Ticket #4476: eventlog-header-fix.patch

File eventlog-header-fix.patch, 170.5 KB (added by scpmw, 3 years ago)
Line 
11 patch for repository http://darcs.haskell.org/ghc:
2
3Fri Nov  5 15:16:55 GMT 2010  scpmw@leeds.ac.uk
4  * Eventlog: Put correct size for startup event
5
6New patches:
7
8[Eventlog: Put correct size for startup event
9scpmw@leeds.ac.uk**20101105151655
10 Ignore-this: 8b6eb4fa2137c8dfe50c5917e3a609a7
11] {
12hunk ./rts/eventlog/EventLog.c 238
13                 sizeof(EventThreadID) + sizeof(StgWord16);
14             break;
15 
16+        case EVENT_STARTUP:         // (cap count)
17+            eventTypes[t].size = sizeof(EventCapNo);
18+            break;
19+
20         case EVENT_SHUTDOWN:        // (cap)
21         case EVENT_REQUEST_SEQ_GC:  // (cap)
22         case EVENT_REQUEST_PAR_GC:  // (cap)
23hunk ./rts/eventlog/EventLog.c 247
24         case EVENT_GC_START:        // (cap)
25         case EVENT_GC_END:          // (cap)
26-        case EVENT_STARTUP:
27         case EVENT_GC_IDLE:
28         case EVENT_GC_WORK:
29         case EVENT_GC_DONE:
30}
31
32Context:
33
34[Fix IPRun by fixing the inferred quantification mechanism
35simonpj@microsoft.com**20101019090220
36 Ignore-this: e7dc8c35979136237fa9a572b70f298e
37]
38[Recover after an error in an implication constraint
39simonpj@microsoft.com**20101019090202
40 Ignore-this: fbbbdc2c3fc0bac0662d6b40e5d57d47
41]
42[Reject programs with equality superclasses for now
43simonpj@microsoft.com**20101019090140
44 Ignore-this: 3e3eb8c876fa95290f0278e800fce5bc
45]
46[Layout and tiny refactoring only
47simonpj@microsoft.com**20101019090124
48 Ignore-this: ee21565fde8cd45514d83d0ca9af3253
49]
50[Clean up the debugger code
51simonpj@microsoft.com**20101019090100
52 Ignore-this: 8677a9084d787f45542419a58e5d8866
53 
54 In particular there is much less fiddly skolemisation now
55 Things are not *quite* right (break001 and 006 still fail),
56 but they are *much* better than before.
57]
58[Add new VarEnv functions minusVarEnv, intersectsVarEnv, unionInScope
59simonpj@microsoft.com**20101019085609
60 Ignore-this: a3a8e23fe522079e3eb9c34ea8032602
61]
62[Major pass through type checker:(1) prioritizing equalities, (2) improved Derived mechanism, (3) bugfixes
63dimitris@microsoft.com**20101018151510]
64[(1) Caching FD improvements for efficiency, (2) preventing cascading deriveds from entering the inert, (3) Fixing bugs in the creation of FlexiTcS variables
65dimitris@microsoft.com**20101015164421]
66[Midstream changes for performance improvement related to superclasses and functional dependencies.
67dimitris@microsoft.com**20101014143811]
68[Minor
69dimitris@microsoft.com**20101012080951]
70[Commentary changes
71dimitris@microsoft.com**20101011142235]
72[Kind checking bugfix (#4356) and preventing wanteds from rewriting wanteds
73dimitris@microsoft.com**20101008173700]
74[Fix a retainer profiling segfault
75Ian Lynagh <igloo@earth.li>**20101019132727
76 Ignore-this: 19f9e2a6dc8a64b30f3769f4101f82db
77 The bitmap type wasn't big enough to hold large bitmaps on 64 bit
78 platforms. Profiling GHC was segfaulting when retainStack was handling a
79 size 33 bitmap.
80]
81[Fix -auto-all: Add SCCs to IDs which have a monotype too
82Ian Lynagh <igloo@earth.li>**20101018153957]
83[Define SpecConstrAnnotation in GHC.Exts, and import it from there
84simonpj@microsoft.com**20101018135746
85 Ignore-this: a6ade815584a1d03be55a6c417de0ab0
86 
87 Reason: avoid having to link the entire ghc package in modules
88 that use compile-time annotations:
89 
90      import GHC.Exts( SpecConstrAnnotation )
91      {-# ANN type T ForceSpecConstr #-}
92 
93 It's a kind of bug that the package exporting SpecConstrAnnotation
94 is linked even though it is only needed at compile time, but putting
95 the data type declaration in GHC.Exts is a simple way to sidestep
96 the problem
97 
98 See See Note [SpecConstrAnnotation] in SpecConstr
99]
100[Fix warnings in AsmCodeGen
101David Terei <davidterei@gmail.com>**20101007143559
102 Ignore-this: 1c34b3a39968a5ca123862726f36a8ee
103]
104[LLVM: Fix compilation of writebarrier, #4308
105David Terei <davidterei@gmail.com>**20101004153843
106 Ignore-this: df256e78b90e3fc263a88d4f0af6ea96
107]
108[Change how the OS X installer's create-links finds the versin number
109Ian Lynagh <igloo@earth.li>**20101017122352
110 Ignore-this: 543d3f40c2ba5daf8b2b4ea22eb4bfca
111 It now gets created by configure, rather than trying to work out the
112 version number at runtime.
113]
114[Add more quoting to distrib/MacOS/installer-scripts/create-links
115Ian Lynagh <igloo@earth.li>**20101017112648
116 Ignore-this: 5119ea2efc925d568d96c616af550c7c
117]
118[change os x installer to allow multiple installed versions
119Evan Laforge <qdunkan@gmail.com>**20100929234538
120 Ignore-this: 4f70551293f53a52c39c5962d8a5bd5b
121 This puts the ghc version into the package name so they are considered separate
122 packages.
123]
124[Only put the boot packages in the haddock contents/index
125Ian Lynagh <igloo@earth.li>**20101016180031
126 Ignore-this: 3dc276dc734fec207def24a1d10eb369
127 We don't install dph etc, so don't put them in the doc index.
128]
129[Correct the regexp used to search for extra packages
130Ian Lynagh <igloo@earth.li>**20101016123421
131 Ignore-this: 98e3f8fda1c5678730a00ec1ac25fde3
132 We weren't ignoring comment lines
133]
134[New member "archiveMemberName" for struct _ObjectCode
135pho@cielonegro.org**20100927224145
136 Ignore-this: 628bc605e7dc4f0c4856c6f7ad23d9ee
137 
138 struct _ObjectCode should be able to retain the name of archive members.
139 Though currently the only use of those names are for debugging outputs.
140]
141[Add a -fghci-sandbox flag so that we can en/disable the ghci sandbox
142Ian Lynagh <igloo@earth.li>**20101015172746
143 Ignore-this: 52bc5729437a93a925d3586d9803623c
144 It's on by default (which matches the previous behaviour).
145 
146 Motivation:
147 GLUT on OS X needs to run on the main thread. If you
148 try to use it from another thread then you just get a
149 white rectangle rendered. For this, or anything else
150 with such restrictions, you can turn the GHCi sandbox off
151 and things will be run in the main thread.
152]
153[Fix boot; it was failing if darcs-all or validate were missing
154Ian Lynagh <igloo@earth.li>**20101015164549
155 Ignore-this: 14b3e98836647aff345e35bca2102f93
156 (which is the case in sdists)
157]
158[Comments and layout
159simonpj@microsoft.com**20101015131924
160 Ignore-this: 126fbdb629a08c1380c7a1f5cd967d97
161]
162[Make (Located a) an instance of Eq, Ord
163simonpj@microsoft.com**20101015131857
164 Ignore-this: 5da50f8dab06fcbc23ce149cd5080062
165 
166 Fulfils Trac #4369
167]
168[Give user-defined rules precedence over built-in rules
169simonpj@microsoft.com**20101015131814
170 Ignore-this: 76fbf36cb1a09e31c10c68c06b98937e
171 
172 This fixes Trac #4397.  See comments with 'isMoreSpecific'.
173]
174[Fix Trac #4401: meta-tyvars allocated by the constraint solver are always touchable
175simonpj@microsoft.com**20101015130818
176 Ignore-this: 7fcdaee825d9cadcf69c8c9b8967a446
177 
178   See Note [Touchable meta type variables] in TcSMonad
179]
180[Remove GHC.extendGlobalRdrScope, GHC.extendGlobalTypeScope
181simonpj@microsoft.com**20101013091107
182 Ignore-this: df65a43d261353ad53811b25507e913e
183 
184 These functions were added by
185 
186    Tue Apr 18 03:36:06 BST 2006  Lemmih <lemmih@gmail.com>
187    * Make the initial rdr and type scope available in the ghc-api
188 
189 The are extremely dubious, because they extend the Rdr and Type
190 env for every compilation.  The right thing to do is to use
191 the InteractiveContext for temporary extensions.
192 
193 So far as we know, no one uses them.  And if they are being used
194 it's probably a mistake.  So we're backing them out.
195]
196[InlinePrag needs an arity only for INLINE, not INLINABLE
197Simon Marlow <marlowsd@gmail.com>**20101015094925
198 Ignore-this: 3b338f0b2b49fa714b195067e0026ef7
199 This doesn't fix anything (we think), but it's morally correct.
200]
201[Fix #4346 (INLINABLE pragma not behaving consistently)
202Simon Marlow <marlowsd@gmail.com>**20101015094836
203 Ignore-this: 9a9d8ad42cfdf7f619adfac284bae021
204 Debugged thanks to lots of help from Simon PJ: we weren't updating the
205 UnfoldingGuidance when the unfolding changed.
206 Also, a bit of refactoring and additinoal comments.
207]
208[Have boot check that we have the dph packages when validating
209Ian Lynagh <igloo@earth.li>**20101014140556
210 Ignore-this: 61e365127933f4dbfb62be66115455bd
211]
212[Add more documentation for interruptible foreign calls
213Simon Marlow <marlowsd@gmail.com>**20101014084253
214 Ignore-this: 28ec0ddd958926a08ab816b6975da344
215]
216[minor refactoring
217Simon Marlow <marlowsd@gmail.com>**20100926105819
218 Ignore-this: 70d38e0a31a096c94dad5f346b0d91f6
219]
220[Fix for interruptible FFI handling
221Simon Marlow <marlowsd@gmail.com>**20100925193442
222 Ignore-this: a63ba3c60f577002a1d32b30bb45090c
223 Set tso->why_blocked before calling maybePerformBlockedException(), so
224 that throwToSingleThreaded() doesn't try to unblock the current thread
225 (it is already unblocked).
226]
227[interruptible FFI: more robust handling of the exception case in the interpreter
228Simon Marlow <marlowsd@gmail.com>**20100925193317
229 Ignore-this: 7f9e67835a3bd096154db2dcf533ec66
230]
231[Don't interrupt when task blocks exceptions, don't immediately start exception.
232Edward Z. Yang <ezyang@mit.edu>**20100925033026
233 Ignore-this: 6c0669995a2b66abf29d68b3711cb78e
234]
235[Interruptible FFI calls with pthread_kill and CancelSynchronousIO. v4
236Edward Z. Yang <ezyang@mit.edu>**20100919002905
237 Ignore-this: 43c260f90eeb9c03413f6e749e23101d
238 
239 This is patch that adds support for interruptible FFI calls in the form
240 of a new foreign import keyword 'interruptible', which can be used
241 instead of 'safe' or 'unsafe'.  Interruptible FFI calls act like safe
242 FFI calls, except that the worker thread they run on may be interrupted.
243 
244 Internally, it replaces BlockedOnCCall_NoUnblockEx with
245 BlockedOnCCall_Interruptible, and changes the behavior of the RTS
246 to not modify the TSO_ flags on the event of an FFI call from
247 a thread that was interruptible.  It also modifies the bytecode
248 format for foreign call, adding an extra Word16 to indicate
249 interruptibility.
250 
251 The semantics of interruption vary from platform to platform, but the
252 intent is that any blocking system calls are aborted with an error code.
253 This is most useful for making function calls to system library
254 functions that support interrupting.  There is no support for pre-Vista
255 Windows.
256 
257 There is a partner testsuite patch which adds several tests for this
258 functionality.
259]
260[Remove ghc-pkg's dependency on haskell98
261Ian Lynagh <igloo@earth.li>**20101013194356
262 Ignore-this: 273fdede3f21e682faa4dc6e61f4e7aa
263]
264[Build haskell98 and haskell2010 with stage2
265Ian Lynagh <igloo@earth.li>**20101013182759
266 Stops us accidentally depending on them
267]
268[Fix warning: Remove unused import
269Ian Lynagh <igloo@earth.li>**20101013141224
270 Ignore-this: ae8891440811f4676d9bb8e721ad22ca
271]
272[Fix warnings
273benl@ouroborus.net**20101013040335
274 Ignore-this: 4e5875827d2840de863000cdb35afb0e
275]
276[RegAlloc: Track slot liveness over jumps in spill cleaner
277benl@ouroborus.net**20101013015414
278 Ignore-this: ccd4a148908b7fbdc6ea76acf527c16b
279]
280[Bump Cabal dep
281Ian Lynagh <igloo@earth.li>**20101012154528
282 Ignore-this: da8539b957b74c7da91efff8ac25872
283]
284[Remove __HASKELL1__, __HASKELL98__, __CONCURRENT_HASKELL__
285Ian Lynagh <igloo@earth.li>**20101012134700
286 Ignore-this: d0db531be58537c52802de1c62694eb1
287 We used to define these CPP symbols, but nothing on hackage uses them
288 and the first 2 are no longer correct (as we support multiple Haskell
289 versions).
290]
291[Follow Cabal changes: Cabal no longer has a docbook userguide
292Ian Lynagh <igloo@earth.li>**20101012130538
293 Ignore-this: bf0b30e465c13f82725d72aec145e939
294 For now we don't build the Cabal userguide, but we should add markdown
295 support so that we can do so.
296]
297[Fix build on Windows: ghc-pkg/Main.hs needs ForeignFunctionInterface
298Ian Lynagh <igloo@earth.li>**20101012112111]
299[Remove unnecessary import
300Ian Lynagh <igloo@earth.li>**20101010222231
301 Ignore-this: 6c7d5cee72837e2af43c2807e10ad602
302]
303[Make "./validate --slow" run the full testsuite
304Ian Lynagh <igloo@earth.li>**20101007004327
305 Ignore-this: 2dee8262c19fd5d5034a186203e20d0f
306]
307[Fix build following haskell98 and -fglasgow-exts changes
308Ian Lynagh <igloo@earth.li>**20101006160656
309 Ignore-this: c56bdb0f01da897b3de75bcab1e2887c
310]
311[Don't automatically link the haskell98 package
312Ian Lynagh <igloo@earth.li>**20101006130235
313 The default language is now Haskell2010, so this was a little odd.
314 Also, --make is now on by default, so this was largely irrelevant.
315]
316[Deprecate -fglasgow-exts
317Ian Lynagh <igloo@earth.li>**20101006124413
318 Ignore-this: 70be67898a633ab31d76f6fca557d55
319]
320[Remove Opt_GADTs and Opt_TypeFamilies from -fglasgow-exts
321Ian Lynagh <igloo@earth.li>**20101006122000
322 Ignore-this: 7153ec880b2b5f9d332ae2f57b97afcc
323 This means most code doesn't get caught by monomorphic local bindings.
324]
325[Fix Trac #4360: omitted case in combineCtLoc
326simonpj@microsoft.com**20101008135747
327 Ignore-this: 834636a97af6469862a822809253db41
328]
329[Beautiful new approach to the skolem-escape check and untouchable
330simonpj@microsoft.com**20101008133751
331 Ignore-this: 33517a772cfdfbf4aa4678609f3dcd71
332 
333 Instead of keeping a *set* of untouchable variables in each
334 implication contraints, we keep a *range* of uniques for the
335 *touchable* variables of an implication.  This are precisely
336 the ones we would call the "existentials" if we were French.
337 
338 It turns out that the code is more efficient, and vastly easier
339 to get right, than the set-based approach.
340 
341 Fixes Trac #4355 among others
342]
343[Do less simplification when doing let-generalisation
344simonpj@microsoft.com**20101008133542
345 Ignore-this: 71366d0de37f10ffba2edc9f3927ddbe
346 
347 This fixes Trac #4361.  In a rather delicate way, but
348 no more delicate than before.  A more remoseless typechecker
349 would reject #4361 altogether.
350 
351 See Note [Avoid unecessary constraint simplification]
352]
353[Suppress ambiguity errors if any other errors occur
354simonpj@microsoft.com**20101008111318
355 Ignore-this: 40f014265c1ab15fe172baaf76c23c87
356]
357[Fix Trac #4361: be more discerning when inferring types
358simonpj@microsoft.com**20101008111227
359 Ignore-this: 33656f9a151f494b26b6318b5fcffef
360 
361 Note [Avoid unecessary constraint simplification] in TcSimplify
362]
363[Float out partial applications
364Simon Marlow <marlowsd@gmail.com>**20101008092709
365 Ignore-this: 2dc9d10597c19d0598ef2fc3cf74156d
366 
367 This fixes at least one case of performance regression in 7.0, and
368 is nice win on nofib:
369 
370         Program           Size    Allocs   Runtime   Elapsed
371             Min          +0.3%    -63.0%    -38.5%    -38.7%
372             Max          +1.2%     +0.2%     +0.9%     +0.9%
373  Geometric Mean          +0.6%     -3.0%     -6.4%     -6.6%
374]
375[Suppress knock-on typechecker errors
376simonpj@microsoft.com**20101008094348
377 Ignore-this: 8d125926286a7614fa1ce998e3b26d04
378 
379 The error cascade caused puzzling errors in T4093b, and
380 suppressing some seems like a good plan.  Very few test
381 outputs change.
382]
383[Some refactoring and simplification in TcInteract.occurCheck
384simonpj@microsoft.com**20101007163500
385 Ignore-this: d43d09370ab27b8796062e2e98ce7e9
386]
387[Comments only
388simonpj@microsoft.com**20101007130301
389 Ignore-this: ab46592edd3d24786bbce42c50feb4fd
390]
391[Implement auto-specialisation of imported Ids
392simonpj@microsoft.com**20101007111051
393 Ignore-this: 45257ff6e9597e4fa4de10b0657e27d6
394 
395 This big-ish patch arranges that if an Id 'f' is
396   * Type-class overloaded
397        f :: Ord a => [a] -> [a]
398   * Defined with an INLINABLE pragma
399        {-# INLINABLE f #-}
400   * Exported from its defining module 'D'
401 
402 then in any module 'U' that imports D
403 
404 1. Any call of 'f' at a fixed type will generate
405    (a) a specialised version of f in U
406    (b) a RULE that rewrites unspecialised calls to the
407        specialised on
408 
409   e.g. if the call is (f Int dOrdInt xs) then the
410   specialiser will generate
411      $sfInt :: [Int] -> [Int]
412      $sfInt = <code for f, imported from D, specialised>
413      {-# RULE forall d.  f Int d = $sfInt #-}
414 
415 2. In addition, you can give an explicit {-# SPECIALISE -#}
416    pragma for the imported Id
417      {-# SPECIALISE f :: [Bool] -> [Bool] #-}
418    This too generates a local specialised definition,
419    and the corresponding RULE
420 
421 The new RULES are exported from module 'U', so that any module
422 importing U will see the specialised versions of 'f', and will
423 not re-specialise them.
424 
425 There's a flag -fwarn-auto-orphan that warns you if the auto-generated
426 RULES are orphan rules. It's not in -Wall, mainly to avoid lots of
427 error messages with existing packages.
428 
429 Main implementation changes
430 
431  - A new flag on a CoreRule to say if it was auto-generated.
432    This is persisted across interface files, so there's a small
433    change in interface file format.
