099 provided we express them in equality-predicate form.  So for
2100 example
2101 
2102   data T a where
2103      MkT1 :: a -> T [a]
2104      MkT2 :: T Int
2105 
2106 will appear in TH syntax like this
2107 
2108   data T a = forall b. (a ~ [b]) => MkT1 b
2109            | (a ~ Int) => MkT2
2110 
2111 While I was at it I also improved the reification of types,
2112 so that we use TH.TupleT and TH.ListT when we can.
2113]
2114[add numSparks# primop (#4167)
2115Simon Marlow <marlowsd@gmail.com>**20100720153746
2116 Ignore-this: f3f925e7de28f3f895213aefbdbe0b0f
2117]
2118[LLVM: Decrease max opt level used under OSX to avoid bug
2119David Terei <davidterei@gmail.com>**20100720160938
2120 Ignore-this: 34b0b3550f00b27b00ad92f8232745e5
2121 
2122 Currently, many programs compiled with GHC at -O2 and LLVM
2123 set to -O3 will segfault (only under OSX). Until this issue
2124 is fixed I have simply 'solved' the segfault by lowering
2125 the max opt level for LLVM used to -O2 under OSX.
2126 
2127 All these recent changes to OSX should mean its finally as
2128 stable as Linux and Windows.
2129]
2130[LLVM: Fix OSX to work again with TNTC disabled.
2131David Terei <davidterei@gmail.com>**20100720160845
2132 Ignore-this: 8dc98139cfa536b2a64aa364d040b581
2133]
2134[LLVM: Fix printing of local vars so LLVM works with -fnew-codegen
2135David Terei <davidterei@gmail.com>**20100720160302
2136 Ignore-this: d883c433dfaed67921a8c5360e1f9f6a
2137]
2138[Use a separate mutex to protect all_tasks, avoiding a lock-order-reversal
2139Simon Marlow <marlowsd@gmail.com>**20100716150832
2140 Ignore-this: ffbdb4ee502e0f724d57acb9bfbe9d92
2141 In GHC 6.12.x I found a rare deadlock caused by this
2142 lock-order-reversal:
2143 
2144 AQ cap->lock
2145   startWorkerTask
2146     newTask
2147       AQ sched_mutex
2148 
2149 scheduleCheckBlackHoles
2150   AQ sched_mutex
2151    unblockOne_
2152     wakeupThreadOnCapabilty
2153       AQ cap->lock
2154 
2155 so sched_mutex and cap->lock are taken in a different order in two
2156 places.
2157 
2158 This doesn't happen in the HEAD because we don't have
2159 scheduleCheckBlackHoles, but I thought it would be prudent to make
2160 this less likely to happen in the future by using a different mutex in
2161 newTask.  We can clearly see that the all_tasks mutex cannot be
2162 involved in a deadlock, becasue we never call anything else while
2163 holding it.
2164]
2165['make fast' in a package does not build any compilers
2166Simon Marlow <marlowsd@gmail.com>**20100715125904
2167 Ignore-this: f27e70faf3944831dad16e89a4e273da
2168]
2169[LLVM: Fix up botched last commit
2170David Terei <davidterei@gmail.com>**20100719104823
2171 Ignore-this: a32e0f6a38cb9e02527eb8ca69b3eb59
2172]
2173[LLVM: Fix warning introduce in last commit.
2174David Terei <davidterei@gmail.com>**20100719103411
2175 Ignore-this: e9c92a9402aff50d60ab26e6ad441bfc
2176]
2177[LLVM: Use mangler to fix up stack alignment issues on OSX
2178David Terei <davidterei@gmail.com>**20100718231000
2179 Ignore-this: 9f6e8cb855269cb3a5ac1a23480d0e71
2180]
2181[Fix #4195 (isGadtSyntaxTyCon returns opposite result)
2182illissius@gmail.com**20100715134134
2183 Ignore-this: a90403f893030432b5c15d743647f350
2184]
2185[Update to time 1.2.0.3
2186Ian Lynagh <igloo@earth.li>**20100717181810
2187 Ignore-this: 1ccb4801a73f399e6718ce556543ede1
2188]
2189[Reorder RTS --info output
2190Ian Lynagh <igloo@earth.li>**20100717162356]
2191[Fix unreg prof build: Define CCS_SYSTEM in stg/MiscClosures.h
2192Ian Lynagh <igloo@earth.li>**20100717142832
2193 Ignore-this: 9675f3f51b6dac40483155344e7f45b6
2194]
2195[Make mkDerivedConstants as a stage 1 program
2196Ian Lynagh <igloo@earth.li>**20100717000827
2197 Ignore-this: 5357403461b209b8606f1d33defb51cf
2198 This way it gets the defines for the right platform when cross-compiling
2199]
2200[Don't generate Haskell dependencies if we don't have any Haskell sources
2201Ian Lynagh <igloo@earth.li>**20100717000800
2202 Ignore-this: 454abd0358f535b7e789327125c9206c
2203]
2204[Link programs that have no Haskell objects with gcc rather than ghc
2205Ian Lynagh <igloo@earth.li>**20100716235303
2206 Ignore-this: f65588b69675edea616cc434e769b0a4
2207]
2208[Use gcc to build C programs for stages >= 1
2209Ian Lynagh <igloo@earth.li>**20100716223703
2210 Ignore-this: 9f843a4e17285cda582117504707f9e7
2211]
2212[Add platform info to "ghc --info" output
2213Ian Lynagh <igloo@earth.li>**20100716141953]
2214[Tidy up Config.hs generation
2215Ian Lynagh <igloo@earth.li>**20100716140630]
2216[Fix HC porting test in makefiles
2217Ian Lynagh <igloo@earth.li>**20100716010808
2218 Ignore-this: 6052c1dd022a6108ab2236a299ee1d84
2219 Now that we are trying to support cross compilation, we can't use
2220     "$(TARGETPLATFORM)" != "$(HOSTPLATFORM)"
2221 as a test for HC-porting.
2222]
2223[Change a BUILD var to a HOST var
2224Ian Lynagh <igloo@earth.li>**20100716002558]
2225[Remove an unnecessary #include
2226Ian Lynagh <igloo@earth.li>**20100715233930
2227 Ignore-this: dcede249de6be7e3c9305c9279c2ca07
2228]
2229[Split up some make commands, so that errors aren't overlooked
2230Ian Lynagh <igloo@earth.li>**20100715152237
2231 Ignore-this: fb69b0a25d9ca71dae5e75d38db675cd
2232 When we ask make to run "a | b", if a fails then the pipeline might
2233 still exit successfuly.
2234]
2235[Remove an unnecessary #include
2236Ian Lynagh <igloo@earth.li>**20100715143000
2237 Ignore-this: 4e098cac5dda2dd595ca0a0f5121853c
2238]
2239[Simplify some more CPP __GLASGOW_HASKELL__ tests
2240Ian Lynagh <igloo@earth.li>**20100715142500]
2241[Remove some code only used with GHC 6.11.*
2242Ian Lynagh <igloo@earth.li>**20100715141720]
2243[__GLASGOW_HASKELL__ >= 609 is now always true
2244Ian Lynagh <igloo@earth.li>**20100715141544]
2245[Correct the values in ghc_boot_platform.h
2246Ian Lynagh <igloo@earth.li>**20100714223717
2247 Ignore-this: 4c99116f7ac73fadbd6d16807f57a693
2248]
2249[Change some TARGET checks to HOST checks
2250Ian Lynagh <igloo@earth.li>**20100714184715]
2251[LLVM: Add inline assembly to binding.
2252David Terei <davidterei@gmail.com>**20100714152530
2253 Ignore-this: 72a7b5460c128ed511e8901e5889fe2b
2254]
2255[LLVM: Fix mistype in last commit which broke TNTC under win/linux.
2256David Terei <davidterei@gmail.com>**20100714153339
2257 Ignore-this: 302d7957e3dded80368ebade5312ab35
2258]
2259[Remove unnecessary #include
2260Ian Lynagh <igloo@earth.li>**20100713153704
2261 Ignore-this: c37d3127b1dc68f59270c07173994c28
2262]
2263[Change some TARGET tests to HOST tests in the RTS
2264Ian Lynagh <igloo@earth.li>**20100713141034
2265 Which was being used seemed to be random
2266]
2267[LLVM: Add in new LLVM mangler for implementing TNTC on OSX
2268David Terei <davidterei@gmail.com>**20100713183243
2269 Ignore-this: 394fb74d7f9657d8b454bd0148d24bf7
2270]
2271[Refactor where an error message is generated
2272simonpj@microsoft.com**20100713115733
2273 Ignore-this: f94467856238586fcbbe48537141cf78
2274]
2275[Comments only
2276simonpj@microsoft.com**20100713115703
2277 Ignore-this: 5815442c4e69b9ec331b34242a596253
2278]
2279[Comments on data type families
2280simonpj@microsoft.com**20100713115640
2281 Ignore-this: 90a333bb7f7d64a49fb7dd180d893f6b
2282]
2283[Fix Trac #T4136: take care with nullary symbol constructors
2284simonpj@microsoft.com**20100707135945
2285 Ignore-this: 2a717a24fefcd593ea41c23dad351db0
2286 
2287 When a nullary constructor is a symbol eg (:=:) we need
2288 to take care.  Annoying.
2289]
2290[Fix Trac #4127 (and hence #4173)
2291simonpj@microsoft.com**20100707123125
2292 Ignore-this: 98bb6d0f7182b59f8c93596c61f9785d
2293 
2294 The change involves a little refactoring, so that the default
2295 method Ids are brought into scope earlier, before the value
2296 declarations are compiled.  (Since a value decl may contain
2297 an instance decl in a quote.)
2298 
2299 See Note [Default method Ids and Template Haskell] in
2300 TcTyClsDcls.
2301]
2302[Fix second bug in Trac #4127
2303simonpj@microsoft.com**20100701140124
2304 Ignore-this: c8d1cc27364fe9ee5a52acb1ecb5cdd9
2305 
2306 This bug concerned the awkward shadowing we do for
2307 Template Haskell declaration brackets.  Lots of
2308 comments in
2309 
2310   Note [Top-level Names in Template Haskell decl quotes]
2311]
2312[ia64: switch handling of 'foreign import wrapper' (FIW) to libffi
2313Sergei Trofimovich <slyfox@community.haskell.org>**20100709213922
2314 Ignore-this: fd07687e0089aebabf62de85d2be693
2315 
2316 I tried to build darcs-2.4.4 with ghc-6.12.3 and got coredumps when darcs is used
2317 in interactive mode. I tried test from ticket #3516 and found out FIW code is broken.
2318 Instead of fixing it I just switched to libffi. Result built successfully, passed
2319 'foreign import wrapper' test from ticket #3516 and builds working darcs.
2320]
2321[* storage manager: preserve upper address bits on 64bit machines (thanks to zygoloid)
2322Sergei Trofimovich <slyfox@community.haskell.org>**20100709115917
2323 Ignore-this: 9f1958a19992091ddc2761c389ade940
2324 
2325 Patch does not touch amd64 as it's address lengts is 48 bits at most, so amd64 is unaffected.
2326 
2327 the issue: during ia64 ghc bootstrap (both 6.10.4 and 6.12.3) I
2328 got the failure on stage2 phase:
2329     "inplace/bin/ghc-stage2"   -H32m -O -H64m -O0 -w ...
2330     ghc-stage2: internal error: evacuate: strange closure type 15
2331         (GHC version 6.12.3 for ia64_unknown_linux)
2332         Please report this as a GHC bug:  http://www.haskell.org/ghc/reportabug
2333     make[1]: *** [libraries/dph/dph-base/dist-install/build/Data/Array/Parallel/Base/Hyperstrict.o] Aborted
2334 
2335 gdb backtrace (break on 'barf'):
2336 Breakpoint 1 at 0x400000000469ec31: file rts/RtsMessages.c, line 39.
2337 (gdb) run -B/var/tmp/portage/dev-lang/ghc-6.12.3/work/ghc-6.12.3/inplace/bin --info
2338 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
2339 [Thread debugging using libthread_db enabled]
2340 
2341 Breakpoint 1, barf (s=0x40000000047915b0 "evacuate: strange closure type %d") at rts/RtsMessages.c:39
2342 39        va_start(ap,s);
2343 (gdb) bt
2344 #0  barf (s=0x40000000047915b0 "evacuate: strange closure type %d") at rts/RtsMessages.c:39
2345 #1  0x400000000474a1e0 in evacuate (p=0x6000000000147958) at rts/sm/Evac.c:756
2346 #2  0x40000000046d68c0 in scavenge_srt (srt=0x6000000000147958, srt_bitmap=7) at rts/sm/Scav.c:348
2347 ...
2348 
2349 > 16:52:53 < zygoloid> slyfox: i'm no ghc expert but it looks like HEAP_ALLOCED_GC(q)
2350 >                      is returning true for a FUN_STATIC closure
2351 > 17:18:43 < zygoloid> try: p HEAP_ALLOCED_miss((unsigned long)(*p) >> 20, *p)
2352 > 17:19:12 < slyfox> (gdb) p HEAP_ALLOCED_miss((unsigned long)(*p) >> 20, *p)
2353 > 17:19:12 < slyfox> $1 = 0
2354 > 17:19:40 < zygoloid> i /think/ that means the mblock_cache is broken
2355 > 17:22:45 < zygoloid> i can't help further. however i am suspicious that you seem to have pointers with similar-looking low 33
2356 >                      bits and different high 4 bits, and it looks like such pointers get put into the same bucket in
2357 >                      mblock_cache.
2358 ...
2359 > 17:36:16 < zygoloid> slyfox: try changing the definition of MbcCacheLine to StgWord64, see if that helps
2360 > 17:36:31 < zygoloid> that's in includes/rts/storage/MBlock.h
2361 And it helped!
2362]
2363[Fixing link failure of compiler on ia64 ('-Wl,' prefixed value passed directly to ld)
2364Sergei Trofimovich <slyfox@community.haskell.org>**20100708180943
2365 Ignore-this: ced99785e1f870ee97e5bec658e2504f
2366 
2367     /usr/bin/ld -Wl,--relax -r -o dist-stage1/build/HSghc-6.10.4.o \
2368                                   dist-stage1/build/BasicTypes.o dist-stage1/build/DataCon.o ...
2369     /usr/bin/ld: unrecognized option '-Wl,--relax'
2370 
2371 If we just drop '-Wl,' part it will not help as '-r' and '--relax' are incompatible.
2372 
2373 Looks like '-Wl,--relax' was skipped by earlier binutils' ld as unknown option.
2374 Removing ia64 specific path.
2375]
2376[LLVM: Allow getelementptr to use LlvmVar for indexes.
2377David Terei <davidterei@gmail.com>**20100712152529
2378 Ignore-this: 9e158d9b89a86bca8abf11d082328278
2379]
2380[Move all the warning workarounds to one place
2381Ian Lynagh <igloo@earth.li>**20100710161723]
2382[xhtml is now warning-free
2383Ian Lynagh <igloo@earth.li>**20100710144635]
2384[Move a bit of build system code
2385Ian Lynagh <igloo@earth.li>**20100709224534]
2386[adapt to the new async exceptions API
2387Simon Marlow <marlowsd@gmail.com>**20100709125238
2388 Ignore-this: 55d845e40b9daed3575c1479d8dda1d5
2389]
2390[quiet some new spewage
2391Simon Marlow <marlowsd@gmail.com>**20100709091521
2392 Ignore-this: de7f91976bbc9789e6fd7091f05c25c0
2393]
2394[New asynchronous exception control API (ghc parts)
2395Simon Marlow <marlowsd@gmail.com>**20100708144851
2396 Ignore-this: 56320c5fc61ae3602d586609387aae22
2397 
2398 As discussed on the libraries/haskell-cafe mailing lists
2399   http://www.haskell.org/pipermail/libraries/2010-April/013420.html
2400 
2401 This is a replacement for block/unblock in the asychronous exceptions
2402 API to fix a problem whereby a function could unblock asynchronous
2403 exceptions even if called within a blocked context.
2404 
2405 The new terminology is "mask" rather than "block" (to avoid confusion
2406 due to overloaded meanings of the latter).
2407 
2408 In GHC, we changed the names of some primops:
2409 
2410   blockAsyncExceptions#   -> maskAsyncExceptions#
2411   unblockAsyncExceptions# -> unmaskAsyncExceptions#
2412   asyncExceptionsBlocked# -> getMaskingState#
2413 
2414 and added one new primop:
2415 
2416   maskUninterruptible#
2417 
2418 See the accompanying patch to libraries/base for the API changes.
2419]
2420[remove outdated comment
2421Simon Marlow <marlowsd@gmail.com>**20100708100840
2422 Ignore-this: afb2e9f6fe1f1acda51b0cbdf2637176
2423]
2424[remove 'mode: xml' emacs settings (#2208)
2425Simon Marlow <marlowsd@gmail.com>**20100708100817
2426 Ignore-this: 3a8d997fb90e01ca88dc47fb95feeba0
2427]
2428[typo in comment
2429Simon Marlow <marlowsd@gmail.com>**20100616111359
2430 Ignore-this: d3ef9288d6d6b9ab3bacbe09e0d9801c
2431]
2432[Win32 getProcessElapsedTime: use a higher-resolution time source
2433Simon Marlow <marlowsd@gmail.com>**20100708093223
2434 Ignore-this: 821989d4ff7ff2bff40cee71a881521c
2435 QueryPerformanceCounter() on Windows gives much better resolution than
2436 GetSystemTimeAsFileTime().
2437]
2438[alpha: switch handling of 'foreign import wrapper' (FIW) to libffi
2439Sergei Trofimovich <slyfox@community.haskell.org>**20100708065318
2440 Ignore-this: ddee15876737a6aa7f6dabc8ff79ce0d
2441 
2442 I tried to build ghc-6.12.3 and found out FIW part of code
2443 does not compile anymore. It uses absent functions under #ifdef.
2444 Instead of fixing it I just switched to libffi. Result built successfully
2445 and passed 'foreign import wrapper' test I wrote for trac ticket #3516.
2446 
2447 I didn't try to build -HEAD yet, but this patch only removes code, so
2448 it should not make -HEAD worse.
2449]
2450[Reorder the CPP flags so -optP can override the platform defines
2451Ian Lynagh <igloo@earth.li>**20100708203523]
2452[Add docs for DatatypeContexts extension
2453Ian Lynagh <igloo@earth.li>**20100707230907
2454 Ignore-this: 8158f03b35a2d7442a75fe85d6f1b1c7
2455]
2456[Make datatype contexts an extension (on by default) (DatatypeContexts)
2457Ian Lynagh <igloo@earth.li>**20100707212529
2458 Ignore-this: 6885ff510a0060610eeeba65122caef5
2459]
2460[LLVM: Fix various typos in comments
2461David Terei <davidterei@gmail.com>**20100707220448
2462 Ignore-this: 1ba3e722f150492da2f9d485c5795e80
2463]
2464[Handle haddock headers when looking for LANGUAGE/OPTIONS_GHC pragmas
2465Ian Lynagh <igloo@earth.li>**20100707120423
2466 Ignore-this: a75aa67690284a6cee3e62c943d4fd01
2467]
2468[Make pragState call mkPState, rather than duplicating everything
2469Ian Lynagh <igloo@earth.li>**20100706173007
2470 Ignore-this: 61fe24b99dbe7a42efff1a9dd703a75c
2471 This also means that extsBitmap gets set, whereas is was just being set
2472 to 0 before.
2473]
2474[LLVM: Add alias type defenitions to LlvmModule.
2475David Terei <davidterei@gmail.com>**20100707142053
2476 Ignore-this: eee6ad5385563ccf08e664d2634a03f2
2477]
2478[LLVM: Use packed structure type instead of structure type
2479David Terei <davidterei@gmail.com>**20100707120320
2480 Ignore-this: a06e0359d182291b81cae56993ca385e
2481 
2482 The regular structure type adds padding to conform to the platform ABI,
2483 which causes problems with structures storing doubles under windows since
2484 we don't conform to the platform ABI there. So we use packed structures
2485 instead now that don't do any padding.
2486]
2487[Make mkPState and pragState take their arguments in the same order
2488Ian Lynagh <igloo@earth.li>**20100706172611]
2489[Remove an out-of-date comment
2490Ian Lynagh <igloo@earth.li>**20100706172217
2491 Ignore-this: 710ebd7d2dc01c1b0f1e58a5b6f85701
2492]
2493[LLVM: Stop llvm saving stg caller-save regs across C calls
2494David Terei <davidterei@gmail.com>**20100705162629
2495 Ignore-this: 28b4877b31b9358e682e38fc54b88658
2496 
2497 This is already handled by the Cmm code generator so LLVM is simply
2498 duplicating work. LLVM also doesn't know which ones are actually live
2499 so saves them all which causes a fair performance overhead for C calls
2500 on x64. We stop llvm saving them across the call by storing undef to
2501 them just before the call.
2502]
2503[LLVM: Add in literal undefined value to binding
2504David Terei <davidterei@gmail.com>**20100705161544
2505 Ignore-this: 95d8361b11584aaeec44c30e76916470
2506]
2507[LLVM: Add a literal NULL value to binding
2508David Terei <davidterei@gmail.com>**20100705161308
2509 Ignore-this: 9507b4b12c1157498704a9d1e5860f12
2510 
2511 Patch from Erik de Castro Lopo <erikd@mega-nerd.com>.
2512]
2513[refactor import declaration support (#2362)
2514Simon Marlow <marlowsd@gmail.com>**20100705104557
2515 Ignore-this: ee034ac377078a7e92bfada1907c86a0
2516]
2517[Disable dynamic linking optimisations on OS X
2518Simon Marlow <marlowsd@gmail.com>**20100705103014
2519 Ignore-this: b04420d3705c51112797758d17b2e40c
2520 To improve performance of the RTS when dynamically linked on x86, I
2521 previously disabled -fPIC for certain critical modules (the GC, and a
2522 few others).  However, build reports suggest that the dynamic linker
2523 on OS X doesn't like this, so I'm disabling this optimsation on that
2524 platform.
2525]
2526[trac #2362 (full import syntax in ghci)
2527amsay@amsay.net**20100625032632
2528 Ignore-this: a9d0859d84956beb74e27b797431bf9c
2529 'import' syntax is seperate from ':module' syntax
2530]
2531[Simplify ghc-pkg's Cabal dependencies
2532Ian Lynagh <igloo@earth.li>**20100704184155
2533 We no longer support building with a compiler that doesn't come with
2534 base 4.
2535]
2536[Use Cabal to configure the dist-install ghc-pkg; fixes trac #4156
2537Ian Lynagh <igloo@earth.li>**20100704132612]
2538[Remove dead code (standalone deriving flag no longer used in parser)
2539Ian Lynagh <igloo@earth.li>**20100701162058]
2540[LLVM: Use the inbounds keyword for getelementptr instructions.
2541David Terei <davidterei@gmail.com>**20100702160511
2542 Ignore-this: 3708e658a4c82b78b1402393f4405541
2543]
2544[threadPaused: fix pointer arithmetic
2545Simon Marlow <marlowsd@gmail.com>**20100701085046
2546 Ignore-this: b78210e5d978f18ffd235f1c78a55a23
2547 Noticed by Henrique Ferreiro <hferreiro@udc.es>, thanks!
2548]
2549[LLVM: Change more operations to use getelementptr
2550David Terei <davidterei@gmail.com>**20100701161856
2551 Ignore-this: fb24eb124e203f50680c6fec3ff9fe7d
2552]
2553[Add the haskell2010 package
2554Simon Marlow <marlowsd@gmail.com>**20100630125532
2555 Ignore-this: e9b011313f283a8ff2fcda7d029a01f
2556]
2557[LLVM: Use getelementptr instruction for a lot of situations
2558David Terei <davidterei@gmail.com>**20100630181157
2559 Ignore-this: 34d314dd8dffad9bdcffdc525261a49d
2560 
2561 LLVM supports creating pointers in two ways, firstly through
2562 pointer arithmetic (by casting between pointers and ints)
2563 and secondly using the getelementptr instruction. The second way
2564 is preferable as it gives LLVM more information to work with.
2565 
2566 This patch changes a lot of pointer related code from the first
2567 method to the getelementptr method.
2568]
2569[remove out of date comments; point to the wiki
2570Simon Marlow <marlowsd@gmail.com>**20100625100313
2571 Ignore-this: 95f363a373534b9471b1818102ec592d
2572]
2573[NCG: allocatableRegs is only giving us 8 SSE regs to allocate to
2574benl@ouroborus.net**20100629054321
2575 Ignore-this: b3e0fa0b4ce988a0258dc12261989ee0
2576]
2577[LLVM: Use intrinsic functions for pow, sqrt, sin, cos
2578David Terei <davidterei@gmail.com>**20100628182949
2579 Ignore-this: 98a0befaca3fe2b36d710d8ff9f062c4
2580 
2581 Instead of calling the C library for these Cmm functions
2582 we use intrinsic functions provided by llvm. LLVM will
2583 then either create a compile time constant if possible, or
2584 use a cpu instruction or as a last resort call the C
2585 library.
2586]
2587[LLVM: Fix test '2047' under linux-x64
2588David Terei <davidterei@gmail.com>**20100628165256
2589 Ignore-this: 41735d4f431a430db636621650ccd71e
2590]
2591[LLVM: Fix test 'ffi005' under linux-x64
2592David Terei <davidterei@gmail.com>**20100628155355
2593 Ignore-this: 841f3142c63cc898ac4c3f89698a837e
2594]
2595[LLVM: Update to use new fp ops introduced in 2.7
2596David Terei <davidterei@gmail.com>**20100628144037
2597 Ignore-this: 5dd2e5964e3c039d297ed586841e706b
2598]
2599[Add noalias and nocapture attributes to pointer stg registers
2600David Terei <davidterei@gmail.com>**20100628115120
2601 Ignore-this: 492a1e723cb3a62498d240d7de92dd7
2602 
2603 At the moment this gives a very slight performance boost of around 1 - 2%.
2604 Future changes to the generated code though so that pointers are kept as
2605 pointers more often instead of being cast to integer types straight away
2606 should hopefully improve the benefit this brings.
2607 
2608]
2609[during shutdown, only free the heap if we waited for foreign calls to exit
2610Simon Marlow <marlowsd@gmail.com>**20100628090536
2611 Ignore-this: d545384a4f641d701455d08ef1217479
2612]
2613[Fix typo in -ddump-pass's document.
2614shelarcy <shelarcy@gmail.com>**20100620070759
2615 Ignore-this: f4f1ddb53f147949e948147d89190c37
2616]
2617[Add #undefs for posix source symbols when including papi.h
2618dmp@rice.edu**20100624163514
2619 Ignore-this: 8a1cba21b880d12a75a75f7e96882053
2620 
2621 Validation fails when validating with PAPI support (i.e. GhcRtsWithPapi  = YES
2622 in validate.mk).  The problem is that the posix symbols are defined by a header
2623 included from papi.h. Compilation then fails because these symbols are
2624 redefined in PosixSource.h.
2625 
2626 This patch adds an undefine for the posix symbols after including papi.h and
2627 before including PosixSource.h. The #undefines are localized to Papi.c since
2628 that is the only case where they are getting defined twice.
2629]
2630[Use machdepCCOpts in runPhase_MoveBinary; fixes trac #3952
2631Ian Lynagh <igloo@earth.li>**20100625220953]
2632[LLVM: Fix bug with calling tail with empty list
2633David Terei <davidterei@gmail.com>**20100625115729
2634 Ignore-this: 46b4b32c8d92372a2d49794a96fe1613
2635]
2636[Fix warnings
2637benl@ouroborus.net**20100624091339
2638 Ignore-this: 5ba4bbd6abb9c9d1fb8c5d21ab73f218
2639]
2640[NCG: Comments and formatting only
2641benl@ouroborus.net**20100624083121
2642 Ignore-this: 86002e72c30d06bcc876d8c49f4caa5a
2643]
2644[NCG: Do the actual reversing of SCCs
2645benl@ouroborus.net**20100624082717
2646 Ignore-this: 12d2027ea118e751fbb48b27126150ef
2647]
2648[NCG: Fix dumping of graphs in regalloc stats for graph allocator
2649benl@ouroborus.net**20100624082625
2650 Ignore-this: 2b971bc9e0318099a9afb0e0db135730
2651]
2652[NCG: Reverse SCCs after each round in the graph allocator
2653benl@ouroborus.net**20100624082437
2654 Ignore-this: f0152e4039d6f16f7b5a99b286538116
2655]
2656[NCG: Don't actually complain on unreachable code blocks
2657benl@ouroborus.net**20100624081445
2658 Ignore-this: e7335ae6120917cb858c38c7c6da8e24
2659]
2660[NCG: Do explicit check for precondition of computeLiveness
2661benl@ouroborus.net**20100624080747
2662 Ignore-this: e7053c4e5e4c3c746b5ebf016913424a
2663 
2664  computeLiveness requires the SCCs of blocks to be in reverse dependent
2665  order, and if they're not it was silently giving bad liveness info,
2666  yielding a bad allocation.
2667 
2668  Now it complains, loudly.
2669]
2670[NCG: Fix off-by-one error in realRegSqueeze
2671benl@ouroborus.net**20100623095813
2672 Ignore-this: ab0698686d4c250da8e207f734f8252d
2673]
2674[NCG: Handle stripping of liveness info from procs with no blocks (like stg_split_marker)
2675benl@ouroborus.net**20100623091209
2676 Ignore-this: c0319b6cc62ec713afe4eb03790406e3
2677]
2678[NCG: Emit a warning on unreachable code block instead of panicing
2679benl@ouroborus.net**20100623085002
2680 Ignore-this: d20314b79e3c31e764ed4cd97290c696
2681]
2682[NCG: Remember to keep the entry block first when erasing liveness info
2683Ben.Lippmeier@anu.edu.au**20090917104429
2684 Ignore-this: 1b0c1df19d622858d50ffb6a01f2cef0
2685]
2686[NCG: Refactor representation of code with liveness info
2687Ben.Lippmeier@anu.edu.au**20090917090730
2688 Ignore-this: 2aebb3b02ebd92e547c5abad9feb0f0d
2689 
2690  * I've pushed the SPILL and RELOAD instrs down into the
2691    LiveInstr type to make them easier to work with.
2692 
2693  * When the graph allocator does a spill cycle it now just
2694    re-annotates the LiveCmmTops instead of converting them
2695    to NatCmmTops and back.