434 
435  - Quite a bit of fiddling with plumbing, to get the
436    {-# SPECIALISE #-} pragmas for imported Ids.  In particular, a
437    new field tgc_imp_specs in TcGblEnv, to keep the specialise
438    pragmas for imported Ids between the typechecker and the desugarer.
439 
440  - Some new code (although surprisingly little) in Specialise,
441    to deal with calls of imported Ids
442]
443[Make NameEnv back into type NameEnv a = UniqFM a
444simonpj@microsoft.com**20101007104638
445 Ignore-this: 24857c013461788be354520e84f4c286
446 
447 I don't think the type distinction of declaring NameEnv with a newtype
448 (as it was) is really useful to us. Moreover, VarEnv is a UniqFM, and
449 I do sometimes want to build an envt with Ids and look up with Names.
450 
451 This may not be the last word on the subject.
452]
453[Improve the rule-matcher
454simonpj@microsoft.com**20101007103700
455 Ignore-this: 9de96237dc4b73a43326bd568e34b53b
456 
457 Previously it was rejecting the match
458 
459   Template: forall s t. map s t
460   Actual:   map Int t
461 
462 which should obviously be fine.  It turns out that this kind of match
463 comes up when specialising.  By freshening that t we could avoid the
464 difficulty, but morally the (forall t) binds t and the rule should
465 be alpha-equivalent regardless of the forall'd variables.
466 
467 This patch makes it so, and incidentally makes matching a little
468 more efficient.  See Note [Eta expansion] in VarEnv.
469]
470[Fix Trac #4345: simplifier bug
471simonpj@microsoft.com**20101007102720
472 Ignore-this: 261c1c9f094df344ce34de814f8b60c5
473 
474 This is another long-standing bug, in which there was a possibility
475 that a loop-breaker could lose its loop-breaker-hood OccInfo,
476 and then the simplifer re-simplified the expression. Result, either
477 non-termination or, in the case of #4345, an unbound identifier.
478 
479 The fix is very simple, in Id.transferPolyIdInfo.
480 See Note [transferPolyIdInfo].
481]
482[Avoid redundant simplification
483simonpj@microsoft.com**20101007095935
484 Ignore-this: 61bd1a2c508260f558866e6a88c29fa3
485 
486 When adding specialisation for imported Ids, I noticed that the
487 Glorious Simplifier was repeatedly (and fruitlessly) simplifying the
488 same term.  It turned out to be easy to fix this, because I already
489 had a flag in the ApplyTo and Select constructors of SimplUtils.SimplCont.
490 
491 See Note [Avoid redundant simplification]
492]
493[Make the occurrence analyser deal correctly with RULES for imported Ids
494simonpj@microsoft.com**20101007094100
495 Ignore-this: 335b1cad013524e42b31e88c0a7a00f6
496 
497 This patch fixes a long-standing lurking bug, but it surfaced when I
498 was adding specialisation for imported Ids.
499 
500 See Note [ImpRuleUsage], which explains the issue.   The solution
501 seems more complicated than the problem really deserves, but I
502 could not think of a simpler way, so I just bit the bullet and
503 wrote the code.  Improvements welcome.
504]
505[Make warning-free
506simonpj@microsoft.com**20101007092007
507 Ignore-this: 4bae0c470a8a1f96d21990d1f3cf1f93
508]
509[This is just white-space and layout
510simonpj@microsoft.com**20101007091618
511 Ignore-this: 759c0335df70fce32558e967f140803a
512 
513 (At least, I don't think there is anything else.)
514]
515[Fix an ASSERT failure in FamInstEnv
516simonpj@microsoft.com**20101007091327
517 Ignore-this: a8c08ccb7ec2bc65864a674b5441539
518 
519 I added a lot of comments too, to explain the preconditions;
520 esp Note [FamInstEnv]
521]
522[Fix a looping bug in the new occur-check code
523simonpj@microsoft.com**20101007084104
524 Ignore-this: a02a2deafb9ec986ef1565f4596049ed
525]
526[Fix test T4235 with -O
527simonpj@microsoft.com**20101006155223
528 Ignore-this: f0fa0fe2f0c493e362d528d71f7a64e1
529 
530 The tag2Enum rule wasn't doing the right thing for
531 enumerations with a phantom type parameter, like
532    data T a = A | B
533]
534[Make warning-free
535simonpj@microsoft.com**20101006155033
536 Ignore-this: 221a3c95a6079c6ecc0468996a38b048
537]
538[Major bugfixing pass through the type checker
539dimitris@microsoft.com**20101006152854]
540[Typechecker performance fixes and flatten skolem bugfixing
541dimitris@microsoft.com**20101004130200
542 Ignore-this: 86721ba3f09479c146a0710796b43459
543]
544[Performance bug fixes
545dimitris@microsoft.com**20100923143918]
546[Fix Trac #4371: matching of view patterns
547simonpj@microsoft.com**20101006115316
548 Ignore-this: 494b28b91f1e6392b2f1521cda0e83b1
549]
550[Remove unused NoMatchContext construtor
551simonpj@microsoft.com**20101006115251
552 Ignore-this: 8985ff1dac51fb652bd65657a630a792
553]
554[Refactoring: mainly rename ic_env_tvs to ic_untch
555simonpj@microsoft.com**20101006102830
556 Ignore-this: 32999403a3f447e14b59cec7896027ff
557 
558 Plus remember to zonk the free_tvs in TcUnify.newImplication
559]
560[remove unnecessary/broken definition of mask_
561Simon Marlow <marlowsd@gmail.com>**20101002195118
562 Ignore-this: 4cdc9c95d40e01cdfe2d0ac411476603
563]
564[-fwarn-tabs: add "Warning" to the message
565Simon Marlow <marlowsd@gmail.com>**20101002195100
566 Ignore-this: 589a36daa3426ab51f2fb140e38df6c
567]
568[give a better error message in the non-threaded RTS for out-of-range FDs
569Simon Marlow <marlowsd@gmail.com>**20100929212916
570 Ignore-this: e94c9f390b8f79d24895a80f9d16c8d9
571 
572 # ./aw
573 aw: file descriptor 1027 out of range for select (0--1024).
574 Recompile with -threaded to work around this.
575]
576[Fix a very rare crash in GHCi
577Simon Marlow <marlowsd@gmail.com>**20101005144735
578 Ignore-this: dad1cd08934bae2ba47e72c0c000acfa
579 When a BCO with a zero-length bitmap was right at the edge of
580 allocated memory, we were reading a word of non-existent memory.
581 
582 This showed up as a segfault in T789(ghci) for me, but the crash was
583 extremely sensitive and went away with most changes.
584 
585 Also, optimised scavenge_large_bitmap a bit while I was in there.
586]
587[Using 'stdcall' when it is not supported is only a warning now (#3336)
588Simon Marlow <marlowsd@gmail.com>**20100924152445
589 Ignore-this: 66c5903a600a47485a7583535bb38455
590]
591[remove unnecessary stg_noForceIO (#3508)
592Simon Marlow <marlowsd@gmail.com>**20100924150202
593 Ignore-this: dec52de9cd9da7dcedae12b20691aba9
594]
595[Replace an outputStr with putStrLn calls; fixes #4332
596Ian Lynagh <igloo@earth.li>**20101003125707
597 Ignore-this: eb4b5f60d9d9d3f7dc203869927b28ba
598]
599[make test and fulltest targets in the main Makefile; fixes #4297
600Ian Lynagh <igloo@earth.li>**20100930224741
601 You can now run "make test" in the root, and the fast testsuite will be
602 run with cleaning enabled. It will also put the summary in
603 testsuite_summary.txt.
604]
605[Don't show the loaded packages in ":show packages"; fixes #4300
606Ian Lynagh <igloo@earth.li>**20100930210128
607 It's never worked properly, and the information is in ":show linker".
608]
609[Handle EXTRA_LIBRARIES when building programs
610Ian Lynagh <igloo@earth.li>**20100930192552
611 Ignore-this: 401a26e18d25dcaee010b13eaed8f011
612 And set hp2ps's EXTRA_LIBRARIES. Based on a patch from Sergei Trofimovich.
613]
614[Fix the doc directory on Windows
615Ian Lynagh <igloo@earth.li>**20100929133328]
616[Remove an unused import on Windows
617Ian Lynagh <igloo@earth.li>**20100929000024
618 Ignore-this: 2899e0e5a47122e637fb5c8aa0df52ab
619]
620[Use showCommandForUser when showing tracing commands
621Ian Lynagh <igloo@earth.li>**20100928235844
622 Ignore-this: 8a4a9c9f8a8029e708c4297b096b6ef1
623]
624[Fix hsc2hs docs: 'gcc' is now the default compiler, not 'ghc'; fixes #4341
625Ian Lynagh <igloo@earth.li>**20100928201938]
626[Use an empty signal handler for SIGPIPE instead of SIG_IGN
627Simon Marlow <marlowsd@gmail.com>**20100925193548
628 Ignore-this: b4dc5de32740a7c5fd8fe4b3bfb1300f
629 
630 This is so that the SIGPIPE handler gets reset to the default
631 automatically on exec().
632]
633[Fix the TH deps
634Ian Lynagh <igloo@earth.li>**20100925210029
635 Ignore-this: 32b832301a3625d4ba70f84c5c4f94d2
636]
637[Check inplace doesn't exist before we try to create it
638Ian Lynagh <igloo@earth.li>**20100924191858
639 This fixes rerunning configure in a tree which already has an inplace
640 directory. Edward Z Yang ran into this; I guess whether it actually
641 fails depends on details of your installation, or we'd have run into
642 it sooner.
643]
644[Fix an egregious bug: INLINE pragmas on monomorphic Ids were being ignored
645simonpj@microsoft.com**20100924155815
646 Ignore-this: 38c6eec6710a92df7662a55fc5132c15
647 
648 I had do to some refactoring to make this work nicely
649 but now it does. I can't think how this escaped our
650 attention for so long!
651]
652[Eta expand only lambdas that bind a non-dictionary Id
653simonpj@microsoft.com**20100924155707
654 Ignore-this: 7cc265eaf6c0bb3fa12eb311d92594ac
655 
656 See Note [When to eta expand]. The idea is that dictionary
657 lambdas are invisible to the user, so we shouldn't eta
658 expand them.
659]
660[Add a comment
661simonpj@microsoft.com**20100924155620
662 Ignore-this: de210a1afdd40328824803e1d77b4d7f
663]
664[Add a debug print
665simonpj@microsoft.com**20100924155614
666 Ignore-this: 1a58b6d297fc77d6ded8eec7ea9f895d
667]
668[Just moving comments around
669simonpj@microsoft.com**20100924155600
670 Ignore-this: 96635b8e8c9d88b50d82938568152ef8
671]
672[use putStrLn instead of Haskeline's outputStrLn
673Simon Marlow <marlowsd@gmail.com>**20100924133154
674 Ignore-this: 7581ae11714a9a52e78ba098c3c216b3
675 use of the latter caused problems for Claus Reinke's macros that
676 redirect stdout.
677]
678[Change "OPTIONS" to "OPTIONS_GHC" in error messages; fixes #4327
679Ian Lynagh <igloo@earth.li>**20100924120423
680 Ignore-this: 1697c83a5c346db640c0a2e22c69ff55
681]
682[Add deps for TH uses in vector
683Ian Lynagh <igloo@earth.li>**20100923220244
684 Ignore-this: 54c3386b1c268821fcdd34b84bc8c6a4
685]
686[Bump Cabal dep
687Ian Lynagh <igloo@earth.li>**20100923143241]
688[Update Cabal's version number
689Ian Lynagh <igloo@earth.li>**20100923141719]
690[Build primitive with stage2
691Ian Lynagh <igloo@earth.li>**20100923140525
692 Ignore-this: 110a819b78a57629a7edf1d4facdc191
693]
694[Fix the Windows __chkstk build error (missing Linker symbol)
695Simon Marlow <marlowsd@gmail.com>**20100924113837
696 Ignore-this: 48f0907bb1bd5eaa0730b94a6bd94ea
697]
698[emit a helpful error message for missing DPH packages
699Simon Marlow <marlowsd@gmail.com>**20100923141957
700 Ignore-this: 55ff2ee90c94524e023e014243bfe5df
701]
702[Fix computation of installed packages
703simonpj@microsoft.com**20100924084737
704 Ignore-this: a597d2fa8be5135ba8ead6d2624b3d71
705 
706 This is a follow-on to Simon's patch yesterday, developed
707 with him.  It cleans up the computation of how packages
708 are installed, and installs the right ones.
709]
710[Fix braino in WwLib/Literal patch
711simonpj@microsoft.com**20100924070914
712 Ignore-this: f6eb3a42e10f8aa7920de541cdfe76d8
713]
714[For now, switch off incomplete-pattern warnings in containers
715simonpj@microsoft.com**20100923130117
716 Ignore-this: 7ffa58567f7a33aafe256492999da325
717 
718 Put it back on when my patch is applied to the containers repo.
719 (the one that removes two refuable lambdas)
720]
721[Make -funfolding-dict-threshold work properly
722simonpj@microsoft.com**20100923130032
723 Ignore-this: 417788f5b09d1d624f6b6371852c80c7
724 
725 and increase its default value. This makes overloaded functions
726 a bit keener to inline.  Which fixes Trac #4321
727]
728[Impredicative types is no longer deprecated
729simonpj@microsoft.com**20100923125910
730 Ignore-this: 2bbaeb38b5e8424551677c0add627683
731]
732[Do not make FunctionalDependencies force MonoLocalBinds
733simonpj@microsoft.com**20100923125900
734 Ignore-this: f4ae1fd07c87ec14f60bdfe3863ba7a9
735]
736[move CHECKED settings to the right place
737Simon Marlow <marlowsd@gmail.com>**20100923123558
738 Ignore-this: e00a0eb5855463cc9b953670b3bbf211
739]
740[turn off -Werror for primitive and vector
741Simon Marlow <marlowsd@gmail.com>**20100923122055
742 Ignore-this: 54d7b80f3f893385e1c3ef431e2a8a7b
743]
744[Add primitive and vector packages for DPH support
745Simon Marlow <marlowsd@gmail.com>**20100923104542
746 Ignore-this: c070d015385b0a0797394132dcbb7670
747 DPH is now using the public vector package instead of its internal
748 version.
749 
750 vector and primitive are not "boot" packages; they aren't required to
751 build GHC, but they are required to validate (because we include DPH
752 when validating).
753 
754 If you say './darcs-all get --no-dph' then you don't get DPH, vector,
755 or primitive.
756]
757[Refactoring and tidy up in the build system
758Simon Marlow <marlowsd@gmail.com>**20100923095642
759 Ignore-this: f7bf3a1fd160149d89b26f464b064fb1
760 
761 Instead of the ghc-stage and ghc-stage2-package files in a package, we
762 now have a list of these in ghc.mk.  There are other similar lists (of
763 boot-packages and non-installable packages), so this is not too bad,
764 and is simpler.
765 
766 While poking around in the top-level ghc.mk file I spotted various
767 opportunities to clean up and re-order some of the cruft that has
768 accumulated over time.
769]
770[Allow absent State# RealWorld arguments
771simonpj@microsoft.com**20100923111356
772 Ignore-this: c2d57633dec0293ebe6723ea3a4bb5df
773]
774[Add notSCCNote, and use it
775simonpj@microsoft.com**20100923105949
776 Ignore-this: c8cc758656558a7f366bf784d75f0304
777 
778 The point here is that SCCs get in the way of eta
779 expansion and we must treat them uniformly.
780]
781[Remove use of lambda with a refutable pattern
782simonpj@microsoft.com**20100923105901
783 Ignore-this: d7d48b94e5744717a838591a1cc79cf0
784]
785[Avoid ASSERT black hole
786simonpj@microsoft.com**20100923105820
787 Ignore-this: 5419d450871be22c8781ac3f0f40d76a
788 
789 When this ASSERT tripped in CoreToStg it tried to print out
790 too much, which tripped the asssertion again.  Result: an
791 infinite loop with no output at all.  Hard to debug!
792]
793[Rejig the absent-arg stuff for unlifted types
794simonpj@microsoft.com**20100923105732
795 Ignore-this: 69daa35816b948b0c4d259c73a5e928e
796 
797 This is what was giving the "absent entered" messages
798 See Note [Absent errors] in WwLib.  We now return a
799 suitable literal for absent values of unlifted type.
800]
801[Remove -fwarn-simple-patterns, and make -fwarn-incomplete-patterns include lambdas
802simonpj@microsoft.com**20100922133934
803 Ignore-this: e851a2fb0377e10c28c506f0bf14cc85
804 
805 This makes
806      \(x:xs) -> e
807 want when you have -fwarn-incomplete-patterns, which is consistent.
808]
809[Get rid of non-exhaustive lambda
810simonpj@microsoft.com**20100922133801
811 Ignore-this: 748b2d5b43b02b6591b81abe7c105cd6
812]
813[Fix an ASSERT failure with profiling
814simonpj@microsoft.com**20100922133741
815 Ignore-this: 170b2e94d6ee8cc7444cc4bb515328a0
816 
817 The problem arose with this kind of thing
818 
819    x = (,) (scc "blah" Nothing)
820 
821 Then 'x' is marked NoCafRefs by CoreTidy, becuase it has
822 arity 1, and doesn't mention any caffy things.
823 
824 That in turns means that CorePrep must not float out the
825 sat binding to give
826 
827   sat = scc "blah" Nothing
828   x = (,) sat
829 
830 Rather we must generate
831 
832   x = \eta. let sat = scc "blah" Nothing
833             in (,) sat eta
834 
835 URGH! This Caf stuff is such a mess.
836]
837[Remove an out of date paragraph from the user guide; fixes #4331
838Ian Lynagh <igloo@earth.li>**20100922225239]
839[Fix bindisttest when GhcProfiled = YES
840Ian Lynagh <igloo@earth.li>**20100921222634
841 Ignore-this: 47c620fd6bec745e3eb699d9f53441d8
842]
843[Fixes for when HADDOCK_DOCS=NO
844Ian Lynagh <igloo@earth.li>**20100921213916
845 Ignore-this: e0e069555c6db9d01a8ac70ba4dde591
846]
847[Bump version to 7.1
848Ian Lynagh <igloo@earth.li>**20100921195935
849 Ignore-this: 4563987e6885d5ef55995ec0fa0d5ae8
850]
851[Don't use -march=i686 on powerpc-apple-darwin
852Ian Lynagh <igloo@earth.li>**20100921193721
853 Thorikil ran into this when doing a PPC OS X build. We now also don't
854 use -m32 on PPC/OSX, but I don't think it should be necessary. We can
855 add it back if it does turn out to be.
856]
857[add a simple trace facility to the build system
858Simon Marlow <marlowsd@gmail.com>**20100921134729
859 Ignore-this: d23ea2d62a648d0711b4b07d98e1b79f
860 
861 saying
862 
863   make TRACE=1
864 
865 prints most of the macro calls and their arguments.  It's easy to
866 trace new macros; see rules/trace.mk.
867]
868[fix building with extra packages (packages were added to BUILD_DIRS twice)
869Simon Marlow <marlowsd@gmail.com>**20100921100153
870 Ignore-this: 4b425dff9777871ad5ba3e05e1d14483
871 Also add some comments about what extra-packages is doing
872]
873[add extra packages to $(EXTRA_PACKAGES), so we avoid installing them by default
874Simon Marlow <marlowsd@gmail.com>**20100920144307
875 Ignore-this: 3395825d911a8bf7ba8385518d8b517b
876]
877[Fix indexing error in archive loader
878Ian Lynagh <igloo@earth.li>**20100921121642]
879[Add some -Dl belches
880Ian Lynagh <igloo@earth.li>**20100921121624]
881[Add casts to fix warnings
882Ian Lynagh <igloo@earth.li>**20100921121714]
883[Add support for BSD-variant large filenames in .a archives
884Ian Lynagh <igloo@earth.li>**20100921000451]
885[Tell Cabal that we're not building GHCi libs if UseArchivesForGhci=YES
886Ian Lynagh <igloo@earth.li>**20100920230449]
887["UseArchivesForGhci = YES" on darwin
888Ian Lynagh <igloo@earth.li>**20100920211538]
889[Add a dependency that my OS X build has recently started tripping up over
890Ian Lynagh <igloo@earth.li>**20100920210239]
891[Add "Use archives for ghci" to --info output
892Ian Lynagh <igloo@earth.li>**20100920210523]
893[Implement archive loading for ghci
894Ian Lynagh <igloo@earth.li>**20100920201620]
895[Tweak gen_contents_index now dph may not be there
896Ian Lynagh <igloo@earth.li>**20100920201513]
897[Filter out the FFI library when loading package in ghci
898Ian Lynagh <igloo@earth.li>**20100920181032
899 The FFI GHCi import lib isn't needed as
900 compiler/ghci/Linker.lhs + rts/Linker.c link the
901 interpreted references to FFI to the compiled FFI.