2696 
2697  * This saves working out the SCCS again, and avoids rewriting
2698    the SPILL and RELOAD meta instructions into real machine
2699    instructions.
2700]
2701[NCG: Add sanity checking to linear allocator
2702Ben.Lippmeier@anu.edu.au**20090917090335
2703 Ignore-this: 5a442be8b5087d04bc8b58dffa9ea080
2704 If there are are unreachable basic blocks in the native code then the
2705 linear allocator might loop. Detect this case and panic instead.
2706]
2707[NCG: Refactor LiveCmmTop to hold a list of SCCs instead of abusing ListGraph
2708Ben.Lippmeier@anu.edu.au**20090917060332
2709 Ignore-this: 3fec8d69ed0f760e53a202f873d5d9cb
2710]
2711[NCG: Allow the liveness map in a LiveInfo to be Nothing
2712Ben.Lippmeier@anu.edu.au**20090917043937
2713 Ignore-this: 5f82422d54d1b0ffc0589eb7e82fb7a4
2714]
2715[NCG: Also show the result of applying coalesings with -ddump-asm-regalloc-stages
2716Ben.Lippmeier.anu.edu.au**20090917034427
2717 Ignore-this: 76bd6d5ca43adb2167cb25832cbaa80b
2718]
2719[Fix panic when running "ghc -H"; trac #3364
2720Ian Lynagh <igloo@earth.li>**20100624234011
2721 The problem is that showing SDoc's looks at the static flags global
2722 variables, but those are panics while we are parsing the static flags.
2723 We work around this by explicitly using a fixed prettyprinter style.
2724]
2725[Allow for stg registers to have pointer type in llvm BE.
2726David Terei <davidterei@gmail.com>**20100621175839
2727 Ignore-this: fc09b1a8314aef0bde945c77af1124fb
2728 
2729 Before all the stg registers were simply a bit type or
2730 floating point type but now they can be declared to have
2731 a pointer type to one of these. This will allow various
2732 optimisations in the future in llvm since the type is
2733 more accurate.
2734]
2735[Add support for parameter attributes to the llvm BE binding
2736David Terei <davidterei@gmail.com>**20100624111744
2737 Ignore-this: 77f3c0c7bf8f81c4a154dc835ae7bcba
2738 
2739 These allow annotations of the code produced by the backend
2740 which should bring some perforamnce gains. At the moment
2741 the attributes aren't being used though.
2742]
2743[Cast some more nats to StgWord to be on the safe side
2744Simon Marlow <marlowsd@gmail.com>**20100624105700
2745 Ignore-this: e6176683856f9872fdeb2358bb065bb8
2746 And add a comment about the dangers of int overflow
2747]
2748[comments only
2749Simon Marlow <marlowsd@gmail.com>**20100624105105
2750 Ignore-this: fc8f762f4c3a5ffca2f8da2bc63ac2a4
2751]
2752[Fix an arithmetic overflow bug causing crashes with multi-GB heaps
2753Simon Marlow <marlowsd@gmail.com>**20100624104654
2754 Ignore-this: 67210755aa098740ff5230347be0fd5d
2755]
2756[Add support for collecting PAPI native events
2757dmp@rice.edu**20100622195953
2758 Ignore-this: 7269f9c4dfb2912a024eb632200fcd1
2759 
2760 This patch extends the PAPI support in the RTS to allow collection of native
2761 events. PAPI can collect data for native events that are exposed by the
2762 hardware beyond the PAPI present events. The native events supported on your
2763 hardware can found by using the papi_native_avail tool.
2764 
2765 The RTS already allows users to specify PAPI preset events from the command
2766 line. This patch extends that support to allow users to specify native events.
2767 The changes needed are:
2768 
2769 1) New option (#) for the RTS PAPI flag for native events. For example, to
2770    collect the native event 0x40000000, use ./a.out +RTS -a#0x40000000 -sstderr
2771 
2772 2) Update the PAPI_FLAGS struct to store whether the user specified event is a
2773    papi preset or a native event
2774 
2775 3) Update init_countable_events function to add the native events after parsing
2776    the event code and decoding the name using PAPI_event_code_to_name
2777 
2778]
2779[Don't warn about unused bindings with parents in .hs-boot files; trac #3449
2780Ian Lynagh <igloo@earth.li>**20100624110351]
2781[fix the home_imps filter to allow for 'import "this" <module>'
2782Simon Marlow <marlowsd@gmail.com>**20100621125535
2783 Ignore-this: da4e605b0513afc32a4e7caa921a2c76
2784 In the PackageImports extension, import "this" means "import from the
2785 current package".
2786]
2787[Use the standard C wrapper code for the ghc-$version.exe wrapper
2788Ian Lynagh <igloo@earth.li>**20100622202859
2789 Ignore-this: 60cd3e6db3afb63e6ba9e2db3b033580
2790]
2791[Don't rely on "-packagefoo" working; use "-package foo" instead
2792Ian Lynagh <igloo@earth.li>**20100622202547]
2793[Remove unnecessary C #includes
2794Ian Lynagh <igloo@earth.li>**20100622172919]
2795[Make the ghci.exe wrapper call the right ghc.exe
2796Ian Lynagh <igloo@earth.li>**20100622172247]
2797[More updates to datalayout description in llvm BE
2798David Terei <davidterei@gmail.com>**20100622165339
2799 Ignore-this: b0c604fe7673b0aa7c7064694d574437
2800]
2801[Remove LlvmAs phase as the llvm opt tool now handles this phase
2802David Terei <davidterei@gmail.com>**20100622144044
2803 Ignore-this: b9fd8f959702b6af014e2fa654bede3
2804 
2805 This phase originally invoked the llvm-as tool that turns a textual
2806 llvm assembly file into a bit code file for the rest of llvm to deal
2807 with. Now the llvm opt tool can do this itself, so we don't need to
2808 use llvm-as anymore.
2809]
2810[Update datalayout info in llvm BE
2811David Terei <davidterei@gmail.com>**20100622123457
2812 Ignore-this: 89b043d211225dcd819f30549afe1840
2813]
2814[Fix handling of float literals in llvm BE
2815David Terei <davidterei@gmail.com>**20100622121642
2816 Ignore-this: a3b5f382ad4b5a426ad4b581664506fa
2817]
2818[Declare some top level globals to be constant when appropriate
2819David Terei <davidterei@gmail.com>**20100621174954
2820 Ignore-this: 44832f65550d4f995d11c01cc1affef5
2821 
2822 This involved removing the old constant handling mechanism
2823 which was fairly hard to use. Now being constant or not is
2824 simply a property of a global variable instead of a separate
2825 type.
2826]
2827[Reduce the number of passes over the cmm in llvm BE
2828David Terei <davidterei@gmail.com>**20100621125220
2829 Ignore-this: cb2f4e54e8d0f982d5087fbeee35c73c
2830]
2831[Fix negate op not working for -0 in llvm backend
2832David Terei <davidterei@gmail.com>**20100621123606
2833 Ignore-this: c5d76e5cffa781fed074137851b1347f
2834]
2835[ROLLBACK: picCCOpts: -dynamic should not entail -optc-fPIC
2836Simon Marlow <marlowsd@gmail.com>**20100621100409
2837 Ignore-this: f2fac7df33d3919199befc59bd455414
2838 and add a comment to explain why it was wrong.  This fixes the dyn
2839 test failures that sprang up recently.
2840]
2841[Check files are really created in libffi
2842Ian Lynagh <igloo@earth.li>**20100620163724
2843 when we think that the libffi build creates them, so they just depend
2844 on the libffi build stamp.
2845]
2846[Improve the missing-import-list warning
2847Ian Lynagh <igloo@earth.li>**20100620124320
2848 Ignore-this: 551e5fdf2dfb56b49d249e0cebaa6115
2849]
2850[Tweak missing-import-list warning
2851Ian Lynagh <igloo@earth.li>**20100620122622
2852 Ignore-this: 360cdf59ae13d66ded181129325506c4
2853]
2854[trac #1789 (warnings for missing import lists)
2855amsay@amsay.net**20100618150649
2856 Ignore-this: b0b0b1e048fbca0817c1e6fade1153fa
2857]
2858[Refix docs for sizeofByteArray#/sizeofMutableByteArray# (#3800)
2859Ian Lynagh <igloo@earth.li>**20100620103749]
2860[Remove some old commented out code
2861Ian Lynagh <igloo@earth.li>**20100620000459]
2862[SET_ARR_HDR's last argument is now a number of bytes, rather than words
2863Ian Lynagh <igloo@earth.li>**20100619235214
2864 This avoids unnecessary work and potential loss of information
2865]
2866[Replace an (incorrect) bytes-to-words calculation with ROUNDUP_BYTES_TO_WDS
2867Ian Lynagh <igloo@earth.li>**20100619234310]
2868[FIX #38000 Store StgArrWords payload size in bytes
2869Antoine Latter <aslatter@gmail.com>**20100101183346
2870 Ignore-this: 7bf3ab4fc080c46311fc10b179361bb6
2871]
2872[Add win32 datalayout support to llvm backend
2873David Terei <davidterei@gmail.com>**20100618131733
2874 Ignore-this: 4b7bffaa8ef38c628ab852c1a6c1c009
2875]
2876[Remove unused 'ddump-opt-llvm' flag
2877David Terei <davidterei@gmail.com>**20100618101237
2878 Ignore-this: f78467496d986897e49d82646ee2907e
2879]
2880[generate "movl lbl(%reg1), %reg2" instructions, better codegen for -fPIC
2881Simon Marlow <marlowsd@gmail.com>**20100618082258
2882 Ignore-this: a25567ebff9f575303ddc8f2deafebbf
2883]
2884[joinToTargets: fix a case of panic "handleComponent cyclic"
2885Simon Marlow <marlowsd@gmail.com>**20100618082147
2886 Ignore-this: 765baeefbb5a41724004acd92405cecc
2887]
2888[comment typo
2889Simon Marlow <marlowsd@gmail.com>**20100618082102
2890 Ignore-this: e495610b7dd5ec30b02938638b56cb7
2891]
2892[Add support of TNTC to llvm backend
2893David Terei <davidterei@gmail.com>**20100618093205
2894 Ignore-this: 2c27d21668374a5b0d5e844882c69439
2895 
2896 We do this through a gnu as feature called subsections,
2897 where you can put data/code into a numbered subsection
2898 and those subsections will be joined together in descending
2899 order by gas at compile time.
2900]
2901[Don't automatically insert a -fvia-C flag in an unregisterised compiler
2902Ian Lynagh <igloo@earth.li>**20100617190901
2903 Ignore-this: eb25a9a338fade9e17c153da7c5f27e9
2904 The default object mode is already HscC, so it's unnecessary, and
2905 -fvia-C generates a deprecated flag warning now.
2906]
2907[In PosixSource.h, conditionally define things based on platform
2908Ian Lynagh <igloo@earth.li>**20100617174912
2909 This may not be ideal, but it should get GHC building on all platforms
2910 again.
2911]
2912[disable -fPIC for the GC for performance reasons
2913Simon Marlow <marlowsd@gmail.com>**20100617140025
2914 Ignore-this: c7c152bbff71ef7891eaee8ff39fc281
2915 see comment for details
2916]
2917[picCCOpts: -dynamic should not entail -optc-fPIC
2918Simon Marlow <marlowsd@gmail.com>**20100617115259
2919 Ignore-this: d71e42bd56e4bd107d2c431b801855e5
2920]
2921[Make getAllocations() visible
2922Simon Marlow <marlowsd@gmail.com>**20100617113259
2923 Ignore-this: 1b7fb38a01358c0acbe8987df07d23f2
2924]
2925[Fix the symbol visibility pragmas
2926Simon Marlow <marlowsd@gmail.com>**20100617105758
2927 Ignore-this: 76552500865473a1dbebbc1cc2def9f0
2928]
2929[pick up changes to $(GhcStage1HcOpts) without re-configuring the ghc package
2930Simon Marlow <marlowsd@gmail.com>**20100616124718
2931 Ignore-this: afb56d5560c813051285607fefb15493
2932]
2933[Fix bindisttest Makefile
2934Ian Lynagh <igloo@earth.li>**20100616205611
2935 Ignore-this: 39cd352152422f378572fc3859c5a377
2936]
2937[Remove some more unused make variables
2938Ian Lynagh <igloo@earth.li>**20100616180519]
2939[Convert some more variable names to FOO_CMD, for consistency
2940Ian Lynagh <igloo@earth.li>**20100616175916]
2941[Rename some variables from FOO to FOO_CMD
2942Ian Lynagh <igloo@earth.li>**20100616161108
2943 This fixes a problem with commands like gzip, where if $GZIP is exported
2944 in the environment, then when make runs a command it'll put the Makefile
2945 variable's value in the environment. But gzip treats $GZIP as arguments
2946 for itself, so when we run gzip it thinks we're giving it "gzip" as an
2947 argument.
2948]
2949[Make the "show" target work anywhere in the build tree
2950Ian Lynagh <igloo@earth.li>**20100616122910
2951 Ignore-this: 299d40cbe16112accd9f14e56fa12158
2952]
2953[Change ghc-pwd's license to a string Cabal recognises
2954Ian Lynagh <igloo@earth.li>**20100615204015
2955 Ignore-this: c935b6ad7f605aab0168997a90b40fc6
2956]
2957[fix warning
2958Simon Marlow <marlowsd@gmail.com>**20100604205933
2959 Ignore-this: 2aaa4ed6a8b9ae1e39adc4696aaf14a3
2960]
2961[--install-signal-handles=no does not affect the timer signal (#1908)
2962Simon Marlow <marlowsd@gmail.com>**20100527214627
2963 Ignore-this: b0c51f1abdb159dc360662485095a11a
2964]
2965[Small optimisation: allocate nursery blocks contiguously
2966Simon Marlow <marlowsd@gmail.com>**20100509194928
2967 Ignore-this: e650e99e9ea9493d2efb245d565beef4
2968 This lets automatic prefetching work better, for a tiny performance boost
2969]
2970[fix -fforce-recomp setting: module is PrimOp, not PrimOps
2971Simon Marlow <marlowsd@gmail.com>**20100507084507
2972 Ignore-this: f76e0d9b643682ec0e8fb7d91afdea68
2973]
2974[it should be an error to use relative directories (#4134)
2975Simon Marlow <marlowsd@gmail.com>**20100615151740
2976 Ignore-this: 2068021701832e018ca41b22877921d5
2977]
2978[missing include-dirs or library-dirs is only a warning now (#4104)
2979Simon Marlow <marlowsd@gmail.com>**20100615151702
2980 Ignore-this: e3114123cef147bbd28ccb64581a1afb
2981]
2982[fix #3822: desugaring case command in arrow notation
2983Ross Paterson <ross@soi.city.ac.uk>**20100615225110
2984 Ignore-this: 477d6c460b4174b94b4cd113fa5b9d19
2985 
2986 Get the set of free variables from the generated case expression:
2987 includes variables in the guards and decls that were missed before,
2988 and is also a bit simpler.