902 We therefore filter it out so that we don't get
903 duplicate symbol errors.
904]
905[Loosen the conditions for -XUndecidableInstances; fixes Trac #4200
906simonpj@microsoft.com**20100919162623
907 Ignore-this: 2f4323e278b1ce9250549727ffd0aa1b
908]
909[Further improvements in error messages
910simonpj@microsoft.com**20100919153355
911 Ignore-this: b6fa0b11ae893df1a3ca68f78e427fa
912]
913[Add a flag -fwarn-missing-local-sigs, and improve -fwarn-mising-signatures
914simonpj@microsoft.com**20100919153327
915 Ignore-this: fda8dfca450054ea692be0ee30b01885
916 
917 The new flag prints out a warning if you have a local,
918 polymorphic binding that lacks a type signature. It's meant
919 to help with the transition to the new typechecker, which
920 discourages local let-generalisation.
921 
922 At the same time I moved the missing-signature code to TcHsSyn,
923 where it takes place as part of zonking.  That way the
924 types are reported after all typechecking is complete,
925 thereby fixing Trac #3696.  (It's even more important for
926 local bindings, which is why I made the change.)
927]
928[Include the "stupid theta" in the type of $con2tag
929simonpj@microsoft.com**20100919152201
930 Ignore-this: d95fae78a0e66f48bbd5862573a11f4d
931]
932[Add a release note about the typechecker
933Ian Lynagh <igloo@earth.li>**20100919132927]
934[Enable shared libs on OpenBSD
935Matthias Kilian <kili@outback.escape.de>**20100918205040
936 Ignore-this: 729dd7ac0bba5d758f43bc31b541896
937]
938[Add separate functions for querying DynFlag and ExtensionFlag options
939Ian Lynagh <igloo@earth.li>**20100918163815
940 and remove the temporary DOpt class workaround.
941]
942[Fix mkUserGuidePart deps
943Ian Lynagh <igloo@earth.li>**20100918145904
944 We need to directly depend on the stage1 libs. The stage1 compiler lib
945 doesn't depend on them.
946]
947[Fix build on cygwin: Normalise slashes in .depend files to be /
948Ian Lynagh <igloo@earth.li>**20100918132328
949 Ignore-this: 664f5ef4a41a4461eb34fe2ca7f2729a
950]
951[extra packages info is now read from packages file
952Ian Lynagh <igloo@earth.li>**20100917224409
953 rather than being repeated in the build system
954]
955[Tweak darcs-all
956Ian Lynagh <igloo@earth.li>**20100917194435]
957[Bump dependencies
958Ian Lynagh <igloo@earth.li>**20100917183609]
959[Library release notes for 7.0.1
960Ian Lynagh <igloo@earth.li>**20100917174850]
961[Fix overriding of implicit parameters in the solver
962simonpj@microsoft.com**20100917140403
963 Ignore-this: af76732309c7e2ca6b04f49327e9c14b
964]
965[Minor type printing amomaly
966simonpj@microsoft.com**20100917140204
967 Ignore-this: c90cb2e51421b4543a827e096051772e
968]
969[Spaces only
970simonpj@microsoft.com**20100917140156
971 Ignore-this: 7e34479502f7cb87d762355e40cbd012
972]
973[Minor refactoring
974simonpj@microsoft.com**20100917140150
975 Ignore-this: 6c0648b949b91b7e2f23c136b124faf2
976]
977[Add types of implicit parameters as untouchable
978simonpj@microsoft.com**20100917140138
979 Ignore-this: ba80740a557a9ba062dc7756e2320d17
980 
981 This is a tricky point:
982    see Note [Implicit parameter untouchables]
983]
984[Better pretty printing of implicit parameters
985simonpj@microsoft.com**20100917140054
986 Ignore-this: 867dd67818a5bd687b2b6a1b59e15775
987]
988[Yet more error message improvement
989simonpj@microsoft.com**20100917121206
990 Ignore-this: 647fe8129d1d39d81e8249debd8df94e
991]
992[More error message wibbles
993simonpj@microsoft.com**20100917094721
994 Ignore-this: 8ec2f150b96b26af2e9ab7ac2b371fc7
995]
996[More error refactoring
997simonpj@microsoft.com**20100917092834
998 Ignore-this: 2d570ac0b9cc11305ddd33d093d11324
999]
1000[Refactor type errors a bit
1001simonpj@microsoft.com**20100917080726
1002 Ignore-this: 33da4549373f585064e2ee22b50ad6ac
1003 
1004 Improves kind error messages in paticular
1005]
1006[Fix a very subtle shadowing bug in optCoercion
1007simonpj@microsoft.com**20100916170452
1008 Ignore-this: 9041cfb3c93e27a5e644e57815313aae
1009 
1010 See Note [Subtle shadowing in coercions]
1011 
1012 This is what was going wrong in Trac 4160.
1013]
1014[Fix bad error in tyVarsOfType
1015simonpj@microsoft.com**20100916170348
1016 Ignore-this: 67c8ce96a668cf6e3a38b82c893bcd81
1017 
1018 We weren't gathering the type variables free in the kind
1019 of a coercion binder!
1020]
1021[More assertions
1022simonpj@microsoft.com**20100916170310
1023 Ignore-this: 7fdcb53c99d791621a3d7e01ef454404
1024]
1025[Add more location info in CoreLint
1026simonpj@microsoft.com**20100916170229
1027 Ignore-this: 6558bab544b4f30189e0430668db87c3
1028]
1029[Print coercion variables as such (debugging change only)
1030simonpj@microsoft.com**20100916165944
1031 Ignore-this: c6d2001c1d8279a2288cb63bc339577d
1032]
1033[Remove pprTrace
1034simonpj@microsoft.com**20100915225935
1035 Ignore-this: 28185bbfa9732386f3c0f3eb4781a637
1036]
1037[Remove dead code dealing with type refinement
1038simonpj@microsoft.com**20100915223230
1039 Ignore-this: 62824b5c2ec1077c7642163352559621
1040]
1041[Use mkAppTy
1042simonpj@microsoft.com**20100915223205
1043 Ignore-this: e79e087b6a49219e9088846a1253a153
1044 
1045 Using AppTy in CoreLint was giving a bogus Lint failure
1046]
1047[Comments only
1048simonpj@microsoft.com**20100915221253
1049 Ignore-this: 3a45ea614188ccbb4a462de5cac96eda
1050]
1051[Extend eta reduction to work with casted arguments
1052simonpj@microsoft.com**20100915221229
1053 Ignore-this: 24b103dcdf70331211507af929789f86
1054 
1055 See Trac #4201, and
1056 Note [Eta reduction with casted arguments]
1057 
1058 Thanks to Louis Wasserman for suggesting this, and
1059 implementing an early version of the patch
1060]
1061[Allow "INLINEABLE" as a synonym
1062simonpj@microsoft.com**20100915154249
1063 Ignore-this: f41f80cb769e9acd5b463b170df698d0
1064]
1065[Documentation for INLINABLE
1066simonpj@microsoft.com**20100915154235
1067 Ignore-this: f942c02bcadc0d2d2f05b9369f93e280
1068]
1069[Implement TH reification of instances (Trac #1835)
1070simonpj@microsoft.com**20100915151242
1071 Ignore-this: 97dfa83db7da8f6cbd1b96801a57f8c5
1072 
1073 Accompanying patch for template-haskell package is reqd
1074]
1075[errno corresponding to ERROR_NO_DATA should be EPIPE (non-threaded RTS)
1076Simon Marlow <marlowsd@gmail.com>**20100915141809
1077 Ignore-this: 709c7280fbaa762e7071fb8796e8c01e
1078]
1079[Windows: use a thread-local variable for myTask()
1080Simon Marlow <marlowsd@gmail.com>**20100915120627
1081 Ignore-this: 13ffa4f19ebd319fe672af53af8d0b9a
1082 Which entailed fixing an incorrect #ifdef in Task.c
1083]
1084[Fix typo
1085Ian Lynagh <igloo@earth.li>**20100915140814]
1086[Add quotes in error message
1087simonpj@microsoft.com**20100915144724
1088 Ignore-this: c5158047c0aa41947a79e4c8edbe54f4
1089]
1090[Fix isDefaultInlinePragma
1091simonpj@microsoft.com**20100915144710
1092 Ignore-this: c9addf6bf811b23dc12603cf8521aa6c
1093]
1094[Implement INLINABLE pragma
1095simonpj@microsoft.com**20100915124442
1096 Ignore-this: 80a4ab2c2d65b27868dc9b2e954d6c6f
1097 
1098 Implements Trac #4299.  Documentation to come.
1099]
1100[Less voluminous error when derived code doesn't typecheck
1101simonpj@microsoft.com**20100915072301
1102 Ignore-this: eca7871dcc50c1070a0b530711adea27
1103]
1104[Improve pretty-printing of family instances
1105simonpj@microsoft.com**20100915123219
1106 Ignore-this: 25ec6bcc7e8a7f7c303b38ca201db90e
1107 
1108 Fixed Trac #4246
1109]
1110[Fix Trac #4240: -ddump-minimal-imports
1111simonpj@microsoft.com**20100915121937
1112 Ignore-this: ab85057cb829a42ea44a92f7b4af24a3
1113 
1114 See Note [Partial export] for the details.
1115 I also fixed one egregious bug that was just
1116 waiting to bite: we were using loadSysInterface
1117 instead of loadSrcInterface.
1118]
1119[Comments only
1120simonpj@microsoft.com**20100915105707
1121 Ignore-this: ab3a5f16f8260b7b8570e748bf97998a
1122]
1123[implement setThreadAffinity on Windows (#1741)
1124Simon Marlow <marlowsd@gmail.com>**20100914155844
1125 Ignore-this: a14c7b4ef812007042342d0a25478f0b
1126]
1127[COFF: cope with new debug sections in gcc 4.x (fixes ghciprog004)
1128Simon Marlow <marlowsd@gmail.com>**20100914153026
1129 Ignore-this: f340e40a2b0390836bc61bba144a04ed
1130 Also updated the object file parser to properly handle the overflow
1131 case for section names longer than 8 chars.
1132]
1133[eliminate clutter from make output
1134Simon Marlow <marlowsd@gmail.com>**20100915105712
1135 Ignore-this: bfa4480dd239dda2a02ac391b6a9219c
1136]
1137[rts_isProfiled should be a visible API (fixes T2615(dyn))
1138Simon Marlow <marlowsd@gmail.com>**20100915083941
1139 Ignore-this: b8ac09bb9d1a929bf45c6122f8485561
1140]
1141[Fix the "lost due to fragmentation" calculation
1142Simon Marlow <marlowsd@gmail.com>**20100914145945
1143 Ignore-this: cdffcc9f3061c3a33da5171be111fc43
1144 It was counting the space used by block descriptors as "lost"
1145]
1146[fix +RTS -S output: use peak_mblocks_allocated, now that mblocks can be freed
1147Simon Marlow <marlowsd@gmail.com>**20100914135030
1148 Ignore-this: 65d21e5f86d3ab6ab4d6c255180b6968
1149]
1150[Fix egregious bug in deeplyInstantiate
1151simonpj@microsoft.com**20100915070325
1152 Ignore-this: 22ede973038877af2673339aaf5de6cf
1153 
1154 This resulted in an infinite loop in applyTypeToArgs, in syb
1155]
1156[Improve HsSyn pretty printing
1157simonpj@microsoft.com**20100915070255
1158 Ignore-this: 7c8e2d86a482453c7e69e22bc31cb03f
1159]
1160[Remove (most of) the FiniteMap wrapper
1161Ian Lynagh <igloo@earth.li>**20100914201703
1162 We still have
1163     insertList, insertListWith, deleteList
1164 which aren't in Data.Map, and
1165     foldRightWithKey
1166 which works around the fold(r)WithKey addition and deprecation.
1167]
1168[Improve ASSERT
1169simonpj@microsoft.com**20100914113900
1170 Ignore-this: dbc0363be5924f543316e77f7d18dd77
1171]
1172[Comment on what an "enumeration" type is
1173simonpj@microsoft.com**20100914113850
1174 Ignore-this: c09c8591e3140f305d55fbf945adbf95
1175]
1176[Make absent-arg wrappers work for unlifted types (fix Trac #4306)
1177simonpj@microsoft.com**20100914113827
1178 Ignore-this: 1945e56779329e8b79780403710aba98
1179 
1180 Previously we were simply passing arguments of unlifted
1181 type to a wrapper, even if they were absent, which was
1182 stupid.
1183 
1184 See Note [Absent error Id] in WwLib.
1185]
1186[Comments only
1187simonpj@microsoft.com**20100914113641
1188 Ignore-this: 3191ce856c9b5d9700cedc9b149b8097
1189]
1190[Move error-ids to MkCore (from PrelRules)
1191simonpj@microsoft.com**20100914113635
1192 Ignore-this: c3d820db62ba6139dd7c96bf97e51bb5
1193 
1194 and adjust imports accordingly
1195]
1196[More wibbles to deriving error messages
1197simonpj@microsoft.com**20100914113523
1198 Ignore-this: bd2df662644961138fa209aec843a2aa
1199]
1200[Fix getThreadCPUTime()
1201Simon Marlow <marlowsd@gmail.com>**20100913153838
1202 Ignore-this: 950e048a5724086534b74c609c7d5ed
1203 ever since the patch "Check with sysconf _POSIX_THREAD_CPUTIME", it
1204 has been returning incorrect results, because the sysconf variable to
1205 check should have been _SC_THREAD_CPUTIME, not _POSIX_THREAD_CPUTIME.
1206]
1207[filter out the gcc-lib directory from the rts package's library-dirs
1208Simon Marlow <marlowsd@gmail.com>**20100913101259
1209 Ignore-this: 46dc1dccbfee8a65f9243e125eee117f
1210 fixes problems when building with GHC 6.10 on Windows
1211]
1212[Don't include GC time in heap profiles (#4225)
1213Simon Marlow <marlowsd@gmail.com>**20100913133852
1214 Ignore-this: 68ac48b004b311384b5996c6b33ba5cc
1215]
1216[Use clock_gettime (if available) to measure the process CPU time
1217Simon Marlow <marlowsd@gmail.com>**20100913133818
1218 Ignore-this: 8c9300df9b929bfc1db4713c9b6065b3
1219 This is much more accurate than getrusage, which was giving misleading
1220 results when trying to time very quick operations like a minor GC.
1221]
1222[make stg_arg_bitmaps public, and available via the GHCi linker (#3672)
1223Simon Marlow <marlowsd@gmail.com>**20100913105235
1224 Ignore-this: e18efd0bd77c521e5530fb59e93b5a42
1225]
1226[fix typo
1227Simon Marlow <marlowsd@gmail.com>**20100913105100
1228 Ignore-this: 6049eea21208864203b2d79db2edd143
1229]
1230[Update release notes and docs with LLVM info.
1231David Terei <davidterei@gmail.com>**20100914072135
1232 Ignore-this: 5b3d0e5c9d5da98ed6ae9c2e8e1f6f30
1233]
1234[Remove defaultExtensionFlags
1235Ian Lynagh <igloo@earth.li>**20100913165949
1236 The default should do into languageExtensions instead
1237]
1238[Improve crash message
1239simonpj@microsoft.com**20100913170407
1240 Ignore-this: 5c26a9979f18be8cd12cea823c9f4b5a
1241]
1242[Fix Trac #4302, plus a little refactoring
1243simonpj@microsoft.com**20100913170355
1244 Ignore-this: cf6886b587aa0e8d723362183625d946
1245]
1246[Fix build with 6.10
1247Ian Lynagh <igloo@earth.li>**20100913160048]
1248[Haddock fixes
1249simonpj@microsoft.com**20100913120510
1250 Ignore-this: f3157d6969f10d4cbd593000a477138b
1251]
1252[Remove two old junk files
1253simonpj@microsoft.com**20100913103426
1254 Ignore-this: ed7af5ef1b9592178909a8d876345302
1255]
1256[Super-monster patch implementing the new typechecker -- at last
1257simonpj@microsoft.com**20100913095048
1258 Ignore-this: 14d14a1e4d7a414f5ae8d9d89d1c6a4b
1259 
1260 This major patch implements the new OutsideIn constraint solving
1261 algorithm in the typecheker, following our JFP paper "Modular type
1262 inference with local assumptions". 
1263 
1264 Done with major help from Dimitrios Vytiniotis and Brent Yorgey.
1265 
1266]
1267[Fix simplifier statistics
1268simonpj@microsoft.com**20100909085441
1269 Ignore-this: 48e383655aafc912dea15c4d94382863
1270]
1271[Trace output
1272simonpj@microsoft.com**20100908170056
1273 Ignore-this: 4b67fa4b310fbf0a16b852686d2d3294
1274]
1275[Better debug output
1276simonpj@microsoft.com**20100908170047
1277 Ignore-this: 410cef00616dda7c0c162f65216e8ca3
1278]
1279[Add Outputable instance for OccEncl
1280simonpj@microsoft.com**20100908150510
1281 Ignore-this: 6362ef9028287d84f070eaf8963c1bfc
1282]
1283[Better simplifier counting
1284simonpj@microsoft.com**20100907214840
1285 Ignore-this: 9d4722703f8f47447e86a28c8c50e0ea
1286]
1287[Put liftStringName into the known-key names
1288simonpj@microsoft.com**20100906112415
1289 Ignore-this: 287064d14ff484da1a6dea6924bc9235
1290]
1291[Deprecate NoRelaxedPolyRec
1292simonpj@microsoft.com**20100903234519
1293 Ignore-this: 607217e77f6bc1b91bf57dfd8dd2b967
1294]
1295[Buglet in Core Lint
1296simonpj@microsoft.com**20100903234457
1297 Ignore-this: 277535d51b396d3b4b0265a0939c2d4
1298]
1299[Give seqId the right type
1300simonpj@microsoft.com**20100903093556
1301 Ignore-this: d1fc7a73dea160614a8d4ddc930f99cd
1302]
1303[Remove dead code
1304simonpj@microsoft.com**20100903093548
1305 Ignore-this: 92cc3f7651445aa349ee7f114d3ec758
1306]
1307[Comments and layout
1308simonpj@microsoft.com**20100903093502
1309 Ignore-this: 9987d1409e654992c1cb1be35cb87728
1310]
1311[Remove checkFreeness; no longer needed
1312simonpj@microsoft.com**20100902233227
1313 Ignore-this: c96a12ac9794290aa30402317d88c095
1314]
1315[Assert
1316simonpj@microsoft.com**20100902073642
1317 Ignore-this: 4be1ab2f6096665ae5ec7fdd1f025a67
1318]
1319[Add aserts
1320simonpj@microsoft.com**20100902073211
1321 Ignore-this: e1409441217fd070c5a7f9ee4cca99ab
1322]
1323[Wibbles
1324simonpj@microsoft.com**20100831113540
1325 Ignore-this: 903811ab493a7b560a62eb86fcf3ee25
1326]
1327[Wibble to allow phantom types in Enum
1328simonpj@microsoft.com**20100825112711
1329 Ignore-this: fdef1c50d92b4a3d46bbe4cbfd8a83ea
1330]
1331[Add HsCoreTy to HsType
1332simonpj@microsoft.com**20100824141845
1333 Ignore-this: 4ca742b099f9cc90af3167f1012dbba6
1334 
1335 The main thing here is to allow us to provide type
1336 signatures for 'deriving' bindings without pain.