2989]
2990[Deprecate the -fvia-C flag; trac #3232
2991Ian Lynagh <igloo@earth.li>**20100615151836
2992 Ignore-this: c2452b2648bf7e44546465c1b964fce
2993]
2994[Avoid using the new ~~ perl operator in the mangler
2995Ian Lynagh <igloo@earth.li>**20100615151236
2996 Ignore-this: 709a7ba4e514b1596841b3ba7e5c6cc
2997]
2998[stmAddInvariantToCheck: add missing init of invariant->lock (#4057)
2999Simon Marlow <marlowsd@gmail.com>**20100615123643
3000 Ignore-this: 3b132547fa934cecf71a846db2a5f70e
3001]
3002[Add new LLVM code generator to GHC. (Version 2)
3003David Terei <davidterei@gmail.com>**20100615094714
3004 Ignore-this: 4dd2fe5854b64a3f0339d484fd5c238
3005 
3006 This was done as part of an honours thesis at UNSW, the paper describing the
3007 work and results can be found at:
3008 
3009 http://www.cse.unsw.edu.au/~pls/thesis/davidt-thesis.pdf
3010 
3011 A Homepage for the backend can be found at:
3012 
3013 http://hackage.haskell.org/trac/ghc/wiki/Commentary/Compiler/Backends/LLVM
3014 
3015 Quick summary of performance is that for the 'nofib' benchmark suite, runtimes
3016 are within 5% slower than the NCG and generally better than the C code
3017 generator.  For some code though, such as the DPH projects benchmark, the LLVM
3018 code generator outperforms the NCG and C code generator by about a 25%
3019 reduction in run times.
3020 
3021]
3022[Fix Trac #4127: build GlobalRdrEnv in GHCi correctly
3023simonpj@microsoft.com**20100615070626
3024 Ignore-this: d907e3bfa7882878cea0af172aaf6e84
3025 
3026 GHCi was building its GlobalRdrEnv wrongly, so that the
3027 gre_par field was bogus.  That in turn fooled the renamer.
3028 The fix is easy: use the right function!  Namely, call
3029 RnNames.gresFromAvail rather than availsToNameSet.
3030]
3031[Comments, and improvement to pretty-printing of HsGroup
3032simonpj@microsoft.com**20100615070409
3033 Ignore-this: ec8358f2485370b20226a97ec84e9024
3034]
3035[Don't reverse bindings in rnMethodBinds (fix Trac #4126)
3036simonpj@microsoft.com**20100614163935
3037 Ignore-this: a6ffbb5af6f51b142ed0aeae8ee5e3a9
3038]
3039[Fix Trac #4120: generate a proper coercion when unifying forall types
3040simonpj@microsoft.com**20100614134311
3041 Ignore-this: 601592bb505305f1954cbe730f168da4
3042 
3043 This was just a blatant omission, which hasn't come up before.
3044 Easily fixed, happily.
3045]
3046[Use mkFunTy to ensure that invariants are respected
3047simonpj@microsoft.com**20100614134159
3048 Ignore-this: 67dcada7a4e8d9927581cd77af71b6f
3049]
3050[Remove redundant debug code
3051simonpj@microsoft.com**20100601154151
3052 Ignore-this: e6ff11c04c631cf6aac73788cbcf02b5
3053]
3054[Fix Trac #4099: better error message for type functions
3055simonpj@microsoft.com**20100531140413
3056 Ignore-this: 3f53ca98cf770577818b9c0937482577
3057 
3058 Now we only want about "T is a type function and might not be
3059 injective" when matchin (T x) against (T y), which is the case
3060 that is really confusing.
3061]
3062[Gruesome fix in CorePrep to fix embarassing Trac #4121
3063simonpj@microsoft.com**20100614132726
3064 Ignore-this: fe82d15474afaac3e6133adfd7a7e055
3065 
3066 This is a long-lurking bug that has been flushed into
3067 the open by other arity-related changes.  There's a
3068 long comment
3069 
3070      Note [CafInfo and floating]
3071 
3072 to explain. 
3073 
3074 I really hate the contortions we have to do through to keep correct
3075 CafRef information on top-level binders.  The Right Thing, I believe,
3076 is to compute CAF and arity information later, and merge it into the
3077 interface-file information when the latter is generated.
3078 
3079 But for now, this hackily fixes the problem.
3080]
3081[Fix a bug in CorePrep that meant output invariants not satisfied
3082simonpj@microsoft.com**20100531150013
3083 Ignore-this: d34eb36d8877d3caf1cf2b20de426abd
3084 
3085 In cpePair I did things in the wrong order so that something that
3086 should have been a CprRhs wasn't.  Result: a crash in CoreToStg.
3087 Fix is easy, and I added more informative type signatures too.
3088]
3089[Robustify the treatement of DFunUnfolding
3090simonpj@microsoft.com**20100531145332
3091 Ignore-this: 8f5506ada4d89f6ab8ad1e8c3ffb09ba
3092 
3093 See Note [DFun unfoldings] in CoreSyn.  The issue here is that
3094 you can't tell how many dictionary arguments a DFun needs just
3095 from looking at the Arity of the DFun Id: if the dictionary is
3096 represented by a newtype the arity might include the dictionary
3097 and value arguments of the (single) method.
3098 
3099 So we need to record the number of arguments need by the DFun
3100 in the DFunUnfolding itself.  Details in
3101    Note [DFun unfoldings] in CoreSyn
3102]
3103[Fix spelling in comment
3104simonpj@microsoft.com**20100614132259
3105 Ignore-this: bbf0d55f2e5f10ef9c74592c12f9201c
3106]
3107[Update docs on view patterns
3108simonpj@microsoft.com**20100614074801
3109 Ignore-this: 8617b9078800d4942d71f142a5b6c831
3110]
3111[Fix printing of splices; part of #4124
3112Ian Lynagh <igloo@earth.li>**20100613154838
3113 Just putting parens around non-atomic expressions isn't sufficient
3114 for splices, as only the $x and $(e) forms are valid input.
3115]
3116[In ghci, catch IO exceptions when calling canonicalizePath
3117Ian Lynagh <igloo@earth.li>**20100613134627
3118 We now get an exception if the path doesn't exist
3119]
3120[Whitespace only
3121Ian Lynagh <igloo@earth.li>**20100612213119]
3122[Whitespace only
3123Ian Lynagh <igloo@earth.li>**20100612165450]
3124[Update ghci example output in user guide; patch from YitzGale in #4111
3125Ian Lynagh <igloo@earth.li>**20100612162250]
3126[Fix #4131 missing UNTAG_CLOSURE in messageBlackHole()
3127benl@ouroborus.net**20100611044614]
3128[messageBlackHole: fix deadlock bug caused by a missing 'volatile'
3129Simon Marlow <marlowsd@gmail.com>**20100610080636
3130 Ignore-this: 3cda3054bb45408aa9bd2d794b69c938
3131]
3132[Pass --no-tmp-comp-dir to Haddock (see comment)
3133Simon Marlow <marlowsd@gmail.com>**20100604083214
3134 Ignore-this: bfa4d74038637bd149f4d878b4eb8a87
3135]
3136[Track changes to DPH libs
3137Roman Leshchinskiy <rl@cse.unsw.edu.au>**20100607052903
3138 Ignore-this: 4dbc3f8418af3e74b3fc4f9a9dfe7764
3139]
3140[Track changes to DPH libs
3141Roman Leshchinskiy <rl@cse.unsw.edu.au>**20100607012642
3142 Ignore-this: 5d4e498171a3c57ab02621bfaea82cff
3143]
3144[In ghc-pkg, send warnings to stderr
3145Ian Lynagh <igloo@earth.li>**20100606161726
3146 Ignore-this: 56927d13b5e1c1ce2752734f0f9b665b
3147]
3148[Re-add newlines to enable layout for multi-line input.
3149Ian Lynagh <igloo@earth.li>**20100602180737
3150 Patch from Adam Vogt <vogt.adam@gmail.com>
3151 Partial fix for #3984
3152]
3153[Don't use unnecessary parens when printing types (Fix Trac 4107)
3154simonpj@microsoft.com**20100604110143
3155 Ignore-this: a833714ab13013c4345b222f4e87db1d
3156 
3157    f :: Eq a => a -> a
3158 rather than
3159    f :: (Eq a) => a -> a
3160]
3161[Track DPH library changes
3162Roman Leshchinskiy <rl@cse.unsw.edu.au>**20100604005728
3163 Ignore-this: 32bc2fbea6ad975e89545d4c42fd7c30
3164]
3165[fix --source-entity option passed to Haddock: we needed to escape a #
3166Simon Marlow <marlowsd@gmail.com>**20100603125459
3167 Ignore-this: d52ae6188b510c482bcebb23f0e553ae
3168]
3169[__stg_EAGER_BLACKHOLE_INFO -> __stg_EAGER_BLACKHOLE_info (#4106)
3170Simon Marlow <marlowsd@gmail.com>**20100602091419
3171 Ignore-this: 293315ac8f86fd366b8d61992ecc7961
3172]
3173[Add xhtml package (a new dependency of Haddock; not installed/shipped)
3174Simon Marlow <marlowsd@gmail.com>**20100602090101
3175 Ignore-this: af0ac8b91abe98f7fdb624ea0a4dee20
3176]
3177[Use UserInterrupt rather than our own Interrupted exception (#4100)
3178Simon Marlow <marlowsd@gmail.com>**20100602082345
3179 Ignore-this: 1909acf2f452593138b9f85024711714
3180]
3181[Add the global package DB to ghc --info (#4103)
3182Simon Marlow <marlowsd@gmail.com>**20100602082233
3183 Ignore-this: fd5c0e207e70eb0f62606c45dc5b8124
3184]
3185[rts/sm/GC.c: resize_generations(): Remove unneeded check of number of generations.
3186Marco Túlio Gontijo e Silva <marcot@debian.org>**20100528115612
3187 Ignore-this: 6f1bea62917c01c7adac636146132c97
3188 
3189 This "if" is inside another "if" which checks for RtsFlags.GcFlags.generations
3190 > 1, so testing this again is redundant, assuming the number of generations
3191 won't change during program execution.
3192]
3193[rts/sm/BlockAlloc.c: Small comment correction.
3194Marco Túlio Gontijo e Silva <marcot@debian.org>**20100526205839
3195 Ignore-this: bd2fcd4597cc872d80b0e2eeb1c3998a
3196]
3197[rts/sm/GC.c: Annotate constants.
3198Marco Túlio Gontijo e Silva <marcot@debian.org>**20100526205707
3199 Ignore-this: f232edb89383564d759ed890a18f602f
3200]
3201[includes/rts/storage/GC.h: generation_: n_words: Improve comment.
3202Marco Túlio Gontijo e Silva <marcot@debian.org>**20100526204615
3203 Ignore-this: f5d5feefa8f7b552303978f1804fea23
3204]
3205[Add PPC_RELOC_LOCAL_SECTDIFF support; patch from PHO in #3654
3206Ian Lynagh <igloo@earth.li>**20100601204211
3207 Ignore-this: 51293b7041cdce3ce7619ef11cf7ceb
3208]
3209[powerpc-apple-darwin now supports shared libs
3210Ian Lynagh <igloo@earth.li>**20100601173325]
3211[Vectoriser: only treat a function as scalar if it actually computes something
3212Roman Leshchinskiy <rl@cse.unsw.edu.au>**20100601045630
3213 Ignore-this: e5d99a6ddb62052e3520094a5af47552
3214]
3215[Add a release notes file for 6.14.1
3216Ian Lynagh <igloo@earth.li>**20100530171117
3217 Ignore-this: 1941e6d3d1f4051b69ca2f17a1cf84d6
3218]
3219[Check dblatex actually creates the files we tell it to
3220Ian Lynagh <igloo@earth.li>**20100530171043
3221 Ignore-this: ccc72caea2313be05cbac59bb54c0603
3222 If it fails, it still exits successfully.
3223]
3224[Add darwin to the list of OSes for which we use mmap
3225Ian Lynagh <igloo@earth.li>**20100529145016
3226 Ignore-this: a86d12a3334aaaafc86f7af9dbb0a7ae
3227 Patch from Barney Stratford
3228]
3229[Simplify the CPP logic in rts/Linker.c
3230Ian Lynagh <igloo@earth.li>**20100529144929
3231 Ignore-this: 1288f5b752cc1ab8b1c90cfd0ecfdf68
3232]
3233[Fix validate on OS X
3234Ian Lynagh <igloo@earth.li>**20100529154726]
3235[OS X x86_64 fix from Barney Stratford
3236Ian Lynagh <igloo@earth.li>**20100529122440]
3237[OS X 64 installer fixes from Barney Stratford
3238Ian Lynagh <igloo@earth.li>**20100528234935]
3239[fix warning
3240Simon Marlow <marlowsd@gmail.com>**20100525155812
3241 Ignore-this: f34eee3fe3d89579fd8d381c91ced750
3242]
3243[Fix doc bugs (#4071)
3244Simon Marlow <marlowsd@gmail.com>**20100525155728
3245 Ignore-this: aa25be196de567de360075022a1942f7
3246]
3247[Make sparks into weak pointers (#2185)
3248Simon Marlow <marlowsd@gmail.com>**20100525150435
3249 Ignore-this: feea0bb5006007b82c932bc3006124d7
3250 The new strategies library (parallel-2.0+, preferably 2.2+) is now
3251 required for parallel programming, otherwise parallelism will be lost.
3252]
3253[If you say 'make' or 'make stage=2' here, pretend we're in the ghc dir
3254Simon Marlow <marlowsd@gmail.com>**20100525085301
3255 Ignore-this: 78b740337aa460915c812cbbcdae5321
3256]
3257[Another attempt to get these #defines right
3258Simon Marlow <marlowsd@gmail.com>**20100525154313
3259 Ignore-this: 460ca0c47d81cd25eae6542114f67899
3260 Apparently on Solaris it is an error to omit _ISOC99_SOURCE when using
3261 _POSIX_C_SOURCE==200112L.
3262]
3263[Add configure flags for the location of GMP includes/library; fixes #4022
3264Ian Lynagh <igloo@earth.li>**20100525221616
3265 Ignore-this: fc3060caf995d07274ec975eeefbdf3e
3266]
3267[Refactor pretty printing of TyThings to fix Trac #4015
3268simonpj@microsoft.com**20100525153126
3269 Ignore-this: 8f15053b7554f62caa84201d2e4976d2
3270]
3271[When haddocking, we need the dependencies to have been built
3272Ian Lynagh <igloo@earth.li>**20100525145830
3273 as haddock loads the .hi files with the GHC API.