1337]
1338[Comments
1339simonpj@microsoft.com**20100823223654
1340 Ignore-this: dd412a55839430c436902d8699d6900b
1341]
1342[Wibbles to error message
1343simonpj@microsoft.com**20100823163308
1344 Ignore-this: 4d6cd8e613762dca8135c2e3b09264ec
1345]
1346[Correct type signatures
1347simonpj@microsoft.com**20100823153045
1348 Ignore-this: 42942309221a443258246098f9c0a13b
1349]
1350[Add missing signatures
1351simonpj@microsoft.com**20100823112413
1352 Ignore-this: 8ee1ce40456306de469938c02df4fed5
1353]
1354[Add type signatures in "deriving" bindings
1355simonpj@microsoft.com**20100820234230
1356 Ignore-this: 4726b28968cf65ec16cb65b7e0e7303e
1357]
1358[Minor
1359dimitris@microsoft.com**20100820131021]
1360[Be a bit less aggressive in mark-many inside a cast
1361simonpj@microsoft.com**20100819104804
1362 Ignore-this: 3fd48fe7647ec7a58c2032cd86ca4d4f
1363]
1364[Wibble
1365simonpj@microsoft.com**20100818185738
1366 Ignore-this: d5c939311377c0d0c7244aa339193315
1367]
1368[Pretty printing change
1369simonpj@microsoft.com**20100818065436
1370 Ignore-this: 4f7e70976dbe52f95effb3e634dfef5d
1371]
1372[Remember to zonk FlatSkols!
1373simonpj@microsoft.com**20100811143555
1374 Ignore-this: 84f7f9dbda97f561a918c69308ddef9a
1375]
1376[De-polymorphise
1377simonpj@microsoft.com**20100730151217
1378 Ignore-this: a9304487b983e517a9083fd697f77576
1379]
1380[Work around missing type signature in Happy
1381simonpj@microsoft.com**20100730122405
1382 Ignore-this: 7f241a655d93c5ad7763a7ffe8db0c7a
1383 
1384 Happy generates
1385 
1386       notHappyAtAll = error "Blah"
1387 
1388 without a type signature, and currently the new
1389 typechecker doesn't generalise it.  This patch
1390 says "no monomorphism restriction" which makes it
1391 generalise again.
1392 
1393 Better would be to add a type sig to Happy's template
1394]
1395[Add two local type signatures
1396simonpj@microsoft.com**20100729152611
1397 Ignore-this: afa99bcc515469aa0990d44d8c18a9e6
1398]
1399[Second test from Simon's laptop
1400simonpj@microsoft.com**20100729091703
1401 Ignore-this: 4dc64cadae314a5a1b05cc5326918a83
1402]
1403[Test commit from Simon's laptop
1404simonpj@microsoft.com**20100729091344
1405 Ignore-this: 109eff835cc19e9f93799d12f09b0ba7
1406]
1407[Add OutsideIn flag
1408simonpj@microsoft.com**20100728075525
1409 Ignore-this: 69c2f5c3a15fa653f6da80598aa8d74d
1410]
1411[Layout only
1412simonpj@microsoft.com**20100727141539
1413 Ignore-this: 1a58a8fe80ba8bced18ae81a2efb9495
1414]
1415[Improvement to SimplUtils.mkLam
1416simonpj@microsoft.com**20100727131659
1417 Ignore-this: 739beaefa79baa7e0ebeb5b2b6d1ea91
1418]
1419[Give the correct kind to unsafeCoerce#
1420simonpj@microsoft.com**20100727131538
1421 Ignore-this: 6b787de3b398c6d7a61fa04fccd15fd6
1422]
1423[Suppress warnings about recursive INLINE in output of desugarer
1424simonpj@microsoft.com**20100727094549
1425 Ignore-this: a361f7238c0fcba526d46326722c42e
1426]
1427[Rename CorePrep.tryEtaReduce to tryEtaReducePrep
1428simonpj@microsoft.com**20100726231253
1429 Ignore-this: 4375ddace205745244ba224ae012252
1430 
1431 This avoids the name clash with the similar but
1432 not identical CoreUtils.tryEtaReduce
1433]
1434[Add a trace message
1435simonpj@microsoft.com**20100719211111
1436 Ignore-this: b5daebe46e50c8cf28cc693f84bbf099
1437]
1438[Don't use RelaxedPolyRec in the compiler; it's built in now
1439simonpj@microsoft.com**20100719170441
1440 Ignore-this: a2e4489cdf63478e46282a421ee7aec3
1441]
1442[Remove duplicated #defines for FreeBSD
1443Matthias Kilian <kili@outback.escape.de>**20100912181518
1444 Ignore-this: d16214fef8635c7c9ef4edec4e8e7896
1445]
1446[Don't fail with absolute silence
1447Matthias Kilian <kili@outback.escape.de>**20100912150506
1448 Ignore-this: 479e2321f39b263fa2d9f80491e5e9f7
1449]
1450[Add a release note: "-dynload wrapper" removed
1451Ian Lynagh <igloo@earth.li>**20100911195809]
1452[put back the conversion of warn-lazy-unlifted-bindings into an error until 7.2
1453Ian Lynagh <igloo@earth.li>**20100911193434
1454 I think we'll currently still have too many people with old versions of
1455 alex/happy to want to make this an error now.
1456]
1457[6.14 -> 7.0
1458Ian Lynagh <igloo@earth.li>**20100911192837]
1459[Add a couple more release notes
1460Ian Lynagh <igloo@earth.li>**20100911162059]
1461[Document -dsuppress-module-prefixes
1462Ian Lynagh <igloo@earth.li>**20100911162005]
1463[Enable -fregs-graph with -O2; fixes #2790
1464Ian Lynagh <igloo@earth.li>**20100910191301]
1465[Remove unused code
1466Ian Lynagh <igloo@earth.li>**20100909170207]
1467[Fix warnings
1468Ian Lynagh <igloo@earth.li>**20100909154348]
1469[Fix warnings
1470Ian Lynagh <igloo@earth.li>**20100909150957]
1471[Remove context completion
1472lykahb@gmail.com**20100901160153
1473 Ignore-this: dc61b259dcb7063f0c76f56788b5d2af
1474 Now completion suggests to remove only modules added to context before.
1475]
1476[avoid Foreign.unsafePerformIO
1477Ross Paterson <ross@soi.city.ac.uk>**20100909125531
1478 Ignore-this: 5cabeae4cffec8fc17ef7c0cabbea22a
1479]
1480[updates to the release notes
1481Simon Marlow <marlowsd@gmail.com>**20100909111450
1482 Ignore-this: a4d25ad8815c305b7e0f21fd4f6ee37b
1483]
1484[newAlignedPinnedByteArray#: avoid allocating an extra word sometimes
1485Simon Marlow <marlowsd@gmail.com>**20100909110805
1486 Ignore-this: 996a3c0460068ab2835b4920905b3e75
1487]
1488[Finish breaking up vectoriser utils
1489benl@ouroborus.net**20100909061311
1490 Ignore-this: 217fe1d58a3e8bb13200bcb81353a416
1491]
1492[Move VectType module to Vectorise tree
1493benl@ouroborus.net**20100909042451
1494 Ignore-this: 5af8cf394d4835911259ca3ffb6774c5
1495]
1496[Sort all the PADict/PData/PRDict/PRepr stuff into their own modules
1497benl@ouroborus.net**20100909035147
1498 Ignore-this: 53436329773347cad793adbd83e90a9e
1499]
1500[Break out Repr and PADict stuff for vectorisation of ADTs to their own modules
1501benl@ouroborus.net**20100909025759
1502 Ignore-this: d2b7d2f79332eda13416449742f7cf1c
1503]
1504[Break out conversion functions to own module
1505benl@ouroborus.net**20100909023332
1506 Ignore-this: 613f2666b6ca7f2f8876fcc1e4a59593
1507]
1508[Comments and formatting only
1509benl@ouroborus.net**20100909022117
1510 Ignore-this: c8e30139d730669e5db44f0ef491a588
1511]
1512[Remove "-dynload wrapper"; fixes trac #4275
1513Ian Lynagh <igloo@earth.li>**20100908213251]
1514[Don't set visibility on Windows
1515Ian Lynagh <igloo@earth.li>**20100905122442
1516 With gcc 4.5.0-1, using visibility hidden gives:
1517     error: visibility attribute not supported in this configuration; ignored
1518]
1519[Fix warnings on Windows
1520Ian Lynagh <igloo@earth.li>**20100905111201
1521 Ignore-this: c5cce63bb1e0c7a27271bed78d25fbc5
1522]
1523[Fix gcc wrapper for new mingw binaries
1524Ian Lynagh <igloo@earth.li>**20100905001807
1525 Ignore-this: f6acc8c911055ffce632bac138ccc939
1526]
1527[Don't pass our gcc options to stage0 ghc's gcc; they may not be suitable
1528Ian Lynagh <igloo@earth.li>**20100905103129]
1529[Update intree-mingw creation
1530Ian Lynagh <igloo@earth.li>**20100904225559]
1531[Update commands to build in-tree mingw
1532Ian Lynagh <igloo@earth.li>**20100904215112]
1533[Break out hoisting utils into their own module
1534benl@ouroborus.net**20100908074102
1535 Ignore-this: e3ba4ed0252a2def1ed88a9e14c58fea
1536]
1537[Break out closure utils into own module
1538benl@ouroborus.net**20100908072040
1539 Ignore-this: 216172b046ff101cf31a1753667a5383
1540]
1541[Move VectVar module to Vectorise tree
1542benl@ouroborus.net**20100908065904
1543 Ignore-this: 1fba5333d29927dba4275381e1a7f315
1544]
1545[Break out vectorisation of expressions into own module
1546benl@ouroborus.net**20100908065128
1547 Ignore-this: 6a952b80fb024b5291f166477eb1976
1548]
1549[Break out TyCon classifier into own module
1550benl@ouroborus.net**20100908063111
1551 Ignore-this: da754c4ef6960b4e152ea1bf8c04ab6f
1552]
1553[Break out vectorisation of TyConDecls into own module
1554benl@ouroborus.net**20100908052004
1555 Ignore-this: c0ab4fb2a05ca396efe348b384db1ebf
1556]
1557[Break out type vectorisation into own module
1558benl@ouroborus.net**20100907110311
1559 Ignore-this: 67bd70a21d16468daf68dd3ec1ff7d62
1560]
1561[Tidy up the ArchHasAdjustorSupport definition
1562Ian Lynagh <igloo@earth.li>**20100904144234]
1563[ppc: switch handling of 'foreign import wrapper' (FIW) to libffi
1564Sergei Trofimovich <slyfox@community.haskell.org>**20100829192859
1565 Ignore-this: 662ea926681ebea0759e2a04a38e82b7
1566 
1567 Joseph Jezak reported darcs-2.4.4 SIGSEGV in interactive mode in ghc-6.12.3.
1568 So I've concluded ppc also has rotten native adjustor. I don't have hardware
1569 to verify the patch (ticket #3516 should help to test it), but I think it will
1570 help (as similar patch helped for ia64 and ppc64).
1571]
1572[Binary no longer has unusable UNPACK pragmas, so no need to turn of -Werror
1573Ian Lynagh <igloo@earth.li>**20100904133339]
1574[Don't haddock packages that we aren't going to install
1575Ian Lynagh <igloo@earth.li>**20100903231921]
1576[Give haddock per-package source entity paths; fixes #3810
1577Ian Lynagh <igloo@earth.li>**20100903221335]
1578[update for containers-0.4
1579Simon Marlow <marlowsd@gmail.com>**20100903105131
1580 Ignore-this: 556eac0e4926c9b8af6b66d7b069302c
1581]
1582[Fix for nursery resizing: the first block's back pointer should be NULL
1583Simon Marlow <marlowsd@gmail.com>**20100827102818
1584 Ignore-this: fb68938e3f1e291e3c9e5e8047f9dcd2
1585 I'm not sure if this could lead to a crash or not, but it upsets +RTS -DS
1586 Might be related to #4265
1587]
1588[Add some -no-user-package-conf flags
1589Ian Lynagh <igloo@earth.li>**20100902224726
1590 Stops user-installed packages breaking the build
1591]
1592[Fix warnings: Remove unused imports
1593Ian Lynagh <igloo@earth.li>**20100902204342]
1594[Finish breaking up VectBuiltIn and VectMonad, and add comments
1595benl@ouroborus.net**20100831100724
1596 Ignore-this: 65604f3d22d03433abc12f10be40050d
1597]
1598[Fix warnings
1599benl@ouroborus.net**20100830083746
1600 Ignore-this: 2a0e000985f694582a6f9a9261ff2739
1601]
1602[Break up vectoriser builtins module
1603benl@ouroborus.net**20100830070900
1604 Ignore-this: b86bd36a7875abdcf16763902ba2e637
1605]
1606[Move VectCore to Vectorise tree
1607benl@ouroborus.net**20100830053415
1608 Ignore-this: d5763ca6424285b39a58c7792f4a84a1
1609]
1610[Split out vectoriser environments into own module
1611benl@ouroborus.net**20100830050252
1612 Ignore-this: 5319111c74831394d2c29b9aedf5a766
1613]
1614[Comments and formatting to vectoriser, and split out varish stuff into own module
1615benl@ouroborus.net**20100830042722
1616 Ignore-this: d3f0c98ed8124dd0fca9a2ccea3e15fd
1617]
1618[Fix warnings
1619benl@ouroborus.net**20100830040340
1620 Ignore-this: d6cfad803ad4617e7fdaa62e4a895282
1621]
1622[Fix warning about multiply exported name
1623benl@ouroborus.net**20100830035243
1624 Ignore-this: 27ce2c1d22d9f99929d16a426343044e
1625]
1626[Vectorisation of method types
1627benl@ouroborus.net**20100830032941
1628 Ignore-this: 75614571d5c246a4906edb3b39ab1e0b
1629]
1630[Comments and formatting to vectoriser
1631benl@ouroborus.net**20100830032516
1632 Ignore-this: fe665b77108501c7960d858be3290761
1633]
1634[Implement -dsuppress-module-prefixes
1635benl@ouroborus.net**20100830032428
1636 Ignore-this: 2bb8bad9c60ef9044132bba118010687
1637]
1638[Whitespace only
1639benl@ouroborus.net**20100527045629
1640 Ignore-this: 4c160dfa77727e659817b6af9c84684a
1641]
1642[Disambiguate a function name
1643Ian Lynagh <igloo@earth.li>**20100828225827]
1644[users_guide.xml is now generated
1645Ian Lynagh <igloo@earth.li>**20100828225751]
1646[Pass more -pgm flags in the ghc wrapper; fixes #3863
1647Ian Lynagh <igloo@earth.li>**20100827204537]
1648[Add a new-IO manager release note
1649Ian Lynagh <igloo@earth.li>**20100827171616]
1650[Merge a duplicate release note
1651Ian Lynagh <igloo@earth.li>**20100827171511]
1652[Typo, spotted by Johan Tibell
1653Ian Lynagh <igloo@earth.li>**20100827153914]
1654[First pass at 6.14.1 release notes
1655Ian Lynagh <igloo@earth.li>**20100826220811]
1656[Fix typo
1657Ian Lynagh <igloo@earth.li>**20100824201330]
1658[FIX BUILD: add rts_isProfiled to the symbol table
1659Simon Marlow <marlowsd@gmail.com>**20100826094319
1660 Ignore-this: 9536ddb0a94721c8dec03a2a981cfa83
1661]
1662[Fix the DPH package cleaning/profiled mess even more (the build was broken)
1663Simon Marlow <marlowsd@gmail.com>**20100826084436
1664 Ignore-this: 49d7e4db2fb53b856c213c74c8969d82
1665]
1666[Remove the debugging memory allocator - valgrind does a better job
1667Simon Marlow <marlowsd@gmail.com>**20100824113537
1668 Ignore-this: a3731a83dc18b0fd0de49452e695a7ca
1669 
1670 I got fed up with the constant bogus output from the debugging memory
1671 allocator in RtsUtils.c.  One problem is that we allocate memory in
1672 constructors that then isn't tracked, because the debugging allocator
1673 hasn't been initialised yet.
1674 
1675 The bigger problem is that for a given piece of leaking memory it's
1676 impossible to find out where it was allocated; however Valgrind gives
1677 output like this:
1678 
1679 ==6967== 8 bytes in 1 blocks are still reachable in loss record 1 of 7
1680 ==6967==    at 0x4C284A8: malloc (vg_replace_malloc.c:236)
1681 ==6967==    by 0x4C28522: realloc (vg_replace_malloc.c:525)
1682 ==6967==    by 0x6745E9: stgReallocBytes (RtsUtils.c:213)
1683 ==6967==    by 0x68D812: setHeapAlloced (MBlock.c:91)
1684 ==6967==    by 0x68D8E2: markHeapAlloced (MBlock.c:116)
1685 ==6967==    by 0x68DB56: getMBlocks (MBlock.c:240)
1686 ==6967==    by 0x684F55: alloc_mega_group (BlockAlloc.c:305)
1687 ==6967==    by 0x6850C8: allocGroup (BlockAlloc.c:358)
1688 ==6967==    by 0x69484F: allocNursery (Storage.c:390)
1689 ==6967==    by 0x694ABD: allocNurseries (Storage.c:436)
1690 ==6967==    by 0x6944F2: initStorage (Storage.c:217)
1691 ==6967==    by 0x673E3C: hs_init (RtsStartup.c:160)
1692 
1693 which tells us exactly what the leaking bit of memory is.  So I don't
1694 think we need our own debugging allocator.
1695]
1696[free the entries in the thread label table on exit
1697Simon Marlow <marlowsd@gmail.com>**20100824112606
1698 Ignore-this: c9d577c06548cda80791e590e40d35b3
1699]
1700[Panic in the right way
1701simonpj@microsoft.com**20100825091614
1702 Ignore-this: e6ea4f6dfd2aea088828ea7a945ddd5f
1703]
1704[Fix the DPH/profiled make thing (again)
1705simonpj@microsoft.com**20100825091602
1706 Ignore-this: bc58fa48034ac40cf7be4170958ea29e
1707]
1708[Don't test for gcc flags before we've found gcc
1709Ian Lynagh <igloo@earth.li>**20100824131401]
1710[Change how the dblatex/lndir problem is worked around
1711Ian Lynagh <igloo@earth.li>**20100824130938
1712 Hack: dblatex normalises the name of the main input file using
1713 os.path.realpath, which means that if we're in a linked build tree,
1714 it find the real source files rather than the symlinks in our link
1715 tree. This is fine for the static sources, but it means it can't
1716 find the generated sources.
1717 
1718 We therefore also generate the main input file, so that it really
1719 is in the link tree, and thus dblatex can find everything.
1720]
1721[Clean the generated userguide sources
1722Ian Lynagh <igloo@earth.li>**20100824105827
1723 Ignore-this: 39b4f9702c688c053ed3273b20969597
1724]
1725[DPH should not even be built if GhcProfiled
1726simonpj@microsoft.com**20100823133439
1727 Ignore-this: 62acbf83de5b70ff6d27ab38ce9218ae
1728 
1729 It's not just when cleaning!
1730]
1731[The templateHaskellOk check should only run in stage2
1732simonpj@microsoft.com**20100823133353
1733 Ignore-this: f6dc9292923a1ca201953c5f58c0af3c
1734 
1735 Because rtsIsProfiled is only available in stage2
1736]
1737[Add a couple of missing tests for EAGER_BLACKHOLE
1738Simon Marlow <marlowsd@gmail.com>**20100823104654
1739 Ignore-this: 70c981b86370b0c7564b29b057650897
1740 This was leading to looping and excessive allocation, when the
1741 computation should have just blocked on the black hole. 