3274]
3275[Fix profiling output; spotted by jlouis
3276Ian Lynagh <igloo@earth.li>**20100525111217
3277 We were outputing the number of words allocated in a column titled "bytes".
3278]
3279[Improve printing of TyThings; fixes Trac #4087
3280simonpj@microsoft.com**20100525114045
3281 Ignore-this: da2a757a533454bba80b9b77cc5a771
3282]
3283[Spelling in comments
3284simonpj@microsoft.com**20100525114001
3285 Ignore-this: 270f3da655e526cf04e27db7a01e29c0
3286]
3287[Refactor (again) the handling of default methods
3288simonpj@microsoft.com**20100525113910
3289 Ignore-this: 6686f6cdb878d57abf6b49fec64fcbb1
3290 
3291 This patch fixes Trac #4056, by
3292 
3293  a) tidying up the treatment of default method names
3294  b) removing the 'module' argument to newTopSrcBinder
3295 
3296 The details aren't that interesting, but the result
3297 is much tidier. The original bug was a 'nameModule' panic,
3298 caused by trying to find the module of a top-level name.
3299 But TH quotes generate Internal top-level names that don't
3300 have a module, and that is generally a good thing. 
3301 
3302 Fixing that in turn led to the default-method refactoring,
3303 which also makes the Name for a default method be handled
3304 in the same way as other derived names, generated in BuildTyCl
3305 via a call newImplicitBinder.  Hurrah.
3306]
3307[Don't do SpecConstr on NOINLINE things (Trac #4064)
3308simonpj@microsoft.com**20100525112807
3309 Ignore-this: 452be0a2cef0042fb67275c2827b5f72
3310 
3311 Since the RULE from specialising gets the same Activation as
3312 the inlining for the Id itself there's no point in specialising
3313 a NOINLINE thing, because the rule will be permanently switched
3314 off.
3315 
3316 See Note [Transfer activation] in SpecConstr
3317 and Note [Auto-specialisation and RULES] in Specialise.
3318]
3319[Change our #defines to work on FreeBSD too
3320Simon Marlow <marlowsd@gmail.com>**20100524105828
3321 Ignore-this: b23ede46211e67859206c0ec57d6a86f
3322 With glibc, things like _POSIX_C_SOURCE and _ISOC99_SOURCE are
3323 additive, but on FreeBSD they are mutually exclusive.  However, it
3324 turns out we only need to define _POSIX_C_SOURCE and _XOPEN_SOURCE to
3325 get all the C99 stuff we need too, so there's no need for any #ifdefs.
3326 
3327 Submitted by: Gabor PALI <pgj@FreeBSD.org>
3328]
3329[Add a missing UNTAG_CLOSURE, causing bus errors on Sparc
3330Simon Marlow <marlowsd@gmail.com>**20100524105547
3331 Ignore-this: a590b5391d6f05d50c8c088456c3c166
3332 We just about got away with this on x86 which isn't
3333 alignment-sensitive.  The result of the memory load is compared
3334 against a few different values, but there is a fallback case that
3335 happened to be the right thing when the pointer was tagged.  A good
3336 bug to find, nonetheless.
3337]
3338[Add wiki links
3339Simon Marlow <marlowsd@gmail.com>**20100520095953
3340 Ignore-this: c22f126cde166e6207922b2eb51d29e3
3341]
3342[the 'stage=0' trick to disable all compiler builds stopped working; fix it
3343Simon Marlow <marlowsd@gmail.com>**20100520104455
3344 Ignore-this: bb6fae9056471612c8dbf06916188c33
3345]
3346[Comments and formatting only
3347benl@ouroborus.net**20100524014021
3348 Ignore-this: 64579c38154728b632e358bec751cc0b
3349]
3350[Core prettyprinter fixes. Patch from Tim Chevalier. Fixes #4085
3351Ian Lynagh <igloo@earth.li>**20100522225048]
3352[Fix the RTS debug_p build
3353Ian Lynagh <igloo@earth.li>**20100522163127]
3354[Unset $CFLAGS for "GNU non-executable stack" configure test; fixes #3889
3355Ian Lynagh <igloo@earth.li>**20100521165005
3356 With gcc 4.4 we get
3357     Error: can't resolve `.note.GNU-stack' {.note.GNU-stack section} - `.Ltext0' {.text section}
3358 when running gcc with the -g flag. To work around this we unset
3359 CFLAGS when running the test.
3360]
3361[Don't run "set -o igncr" before configuring libffi
3362Ian Lynagh <igloo@earth.li>**20100520162918
3363 Ignore-this: 489fa94df23f2adf4ff63c8ede2c0794
3364 It used to make the build work on cygwin, but now it breaks it instead:
3365     config.status: creating include/Makefile
3366     gawk: ./confLqjohp/subs.awk:1: BEGIN {\r
3367     gawk: ./confLqjohp/subs.awk:1: ^ backslash not last character on line
3368     config.status: error: could not create include/Makefile
3369     make[2]: *** [libffi/stamp.ffi.configure-shared] Error 1
3370     make[1]: *** [all] Error 2
3371]
3372[Stop passing -Wl,-macosx_version_min to gcc
3373Ian Lynagh <igloo@earth.li>**20100520154003
3374 Fixes a build failure on OS X 10.6. When linking
3375     rts/dist/build/libHSrts-ghc6.13.20100519.dylib
3376 we got
3377     ld: symbol dyld_stub_binding_helper not defined (usually in crt1.o/dylib1.o/bundle1.o)
3378     collect2: ld returned 1 exit status
3379]
3380[Fix build on FreeBSD; patch from Gabor PALI
3381Ian Lynagh <igloo@earth.li>**20100519140552]
3382[Fix package shadowing order (#4072)
3383Simon Marlow <marlowsd@gmail.com>**20100519104617
3384 Ignore-this: 26ea5e4bb5dff18618b807a54c7d6ebb
3385 
3386 Later packages are supposed to shadow earlier ones in the stack,
3387 unless the ordering is overriden with -package-id flags.
3388 Unfortunately an earlier fix for something else had sorted the list of
3389 packages so that it was in lexicographic order by installedPackageId,
3390 and sadly our test (cabal/shadow) didn't pick this up because the
3391 lexicographic ordering happened to work for the test.  I've now fixed
3392 the test so it tries both orderings.
3393]
3394[Set more env variables when configuring libffi
3395Ian Lynagh <igloo@earth.li>**20100518185014
3396 We now tell it where to find ld, nm and ar
3397]
3398[Set the location of ar to be the in-tree ar on Windows
3399Ian Lynagh <igloo@earth.li>**20100518181556]
3400[Change another / to </> to avoid building paths containing \/
3401Ian Lynagh <igloo@earth.li>**20100518172015
3402 This will hopefully fix #2889.
3403]
3404[Fix #4074 (I hope).
3405Simon Marlow <marlowsd@gmail.com>**20100518113214
3406 Ignore-this: 73cd70f5bc6f5add5247b61985c03fc1
3407 
3408 1. allow multiple threads to call startTimer()/stopTimer() pairs
3409 2. disable the timer around fork() in forkProcess()
3410 
3411 A corresponding change to the process package is required.
3412]
3413[we don't have a gcc-lib in LIB_DIR any more
3414Simon Marlow <marlowsd@gmail.com>**20100401102351
3415 Ignore-this: f41acd2d8f8e6763aa8bd57a0b44a7e4
3416]
3417[In validate, use gmake if available; based on a patch from Gabor PALI
3418Ian Lynagh <igloo@earth.li>**20100517200654]
3419[Remove duplicate "./configure --help" output; fixes #4075
3420Ian Lynagh <igloo@earth.li>**20100516141206]
3421[Update various 'sh boot's to 'perl boot'
3422Ian Lynagh <igloo@earth.li>**20100516122609
3423 Spotted by Marco Túlio Gontijo e Silva
3424]
3425[add missing initialisation for eventBufMutex
3426Simon Marlow <marlowsd@gmail.com>**20100514094943
3427 Ignore-this: 7f75594a8cb54fbec5aebd46bb959f45
3428]
3429[Undo part of #4003 patch
3430Simon Marlow <marlowsd@gmail.com>**20100513142017
3431 Ignore-this: cb65db86a38a7e5ccee9f779e489d104
3432 We still need the workaround for when compiling HEAD with 6.12.2
3433 
3434]
3435[fix !TABLES_NEXT_TO_CODE
3436Simon Marlow <marlowsd@gmail.com>**20100510151934
3437 Ignore-this: fccb859b114bef1c3122c98e60af51
3438]
3439[looksLikeModuleName: allow apostrophe in module names (#4051)
3440Simon Marlow <marlowsd@gmail.com>**20100510094741
3441 Ignore-this: df9348f3ba90608bec57257b47672985
3442]
3443[add the proper library dependencies for GhcProfiled=YES
3444Simon Marlow <marlowsd@gmail.com>**20100506122118
3445 Ignore-this: 6236993aa308ab5b5e1e5ea5f65982a
3446]
3447[Fix Trac #4003: fix the knot-tying in checkHiBootIface
3448simonpj@microsoft.com**20100511075026
3449 Ignore-this: a9ce2a318386fdc8782848df84592002
3450 
3451 I had incorrectly "optimised" checkHiBootIface so that it forgot
3452 to update the "knot-tied" type environment.
3453 
3454 This patch fixes the HEAD
3455]
3456[Re-engineer the derived Ord instance generation code (fix Trac #4019)
3457simonpj@microsoft.com**20100510133333
3458 Ignore-this: 8fe46e4dad27fbee211a7928acf372c2
3459   
3460 As well as fixing #4019, I rejigged the way that Ord instances are
3461 generated, which should make them faster in general.  See the
3462 Note [Generating Ord instances].
3463 
3464 I tried to measure the performance difference from this change, but
3465 the #4019 fix only removes one conditional branch per iteration, and
3466 I couldn't measure a consistent improvement.  But still, tihs is
3467 better than before.
3468]
3469[Make arity of INLINE things consistent
3470simonpj@microsoft.com**20100510133005
3471 Ignore-this: 15e7abf803d1dcb3f4ca760d2d939d0d
3472 
3473 We eta-expand things with INLINE pragmas;
3474 see Note [Eta-expanding INLINE things].
3475 
3476 But I eta-expanded it the wrong amount when the function
3477 was overloaded.  Ooops.
3478]
3479[Compacting GC fix, we forgot to thread the new bq field of StgTSO.
3480Simon Marlow <marlowsd@gmail.com>**20100510082325
3481 Ignore-this: a079c8446e2ad53efff6fd95d0f3ac80
3482]
3483[Add version constraints for the boot packages; fixes trac #3852
3484Ian Lynagh <igloo@earth.li>**20100509175051
3485 When using the bootstrapping compiler, we now explicitly constrain
3486 the version of boot packages (Cabal, extensible-exceptions, etc) to the
3487 in-tree version, so that the build system is less fragile should the
3488 user have a newer version installed for the bootstrapping compiler.
3489]
3490[Don't include inter-package dependencies when compiling with stage 0; #4031
3491Ian Lynagh <igloo@earth.li>**20100509130511
3492 This fixes a problem when building with GHC 6.12 on Windows, where
3493 dependencies on stage 0 (bootstrapping compiler) packages have absolute
3494 paths c:/ghc/..., and make gets confused by the colon.
3495]
3496[Add a ghc.mk for bindisttest/
3497Ian Lynagh <igloo@earth.li>**20100508223911]
3498[Move some make variables around so they are available when cleaning
3499Ian Lynagh <igloo@earth.li>**20100508212405]
3500[Optimise checkremove a bit
3501Ian Lynagh <igloo@earth.li>**20100508202006]
3502[Improve the bindisttest Makefile
3503Ian Lynagh <igloo@earth.li>**20100508195450]
3504[Add tools to test that cleaning works properly
3505Ian Lynagh <igloo@earth.li>**20100508194105]
3506[Tweak the ghc-pkg finding code
3507Ian Lynagh <igloo@earth.li>**20100508125815
3508 It now understand the ghc-stage[123] names we use in-tree, and it won't
3509 go looking for any old ghc-pkg if it can't find the one that matches
3510 ghc.
3511]
3512[Add a way to show what cleaning would be done, without actually doing it
3513Ian Lynagh <igloo@earth.li>**20100508122438]
3514[Tidy up the "rm" flags in the build system
3515Ian Lynagh <igloo@earth.li>**20100508115745]
3516[Correct install-name for dynamic Darwin rts
3517pho@cielonegro.org**20100508151155
3518 Ignore-this: 6d31716c8c113dcb46e9cb925c4201df
3519]
3520[PIC support for PowerPC
3521pho@cielonegro.org**20100508143900
3522 Ignore-this: 3673859a305398c4acae3f4d7c997615
3523 
3524 PPC.CodeGen.getRegister was not properly handling PicBaseReg.
3525 It seems working with this patch, but I'm not sure this change is correct.
3526]
3527[Fix makefile loop (#4050)
3528pho@cielonegro.org**20100507140707
3529 Ignore-this: 3a1cb13d0600977e74d17ac26cbef83d
3530 
3531 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.
3532]
3533[Fix crash in nested callbacks (#4038)
3534Simon Marlow <marlowsd@gmail.com>**20100507093222
3535 Ignore-this: cade85e361534ce711865a4820276388
3536 Broken by "Split part of the Task struct into a separate struct
3537 InCall".
3538]
3539[Add $(GhcDynamic) knob, set to YES to get stage2 linked with -dynamic
3540Simon Marlow <marlowsd@gmail.com>**20100428205241
3541 Ignore-this: 1db8bccf92099785ecac39aebd27c92d
3542 Default currently NO.
3543 
3544 Validate passed with GhcDynamic=YES on x86/Linux here.
3545 
3546 The compiler is currently slower on x86 when linked -dynamic,
3547 because the GC inner loop has been adversely affected by -fPIC, I'm
3548 looking into how to fix it.
3549]
3550[omit "dyn" from the way appended to the __stginit label
3551Simon Marlow <marlowsd@gmail.com>**20100428204914
3552 Ignore-this: 14183f3defa9f2bde68fda6729b740bc
3553 When GHCi is linked dynamically, we still want to be able to load
3554 non-dynamic object files.
3555]
3556[improvements to findPtr(), a neat hack for browsing the heap in gdb
3557Simon Marlow <marlowsd@gmail.com>**20100506115427
3558 Ignore-this: ac57785bb3e13b97a5945f753f068738
3559]
3560[Fix +RTS -G1
3561Simon Marlow <marlowsd@gmail.com>**20100506110739
3562 Ignore-this: 86a5de39a94d3331a4ee1213f82be497
3563]
3564[Enable the "redundant specialise pragmas" warning; fixes trac #3855
3565Ian Lynagh <igloo@earth.li>**20100506175351]
3566[Find the correct external ids when there's a wrapper
3567simonpj@microsoft.com**20100506164135
3568 Ignore-this: 636266407b174b05b2b8646cc73062c0
3569 
3570 We were failing to externalise the wrapper id for a function
3571 that had one.
3572]
3573[Add a comment about pattern coercions
3574simonpj@microsoft.com**20100506164027
3575 Ignore-this: 17428089f3df439f65d892e23e8ed61a
3576]
3577[Comments only
3578simonpj@microsoft.com**20100506163829
3579 Ignore-this: 169167b6463873ab173cc5750c5be469
3580]
3581[Make a missing name in mkUsageInfo into a panic
3582simonpj@microsoft.com**20100506163813
3583 Ignore-this: b82ff1b8bf89f74f146db7cb5cc4c4d7
3584 
3585 We really want to know about this!
3586]
3587[Refactoring of hsXxxBinders
3588simonpj@microsoft.com**20100506163737
3589 Ignore-this: 97c6667625262b160f9746f7bea1c980
3590 
3591 This patch moves various functions that extract the binders
3592 from a HsTyClDecl, HsForeignDecl etc into HsUtils, and gives
3593 them consistent names.
3594]
3595[Fix Trac #3966: warn about useless UNPACK pragmas
3596simonpj@microsoft.com**20100506163337
3597 Ignore-this: 5beb24b686eda6113b614dfac8490df1
3598 
3599 Warning about useless UNPACK pragmas wasn't as easy as I thought.
3600 I did quite a bit of refactoring, which improved the code by refining
3601 the types somewhat.  In particular notice that in DataCon, we have
3602 
3603     dcStrictMarks   :: [HsBang]
3604     dcRepStrictness :: [StrictnessMarks]
3605 
3606 The former relates to the *source-code* annotation, the latter to
3607 GHC's representation choice.
3608]
3609[Make tcg_dus behave more sanely; fixes a mkUsageInfo panic
3610simonpj@microsoft.com**20100506162719
3611 Ignore-this: d000bca15b0e127e297378ded1bfb81b
3612 
3613 The tcg_dus field used to contain *uses* of type and class decls,
3614 but not *defs*.  That was inconsistent, and it really went wrong
3615 for Template Haskell bracket.  What happened was that
3616  foo = [d| data A = A
3617                   f :: A -> A
3618                   f x = x |]
3619 would find a "use" of A when processing the top level of the module,
3620 which in turn led to a mkUsageInfo panic in MkIface.  The cause was
3621 the fact that the tcg_dus for the nested quote didn't have defs for
3622 A.
3623]
3624[Add a HsExplicitFlag to SpliceDecl, to improve Trac #4042
3625simonpj@microsoft.com**20100506161523
3626 Ignore-this: e4e563bac2fd831cc9e94612f5b4fa9d
3627 
3628 The issue here is that
3629 
3630     g :: A -> A
3631     f
3632     data A = A
3633 
3634 is treated as if you'd written $(f); that is the call of
3635 f is a top-level Template Haskell splice.  This patch
3636 makes sure that we *first* check the -XTemplateHaskellFlag
3637 and bleat about a parse error if it's off.  Othewise we
3638 get strange seeing "A is out of scope" errors.
3639]
3640[Change an assert to a warn
3641simonpj@microsoft.com**20100506161111
3642 Ignore-this: 739a4fb4c7940376b0f2c8ad52a1966c
3643 
3644 This is in the constraint simplifier which I'm about
3645 to rewrite, so I'm hoping the assert isn't fatal!
3646]
3647[Tidy up debug print a little
3648simonpj@microsoft.com**20100506161027
3649 Ignore-this: bd5492878e06bee1cddcbb3fc4df66d8
3650]
3651[Remove useless UNPACK pragmas
3652simonpj@microsoft.com**20100506161012
3653 Ignore-this: 3e5ab1a7cf58107034412a798bc214e5
3654]
3655[Add WARNM2 macro, plus some refactoring
3656simonpj@microsoft.com**20100506160808
3657 Ignore-this: 2ab4f1f0b5d94be683036e77aec09255
3658]
3659[Use -Wwarn for the binary package, becuase it has redundant UNPACK pragmas
3660simonpj@microsoft.com**20100506160750
3661 Ignore-this: cf0d3a11473e28bfce9602e716e69a5f
3662]
3663[Fix Trac #3966: warn about unused UNPACK pragmas
3664simonpj@microsoft.com**20100409201812
3665 Ignore-this: c96412596b39c918b5fb9b3c39ce2119
3666]
3667[Fix Trac #3953: fail earlier when using a bogus quasiquoter
3668simonpj@microsoft.com**20100409201748
3669 Ignore-this: ef48e39aa932caed538643985234f043
3670]
3671[Fix Trac #3965: tighten conditions when deriving Data
3672simonpj@microsoft.com**20100409184420
3673 Ignore-this: 96f7d7d2da11565d26b465d7d0497ac9
3674 
3675 It's tricky to set up the context for a Data instance.  I got it wrong
3676 once, and fixed it -- hence the "extra_constraints" in
3677 TcDeriv.inferConstraints. 
3678 
3679 But it still wasn't right!  The tricky bit is that dataCast1 is only
3680 generated when T :: *->*, and dataCast2 when T :: *->*->*. (See
3681 the code in TcGenDeriv for dataCastX.
3682]
3683[Fix Trac #3964: view patterns in DsArrows
3684simonpj@microsoft.com**20100409165557
3685 Ignore-this: d823c182831d5e2e592e995b16180e2f
3686 
3687 Just a missing case; I've eliminated the catch-all so
3688 that we get a warning next time we extend HsPat
3689]
3690[Fix Trac #3955: renamer and type variables
3691simonpj@microsoft.com**20100409163710
3692 Ignore-this: bd5ec64d76c0f583bf5f224792bf294c
3693 
3694 The renamer wasn't computing the free variables of a type declaration
3695 properly.  This patch refactors a bit, and makes it more robust,
3696 fixing #3955 and several other closely-related bugs.  (We were
3697 omitting some free variables and that could just possibly lead to a
3698 usage-version tracking error.
3699]
3700[Layout only
3701simonpj@microsoft.com**20100409163506
3702 Ignore-this: 1f14990b5aa0b9821b84452fb34e9f41
3703]
3704[Give a better deprecated message for INCLUDE pragmas; fixes #3933
3705Ian Lynagh <igloo@earth.li>**20100506130910
3706 We now have a DeprecatedFullText constructor, so we can override the
3707 "-#include is deprecated: " part of the warning.
3708]
3709[De-haddock a comment that confuses haddock
3710Ian Lynagh <igloo@earth.li>**20100506123607]
3711[Fix comment to not confuse haddock
3712Ian Lynagh <igloo@earth.li>**20100506113642]
3713[Detect EOF when trying to parse a string in hp2ps
3714Ian Lynagh <igloo@earth.li>**20100506000830]
3715[Make the demand analyser sdd demands for strict constructors
3716simonpj@microsoft.com**20100505200936
3717 Ignore-this: eb32632adbc354eb7a5cf884c263e0d3
3718 
3719 This opportunity was spotted by Roman, and is documented in
3720 Note [Add demands for strict constructors] in DmdAnal.
3721]
3722[Fix interaction of exprIsCheap and the lone-variable inlining check
3723simonpj@microsoft.com**20100505200723
3724 Ignore-this: f3cb65085c5673a99153d5d7b6559ab1
3725 
3726 See Note [Interaction of exprIsCheap and lone variables] in CoreUnfold
3727 
3728 This buglet meant that a nullary definition with an INLINE pragma
3729 counter-intuitively didn't get inlined at all.  Roman identified
3730 the bug.
3731]
3732[Matching cases in SpecConstr and Rules
3733simonpj@microsoft.com**20100505200543
3734 Ignore-this: f5c28c780fbf8badce84c6fdc9aa1779
3735 
3736 This patch has zero effect.  It includes comments,
3737 a bit of refactoring, and a tiny bit of commment-out
3738 code go implement the "matching cases" idea below.
3739 
3740 In the end I've left it disabled because while I think
3741 it does no harm I don't think it'll do any good either.
3742 But I didn't want to lose the idea totally. There's
3743 a thread called "Storable and constant memory" on
3744 the libraries@haskell.org list (Apr 2010) about it.
3745 
3746 Note [Matching cases]
3747 ~~~~~~~~~~~~~~~~~~~~~
3748 {- NOTE: This idea is currently disabled.  It really only works if
3749          the primops involved are OkForSpeculation, and, since
3750         they have side effects readIntOfAddr and touch are not.
3751         Maybe we'll get back to this later .  -}
3752   
3753 Consider
3754    f (case readIntOffAddr# p# i# realWorld# of { (# s#, n# #) ->
3755       case touch# fp s# of { _ ->
3756       I# n# } } )
3757 This happened in a tight loop generated by stream fusion that
3758 Roman encountered.  We'd like to treat this just like the let
3759 case, because the primops concerned are ok-for-speculation.
3760 That is, we'd like to behave as if it had been
3761    case readIntOffAddr# p# i# realWorld# of { (# s#, n# #) ->
3762    case touch# fp s# of { _ ->
3763    f (I# n# } } )
3764]
3765[Comments only
3766simonpj@microsoft.com**20100504163629
3767 Ignore-this: 3be12df04714aa820bce706b5dc8a9cb
3768]
3769[Comments only
3770simonpj@microsoft.com**20100504163529
3771 Ignore-this: 791e2fd39c7d880ce1dc80ebdf3a5398
3772]
3773[Comments only
3774simonpj@microsoft.com**20100504163457
3775 Ignore-this: f19e9ffeb3d65770b1595bca5f97a59d
3776]
3777[Comments only (about type families)
3778simonpj@microsoft.com**20100417145032
3779 Ignore-this: dd39425ef2155d52dbf55a4d5fd97cb8
3780]
3781[Fix hp2ps when the .hp file has large string literals
3782Ian Lynagh <igloo@earth.li>**20100505191921]
3783[In build system, call package-config after including package data
3784Ian Lynagh <igloo@earth.li>**20100504225035
3785 Otherwise the $1_$2_HC_OPTS variable gets clobbered.
3786]
3787[runghc: flush stdout/stderr on an exception (#3890)
3788Simon Marlow <marlowsd@gmail.com>**20100505133848
3789 Ignore-this: 224c1898cec64cb1c94e0d7033e7590e
3790]
3791[Remove the Unicode alternative for ".." (#3894)
3792Simon Marlow <marlowsd@gmail.com>**20100505121202
3793 Ignore-this: 2452cd67281667106f9169747b6d784f
3794]
3795[tidyup; no functional changes
3796Simon Marlow <marlowsd@gmail.com>**20100505115015
3797 Ignore-this: d0787e5cdeef1dee628682fa0a46019
3798]
3799[Make the running_finalizers flag task-local
3800Simon Marlow <marlowsd@gmail.com>**20100505114947
3801 Ignore-this: 345925d00f1dca203941b3c5d84c90e1
3802 Fixes a bug reported by Lennart Augustsson, whereby we could get an
3803 incorrect error from the RTS about re-entry from a finalizer,
3804]
3805[add a MAYBE_GC() in killThread#, fixes throwto003(threaded2) looping
3806Simon Marlow <marlowsd@gmail.com>**20100505114746
3807 Ignore-this: efea04991d6feed04683a42232fc85da
3808]
3809[Allow filepath-1.2.*
3810Simon Marlow <marlowsd@gmail.com>**20100505101139
3811 Ignore-this: 1b5580cd9cd041ec48f40cd37603326a
3812]
3813[BlockedOnMsgThrowTo is possible in resurrectThreads (#4030)
3814Simon Marlow <marlowsd@gmail.com>**20100505094534
3815 Ignore-this: ac24a22f95ffeaf480187a1620fdddb2
3816]
3817[Don't raise a throwTo when the target is masking and BlockedOnBlackHole
3818Simon Marlow <marlowsd@gmail.com>**20100505094506
3819 Ignore-this: 302616931f61667030d77ddfbb02374e
3820]
3821[Fix build with GHC 6.10
3822Ian Lynagh <igloo@earth.li>**20100504180302
3823 In GHC 6.10, intersectionWith is (a -> b -> a) instead of (a -> b -> c),
3824 so we need to jump through some hoops to get the more general type.
3825]
3826[The libffi patches are no longer needed
3827Ian Lynagh <igloo@earth.li>**20100504171603]
3828[Use the in-tree windres; fixes trac #4032
3829Ian Lynagh <igloo@earth.li>**20100504170941]
3830[Print unfoldings on lambda-bound variables
3831Simon PJ <simonpj@microsoft.com>**20100503181822
3832 Ignore-this: 2fd5a7502cc6273d96258e0914f0f8cd
3833 
3834 ...in the unusual case where they have one;
3835 see Note [Case binders and join points] in Simplify.lhs
3836]
3837[Replace FiniteMap and UniqFM with counterparts from containers.
3838Milan Straka <fox@ucw.cz>**20100503171315
3839 Ignore-this: a021972239163dbf728284b19928cebb
3840 
3841 The original interfaces are kept. There is small performance improvement:
3842 - when compiling for five nofib, we get following speedups:
3843     Average                -----           -2.5%
3844     Average                -----           -0.6%
3845     Average                -----           -0.5%
3846     Average                -----           -5.5%
3847     Average                -----          -10.3%
3848 - when compiling HPC ten times, we get:
3849     switches                          oldmaps   newmaps
3850     -O -fasm                          117.402s  116.081s (98.87%)
3851     -O -fasm -fregs-graph             119.993s  118.735s (98.95%)
3852     -O -fasm -fregs-iterative         120.191s  118.607s (98.68%)
3853]
3854[Make the demand analyser take account of lambda-bound unfoldings
3855Simon PJ <simonpj@microsoft.com>**20100503151630
3856 Ignore-this: 2ee8e27d4df2debfc79e6b8a17c32bc1
3857 
3858 This is a long-standing lurking bug. See Note [Lamba-bound unfoldings]
3859 in DmdAnal.
3860 
3861 I'm still not really happy with this lambda-bound-unfolding stuff.
3862]
3863[Fix dynamic libs on OS X, and enable them by default
3864Ian Lynagh <igloo@earth.li>**20100503150302]
3865[Switch back to using bytestring from the darcs repo; partially fixes #3855
3866Ian Lynagh <igloo@earth.li>**20100502113458]
3867[Fix some cpp warnings when building on FreeBSD; patch from Gabor PALI
3868Ian Lynagh <igloo@earth.li>**20100428150700]
3869[Fix "make 2"
3870Ian Lynagh <igloo@earth.li>**20100427162212
3871 The new Makefile logic was enabling the stage 1 rules when stage=2,
3872 so "make 2" was rebuilding stage 1.
3873]
3874[Inplace programs depend on their shell wrappers
3875Ian Lynagh <igloo@earth.li>**20100427160038]
3876[--make is now the default (#3515), and -fno-code works with --make (#3783)
3877Simon Marlow <marlowsd@gmail.com>**20100427122851
3878 Ignore-this: 33330474fa4703f32bf9997462b4bf3c
3879 If the command line contains any Haskell source files, then we behave
3880 as if --make had been given.
3881 
3882 The meaning of the -c flag has changed (back): -c now selects one-shot
3883 compilation, but stops before linking.  However, to retain backwards
3884 compatibility, -c is still allowed with --make, and means the same as
3885 --make -no-link.  The -no-link flag has been un-deprecated.
3886 
3887 -fno-code is now allowed with --make (#3783); the fact that it was
3888 disabled before was largely accidental, it seems.  We also had some
3889 regressions in this area: it seems that -fno-code was causing a .hc
3890 file to be emitted in certain cases.  I've tidied up the code, there
3891 was no need for -fno-code to be a "mode" flag, as far as I can tell.