1742 
1743 Reported by Christian Höner zu Siederdissen <choener@tbi.univie.ac.at>
1744 on glasgow-haskell-users.
1745]
1746[Don't check for swept blocks in -DS.
1747Marco Túlio Gontijo e Silva <marcot@marcot.eti.br>**20100718225526
1748 Ignore-this: ad5dcf3c247bc19fbef5122c1142f3b2
1749 
1750 The checkHeap function assumed the allocated part of the block contained only
1751 alive objects and slops.  This was not true for blocks that are collected using
1752 mark sweep.  The code in this patch skip the test for this kind of blocks.
1753]
1754[Fix "darcs get"
1755Ian Lynagh <igloo@earth.li>**20100822183542]
1756[Add "darcs-all upstreampull"
1757Ian Lynagh <igloo@earth.li>**20100822163419
1758 This pulls from the upstream repos, for those packages which have
1759 upstreams
1760]
1761[Generate the bit in the user guide where we say what -fglasgow-exts does
1762Ian Lynagh <igloo@earth.li>**20100822155514
1763 Stops the docs going out of sync with the code.
1764]
1765[Factor out the packages file parsing in darcs-all
1766Ian Lynagh <igloo@earth.li>**20100822154813]
1767[Document --supported-extensions
1768Ian Lynagh <igloo@earth.li>**20100822134530]
1769[fix extraction of command stack of arguments of arrow "forms" (fixes #4236)
1770Ross Paterson <ross@soi.city.ac.uk>**20100822090022
1771 Ignore-this: a93db04ec4f20540642a19cdc67d1666
1772 
1773 The command stack was being extracted (by unscramble) with the outermost
1774 type first, contrary to the comment on the function.
1775]
1776[minor fix to comment
1777Ross Paterson <ross@soi.city.ac.uk>**20100822085838
1778 Ignore-this: 8d203ba2600eaf4cf21b043dcfa96cdc
1779]
1780[Add the RTS library path to the library search path
1781Ian Lynagh <igloo@earth.li>**20100820155523
1782 In case the RTS is being explicitly linked in. For #3807.
1783]
1784[Remove some duplication of C flags
1785Ian Lynagh <igloo@earth.li>**20100819233743
1786 We now use the CONF_CC_OPTS_STAGEn C flags in machdepCCOpts, rather than
1787 repeating them there.
1788]
1789[Set -fno-stack-protector in CONF_CC_OPTS_STAGE* rathre than extra-gcc-opts
1790Ian Lynagh <igloo@earth.li>**20100819233031
1791 The latter is only used when compiling .hc files, whereas we need it for
1792 .c files too.
1793]
1794[Give clearer errors for bad input in the packages file; suggested by pejo
1795Ian Lynagh <igloo@earth.li>**20100819232420]
1796[Set -march=i686 on OS X x86 in the configure variables
1797Ian Lynagh <igloo@earth.li>**20100819230939
1798 We used to set it only in machdepCCOpts, so this is more consistent
1799]
1800[Give each stage its own Config.hs
1801Ian Lynagh <igloo@earth.li>**20100819224709
1802 This also means the file is generated in a dist directory, not a
1803 source directory.
1804]
1805[Fix cleaning when GhcProfiled = YES
1806Ian Lynagh <igloo@earth.li>**20100819131346
1807 We need to include the DPH cleaning rules, even though we don't build DPH
1808 when GhcProfiled = YES.
1809]
1810[stgReallocBytes(DEBUG): don't fail when the ptr passed in is NULL
1811Simon Marlow <marlowsd@gmail.com>**20100817150836
1812 Ignore-this: 4b5063e65e01399f64a33f0d0555ff38
1813]
1814[Use make-command in rules/bindist.mk
1815Ian Lynagh <igloo@earth.li>**20100818191243
1816 Rather than it having its own specialised version
1817]
1818[Use make-command when installing packages
1819Ian Lynagh <igloo@earth.li>**20100818190600]
1820[Add _DATA_FILES to package-data.mk files
1821Ian Lynagh <igloo@earth.li>**20100818185801]
1822[Add a "make-command" utility Makefile function
1823Ian Lynagh <igloo@earth.li>**20100818183055]
1824[LLVM: Nicer format for lack of shared lib warning
1825David Terei <davidterei@gmail.com>**20100817145207
1826 Ignore-this: 753d45762601d87761614937a1bb6716
1827]
1828[fix FP_CHECK_ALIGNMENT for autoconf 2.66 (fixes #4252)
1829Ross Paterson <ross@soi.city.ac.uk>**20100816142442
1830 Ignore-this: cd784b8888d32b3b2cc2cc0969ec40f
1831 
1832 Recent versions of AS_LITERAL_IF don't like *'s.  Fix from
1833 
1834 http://blog.gmane.org/gmane.comp.sysutils.autoconf.general/month=20100701
1835]
1836[Refactor the command-line argument parsing (again)
1837simonpj@microsoft.com**20100816074453
1838 Ignore-this: 26dc9e37a88660a887a2e316ed7a9803
1839 
1840 This change allows the client of CmdLineParser a bit more flexibility,
1841 by giving him an arbitrary computation (not just a deprecation
1842 message) for each flag. 
1843 
1844 There are several clients, so there are lots of boilerplate changes.
1845 
1846 Immediate motivation: if RTS is not profiled, we want to make
1847 Template Haskell illegal.  That wasn't with the old setup.
1848]
1849[Add upstream repo to the packages file
1850Ian Lynagh <igloo@earth.li>**20100815154741]
1851[Make the "tag" column of the packages file always present
1852Ian Lynagh <igloo@earth.li>**20100815151657
1853 It makes the parsing simpler if we always have the same number of columns
1854]
1855[Disable object splitting on OSX; works around #4013
1856Ian Lynagh <igloo@earth.li>**20100815134759]
1857[Return memory to the OS; trac #698
1858Ian Lynagh <igloo@earth.li>**20100813170402]
1859[Reduce the xargs -s value we use on Windows
1860Ian Lynagh <igloo@earth.li>**20100812223721
1861 With 30000 I was getting:
1862     xargs: value for -s option should be < 28153
1863]
1864[LLVM: Enable shared lib support on Linux x64
1865David Terei <davidterei@gmail.com>**20100813191534
1866 Ignore-this: 642ed37af38e5f17d419bf4f09332671
1867]
1868[Re-do the arity calculation mechanism again (fix Trac #3959)
1869simonpj@microsoft.com**20100813161151
1870 Ignore-this: d4a2aa48150b503b20c25351a79decfb
1871 
1872 After rumination, yet again, on the subject of arity calculation,
1873 I have redone what an ArityType is (it's purely internal to the
1874 CoreArity module), and documented it better.  The result should
1875 fix #3959, and I hope the related #3961, #3983.
1876 
1877 There is lots of new documentation: in particular
1878  * Note [ArityType] 
1879    describes the new datatype for arity info
1880 
1881  * Note [State hack and bottoming functions]
1882    says how bottoming functions are dealt with, particularly
1883    covering catch# and Trac #3959
1884 
1885 I also found I had to be careful not to eta-expand single-method
1886 class constructors; see Note [Newtype classes and eta expansion].
1887 I think this part could be done better, but it works ok.
1888]
1889[Comments only
1890simonpj@microsoft.com**20100813161019
1891 Ignore-this: baf68300d8bc630bf0b7ab27647b33a0
1892]
1893[Modify FloatOut to fix Trac #4237
1894simonpj@microsoft.com**20100813163120
1895 Ignore-this: ffc8d00d4b7f0a8a785fcef312900413
1896 
1897 The problem was that a strict binding was getting floated
1898 out into a letrec. This only happened when profiling was
1899 on.  It exposed a fragility in the floating strategy.  This
1900 patch makes it more robust.  See
1901       Note [Avoiding unnecessary floating]
1902]
1903[Fix egregious bug in SetLevels.notWorthFloating
1904simonpj@microsoft.com**20100813161429
1905 Ignore-this: d22865f48d417e6a6b732de3dfba378f
1906 
1907 This bug just led to stupid code, which would
1908 later be optimised away, but better not to generate
1909 stupid code in the first place.
1910]
1911[Delete GhcLibProfiled
1912simonpj@microsoft.com**20100813140152
1913 Ignore-this: 2e1a3f677308be726bd022f45e2fd856
1914 
1915 Simon M and I looked at this, and we think GhcLibProfiled is
1916 (a) not needed (b) confusing.
1917 
1918 Ian should review.
1919 
1920 Really, if GhcProfiled is on we should also
1921 check that 'p' is in the GhcLibWays
1922]
1923[Do not build DPH when GhcProfiled (fixes #4172)
1924simonpj@microsoft.com**20100813140021
1925 Ignore-this: 9e20181643b456e841f845ae0cab0a9a
1926 
1927 Reason: DPH uses Template Haskell and TH doesn't work
1928 in a profiled compiler
1929]
1930[Fix Trac #4220
1931simonpj@microsoft.com**20100812131319
1932 Ignore-this: 33141cfd81627592150a9e5973411ff8
1933 
1934 For deriving Functor, Foldable, Traversable with empty
1935 data cons I just generate a null equation
1936    f _ = error "urk"
1937 
1938 There are probably more lurking (eg Enum) but this will do for now.
1939]
1940[Improve the Specialiser, fixing Trac #4203
1941simonpj@microsoft.com**20100812131133
1942 Ignore-this: 482afbf75165e24a80527a6e52080c07
1943 
1944 Simply fixing #4203 is a tiny fix: in case alterantives we should
1945 do dumpUDs *including* the case binder. 
1946 
1947 But I realised that we can do better and wasted far too much time
1948 implementing the idea.  It's described in
1949     Note [Floating dictionaries out of cases]
1950]
1951[Comments
1952simonpj@microsoft.com**20100812101456
1953 Ignore-this: 6362fe887d25688c12ef2c3cf5554ce4
1954]
1955[Comments only
1956simonpj@microsoft.com**20100812101439
1957 Ignore-this: 7ed2f5fc08811cbe9958c2309a9ed1fa
1958]
1959[Fix bug in linting of shadowed case-alternative binders
1960simonpj@microsoft.com**20100812101413
1961 Ignore-this: 9212a5e2c03421749f5935b3944ecf53
1962]
1963[Comments and spacing only
1964simonpj@microsoft.com**20100812101347
1965 Ignore-this: ed59a7dae7decb24470709dc1c118dbb
1966]
1967[Add more info to more parse error messages (#3811)
1968Ian Lynagh <igloo@earth.li>**20100809233108]
1969[Run finalizers *after* updating the stable pointer table (#4221)
1970Simon Marlow <marlowsd@gmail.com>**20100810133739
1971 Ignore-this: b0462f80dd64eac71e599d8a9f6dd665
1972 Silly bug really, we were running the C finalizers while the StablePtr
1973 table was still in a partially-updated state during GC, but finalizers
1974 are allowed to call freeStablePtr() (via hs_free_fun_ptr(), for
1975 example), and chaos ensues.
1976]
1977[Do the dependency-omitting for 'make 1' in a slightly different way
1978Simon Marlow <marlowsd@gmail.com>**20100810093446
1979 Ignore-this: af15edd3a1492cbd93111316b57e02e4
1980 
1981 I encountered a couple of things that broke after Ian's previous
1982 patch: one was my nightly build scripts that use 'make stage=2' at the
1983 top level, and the other is 'make fast' in libraries/base, which uses
1984 'stage=0' to avoid building any compilers.
1985 
1986 So my version of this patch is more direct: it just turns off the
1987 appropriate dependencies using a variable set by 'make 1', 'make 2',
1988 etc.
1989]
1990[Integrate new I/O manager, with signal support
1991Johan Tibell <johan.tibell@gmail.com>**20100724102355
1992 Ignore-this: eb092857a2a1b0ca966649caffe7ac2b
1993]
1994[Add DoAndIfThenElse support
1995Ian Lynagh <igloo@earth.li>**20100808194625]
1996[Make another parse error more informative
1997Ian Lynagh <igloo@earth.li>**20100808193340]
1998[Make a parse error say what it is failing to parse; part of #3811
1999Ian Lynagh <igloo@earth.li>**20100808155732]
2000[Send ghc progress output to stdout; fixes #3636
2001Ian Lynagh <igloo@earth.li>**20100808142542]
2002[Fix the HsColour test in the build system
2003Ian Lynagh <igloo@earth.li>**20100805155319
2004 Ignore-this: ba2752b04801a253e891b31e1914485d
2005]
2006[Fix the -lm configure test; fixes #4155
2007Ian Lynagh <igloo@earth.li>**20100805142508
2008 Ignore-this: 358b8b1074d2d22fb8d362ea6d8b80d6
2009]
2010[Don't restrict filenames in line pragmas to printable characters; fixes #4207
2011Ian Lynagh <igloo@earth.li>**20100805135011
2012 Ignore-this: e3d32312127165e40e6eaa919193d60b
2013 "printable" is ASCII-only, whereas in other locales we can get things like
2014 # 1 "<línea-de-orden>"
2015]
2016[Ensure extension flags are flattened in the Cmm phase
2017Ian Lynagh <igloo@earth.li>**20100805133614
2018 If we start with a .cmmcpp file then they don't get flattened in
2019 the CmmCpp phase, as we don't run that phase.
2020]
2021[Add "cmmcpp" as a Haskellish source suffix
2022Ian Lynagh <igloo@earth.li>**20100805132555]
2023[On amd64/OSX we don't need to be given memory in the first 31bits
2024Ian Lynagh <igloo@earth.li>**20100805120600
2025 Ignore-this: 42eb64e25ad4b66ae022884305e0297b
2026 as PIC is always on
2027]
2028[NCG: Don't worry about trying to re-freeze missing coalescences
2029benl@ouroborus.net**20100702053319
2030 Ignore-this: ea05cbee19b6c5c410db41292cbb64b0
2031]
2032[Make -rtsopts more flexible
2033Ian Lynagh <igloo@earth.li>**20100805011137
2034 The default is a new "some" state, which allows only known-safe flags
2035 that we want on by default. Currently this is only "--info".
2036]
2037[Test for (fd < 0) before trying to FD_SET it
2038Ian Lynagh <igloo@earth.li>**20100804173636]
2039[Remove "On by default" comments in DynFlags
2040Ian Lynagh <igloo@earth.li>**20100802110803
2041 Ignore-this: 2a51055277b5ce9f0e98e1438b212027
2042 These make less sense now we support multiple languges. The
2043 "languageExtensions" function gives the defaults.
2044]
2045[Fix build: Add newline to end of ghc-pkg/Main.hs
2046Ian Lynagh <igloo@earth.li>**20100801183206]
2047[Add a versions haddock binary for Windows
2048Ian Lynagh <igloo@earth.li>**20100801180917]
2049[ghc-pkg: don't fail, if a file is already removed
2050ich@christoph-bauer.net**20100725162606
2051 Ignore-this: 5501d6812c31f4da525c7fb24f6dcaed
2052]
2053[Remove push-all from file list in boot script (push-all no longer exists)
2054Ian Lynagh <igloo@earth.li>**20100801121841
2055 Ignore-this: eec130f06610d8728a57626682860a1a
2056]
2057[Add error checking to boot-pkgs script
2058Ian Lynagh <igloo@earth.li>**20100801121432
2059 Ignore-this: 8afd6663db443c774bad45d75bbfe950
2060]
2061[Add more error checking to the boot script
2062Ian Lynagh <igloo@earth.li>**20100801113628]
2063[Remove libHSrtsmain.a before creating it
2064Ian Lynagh <igloo@earth.li>**20100801005432
2065 Otherwise it isn't updated properly if rts/Main.c changes
2066]
2067[Expose the functions haddock needs even when haddock is disabled; #3558
2068Ian Lynagh <igloo@earth.li>**20100731115506]
2069[Always haddock by default
2070Ian Lynagh <igloo@earth.li>**20100730235001
2071 Revert this patch:
2072     Matthias Kilian <kili@outback.escape.de>**20090920181319
2073     Don't build haddock if HADDOC_DOCS = NO, and disable HADDOC_DOCS
2074         if GhcWithInterpreter = NO
2075     Haddock uses TcRnDriver.tcRnGetInfo, which is only available if
2076     GHCI is built. Set HADDOC_DOCS to NO if GhcWithInterpreter is NO,
2077     and disable the haddock build if HADDOC_DOCS = NO.
2078]
2079[Add a debugTrace for the phases that we run
2080Ian Lynagh <igloo@earth.li>**20100729201503]
2081[* Add StringPrimL as a constructor for Template Haskell (Trac #4168)
2082simonpj@microsoft.com**20100730131922
2083 Ignore-this: 520d0a0a14b499b299e8b2be8d148ff0
2084   
2085 There are already constructors for IntPrim, FloatPrim etc,
2086 so this makes it more uniform.
2087   
2088 There's a corresponding patch for the TH library
2089]
2090[Add thread affinity support for FreeBSD
2091Gabor Pali <pgj@FreeBSD.org>**20100720001409
2092 Ignore-this: 6c117b8219bfb45445089e82ee470410
2093 - Implement missing functions for setting thread affinity and getting real
2094   number of processors.
2095 - It is available starting from 7.1-RELEASE, which includes a native support
2096   for managing CPU sets.
2097 - Add __BSD_VISIBLE, since it is required for certain types to be visible in
2098   addition to POSIX & C99.
2099]
2100[Disable symbol visibility pragmas for FreeBSD
2101Ian Lynagh <igloo@earth.li>**20100729012507
2102 Do not use GCC pragmas for controlling visibility, because it causes
2103 "undefined reference" errors at link-time.  The true reasons are
2104 unknown, however FreeBSD 8.x includes GCC 4.2.1 in the base system,
2105 which might be buggy.
2106]
2107[Fix numeric escape sequences parsing
2108Anton Nikishaev <anton.nik@gmail.com>**20100721194208
2109 Ignore-this: dd71935b1866b5624f7975c45ad519a1
2110 This fixes trac bug #1344
2111]
2112[Explicitly give the right path to perl when making the OS X installer; #4183
2113Ian Lynagh <igloo@earth.li>**20100728163030]
2114[Set -fno-stack-protector in extra-gcc-opts; fixes #4206
2115Ian Lynagh <igloo@earth.li>**20100728161957
2116 We were using it only when building the RTS, and only on certain
2117 platforms. However, some versions of OS X need the flag, while others
2118 don't support it, so we now test for it properly.
2119]
2120[Make PersistentLinkerState fields strict; fixes #4208
2121Ian Lynagh <igloo@earth.li>**20100727201911
2122 Ignore-this: fc5cfba48cd16624f6bb15a7a03a3b4
2123 We modify fields a lot, so we retain the old value if they aren't forced.
2124]
2125[Don't rebuild dependency files unnecessarily when doing "make 1" etc
2126Ian Lynagh <igloo@earth.li>**20100726211512
2127 Ignore-this: d91a729e5113aa964cc67768e92e57ef
2128]
2129[LLVM: If user specifies optlo, don't use '-O' levels
2130David Terei <davidterei@gmail.com>**20100726105650
2131 Ignore-this: e05e103b09d1de937540ffad7983f88e
2132]
2133[Flatten flags for ghci's :show
2134Ian Lynagh <igloo@earth.li>**20100725135320]
2135[Add support for Haskell98 and Haskell2010 "languages"
2136Ian Lynagh <igloo@earth.li>**20100724230121]
2137[Rename "language" varibles etc to "extension", and add --supported-extensions
2138Ian Lynagh <igloo@earth.li>**20100724223624]
2139[Separate language option handling into 2 phases
2140Ian Lynagh <igloo@earth.li>**20100724212013
2141 We now first collect the option instructions (from the commandline,
2142 from pragmas in source files, etc), and then later flatten them into
2143 the list of enabled options. This will enable us to use different
2144 standards (H98, H2010, etc) as a base upon which to apply the
2145 instructions, when we don't know what the base will be when we start
2146 collecting instructions.
2147]
2148[Separate the language flags from the other DynFlag's
2149Ian Lynagh <igloo@earth.li>**20100724133103
2150 Ignore-this: 47bb8d42e621e47016b66c7472bd6cb5
2151]
2152[Set stage-specific CC/LD opts in the bindist configure.ac
2153Ian Lynagh <igloo@earth.li>**20100724113748
2154 Ignore-this: f06926d185a35ddd05490ca4a257e992
2155]
2156[Use different CC/LD options for different stages
2157Ian Lynagh <igloo@earth.li>**20100723223059]
2158[Add some error belchs to the linker, when we find bad magic numbers
2159Ian Lynagh <igloo@earth.li>**20100723200822]
2160[Add some more linker debugging prints
2161Ian Lynagh <igloo@earth.li>**20100723180237]
2162[When (un)loading an object fails, say which object in teh panic
2163Ian Lynagh <igloo@earth.li>**20100723162649]
2164[Add a release note: GHCi import syntax
2165Ian Lynagh <igloo@earth.li>**20100721193647]
2166[Deprecate NewQualifiedOperators extension (rejected by H')
2167Ian Lynagh <igloo@earth.li>**20100719150909
2168 Ignore-this: 6e7e3bedc5360c5975f73497b3e6cba5
2169]
2170[LLVM: Allow optlc and optlo to override default params for these systools
2171David Terei <davidterei@gmail.com>**20100722181631
2172 Ignore-this: e60af7941996f7170fb3bfb02a002082
2173]
2174[LLVM: Code and speed improvement to dominateAllocs pass.
2175David Terei <davidterei@gmail.com>**20100721143654
2176 Ignore-this: 9fb7058c8a2afc005521298c7b8d0036
2177]
2178[Comments only
2179simonpj@microsoft.com**20100721144257
2180 Ignore-this: b3091ddcd1df271eb85fe90978ab7adc
2181]
2182[Fix inlining for default methods
2183simonpj@microsoft.com**20100721144248
2184 Ignore-this: 61a11a8f741f775000c6318aae4b3191
2185 
2186 This was discombobulated by a patch a week ago;
2187 now fixed, quite straightforwardly.  See
2188 Note [Default methods and instances]
2189]
2190[Allow reification of existentials and GADTs
2191simonpj@microsoft.com**20100721090437
2192 Ignore-this: 20f1ccd336cc25aff4d4d67a9ac2211a
2193 
2194 It turns out that TH.Syntax is rich enough to express even GADTs,
2195 provided we express them in equality-predicate form.  So for
2196 example
2197 
2198   data T a where
2199      MkT1 :: a -> T [a]
2200      MkT2 :: T Int
2201 
2202 will appear in TH syntax like this
2203 
2204   data T a = forall b. (a ~ [b]) => MkT1 b
2205            | (a ~ Int) => MkT2
2206 
2207 While I was at it I also improved the reification of types,
2208 so that we use TH.TupleT and TH.ListT when we can.
2209]
2210[add numSparks# primop (#4167)
2211Simon Marlow <marlowsd@gmail.com>**20100720153746
2212 Ignore-this: f3f925e7de28f3f895213aefbdbe0b0f
2213]
2214[LLVM: Decrease max opt level used under OSX to avoid bug
2215David Terei <davidterei@gmail.com>**20100720160938
2216 Ignore-this: 34b0b3550f00b27b00ad92f8232745e5
2217 
2218 Currently, many programs compiled with GHC at -O2 and LLVM
2219 set to -O3 will segfault (only under OSX). Until this issue
2220 is fixed I have simply 'solved' the segfault by lowering
2221 the max opt level for LLVM used to -O2 under OSX.
2222 
2223 All these recent changes to OSX should mean its finally as
2224 stable as Linux and Windows.
2225]
2226[LLVM: Fix OSX to work again with TNTC disabled.
2227David Terei <davidterei@gmail.com>**20100720160845
2228 Ignore-this: 8dc98139cfa536b2a64aa364d040b581
2229]
2230[LLVM: Fix printing of local vars so LLVM works with -fnew-codegen
2231David Terei <davidterei@gmail.com>**20100720160302
2232 Ignore-this: d883c433dfaed67921a8c5360e1f9f6a
2233]
2234[Use a separate mutex to protect all_tasks, avoiding a lock-order-reversal
2235Simon Marlow <marlowsd@gmail.com>**20100716150832
2236 Ignore-this: ffbdb4ee502e0f724d57acb9bfbe9d92
2237 In GHC 6.12.x I found a rare deadlock caused by this
2238 lock-order-reversal:
2239 
2240 AQ cap->lock
2241   startWorkerTask
2242     newTask
2243       AQ sched_mutex
2244 
2245 scheduleCheckBlackHoles
2246   AQ sched_mutex
2247    unblockOne_
2248     wakeupThreadOnCapabilty
2249       AQ cap->lock
2250 
2251 so sched_mutex and cap->lock are taken in a different order in two
2252 places.
2253 
2254 This doesn't happen in the HEAD because we don't have
2255 scheduleCheckBlackHoles, but I thought it would be prudent to make
2256 this less likely to happen in the future by using a different mutex in
2257 newTask.  We can clearly see that the all_tasks mutex cannot be
2258 involved in a deadlock, becasue we never call anything else while
2259 holding it.
2260]
2261['make fast' in a package does not build any compilers
2262Simon Marlow <marlowsd@gmail.com>**20100715125904
2263 Ignore-this: f27e70faf3944831dad16e89a4e273da
2264]
2265[LLVM: Fix up botched last commit
2266David Terei <davidterei@gmail.com>**20100719104823
2267 Ignore-this: a32e0f6a38cb9e02527eb8ca69b3eb59
2268]
2269[LLVM: Fix warning introduce in last commit.
2270David Terei <davidterei@gmail.com>**20100719103411
2271 Ignore-this: e9c92a9402aff50d60ab26e6ad441bfc
2272]
2273[LLVM: Use mangler to fix up stack alignment issues on OSX
2274David Terei <davidterei@gmail.com>**20100718231000
2275 Ignore-this: 9f6e8cb855269cb3a5ac1a23480d0e71
2276]
2277[Fix #4195 (isGadtSyntaxTyCon returns opposite result)
2278illissius@gmail.com**20100715134134
2279 Ignore-this: a90403f893030432b5c15d743647f350
2280]
2281[Update to time 1.2.0.3
2282Ian Lynagh <igloo@earth.li>**20100717181810
2283 Ignore-this: 1ccb4801a73f399e6718ce556543ede1
2284]
2285[Reorder RTS --info output
2286Ian Lynagh <igloo@earth.li>**20100717162356]
2287[Fix unreg prof build: Define CCS_SYSTEM in stg/MiscClosures.h
2288Ian Lynagh <igloo@earth.li>**20100717142832
2289 Ignore-this: 9675f3f51b6dac40483155344e7f45b6
2290]
2291[Make mkDerivedConstants as a stage 1 program
2292Ian Lynagh <igloo@earth.li>**20100717000827
2293 Ignore-this: 5357403461b209b8606f1d33defb51cf
2294 This way it gets the defines for the right platform when cross-compiling
2295]
2296[Don't generate Haskell dependencies if we don't have any Haskell sources
2297Ian Lynagh <igloo@earth.li>**20100717000800
2298 Ignore-this: 454abd0358f535b7e789327125c9206c
2299]
2300[Link programs that have no Haskell objects with gcc rather than ghc
2301Ian Lynagh <igloo@earth.li>**20100716235303
2302 Ignore-this: f65588b69675edea616cc434e769b0a4
2303]
2304[Use gcc to build C programs for stages >= 1
2305Ian Lynagh <igloo@earth.li>**20100716223703
2306 Ignore-this: 9f843a4e17285cda582117504707f9e7
2307]
2308[Add platform info to "ghc --info" output
2309Ian Lynagh <igloo@earth.li>**20100716141953]
2310[Tidy up Config.hs generation
2311Ian Lynagh <igloo@earth.li>**20100716140630]
2312[Fix HC porting test in makefiles
2313Ian Lynagh <igloo@earth.li>**20100716010808
2314 Ignore-this: 6052c1dd022a6108ab2236a299ee1d84
2315 Now that we are trying to support cross compilation, we can't use
2316     "$(TARGETPLATFORM)" != "$(HOSTPLATFORM)"
2317 as a test for HC-porting.
2318]
2319[Change a BUILD var to a HOST var
2320Ian Lynagh <igloo@earth.li>**20100716002558]
2321[Remove an unnecessary #include
2322Ian Lynagh <igloo@earth.li>**20100715233930
2323 Ignore-this: dcede249de6be7e3c9305c9279c2ca07
2324]
2325[Split up some make commands, so that errors aren't overlooked
2326Ian Lynagh <igloo@earth.li>**20100715152237
2327 Ignore-this: fb69b0a25d9ca71dae5e75d38db675cd
2328 When we ask make to run "a | b", if a fails then the pipeline might
2329 still exit successfuly.
2330]
2331[Remove an unnecessary #include
2332Ian Lynagh <igloo@earth.li>**20100715143000
2333 Ignore-this: 4e098cac5dda2dd595ca0a0f5121853c
2334]
2335[Simplify some more CPP __GLASGOW_HASKELL__ tests
2336Ian Lynagh <igloo@earth.li>**20100715142500]
2337[Remove some code only used with GHC 6.11.*
2338Ian Lynagh <igloo@earth.li>**20100715141720]
2339[__GLASGOW_HASKELL__ >= 609 is now always true
2340Ian Lynagh <igloo@earth.li>**20100715141544]
2341[Correct the values in ghc_boot_platform.h
2342Ian Lynagh <igloo@earth.li>**20100714223717
2343 Ignore-this: 4c99116f7ac73fadbd6d16807f57a693
2344]
2345[Change some TARGET checks to HOST checks
2346Ian Lynagh <igloo@earth.li>**20100714184715]
2347[LLVM: Add inline assembly to binding.
2348David Terei <davidterei@gmail.com>**20100714152530
2349 Ignore-this: 72a7b5460c128ed511e8901e5889fe2b
2350]
2351[LLVM: Fix mistype in last commit which broke TNTC under win/linux.
2352David Terei <davidterei@gmail.com>**20100714153339
2353 Ignore-this: 302d7957e3dded80368ebade5312ab35
2354]
2355[Remove unnecessary #include
2356Ian Lynagh <igloo@earth.li>**20100713153704
2357 Ignore-this: c37d3127b1dc68f59270c07173994c28
2358]
2359[Change some TARGET tests to HOST tests in the RTS
2360Ian Lynagh <igloo@earth.li>**20100713141034
2361 Which was being used seemed to be random
2362]
2363[LLVM: Add in new LLVM mangler for implementing TNTC on OSX
2364David Terei <davidterei@gmail.com>**20100713183243
2365 Ignore-this: 394fb74d7f9657d8b454bd0148d24bf7
2366]
2367[Refactor where an error message is generated
2368simonpj@microsoft.com**20100713115733
2369 Ignore-this: f94467856238586fcbbe48537141cf78
2370]
2371[Comments only
2372simonpj@microsoft.com**20100713115703
2373 Ignore-this: 5815442c4e69b9ec331b34242a596253
2374]
2375[Comments on data type families
2376simonpj@microsoft.com**20100713115640
2377 Ignore-this: 90a333bb7f7d64a49fb7dd180d893f6b
2378]
2379[Fix Trac #T4136: take care with nullary symbol constructors
2380simonpj@microsoft.com**20100707135945
2381 Ignore-this: 2a717a24fefcd593ea41c23dad351db0
2382 
2383 When a nullary constructor is a symbol eg (:=:) we need
2384 to take care.  Annoying.
2385]
2386[Fix Trac #4127 (and hence #4173)
2387simonpj@microsoft.com**20100707123125
2388 Ignore-this: 98bb6d0f7182b59f8c93596c61f9785d
2389 
2390 The change involves a little refactoring, so that the default
2391 method Ids are brought into scope earlier, before the value
2392 declarations are compiled.  (Since a value decl may contain
2393 an instance decl in a quote.)
2394 
2395 See Note [Default method Ids and Template Haskell] in
2396 TcTyClsDcls.
2397]
2398[Fix second bug in Trac #4127
2399simonpj@microsoft.com**20100701140124
2400 Ignore-this: c8d1cc27364fe9ee5a52acb1ecb5cdd9
2401 
2402 This bug concerned the awkward shadowing we do for
2403 Template Haskell declaration brackets.  Lots of
2404 comments in
2405 
2406   Note [Top-level Names in Template Haskell decl quotes]
2407]
2408[ia64: switch handling of 'foreign import wrapper' (FIW) to libffi
2409Sergei Trofimovich <slyfox@community.haskell.org>**20100709213922
2410 Ignore-this: fd07687e0089aebabf62de85d2be693
2411 
2412 I tried to build darcs-2.4.4 with ghc-6.12.3 and got coredumps when darcs is used
2413 in interactive mode. I tried test from ticket #3516 and found out FIW code is broken.
2414 Instead of fixing it I just switched to libffi. Result built successfully, passed
2415 'foreign import wrapper' test from ticket #3516 and builds working darcs.
2416]
2417[* storage manager: preserve upper address bits on 64bit machines (thanks to zygoloid)
2418Sergei Trofimovich <slyfox@community.haskell.org>**20100709115917
2419 Ignore-this: 9f1958a19992091ddc2761c389ade940
2420 
2421 Patch does not touch amd64 as it's address lengts is 48 bits at most, so amd64 is unaffected.
2422 
2423 the issue: during ia64 ghc bootstrap (both 6.10.4 and 6.12.3) I
2424 got the failure on stage2 phase:
2425     "inplace/bin/ghc-stage2"   -H32m -O -H64m -O0 -w ...
2426     ghc-stage2: internal error: evacuate: strange closure type 15
2427         (GHC version 6.12.3 for ia64_unknown_linux)
2428         Please report this as a GHC bug:  http://www.haskell.org/ghc/reportabug
2429     make[1]: *** [libraries/dph/dph-base/dist-install/build/Data/Array/Parallel/Base/Hyperstrict.o] Aborted
2430 
2431 gdb backtrace (break on 'barf'):
2432 Breakpoint 1 at 0x400000000469ec31: file rts/RtsMessages.c, line 39.
2433 (gdb) run -B/var/tmp/portage/dev-lang/ghc-6.12.3/work/ghc-6.12.3/inplace/bin --info
2434 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
2435 [Thread debugging using libthread_db enabled]
2436 
2437 Breakpoint 1, barf (s=0x40000000047915b0 "evacuate: strange closure type %d") at rts/RtsMessages.c:39
2438 39        va_start(ap,s);
2439 (gdb) bt
2440 #0  barf (s=0x40000000047915b0 "evacuate: strange closure type %d") at rts/RtsMessages.c:39
2441 #1  0x400000000474a1e0 in evacuate (p=0x6000000000147958) at rts/sm/Evac.c:756
2442 #2  0x40000000046d68c0 in scavenge_srt (srt=0x6000000000147958, srt_bitmap=7) at rts/sm/Scav.c:348
2443 ...
2444 
2445 > 16:52:53 < zygoloid> slyfox: i'm no ghc expert but it looks like HEAP_ALLOCED_GC(q)
2446 >                      is returning true for a FUN_STATIC closure
2447 > 17:18:43 < zygoloid> try: p HEAP_ALLOCED_miss((unsigned long)(*p) >> 20, *p)
2448 > 17:19:12 < slyfox> (gdb) p HEAP_ALLOCED_miss((unsigned long)(*p) >> 20, *p)
2449 > 17:19:12 < slyfox> $1 = 0
2450 > 17:19:40 < zygoloid> i /think/ that means the mblock_cache is broken
2451 > 17:22:45 < zygoloid> i can't help further. however i am suspicious that you seem to have pointers with similar-looking low 33
2452 >                      bits and different high 4 bits, and it looks like such pointers get put into the same bucket in
2453 >                      mblock_cache.
2454 ...
2455 > 17:36:16 < zygoloid> slyfox: try changing the definition of MbcCacheLine to StgWord64, see if that helps
2456 > 17:36:31 < zygoloid> that's in includes/rts/storage/MBlock.h
2457 And it helped!
2458]
2459[Fixing link failure of compiler on ia64 ('-Wl,' prefixed value passed directly to ld)
2460Sergei Trofimovich <slyfox@community.haskell.org>**20100708180943
2461 Ignore-this: ced99785e1f870ee97e5bec658e2504f
2462 
2463     /usr/bin/ld -Wl,--relax -r -o dist-stage1/build/HSghc-6.10.4.o \
2464                                   dist-stage1/build/BasicTypes.o dist-stage1/build/DataCon.o ...
2465     /usr/bin/ld: unrecognized option '-Wl,--relax'
2466 
2467 If we just drop '-Wl,' part it will not help as '-r' and '--relax' are incompatible.
2468 
2469 Looks like '-Wl,--relax' was skipped by earlier binutils' ld as unknown option.
2470 Removing ia64 specific path.
2471]
2472[LLVM: Allow getelementptr to use LlvmVar for indexes.
2473David Terei <davidterei@gmail.com>**20100712152529
2474 Ignore-this: 9e158d9b89a86bca8abf11d082328278
2475]
2476[Move all the warning workarounds to one place
2477Ian Lynagh <igloo@earth.li>**20100710161723]
2478[xhtml is now warning-free
2479Ian Lynagh <igloo@earth.li>**20100710144635]
2480[Move a bit of build system code
2481Ian Lynagh <igloo@earth.li>**20100709224534]
2482[adapt to the new async exceptions API
2483Simon Marlow <marlowsd@gmail.com>**20100709125238
2484 Ignore-this: 55d845e40b9daed3575c1479d8dda1d5
2485]
2486[quiet some new spewage
2487Simon Marlow <marlowsd@gmail.com>**20100709091521
2488 Ignore-this: de7f91976bbc9789e6fd7091f05c25c0
2489]
2490[New asynchronous exception control API (ghc parts)
2491Simon Marlow <marlowsd@gmail.com>**20100708144851
2492 Ignore-this: 56320c5fc61ae3602d586609387aae22
2493 
2494 As discussed on the libraries/haskell-cafe mailing lists
2495   http://www.haskell.org/pipermail/libraries/2010-April/013420.html
2496 
2497 This is a replacement for block/unblock in the asychronous exceptions
2498 API to fix a problem whereby a function could unblock asynchronous
2499 exceptions even if called within a blocked context.
2500 
2501 The new terminology is "mask" rather than "block" (to avoid confusion
2502 due to overloaded meanings of the latter).
2503 
2504 In GHC, we changed the names of some primops:
2505 
2506   blockAsyncExceptions#   -> maskAsyncExceptions#
2507   unblockAsyncExceptions# -> unmaskAsyncExceptions#
2508   asyncExceptionsBlocked# -> getMaskingState#
2509 
2510 and added one new primop:
2511 
2512   maskUninterruptible#
2513 
2514 See the accompanying patch to libraries/base for the API changes.
2515]
2516[remove outdated comment
2517Simon Marlow <marlowsd@gmail.com>**20100708100840
2518 Ignore-this: afb2e9f6fe1f1acda51b0cbdf2637176
2519]
2520[remove 'mode: xml' emacs settings (#2208)
2521Simon Marlow <marlowsd@gmail.com>**20100708100817
2522 Ignore-this: 3a8d997fb90e01ca88dc47fb95feeba0
2523]
2524[typo in comment
2525Simon Marlow <marlowsd@gmail.com>**20100616111359
2526 Ignore-this: d3ef9288d6d6b9ab3bacbe09e0d9801c
2527]
2528[Win32 getProcessElapsedTime: use a higher-resolution time source
2529Simon Marlow <marlowsd@gmail.com>**20100708093223
2530 Ignore-this: 821989d4ff7ff2bff40cee71a881521c
2531 QueryPerformanceCounter() on Windows gives much better resolution than
2532 GetSystemTimeAsFileTime().
2533]
2534[alpha: switch handling of 'foreign import wrapper' (FIW) to libffi
2535Sergei Trofimovich <slyfox@community.haskell.org>**20100708065318
2536 Ignore-this: ddee15876737a6aa7f6dabc8ff79ce0d
2537 
2538 I tried to build ghc-6.12.3 and found out FIW part of code
2539 does not compile anymore. It uses absent functions under #ifdef.