3892 
3893 -fno-code does not emit interface files, nor does it do recompilation
3894 checking, as suggested in #3783.  This would make Haddock emit
3895 interface files, for example, and I'm fairly sure we don't want to do
3896 that.  Compiling with -fno-code is pretty quick anyway, perhaps we can
3897 get away without recompilation checking.
3898]
3899[remove duplicate docs for -e in --help output (#4010)
3900Simon Marlow <marlowsd@gmail.com>**20100426140642
3901 Ignore-this: 187ff893ba8ffa0ec127867a7590e38d
3902]
3903[workaround for #4003, fixes HEAD build with 6.12.2
3904Simon Marlow <marlowsd@gmail.com>**20100426103428
3905 Ignore-this: c4bc445dc8052d4e6efef3f1daf63562
3906]
3907[Make sure all the clean rules are always included
3908Ian Lynagh <igloo@earth.li>**20100424181823
3909 In particular, this fixes a problem where stage3 bits weren't being cleaned
3910]
3911[Correct the name of the amd64/FreeBSD platform in PlatformSupportsSharedLibs
3912Ian Lynagh <igloo@earth.li>**20100424132830
3913 We weren't getting sharedlibs on amd64/FreeBSD because of this
3914]
3915[Include DPH docs in bindists
3916Ian Lynagh <igloo@earth.li>**20100424123101]
3917[reinstate eta-expansion during SimplGently, to fix inlining of sequence_
3918Simon Marlow <marlowsd@gmail.com>**20100423124853
3919 Ignore-this: 4fa0fd5bafe0d6b58fc81076f50d5f8d
3920]
3921[fix 64-bit value for W_SHIFT, which thankfully appears to be not used
3922Simon Marlow <marlowsd@gmail.com>**20100422213605
3923 Ignore-this: 525c062d2456c224ec8d0e083edd3b55
3924]
3925[Add missing constant folding and optimisation for unsigned division
3926Simon Marlow <marlowsd@gmail.com>**20100422213443
3927 Ignore-this: fb10d1cda0852fab0cbcb47247498fb3
3928 Noticed by Denys Rtveliashvili <rtvd@mac.com>, see #4004
3929]
3930[Fix the GHC API link in the main doc index.html
3931Ian Lynagh <igloo@earth.li>**20100422213226]
3932[Give the right exit code in darcs-all
3933Ian Lynagh <igloo@earth.li>**20100421171339
3934 Our END block was calling system, which alters $?. So now we save and
3935 restore it.
3936]
3937[Use StgWord64 instead of ullong
3938Ian Lynagh <igloo@earth.li>**20100421162336
3939 This patch also fixes ullong_format_string (renamed to showStgWord64)
3940 so that it works with values outside the 32bit range (trac #3979), and
3941 simplifies the without-commas case.
3942]
3943[Implement try10Times in Makefile
3944Ian Lynagh <igloo@earth.li>**20100420165909
3945 Avoid using seq, as FreeBSD has jot instead.
3946]
3947[Fix crash in non-threaded RTS on Windows
3948Simon Marlow <marlowsd@gmail.com>**20100420122125
3949 Ignore-this: 28b0255a914a8955dce02d89a7dfaca
3950 The tso->block_info field is now overwritten by pushOnRunQueue(), but
3951 stg_block_async_info was assuming that it still held a pointer to the
3952 StgAsyncIOResult.  We must therefore save this value somewhere safe
3953 before putting the TSO on the run queue.
3954]
3955[Expand the scope of the event_buf_mutex to cover io_manager_event
3956Simon Marlow <marlowsd@gmail.com>**20100420122026
3957 Ignore-this: 185a6d84f7d4a35997f10803f6dacef1
3958 I once saw a failure that I think was due to a race on
3959 io_manager_event, this should fix it.
3960]
3961[Flags -auto and -auto-all operate only on functions not marked INLINE.
3962Milan Straka <fox@ucw.cz>**20100331191050
3963 Ignore-this: 3b63580cfcb3c33d62ad697c36d94d05
3964]
3965[Spelling correction for LANGUAGE pragmas
3966Max Bolingbroke <batterseapower@hotmail.com>**20100413192825
3967 Ignore-this: 311b51ba8d43f6c7fd32f48db9a88dee
3968]
3969[Update the user guide so it talks about the newer "do rec" notation everywhere
3970Ian Lynagh <igloo@earth.li>**20100416205416
3971 Some of the problems highlighted in trac #3968.
3972]
3973[Fix typo
3974Ian Lynagh <igloo@earth.li>**20100416205412]
3975[Fix Trac #3950: unifying types of different kinds
3976simonpj@microsoft.com**20100412151845
3977 Ignore-this: d145b9de5ced136ef2c39f3ea4a04f4a
3978 
3979 I was assuming that the unifer only unified types of the
3980 same kind, but now we can "defer" unsolved constraints that
3981 invariant no longer holds.  Or at least is's more complicated
3982 to ensure. 
3983 
3984 This patch takes the path of not assuming the invariant, which
3985 is simpler and more robust.  See
3986 Note [Mismatched type lists and application decomposition]
3987]
3988[Fix Trac #3943: incorrect unused-variable warning
3989simonpj@microsoft.com**20100412151630
3990 Ignore-this: 52459f2b8b02c3cb120abe674dc9a060
3991 
3992 In fixing this I did the usual little bit of refactoring
3993]
3994[Convert boot and boot-pkgs to perl
3995Ian Lynagh <igloo@earth.li>**20100415143919
3996 This stops us having to worry about sh/sed/... portability.
3997]
3998[Use $(MAKE), not make, when recursively calling make
3999Ian Lynagh <igloo@earth.li>**20100415121453]
4000[Remove the ghc_ge_609 makefile variables
4001Ian Lynagh <igloo@earth.li>**20100412235658
4002 They are now guaranteed to be YES
4003]
4004[Increase the minimum version number required to 6.10 in configure.ac
4005Ian Lynagh <igloo@earth.li>**20100412235313]
4006[The bootstrapping compiler is now required to be > 609
4007Ian Lynagh <igloo@earth.li>**20100409161046]
4008[Handle IND_STATIC in isRetainer
4009Ian Lynagh <igloo@earth.li>**20100409104207
4010 IND_STATIC used to be an error, but at the moment it can happen
4011 as isAlive doesn't look through IND_STATIC as it ignores static
4012 closures. See trac #3956 for a program that hit this error.
4013]
4014[Add Data and Typeable instances to HsSyn
4015David Waern <david.waern@gmail.com>**20100330011020
4016 Ignore-this: c3f2717207b15539fea267c36b686e6a
4017 
4018 The instances (and deriving declarations) have been taken from the ghc-syb
4019 package.
4020]
4021[Fix for derefing ThreadRelocated TSOs in MVar operations
4022Simon Marlow <marlowsd@gmail.com>**20100407092824
4023 Ignore-this: 94dd7c68a6094eda667e2375921a8b78
4024]
4025[sanity check fix
4026Simon Marlow <marlowsd@gmail.com>**20100407092746
4027 Ignore-this: 9c18cd5f5393e5049015ca52e62a1269
4028]
4029[get the reg liveness right in the putMVar# heap check
4030Simon Marlow <marlowsd@gmail.com>**20100407092724
4031 Ignore-this: b1ba07a59ecfae00e9a1f8391741abc
4032]
4033[initialise the headers of MSG_BLACKHOLE objects properly
4034Simon Marlow <marlowsd@gmail.com>**20100407081712
4035 Ignore-this: 183dcd0ca6a395d08db2be12b02bdd79
4036]
4037[initialise the headers of MVAR_TSO_QUEUE objects properly
4038Simon Marlow <marlowsd@gmail.com>**20100407081514
4039 Ignore-this: 4b4a2f30cf2fb69ca4128c41744687bb
4040]
4041[undo debugging code
4042Simon Marlow <marlowsd@gmail.com>**20100406142740
4043 Ignore-this: 323c2248f817b6717c19180482fc4b00
4044]
4045[putMVar#: fix reg liveness in the heap check
4046Simon Marlow <marlowsd@gmail.com>**20100406135832
4047 Ignore-this: cddd2c7807ac7612c9b2c4c0d384d284
4048]
4049[account for the new BLACKHOLEs in the GHCi debugger
4050Simon Marlow <marlowsd@gmail.com>**20100406133406
4051 Ignore-this: 4d4aeb4bbada3f50dc1fb0123f565e8f
4052]
4053[don't forget to deRefTSO() in tryWakeupThread()
4054Simon Marlow <marlowsd@gmail.com>**20100406130411
4055 Ignore-this: 171d57c4f8653835dec0b69f9be9881c
4056]
4057[Fix bug in popRunQueue
4058Simon Marlow <marlowsd@gmail.com>**20100406091453
4059 Ignore-this: 9d3cec8f18f5c5cbd51751797386eb6f
4060]
4061[fix bug in migrateThread()
4062Simon Marlow <marlowsd@gmail.com>**20100401105840
4063 Ignore-this: 299bcf0d1ea0f8865f3e845eb93d2ad3
4064]
4065[Remove the IND_OLDGEN and IND_OLDGEN_PERM closure types
4066Simon Marlow <marlowsd@gmail.com>**20100401093519
4067 Ignore-this: 95f2480c8a45139835eaf5610217780b
4068 These are no longer used: once upon a time they used to have different
4069 layout from IND and IND_PERM respectively, but that is no longer the
4070 case since we changed the remembered set to be an array of addresses
4071 instead of a linked list of closures.
4072]
4073[Change the representation of the MVar blocked queue
4074Simon Marlow <marlowsd@gmail.com>**20100401091605
4075 Ignore-this: 20a35bfabacef2674df362905d7834fa
4076 
4077 The list of threads blocked on an MVar is now represented as a list of
4078 separately allocated objects rather than being linked through the TSOs
4079 themselves.  This lets us remove a TSO from the list in O(1) time
4080 rather than O(n) time, by marking the list object.  Removing this
4081 linear component fixes some pathalogical performance cases where many
4082 threads were blocked on an MVar and became unreachable simultaneously
4083 (nofib/smp/threads007), or when sending an asynchronous exception to a
4084 TSO in a long list of thread blocked on an MVar.
4085 
4086 MVar performance has actually improved by a few percent as a result of
4087 this change, slightly to my surprise.
4088 
4089 This is the final cleanup in the sequence, which let me remove the old
4090 way of waking up threads (unblockOne(), MSG_WAKEUP) in favour of the
4091 new way (tryWakeupThread and MSG_TRY_WAKEUP, which is idempotent).  It
4092 is now the case that only the Capability that owns a TSO may modify
4093 its state (well, almost), and this simplifies various things.  More of
4094 the RTS is based on message-passing between Capabilities now.
4095]
4096[eliminate some duplication with a bit of CPP
4097Simon Marlow <marlowsd@gmail.com>**20100330154355
4098 Ignore-this: 838f7d341f096ca14c86ab9c81193e36
4099]
4100[Make ioManagerDie() idempotent
4101Simon Marlow <marlowsd@gmail.com>**20100401100705
4102 Ignore-this: a5996b43cdb2e2d72e6e971d7ea925fb
4103 Avoids screeds of "event buffer overflowed; event dropped" in
4104 conc059(threaded1).
4105]
4106[Move a thread to the front of the run queue when another thread blocks on it
4107Simon Marlow <marlowsd@gmail.com>**20100329144521
4108 Ignore-this: c518ff0d41154680edc811d891826a29
4109 This fixes #3838, and was made possible by the new BLACKHOLE
4110 infrastructure.  To allow reording of the run queue I had to make it
4111 doubly-linked, which entails some extra trickiness with regard to
4112 GC write barriers and suchlike.
4113]
4114[remove non-existent MUT_CONS symbols
4115Simon Marlow <marlowsd@gmail.com>**20100330152600
4116 Ignore-this: 885628257a9d03f2ece2a754d993014a
4117]
4118[change throwTo to use tryWakeupThread rather than unblockOne
4119Simon Marlow <marlowsd@gmail.com>**20100329144613
4120 Ignore-this: 10ad4965e6c940db71253f1c72218bbb
4121]
4122[tiny GC optimisation
4123Simon Marlow <marlowsd@gmail.com>**20100329144551
4124 Ignore-this: 9e095b9b73fff0aae726f9937846ba92
4125]
4126[New implementation of BLACKHOLEs
4127Simon Marlow <marlowsd@gmail.com>**20100329144456
4128 Ignore-this: 96cd26793b4e6ab9ddd0d59aae5c2f1d
4129 
4130 This replaces the global blackhole_queue with a clever scheme that
4131 enables us to queue up blocked threads on the closure that they are
4132 blocked on, while still avoiding atomic instructions in the common
4133 case.
4134 
4135 Advantages:
4136 
4137  - gets rid of a locked global data structure and some tricky GC code
4138    (replacing it with some per-thread data structures and different
4139    tricky GC code :)
4140 
4141  - wakeups are more prompt: parallel/concurrent performance should
4142    benefit.  I haven't seen anything dramatic in the parallel
4143    benchmarks so far, but a couple of threading benchmarks do improve
4144    a bit.
4145 
4146  - waking up a thread blocked on a blackhole is now O(1) (e.g. if
4147    it is the target of throwTo).
4148 
4149  - less sharing and better separation of Capabilities: communication
4150    is done with messages, the data structures are strictly owned by a
4151    Capability and cannot be modified except by sending messages.
4152 
4153  - this change will utlimately enable us to do more intelligent
4154    scheduling when threads block on each other.  This is what started
4155    off the whole thing, but it isn't done yet (#3838).
4156 
4157 I'll be documenting all this on the wiki in due course.
4158 
4159]
4160[Fix warnings (allow pushOnRunQueue() to not be inlined)
4161Simon Marlow <marlowsd@gmail.com>**20100401114559
4162 Ignore-this: f40bfbfad70a5165a946d11371605b7d
4163]
4164[remove out of date comment
4165Simon Marlow <marlowsd@gmail.com>**20100401105853
4166 Ignore-this: 26af88dd418ee0bcda7223b3b7e4e8d2
4167]
4168[tidy up spacing in stderr traces
4169Simon Marlow <marlowsd@gmail.com>**20100326163122
4170 Ignore-this: 16558b0433a274be217d4bf39aa4946
4171]
4172[Fix an assertion that was not safe when running in parallel
4173Simon Marlow <marlowsd@gmail.com>**20100325143656
4174 Ignore-this: cad08fb8900eb3a475547af0189fcc47
4175]
4176[Never jump directly to a thunk's entry code, even if it is single-entry
4177Simon Marlow <marlowsd@gmail.com>**20100325114847
4178 Ignore-this: 938da172c06a97762ef605c8fccfedf1
4179 I don't think this fixes any bugs as we don't have single-entry thunks
4180 at the moment, but it could cause problems for parallel execution if
4181 we ever did re-introduce update avoidance.
4182]
4183[Rename forgotten -dverbose-simpl to -dverbose-core2core in the docs.
4184Milan Straka <fox@ucw.cz>**20100331153626
4185 Ignore-this: 2da58477fb96e1cfb80f37dddd7c422c
4186]
4187[Add -pa and -V to the documentation of time profiling options.