2540 Instead of fixing it I just switched to libffi. Result built successfully
2541 and passed 'foreign import wrapper' test I wrote for trac ticket #3516.
2542 
2543 I didn't try to build -HEAD yet, but this patch only removes code, so
2544 it should not make -HEAD worse.
2545]
2546[Reorder the CPP flags so -optP can override the platform defines
2547Ian Lynagh <igloo@earth.li>**20100708203523]
2548[Add docs for DatatypeContexts extension
2549Ian Lynagh <igloo@earth.li>**20100707230907
2550 Ignore-this: 8158f03b35a2d7442a75fe85d6f1b1c7
2551]
2552[Make datatype contexts an extension (on by default) (DatatypeContexts)
2553Ian Lynagh <igloo@earth.li>**20100707212529
2554 Ignore-this: 6885ff510a0060610eeeba65122caef5
2555]
2556[LLVM: Fix various typos in comments
2557David Terei <davidterei@gmail.com>**20100707220448
2558 Ignore-this: 1ba3e722f150492da2f9d485c5795e80
2559]
2560[Handle haddock headers when looking for LANGUAGE/OPTIONS_GHC pragmas
2561Ian Lynagh <igloo@earth.li>**20100707120423
2562 Ignore-this: a75aa67690284a6cee3e62c943d4fd01
2563]
2564[Make pragState call mkPState, rather than duplicating everything
2565Ian Lynagh <igloo@earth.li>**20100706173007
2566 Ignore-this: 61fe24b99dbe7a42efff1a9dd703a75c
2567 This also means that extsBitmap gets set, whereas is was just being set
2568 to 0 before.
2569]
2570[LLVM: Add alias type defenitions to LlvmModule.
2571David Terei <davidterei@gmail.com>**20100707142053
2572 Ignore-this: eee6ad5385563ccf08e664d2634a03f2
2573]
2574[LLVM: Use packed structure type instead of structure type
2575David Terei <davidterei@gmail.com>**20100707120320
2576 Ignore-this: a06e0359d182291b81cae56993ca385e
2577 
2578 The regular structure type adds padding to conform to the platform ABI,
2579 which causes problems with structures storing doubles under windows since
2580 we don't conform to the platform ABI there. So we use packed structures
2581 instead now that don't do any padding.
2582]
2583[Make mkPState and pragState take their arguments in the same order
2584Ian Lynagh <igloo@earth.li>**20100706172611]
2585[Remove an out-of-date comment
2586Ian Lynagh <igloo@earth.li>**20100706172217
2587 Ignore-this: 710ebd7d2dc01c1b0f1e58a5b6f85701
2588]
2589[LLVM: Stop llvm saving stg caller-save regs across C calls
2590David Terei <davidterei@gmail.com>**20100705162629
2591 Ignore-this: 28b4877b31b9358e682e38fc54b88658
2592 
2593 This is already handled by the Cmm code generator so LLVM is simply
2594 duplicating work. LLVM also doesn't know which ones are actually live
2595 so saves them all which causes a fair performance overhead for C calls
2596 on x64. We stop llvm saving them across the call by storing undef to
2597 them just before the call.
2598]
2599[LLVM: Add in literal undefined value to binding
2600David Terei <davidterei@gmail.com>**20100705161544
2601 Ignore-this: 95d8361b11584aaeec44c30e76916470
2602]
2603[LLVM: Add a literal NULL value to binding
2604David Terei <davidterei@gmail.com>**20100705161308
2605 Ignore-this: 9507b4b12c1157498704a9d1e5860f12
2606 
2607 Patch from Erik de Castro Lopo <erikd@mega-nerd.com>.
2608]
2609[refactor import declaration support (#2362)
2610Simon Marlow <marlowsd@gmail.com>**20100705104557
2611 Ignore-this: ee034ac377078a7e92bfada1907c86a0
2612]
2613[Disable dynamic linking optimisations on OS X
2614Simon Marlow <marlowsd@gmail.com>**20100705103014
2615 Ignore-this: b04420d3705c51112797758d17b2e40c
2616 To improve performance of the RTS when dynamically linked on x86, I
2617 previously disabled -fPIC for certain critical modules (the GC, and a
2618 few others).  However, build reports suggest that the dynamic linker
2619 on OS X doesn't like this, so I'm disabling this optimsation on that
2620 platform.
2621]
2622[trac #2362 (full import syntax in ghci)
2623amsay@amsay.net**20100625032632
2624 Ignore-this: a9d0859d84956beb74e27b797431bf9c
2625 'import' syntax is seperate from ':module' syntax
2626]
2627[Simplify ghc-pkg's Cabal dependencies
2628Ian Lynagh <igloo@earth.li>**20100704184155
2629 We no longer support building with a compiler that doesn't come with
2630 base 4.
2631]
2632[Use Cabal to configure the dist-install ghc-pkg; fixes trac #4156
2633Ian Lynagh <igloo@earth.li>**20100704132612]
2634[Remove dead code (standalone deriving flag no longer used in parser)
2635Ian Lynagh <igloo@earth.li>**20100701162058]
2636[LLVM: Use the inbounds keyword for getelementptr instructions.
2637David Terei <davidterei@gmail.com>**20100702160511
2638 Ignore-this: 3708e658a4c82b78b1402393f4405541
2639]
2640[threadPaused: fix pointer arithmetic
2641Simon Marlow <marlowsd@gmail.com>**20100701085046
2642 Ignore-this: b78210e5d978f18ffd235f1c78a55a23
2643 Noticed by Henrique Ferreiro <hferreiro@udc.es>, thanks!
2644]
2645[LLVM: Change more operations to use getelementptr
2646David Terei <davidterei@gmail.com>**20100701161856
2647 Ignore-this: fb24eb124e203f50680c6fec3ff9fe7d
2648]
2649[Add the haskell2010 package
2650Simon Marlow <marlowsd@gmail.com>**20100630125532
2651 Ignore-this: e9b011313f283a8ff2fcda7d029a01f
2652]
2653[LLVM: Use getelementptr instruction for a lot of situations
2654David Terei <davidterei@gmail.com>**20100630181157
2655 Ignore-this: 34d314dd8dffad9bdcffdc525261a49d
2656 
2657 LLVM supports creating pointers in two ways, firstly through
2658 pointer arithmetic (by casting between pointers and ints)
2659 and secondly using the getelementptr instruction. The second way
2660 is preferable as it gives LLVM more information to work with.
2661 
2662 This patch changes a lot of pointer related code from the first
2663 method to the getelementptr method.
2664]
2665[remove out of date comments; point to the wiki
2666Simon Marlow <marlowsd@gmail.com>**20100625100313
2667 Ignore-this: 95f363a373534b9471b1818102ec592d
2668]
2669[NCG: allocatableRegs is only giving us 8 SSE regs to allocate to
2670benl@ouroborus.net**20100629054321
2671 Ignore-this: b3e0fa0b4ce988a0258dc12261989ee0
2672]
2673[LLVM: Use intrinsic functions for pow, sqrt, sin, cos
2674David Terei <davidterei@gmail.com>**20100628182949
2675 Ignore-this: 98a0befaca3fe2b36d710d8ff9f062c4
2676 
2677 Instead of calling the C library for these Cmm functions
2678 we use intrinsic functions provided by llvm. LLVM will
2679 then either create a compile time constant if possible, or
2680 use a cpu instruction or as a last resort call the C
2681 library.
2682]
2683[LLVM: Fix test '2047' under linux-x64
2684David Terei <davidterei@gmail.com>**20100628165256
2685 Ignore-this: 41735d4f431a430db636621650ccd71e
2686]
2687[LLVM: Fix test 'ffi005' under linux-x64
2688David Terei <davidterei@gmail.com>**20100628155355
2689 Ignore-this: 841f3142c63cc898ac4c3f89698a837e
2690]
2691[LLVM: Update to use new fp ops introduced in 2.7
2692David Terei <davidterei@gmail.com>**20100628144037
2693 Ignore-this: 5dd2e5964e3c039d297ed586841e706b
2694]
2695[Add noalias and nocapture attributes to pointer stg registers
2696David Terei <davidterei@gmail.com>**20100628115120
2697 Ignore-this: 492a1e723cb3a62498d240d7de92dd7
2698 
2699 At the moment this gives a very slight performance boost of around 1 - 2%.
2700 Future changes to the generated code though so that pointers are kept as
2701 pointers more often instead of being cast to integer types straight away
2702 should hopefully improve the benefit this brings.
2703 
2704]
2705[during shutdown, only free the heap if we waited for foreign calls to exit
2706Simon Marlow <marlowsd@gmail.com>**20100628090536
2707 Ignore-this: d545384a4f641d701455d08ef1217479
2708]
2709[Fix typo in -ddump-pass's document.
2710shelarcy <shelarcy@gmail.com>**20100620070759
2711 Ignore-this: f4f1ddb53f147949e948147d89190c37
2712]
2713[Add #undefs for posix source symbols when including papi.h
2714dmp@rice.edu**20100624163514
2715 Ignore-this: 8a1cba21b880d12a75a75f7e96882053
2716 
2717 Validation fails when validating with PAPI support (i.e. GhcRtsWithPapi  = YES
2718 in validate.mk).  The problem is that the posix symbols are defined by a header
2719 included from papi.h. Compilation then fails because these symbols are
2720 redefined in PosixSource.h.
2721 
2722 This patch adds an undefine for the posix symbols after including papi.h and
2723 before including PosixSource.h. The #undefines are localized to Papi.c since
2724 that is the only case where they are getting defined twice.
2725]
2726[Use machdepCCOpts in runPhase_MoveBinary; fixes trac #3952
2727Ian Lynagh <igloo@earth.li>**20100625220953]
2728[LLVM: Fix bug with calling tail with empty list
2729David Terei <davidterei@gmail.com>**20100625115729
2730 Ignore-this: 46b4b32c8d92372a2d49794a96fe1613
2731]
2732[Fix warnings
2733benl@ouroborus.net**20100624091339
2734 Ignore-this: 5ba4bbd6abb9c9d1fb8c5d21ab73f218
2735]
2736[NCG: Comments and formatting only
2737benl@ouroborus.net**20100624083121
2738 Ignore-this: 86002e72c30d06bcc876d8c49f4caa5a
2739]
2740[NCG: Do the actual reversing of SCCs
2741benl@ouroborus.net**20100624082717
2742 Ignore-this: 12d2027ea118e751fbb48b27126150ef
2743]
2744[NCG: Fix dumping of graphs in regalloc stats for graph allocator
2745benl@ouroborus.net**20100624082625
2746 Ignore-this: 2b971bc9e0318099a9afb0e0db135730
2747]
2748[NCG: Reverse SCCs after each round in the graph allocator
2749benl@ouroborus.net**20100624082437
2750 Ignore-this: f0152e4039d6f16f7b5a99b286538116
2751]
2752[NCG: Don't actually complain on unreachable code blocks
2753benl@ouroborus.net**20100624081445
2754 Ignore-this: e7335ae6120917cb858c38c7c6da8e24
2755]
2756[NCG: Do explicit check for precondition of computeLiveness
2757benl@ouroborus.net**20100624080747
2758 Ignore-this: e7053c4e5e4c3c746b5ebf016913424a
2759 
2760  computeLiveness requires the SCCs of blocks to be in reverse dependent
2761  order, and if they're not it was silently giving bad liveness info,
2762  yielding a bad allocation.
2763 
2764  Now it complains, loudly.
2765]
2766[NCG: Fix off-by-one error in realRegSqueeze
2767benl@ouroborus.net**20100623095813
2768 Ignore-this: ab0698686d4c250da8e207f734f8252d
2769]
2770[NCG: Handle stripping of liveness info from procs with no blocks (like stg_split_marker)
2771benl@ouroborus.net**20100623091209
2772 Ignore-this: c0319b6cc62ec713afe4eb03790406e3
2773]
2774[NCG: Emit a warning on unreachable code block instead of panicing
2775benl@ouroborus.net**20100623085002
2776 Ignore-this: d20314b79e3c31e764ed4cd97290c696
2777]
2778[NCG: Remember to keep the entry block first when erasing liveness info
2779Ben.Lippmeier@anu.edu.au**20090917104429
2780 Ignore-this: 1b0c1df19d622858d50ffb6a01f2cef0
2781]
2782[NCG: Refactor representation of code with liveness info
2783Ben.Lippmeier@anu.edu.au**20090917090730
2784 Ignore-this: 2aebb3b02ebd92e547c5abad9feb0f0d
2785 
2786  * I've pushed the SPILL and RELOAD instrs down into the
2787    LiveInstr type to make them easier to work with.
2788 
2789  * When the graph allocator does a spill cycle it now just
2790    re-annotates the LiveCmmTops instead of converting them
2791    to NatCmmTops and back.
2792 
2793  * This saves working out the SCCS again, and avoids rewriting
2794    the SPILL and RELOAD meta instructions into real machine
2795    instructions.
2796]
2797[NCG: Add sanity checking to linear allocator
2798Ben.Lippmeier@anu.edu.au**20090917090335
2799 Ignore-this: 5a442be8b5087d04bc8b58dffa9ea080
2800 If there are are unreachable basic blocks in the native code then the
2801 linear allocator might loop. Detect this case and panic instead.
2802]
2803[NCG: Refactor LiveCmmTop to hold a list of SCCs instead of abusing ListGraph
2804Ben.Lippmeier@anu.edu.au**20090917060332
2805 Ignore-this: 3fec8d69ed0f760e53a202f873d5d9cb
2806]
2807[NCG: Allow the liveness map in a LiveInfo to be Nothing
2808Ben.Lippmeier@anu.edu.au**20090917043937
2809 Ignore-this: 5f82422d54d1b0ffc0589eb7e82fb7a4
2810]
2811[NCG: Also show the result of applying coalesings with -ddump-asm-regalloc-stages
2812Ben.Lippmeier.anu.edu.au**20090917034427
2813 Ignore-this: 76bd6d5ca43adb2167cb25832cbaa80b
2814]
2815[Fix panic when running "ghc -H"; trac #3364
2816Ian Lynagh <igloo@earth.li>**20100624234011
2817 The problem is that showing SDoc's looks at the static flags global
2818 variables, but those are panics while we are parsing the static flags.
2819 We work around this by explicitly using a fixed prettyprinter style.
2820]
2821[Allow for stg registers to have pointer type in llvm BE.
2822David Terei <davidterei@gmail.com>**20100621175839
2823 Ignore-this: fc09b1a8314aef0bde945c77af1124fb
2824 
2825 Before all the stg registers were simply a bit type or
2826 floating point type but now they can be declared to have
2827 a pointer type to one of these. This will allow various
2828 optimisations in the future in llvm since the type is
2829 more accurate.
2830]
2831[Add support for parameter attributes to the llvm BE binding
2832David Terei <davidterei@gmail.com>**20100624111744
2833 Ignore-this: 77f3c0c7bf8f81c4a154dc835ae7bcba
2834 
2835 These allow annotations of the code produced by the backend
2836 which should bring some perforamnce gains. At the moment
2837 the attributes aren't being used though.
2838]
2839[Cast some more nats to StgWord to be on the safe side
2840Simon Marlow <marlowsd@gmail.com>**20100624105700
2841 Ignore-this: e6176683856f9872fdeb2358bb065bb8
2842 And add a comment about the dangers of int overflow
2843]
2844[comments only
2845Simon Marlow <marlowsd@gmail.com>**20100624105105
2846 Ignore-this: fc8f762f4c3a5ffca2f8da2bc63ac2a4
2847]
2848[Fix an arithmetic overflow bug causing crashes with multi-GB heaps
2849Simon Marlow <marlowsd@gmail.com>**20100624104654
2850 Ignore-this: 67210755aa098740ff5230347be0fd5d
2851]
2852[Add support for collecting PAPI native events
2853dmp@rice.edu**20100622195953
2854 Ignore-this: 7269f9c4dfb2912a024eb632200fcd1
2855 
2856 This patch extends the PAPI support in the RTS to allow collection of native
2857 events. PAPI can collect data for native events that are exposed by the
2858 hardware beyond the PAPI present events. The native events supported on your
2859 hardware can found by using the papi_native_avail tool.
2860 
2861 The RTS already allows users to specify PAPI preset events from the command
2862 line. This patch extends that support to allow users to specify native events.
2863 The changes needed are:
2864 
2865 1) New option (#) for the RTS PAPI flag for native events. For example, to
2866    collect the native event 0x40000000, use ./a.out +RTS -a#0x40000000 -sstderr
2867 
2868 2) Update the PAPI_FLAGS struct to store whether the user specified event is a
2869    papi preset or a native event
2870 
2871 3) Update init_countable_events function to add the native events after parsing
2872    the event code and decoding the name using PAPI_event_code_to_name
2873 
2874]
2875[Don't warn about unused bindings with parents in .hs-boot files; trac #3449
2876Ian Lynagh <igloo@earth.li>**20100624110351]
2877[fix the home_imps filter to allow for 'import "this" <module>'
2878Simon Marlow <marlowsd@gmail.com>**20100621125535
2879 Ignore-this: da4e605b0513afc32a4e7caa921a2c76
2880 In the PackageImports extension, import "this" means "import from the
2881 current package".
2882]
2883[Use the standard C wrapper code for the ghc-$version.exe wrapper
2884Ian Lynagh <igloo@earth.li>**20100622202859
2885 Ignore-this: 60cd3e6db3afb63e6ba9e2db3b033580
2886]
2887[Don't rely on "-packagefoo" working; use "-package foo" instead
2888Ian Lynagh <igloo@earth.li>**20100622202547]
2889[Remove unnecessary C #includes
2890Ian Lynagh <igloo@earth.li>**20100622172919]
2891[Make the ghci.exe wrapper call the right ghc.exe
2892Ian Lynagh <igloo@earth.li>**20100622172247]
2893[More updates to datalayout description in llvm BE
2894David Terei <davidterei@gmail.com>**20100622165339
2895 Ignore-this: b0c604fe7673b0aa7c7064694d574437
2896]
2897[Remove LlvmAs phase as the llvm opt tool now handles this phase
2898David Terei <davidterei@gmail.com>**20100622144044
2899 Ignore-this: b9fd8f959702b6af014e2fa654bede3
2900 
2901 This phase originally invoked the llvm-as tool that turns a textual
2902 llvm assembly file into a bit code file for the rest of llvm to deal
2903 with. Now the llvm opt tool can do this itself, so we don't need to
2904 use llvm-as anymore.
2905]
2906[Update datalayout info in llvm BE
2907David Terei <davidterei@gmail.com>**20100622123457
2908 Ignore-this: 89b043d211225dcd819f30549afe1840
2909]
2910[Fix handling of float literals in llvm BE
2911David Terei <davidterei@gmail.com>**20100622121642
2912 Ignore-this: a3b5f382ad4b5a426ad4b581664506fa
2913]
2914[Declare some top level globals to be constant when appropriate
2915David Terei <davidterei@gmail.com>**20100621174954
2916 Ignore-this: 44832f65550d4f995d11c01cc1affef5
2917 
2918 This involved removing the old constant handling mechanism
2919 which was fairly hard to use. Now being constant or not is
2920 simply a property of a global variable instead of a separate
2921 type.
2922]
2923[Reduce the number of passes over the cmm in llvm BE
2924David Terei <davidterei@gmail.com>**20100621125220
2925 Ignore-this: cb2f4e54e8d0f982d5087fbeee35c73c
2926]
2927[Fix negate op not working for -0 in llvm backend
2928David Terei <davidterei@gmail.com>**20100621123606
2929 Ignore-this: c5d76e5cffa781fed074137851b1347f
2930]
2931[ROLLBACK: picCCOpts: -dynamic should not entail -optc-fPIC
2932Simon Marlow <marlowsd@gmail.com>**20100621100409
2933 Ignore-this: f2fac7df33d3919199befc59bd455414
2934 and add a comment to explain why it was wrong.  This fixes the dyn
2935 test failures that sprang up recently.