4188Milan Straka <fox@ucw.cz>**20100329191121
4189 Ignore-this: be74d216481ec5a19e5f40f85e6e3d65
4190]
4191[Keep gcc 4.5 happy
4192Simon Marlow <marlowsd@gmail.com>**20100330120425
4193 Ignore-this: 7811878cc2bd1ce9cfbb5bf102fe3454
4194]
4195[Fix warning compiling Linker.c for PPC Mac
4196naur@post11.tele.dk**20100403182355
4197 Ignore-this: e2d2448770c9714ce17dd6cf3e297063
4198 The warning message eliminated is:
4199 > rts/Linker.c:4756:0:
4200 >      warning: nested extern declaration of 'symbolsWithoutUnderscore'
4201]
4202[Fix error compiling AsmCodeGen.lhs for PPC Mac (mkRtsCodeLabel)
4203naur@post11.tele.dk**20100403181656
4204 Ignore-this: deb7524ea7852a15a2ac0849c8c82f74
4205 The error messages eliminated are:
4206 > compiler/nativeGen/AsmCodeGen.lhs:875:31:
4207 >     Not in scope: `mkRtsCodeLabel'
4208 > compiler/nativeGen/AsmCodeGen.lhs:879:31:
4209 >     Not in scope: `mkRtsCodeLabel'
4210 > compiler/nativeGen/AsmCodeGen.lhs:883:31:
4211 >     Not in scope: `mkRtsCodeLabel'
4212]
4213[Fix error compiling AsmCodeGen.lhs for PPC Mac (DestBlockId)
4214naur@post11.tele.dk**20100403180643
4215 Ignore-this: 71e833e94ed8371b2ffabc2cf80bf585
4216 The error message eliminated is:
4217 > compiler/nativeGen/AsmCodeGen.lhs:637:16:
4218 >     Not in scope: data constructor `DestBlockId'
4219]
4220[Fix boot-pkgs's sed usage to work with Solaris's sed
4221Ian Lynagh <igloo@earth.li>**20100401153441]
4222[Pass "-i org.haskell.GHC" to packagemaker when building the OS X installer
4223Ian Lynagh <igloo@earth.li>**20100331144707
4224 This seems to fix this failure:
4225 [...]
4226 ** BUILD SUCCEEDED **
4227 rm -f -f GHC-system.pmdoc/*-contents.xml
4228 /Developer/usr/bin/packagemaker -v --doc GHC-system.pmdoc\
4229              -o /Users/ian/to_release/ghc-6.12.1.20100330/GHC-6.12.1.20100330-i386.pkg
4230 2010-03-31 15:08:15.695 packagemaker[13909:807] Setting to : 0 (null)
4231 2010-03-31 15:08:15.709 packagemaker[13909:807] Setting to : 0 org.haskell.glasgowHaskellCompiler.ghc.pkg
4232 2010-03-31 15:08:15.739 packagemaker[13909:807] relocate: (null) 0
4233 2010-03-31 15:08:15.740 packagemaker[13909:807] *** Terminating app due to uncaught exception 'NSInvalidArgumentException', reason: '*** -[NSXMLDocument initWithXMLString:options:error:]: nil argument'
4234 2010-03-31 15:08:15.741 packagemaker[13909:807] Stack: (
4235     2511962091,
4236     2447007291,
4237     2511961547,
4238     2511961610,
4239     2432803204,
4240     453371,
4241     447720,
4242     436209,
4243     435510,
4244     9986,
4245     9918
4246 )
4247 make[1]: *** [framework-pkg] Trace/BPT trap
4248 make: *** [framework-pkg] Error 2
4249]
4250[Use machdepCCOpts when compiling the file to toggle -(no-)rtsopts
4251Ian Lynagh <igloo@earth.li>**20100331161302
4252 Should fix toggling on OS X "Snow Leopard". Diagnosed by Roman Leshchinskiy.
4253]
4254[Avoid a non-portable use of tar reported by Roman Leshchinskiy
4255Ian Lynagh <igloo@earth.li>**20100330145802]
4256[Don't install EXTRA_PACKAGES by default
4257Simon Marlow <marlowsd@gmail.com>**20100330142714
4258 Ignore-this: d4cc8f87a6de8d9d1d6dc9b77130b3
4259]
4260[fix a non-portable printf format
4261Simon Marlow <marlowsd@gmail.com>**20100330134437
4262 Ignore-this: d41c23c54ec29654cb2049de1e588570
4263]
4264[avoid single quote in #error
4265Simon Marlow <marlowsd@gmail.com>**20100330120346
4266 Ignore-this: 663f39e7a27fead2f648fbf22d345bb4
4267]
4268[use FMT_Word64 instead of locally-defined version
4269Simon Marlow <marlowsd@gmail.com>**20100330114650
4270 Ignore-this: 82697b8095dffb3a8e196c687006ece0
4271]
4272[remove old/unused DotnetSupport and GhcLibsWithUnix
4273Simon Marlow <marlowsd@gmail.com>**20100330123732
4274 Ignore-this: c68814868b3671abdc369105bbeafe6c
4275]
4276[fix return type cast in f.i.wrapper when using libffi (#3516)
4277Simon Marlow <marlowsd@gmail.com>**20100329154220
4278 Ignore-this: f898eb8c9ae2ca2009e539735b92c438
4279 
4280 Original fix submitted by
4281   Sergei Trofimovich <slyfox@community.haskell.org>
4282 modified by me:
4283  - exclude 64-bit types
4284  - compare uniques, not strings
4285  - #include "ffi.h" is conditional
4286]
4287[libffi: install 'ffitarget.h' header as sole 'ffi.h' is unusable
4288Simon Marlow <marlowsd@gmail.com>**20100329135734
4289 Ignore-this: f9b555ea289d8df1aa22cb6faa219a39
4290 Submitted by: Sergei Trofimovich <slyfox@community.haskell.org>
4291 Re-recorded against HEAD.
4292]
4293[avoid a fork deadlock (see comments)
4294Simon Marlow <marlowsd@gmail.com>**20100329132329
4295 Ignore-this: 3377f88b83bb3b21e42d7fc5f0d866f
4296]
4297[tidy up the end of the all_tasks list after forking
4298Simon Marlow <marlowsd@gmail.com>**20100329132253
4299 Ignore-this: 819d679875be5f344e816210274d1c29
4300]
4301[Add a 'setKeepCAFs' external function (#3900)
4302Simon Marlow <marlowsd@gmail.com>**20100329110036
4303 Ignore-this: ec532a18cad4259a09847b0b9ae2e1d2
4304]
4305[Explicitly check whether ar supports the @file syntax
4306Ian Lynagh <igloo@earth.li>**20100329123325
4307 rather than assuming that all GNU ar's do.
4308 Apparently OpenBSD's older version doesn't.
4309]
4310[Fix the format specifier for Int64/Word64 on Windows
4311Ian Lynagh <igloo@earth.li>**20100327182126
4312 mingw doesn't understand %llu/%lld - it treats them as 32-bit rather
4313 than 64-bit. We use %I64u/%I64d instead.
4314]
4315[Fix the ghci startmenu item
4316Ian Lynagh <igloo@earth.li>**20100326235934
4317 I'm not sure what changed, but it now doesn't work for me without
4318 the "Start in" field being set.
4319]
4320[Fix paths to docs in "Start Menu" entries in Windows installer; fixes #3847
4321Ian Lynagh <igloo@earth.li>**20100326155917]
4322[Add a licence file for the Windows installer to use
4323Ian Lynagh <igloo@earth.li>**20100326155130]
4324[Add gcc-g++ to the inplace mingw installation; fixes #3893
4325Ian Lynagh <igloo@earth.li>**20100326154714]
4326[Add the licence file to the Windows installer. Fixes #3934
4327Ian Lynagh <igloo@earth.li>**20100326152449]
4328[Quote the paths to alex and happy in configure
4329Ian Lynagh <igloo@earth.li>**20100325143449
4330 Ignore-this: d6d6e1a250f88985bbeea760e63a79db
4331]
4332[Use </> rather than ++ "/"
4333Ian Lynagh <igloo@earth.li>**20100325133237
4334 This stops us generating paths like
4335     c:\foo\/ghc460_0/ghc460_0.o
4336 which windres doesn't understand.
4337]
4338[Append $(exeext) to utils/ghc-pkg_dist_PROG
4339Ian Lynagh <igloo@earth.li>**20100324233447
4340 Fixes bindist creation
4341]
4342[A sanity check
4343Simon Marlow <marlowsd@gmail.com>**20100325110500
4344 Ignore-this: 3b3b76d898c822456857e506b7531e65
4345]
4346[do_checks: do not set HpAlloc if the stack check fails
4347Simon Marlow <marlowsd@gmail.com>**20100325110328
4348 Ignore-this: 899ac8c29ca975d03952dbf4608d758
4349 
4350 This fixes a very rare heap corruption bug, whereby
4351 
4352  - a context switch is requested, which sets HpLim to zero
4353    (contextSwitchCapability(), called by the timer signal or
4354    another Capability).
4355 
4356  - simultaneously a stack check fails, in a code fragment that has
4357    both a stack and a heap check.
4358 
4359 The RTS then assumes that a heap-check failure has occurred and
4360 subtracts HpAlloc from Hp, although in fact it was a stack-check
4361 failure and retreating Hp will overwrite valid heap objects.  The bug
4362 is that HpAlloc should only be set when Hp has been incremented by the
4363 heap check.  See comments in rts/HeapStackCheck.cmm for more details.
4364 
4365 This bug is probably incredibly rare in practice, but I happened to be
4366 working on a test that triggers it reliably:
4367 concurrent/should_run/throwto001, compiled with -O -threaded, args 30
4368 300 +RTS -N2, run repeatedly in a loop.
4369]
4370[comments and formatting only
4371Simon Marlow <marlowsd@gmail.com>**20100325104617
4372 Ignore-this: c0a211e15b5953bb4a84771bcddd1d06
4373]
4374[Change how perl scripts get installed; partially fixes #3863
4375Ian Lynagh <igloo@earth.li>**20100324171422
4376 We now regenerate them when installing, which means the path for perl
4377 doesn't get baked in
4378]
4379[Pass the location of gcc in the ghc wrapper script; partially fixes #3863
4380Ian Lynagh <igloo@earth.li>**20100324171408
4381 This means we don't rely on baking a path to gcc into the executable
4382]
4383[Quote the ar path in configure
4384Ian Lynagh <igloo@earth.li>**20100324162043]
4385[Remove unused cUSER_WAY_NAMES cUSER_WAY_OPTS
4386Ian Lynagh <igloo@earth.li>**20100324145048]
4387[Remove unused cCONTEXT_DIFF
4388Ian Lynagh <igloo@earth.li>**20100324145013]
4389[Remove unused cEnableWin32DLLs
4390Ian Lynagh <igloo@earth.li>**20100324144841]
4391[Remove unused cGHC_CP
4392Ian Lynagh <igloo@earth.li>**20100324144656]
4393[Fix the build for non-GNU-ar
4394Ian Lynagh <igloo@earth.li>**20100324132907]
4395[Tweak the Makefile code for making .a libs; fixes trac #3642
4396Ian Lynagh <igloo@earth.li>**20100323221325
4397 The main change is that, rather than using "xargs ar" we now put
4398 all the filenames into a file, and do "ar @file". This means that
4399 ar adds all the files at once, which works around a problem where
4400 files with the same basename in a later invocation were overwriting
4401 the existing file in the .a archive.
4402]
4403[Enable shared libraries on Windows; fixes trac #3879
4404Ian Lynagh <igloo@earth.li>**20100320231414
4405 Ignore-this: c93b35ec5b7a7fa6ddb286d17a616216
4406]
4407[Add the external core PDF to the new build system
4408Ian Lynagh <igloo@earth.li>**20100321161909]
4409[Allow specifying $threads directly when validating
4410Ian Lynagh <igloo@earth.li>**20100321112835]
4411[Remove LazyUniqFM; fixes trac #3880
4412Ian Lynagh <igloo@earth.li>**20100320213837]
4413[UNDO: slight improvement to scavenging ...
4414Simon Marlow <marlowsd@gmail.com>**20100319153413
4415 Ignore-this: f0ab581c07361f7b57eae02dd6ec893c
4416 
4417 Accidnetally pushed this patch which, while it validates, isn't
4418 correct.
4419 
4420 rolling back:
4421 
4422 Fri Mar 19 11:21:27 GMT 2010  Simon Marlow <marlowsd@gmail.com>
4423   * slight improvement to scavenging of update frames when a collision has occurred
4424 
4425     M ./rts/sm/Scav.c -19 +15
4426]
4427[slight improvement to scavenging of update frames when a collision has occurred
4428Simon Marlow <marlowsd@gmail.com>**20100319112127
4429 Ignore-this: 6de2bb9614978975f17764a0f259d9bf
4430]
4431[Don't install the utf8-string package
4432Ian Lynagh <igloo@earth.li>**20100317212709]
4433[Don't use -Bsymbolic when linking the RTS
4434Ian Lynagh <igloo@earth.li>**20100316233357
4435 This makes the RTS hooks work when doing dynamic linking
4436]
4437[Fix Trac #3920: Template Haskell kinds
4438simonpj@microsoft.com**20100317123519
4439 Ignore-this: 426cac7920446e04f3cc30bd1d9f76e2
4440 
4441 Fix two places where we were doing foldl instead of foldr
4442 after decomposing a Kind.  Strange that the same bug appears
4443 in two quite different places!
4444]
4445[copy_tag_nolock(): fix write ordering and add a write_barrier()
4446Simon Marlow <marlowsd@gmail.com>**20100316143103
4447 Ignore-this: ab7ca42904f59a0381ca24f3eb38d314
4448 
4449 Fixes a rare crash in the parallel GC.
4450 
4451 If we copy a closure non-atomically during GC, as we do for all
4452 immutable values, then before writing the forwarding pointer we better
4453 make sure that the closure itself is visible to other threads that
4454 might follow the forwarding pointer.  I imagine this doesn't happen
4455 very often, but I just found one case of it: in scavenge_stack, the
4456 RET_FUN case, after evacuating ret_fun->fun we then follow it and look
4457 up the info pointer.
4458]
4459[Add sliceP mapping to vectoriser builtins
4460benl@ouroborus.net**20100316060517
4461 Ignore-this: 54c3cafff584006b6fbfd98124330aa3
4462]
4463[Comments only
4464benl@ouroborus.net**20100311064518
4465 Ignore-this: d7dc718cc437d62aa5b1b673059a9b22
4466]
4467[TAG 2010-03-16
4468Ian Lynagh <igloo@earth.li>**20100316005137
4469 Ignore-this: 234e3bc29e2f26cc59d7b03d780cc352
4470]
4471Patch bundle hash:
447260a473180fb88cf3c5f27c6b461f8034ba9d100e
4473-----BEGIN PGP SIGNATURE-----
4474Version: GnuPG v1.4.10 (Darwin)
4475
4476iD8DBQFMuuDIFOecpxqG73IRAswOAKCJ4/0fDtCMDSs99igZlBxnEnFJUACdFSJa
44775lBYSwqY9P3cP98L6KcBlYU=
4478=6baf
4479-----END PGP SIGNATURE-----