2936]
2937[Check files are really created in libffi
2938Ian Lynagh <igloo@earth.li>**20100620163724
2939 when we think that the libffi build creates them, so they just depend
2940 on the libffi build stamp.
2941]
2942[Improve the missing-import-list warning
2943Ian Lynagh <igloo@earth.li>**20100620124320
2944 Ignore-this: 551e5fdf2dfb56b49d249e0cebaa6115
2945]
2946[Tweak missing-import-list warning
2947Ian Lynagh <igloo@earth.li>**20100620122622
2948 Ignore-this: 360cdf59ae13d66ded181129325506c4
2949]
2950[trac #1789 (warnings for missing import lists)
2951amsay@amsay.net**20100618150649
2952 Ignore-this: b0b0b1e048fbca0817c1e6fade1153fa
2953]
2954[Refix docs for sizeofByteArray#/sizeofMutableByteArray# (#3800)
2955Ian Lynagh <igloo@earth.li>**20100620103749]
2956[Remove some old commented out code
2957Ian Lynagh <igloo@earth.li>**20100620000459]
2958[SET_ARR_HDR's last argument is now a number of bytes, rather than words
2959Ian Lynagh <igloo@earth.li>**20100619235214
2960 This avoids unnecessary work and potential loss of information
2961]
2962[Replace an (incorrect) bytes-to-words calculation with ROUNDUP_BYTES_TO_WDS
2963Ian Lynagh <igloo@earth.li>**20100619234310]
2964[FIX #38000 Store StgArrWords payload size in bytes
2965Antoine Latter <aslatter@gmail.com>**20100101183346
2966 Ignore-this: 7bf3ab4fc080c46311fc10b179361bb6
2967]
2968[Add win32 datalayout support to llvm backend
2969David Terei <davidterei@gmail.com>**20100618131733
2970 Ignore-this: 4b7bffaa8ef38c628ab852c1a6c1c009
2971]
2972[Remove unused 'ddump-opt-llvm' flag
2973David Terei <davidterei@gmail.com>**20100618101237
2974 Ignore-this: f78467496d986897e49d82646ee2907e
2975]
2976[generate "movl lbl(%reg1), %reg2" instructions, better codegen for -fPIC
2977Simon Marlow <marlowsd@gmail.com>**20100618082258
2978 Ignore-this: a25567ebff9f575303ddc8f2deafebbf
2979]
2980[joinToTargets: fix a case of panic "handleComponent cyclic"
2981Simon Marlow <marlowsd@gmail.com>**20100618082147
2982 Ignore-this: 765baeefbb5a41724004acd92405cecc
2983]
2984[comment typo
2985Simon Marlow <marlowsd@gmail.com>**20100618082102
2986 Ignore-this: e495610b7dd5ec30b02938638b56cb7
2987]
2988[Add support of TNTC to llvm backend
2989David Terei <davidterei@gmail.com>**20100618093205
2990 Ignore-this: 2c27d21668374a5b0d5e844882c69439
2991 
2992 We do this through a gnu as feature called subsections,
2993 where you can put data/code into a numbered subsection
2994 and those subsections will be joined together in descending
2995 order by gas at compile time.
2996]
2997[Don't automatically insert a -fvia-C flag in an unregisterised compiler
2998Ian Lynagh <igloo@earth.li>**20100617190901
2999 Ignore-this: eb25a9a338fade9e17c153da7c5f27e9
3000 The default object mode is already HscC, so it's unnecessary, and
3001 -fvia-C generates a deprecated flag warning now.
3002]
3003[In PosixSource.h, conditionally define things based on platform
3004Ian Lynagh <igloo@earth.li>**20100617174912
3005 This may not be ideal, but it should get GHC building on all platforms
3006 again.
3007]
3008[disable -fPIC for the GC for performance reasons
3009Simon Marlow <marlowsd@gmail.com>**20100617140025
3010 Ignore-this: c7c152bbff71ef7891eaee8ff39fc281
3011 see comment for details
3012]
3013[picCCOpts: -dynamic should not entail -optc-fPIC
3014Simon Marlow <marlowsd@gmail.com>**20100617115259
3015 Ignore-this: d71e42bd56e4bd107d2c431b801855e5
3016]
3017[Make getAllocations() visible
3018Simon Marlow <marlowsd@gmail.com>**20100617113259
3019 Ignore-this: 1b7fb38a01358c0acbe8987df07d23f2
3020]
3021[Fix the symbol visibility pragmas
3022Simon Marlow <marlowsd@gmail.com>**20100617105758
3023 Ignore-this: 76552500865473a1dbebbc1cc2def9f0
3024]
3025[pick up changes to $(GhcStage1HcOpts) without re-configuring the ghc package
3026Simon Marlow <marlowsd@gmail.com>**20100616124718
3027 Ignore-this: afb56d5560c813051285607fefb15493
3028]
3029[Fix bindisttest Makefile
3030Ian Lynagh <igloo@earth.li>**20100616205611
3031 Ignore-this: 39cd352152422f378572fc3859c5a377
3032]
3033[Remove some more unused make variables
3034Ian Lynagh <igloo@earth.li>**20100616180519]
3035[Convert some more variable names to FOO_CMD, for consistency
3036Ian Lynagh <igloo@earth.li>**20100616175916]
3037[Rename some variables from FOO to FOO_CMD
3038Ian Lynagh <igloo@earth.li>**20100616161108
3039 This fixes a problem with commands like gzip, where if $GZIP is exported
3040 in the environment, then when make runs a command it'll put the Makefile
3041 variable's value in the environment. But gzip treats $GZIP as arguments
3042 for itself, so when we run gzip it thinks we're giving it "gzip" as an
3043 argument.
3044]
3045[Make the "show" target work anywhere in the build tree
3046Ian Lynagh <igloo@earth.li>**20100616122910
3047 Ignore-this: 299d40cbe16112accd9f14e56fa12158
3048]
3049[Change ghc-pwd's license to a string Cabal recognises
3050Ian Lynagh <igloo@earth.li>**20100615204015
3051 Ignore-this: c935b6ad7f605aab0168997a90b40fc6
3052]
3053[fix warning
3054Simon Marlow <marlowsd@gmail.com>**20100604205933
3055 Ignore-this: 2aaa4ed6a8b9ae1e39adc4696aaf14a3
3056]
3057[--install-signal-handles=no does not affect the timer signal (#1908)
3058Simon Marlow <marlowsd@gmail.com>**20100527214627
3059 Ignore-this: b0c51f1abdb159dc360662485095a11a
3060]
3061[Small optimisation: allocate nursery blocks contiguously
3062Simon Marlow <marlowsd@gmail.com>**20100509194928
3063 Ignore-this: e650e99e9ea9493d2efb245d565beef4
3064 This lets automatic prefetching work better, for a tiny performance boost
3065]
3066[fix -fforce-recomp setting: module is PrimOp, not PrimOps
3067Simon Marlow <marlowsd@gmail.com>**20100507084507
3068 Ignore-this: f76e0d9b643682ec0e8fb7d91afdea68
3069]
3070[it should be an error to use relative directories (#4134)
3071Simon Marlow <marlowsd@gmail.com>**20100615151740
3072 Ignore-this: 2068021701832e018ca41b22877921d5
3073]
3074[missing include-dirs or library-dirs is only a warning now (#4104)
3075Simon Marlow <marlowsd@gmail.com>**20100615151702
3076 Ignore-this: e3114123cef147bbd28ccb64581a1afb
3077]
3078[fix #3822: desugaring case command in arrow notation
3079Ross Paterson <ross@soi.city.ac.uk>**20100615225110
3080 Ignore-this: 477d6c460b4174b94b4cd113fa5b9d19
3081 
3082 Get the set of free variables from the generated case expression:
3083 includes variables in the guards and decls that were missed before,
3084 and is also a bit simpler.
3085]
3086[Deprecate the -fvia-C flag; trac #3232
3087Ian Lynagh <igloo@earth.li>**20100615151836
3088 Ignore-this: c2452b2648bf7e44546465c1b964fce
3089]
3090[Avoid using the new ~~ perl operator in the mangler
3091Ian Lynagh <igloo@earth.li>**20100615151236
3092 Ignore-this: 709a7ba4e514b1596841b3ba7e5c6cc
3093]
3094[stmAddInvariantToCheck: add missing init of invariant->lock (#4057)
3095Simon Marlow <marlowsd@gmail.com>**20100615123643
3096 Ignore-this: 3b132547fa934cecf71a846db2a5f70e
3097]
3098[Add new LLVM code generator to GHC. (Version 2)
3099David Terei <davidterei@gmail.com>**20100615094714
3100 Ignore-this: 4dd2fe5854b64a3f0339d484fd5c238
3101 
3102 This was done as part of an honours thesis at UNSW, the paper describing the
3103 work and results can be found at:
3104 
3105 http://www.cse.unsw.edu.au/~pls/thesis/davidt-thesis.pdf
3106 
3107 A Homepage for the backend can be found at:
3108 
3109 http://hackage.haskell.org/trac/ghc/wiki/Commentary/Compiler/Backends/LLVM
3110 
3111 Quick summary of performance is that for the 'nofib' benchmark suite, runtimes
3112 are within 5% slower than the NCG and generally better than the C code
3113 generator.  For some code though, such as the DPH projects benchmark, the LLVM
3114 code generator outperforms the NCG and C code generator by about a 25%
3115 reduction in run times.
3116 
3117]
3118[Fix Trac #4127: build GlobalRdrEnv in GHCi correctly
3119simonpj@microsoft.com**20100615070626
3120 Ignore-this: d907e3bfa7882878cea0af172aaf6e84
3121 
3122 GHCi was building its GlobalRdrEnv wrongly, so that the
3123 gre_par field was bogus.  That in turn fooled the renamer.
3124 The fix is easy: use the right function!  Namely, call
3125 RnNames.gresFromAvail rather than availsToNameSet.
3126]
3127[Comments, and improvement to pretty-printing of HsGroup
3128simonpj@microsoft.com**20100615070409
3129 Ignore-this: ec8358f2485370b20226a97ec84e9024
3130]
3131[Don't reverse bindings in rnMethodBinds (fix Trac #4126)
3132simonpj@microsoft.com**20100614163935
3133 Ignore-this: a6ffbb5af6f51b142ed0aeae8ee5e3a9
3134]
3135[Fix Trac #4120: generate a proper coercion when unifying forall types
3136simonpj@microsoft.com**20100614134311
3137 Ignore-this: 601592bb505305f1954cbe730f168da4
3138 
3139 This was just a blatant omission, which hasn't come up before.
3140 Easily fixed, happily.
3141]
3142[Use mkFunTy to ensure that invariants are respected
3143simonpj@microsoft.com**20100614134159
3144 Ignore-this: 67dcada7a4e8d9927581cd77af71b6f
3145]
3146[Remove redundant debug code
3147simonpj@microsoft.com**20100601154151
3148 Ignore-this: e6ff11c04c631cf6aac73788cbcf02b5
3149]
3150[Fix Trac #4099: better error message for type functions
3151simonpj@microsoft.com**20100531140413
3152 Ignore-this: 3f53ca98cf770577818b9c0937482577
3153 
3154 Now we only want about "T is a type function and might not be
3155 injective" when matchin (T x) against (T y), which is the case
3156 that is really confusing.
3157]
3158[Gruesome fix in CorePrep to fix embarassing Trac #4121
3159simonpj@microsoft.com**20100614132726
3160 Ignore-this: fe82d15474afaac3e6133adfd7a7e055
3161 
3162 This is a long-lurking bug that has been flushed into
3163 the open by other arity-related changes.  There's a
3164 long comment
3165 
3166      Note [CafInfo and floating]
3167 
3168 to explain. 
3169 
3170 I really hate the contortions we have to do through to keep correct
3171 CafRef information on top-level binders.  The Right Thing, I believe,
3172 is to compute CAF and arity information later, and merge it into the
3173 interface-file information when the latter is generated.
3174 
3175 But for now, this hackily fixes the problem.
3176]
3177[Fix a bug in CorePrep that meant output invariants not satisfied
3178simonpj@microsoft.com**20100531150013
3179 Ignore-this: d34eb36d8877d3caf1cf2b20de426abd
3180 
3181 In cpePair I did things in the wrong order so that something that
3182 should have been a CprRhs wasn't.  Result: a crash in CoreToStg.
3183 Fix is easy, and I added more informative type signatures too.
3184]
3185[Robustify the treatement of DFunUnfolding
3186simonpj@microsoft.com**20100531145332
3187 Ignore-this: 8f5506ada4d89f6ab8ad1e8c3ffb09ba
3188 
3189 See Note [DFun unfoldings] in CoreSyn.  The issue here is that
3190 you can't tell how many dictionary arguments a DFun needs just
3191 from looking at the Arity of the DFun Id: if the dictionary is
3192 represented by a newtype the arity might include the dictionary
3193 and value arguments of the (single) method.
3194 
3195 So we need to record the number of arguments need by the DFun
3196 in the DFunUnfolding itself.  Details in
3197    Note [DFun unfoldings] in CoreSyn
3198]
3199[Fix spelling in comment
3200simonpj@microsoft.com**20100614132259
3201 Ignore-this: bbf0d55f2e5f10ef9c74592c12f9201c
3202]
3203[Update docs on view patterns
3204simonpj@microsoft.com**20100614074801
3205 Ignore-this: 8617b9078800d4942d71f142a5b6c831
3206]
3207[Fix printing of splices; part of #4124
3208Ian Lynagh <igloo@earth.li>**20100613154838
3209 Just putting parens around non-atomic expressions isn't sufficient
3210 for splices, as only the $x and $(e) forms are valid input.
3211]
3212[In ghci, catch IO exceptions when calling canonicalizePath
3213Ian Lynagh <igloo@earth.li>**20100613134627
3214 We now get an exception if the path doesn't exist
3215]
3216[Whitespace only
3217Ian Lynagh <igloo@earth.li>**20100612213119]
3218[Whitespace only
3219Ian Lynagh <igloo@earth.li>**20100612165450]
3220[Update ghci example output in user guide; patch from YitzGale in #4111
3221Ian Lynagh <igloo@earth.li>**20100612162250]
3222[Fix #4131 missing UNTAG_CLOSURE in messageBlackHole()
3223benl@ouroborus.net**20100611044614]
3224[messageBlackHole: fix deadlock bug caused by a missing 'volatile'
3225Simon Marlow <marlowsd@gmail.com>**20100610080636
3226 Ignore-this: 3cda3054bb45408aa9bd2d794b69c938
3227]
3228[Pass --no-tmp-comp-dir to Haddock (see comment)
3229Simon Marlow <marlowsd@gmail.com>**20100604083214
3230 Ignore-this: bfa4d74038637bd149f4d878b4eb8a87
3231]
3232[Track changes to DPH libs
3233Roman Leshchinskiy <rl@cse.unsw.edu.au>**20100607052903
3234 Ignore-this: 4dbc3f8418af3e74b3fc4f9a9dfe7764
3235]
3236[Track changes to DPH libs
3237Roman Leshchinskiy <rl@cse.unsw.edu.au>**20100607012642
3238 Ignore-this: 5d4e498171a3c57ab02621bfaea82cff
3239]
3240[In ghc-pkg, send warnings to stderr
3241Ian Lynagh <igloo@earth.li>**20100606161726
3242 Ignore-this: 56927d13b5e1c1ce2752734f0f9b665b
3243]
3244[Re-add newlines to enable layout for multi-line input.
3245Ian Lynagh <igloo@earth.li>**20100602180737
3246 Patch from Adam Vogt <vogt.adam@gmail.com>
3247 Partial fix for #3984
3248]
3249[Don't use unnecessary parens when printing types (Fix Trac 4107)
3250simonpj@microsoft.com**20100604110143
3251 Ignore-this: a833714ab13013c4345b222f4e87db1d
3252 
3253    f :: Eq a => a -> a
3254 rather than
3255    f :: (Eq a) => a -> a
3256]
3257[Track DPH library changes
3258Roman Leshchinskiy <rl@cse.unsw.edu.au>**20100604005728
3259 Ignore-this: 32bc2fbea6ad975e89545d4c42fd7c30
3260]
3261[fix --source-entity option passed to Haddock: we needed to escape a #
3262Simon Marlow <marlowsd@gmail.com>**20100603125459
3263 Ignore-this: d52ae6188b510c482bcebb23f0e553ae
3264]
3265[__stg_EAGER_BLACKHOLE_INFO -> __stg_EAGER_BLACKHOLE_info (#4106)
3266Simon Marlow <marlowsd@gmail.com>**20100602091419
3267 Ignore-this: 293315ac8f86fd366b8d61992ecc7961
3268]
3269[Add xhtml package (a new dependency of Haddock; not installed/shipped)
3270Simon Marlow <marlowsd@gmail.com>**20100602090101
3271 Ignore-this: af0ac8b91abe98f7fdb624ea0a4dee20
3272]
3273[Use UserInterrupt rather than our own Interrupted exception (#4100)
3274Simon Marlow <marlowsd@gmail.com>**20100602082345
3275 Ignore-this: 1909acf2f452593138b9f85024711714
3276]
3277[Add the global package DB to ghc --info (#4103)
3278Simon Marlow <marlowsd@gmail.com>**20100602082233
3279 Ignore-this: fd5c0e207e70eb0f62606c45dc5b8124
3280]
3281[rts/sm/GC.c: resize_generations(): Remove unneeded check of number of generations.
3282Marco Túlio Gontijo e Silva <marcot@debian.org>**20100528115612
3283 Ignore-this: 6f1bea62917c01c7adac636146132c97
3284 
3285 This "if" is inside another "if" which checks for RtsFlags.GcFlags.generations
3286 > 1, so testing this again is redundant, assuming the number of generations
3287 won't change during program execution.
3288]
3289[rts/sm/BlockAlloc.c: Small comment correction.
3290Marco Túlio Gontijo e Silva <marcot@debian.org>**20100526205839
3291 Ignore-this: bd2fcd4597cc872d80b0e2eeb1c3998a
3292]
3293[rts/sm/GC.c: Annotate constants.
3294Marco Túlio Gontijo e Silva <marcot@debian.org>**20100526205707
3295 Ignore-this: f232edb89383564d759ed890a18f602f
3296]
3297[includes/rts/storage/GC.h: generation_: n_words: Improve comment.
3298Marco Túlio Gontijo e Silva <marcot@debian.org>**20100526204615
3299 Ignore-this: f5d5feefa8f7b552303978f1804fea23
3300]
3301[Add PPC_RELOC_LOCAL_SECTDIFF support; patch from PHO in #3654
3302Ian Lynagh <igloo@earth.li>**20100601204211
3303 Ignore-this: 51293b7041cdce3ce7619ef11cf7ceb
3304]
3305[powerpc-apple-darwin now supports shared libs
3306Ian Lynagh <igloo@earth.li>**20100601173325]
3307[PIC support for PowerPC
3308pho@cielonegro.org**20100508143900
3309 Ignore-this: 3673859a305398c4acae3f4d7c997615
3310 
3311 PPC.CodeGen.getRegister was not properly handling PicBaseReg.
3312 It seems working with this patch, but I'm not sure this change is correct.
3313]
3314[Vectoriser: only treat a function as scalar if it actually computes something
3315Roman Leshchinskiy <rl@cse.unsw.edu.au>**20100601045630
3316 Ignore-this: e5d99a6ddb62052e3520094a5af47552
3317]
3318[Add a release notes file for 6.14.1
3319Ian Lynagh <igloo@earth.li>**20100530171117
3320 Ignore-this: 1941e6d3d1f4051b69ca2f17a1cf84d6
3321]
3322[Check dblatex actually creates the files we tell it to
3323Ian Lynagh <igloo@earth.li>**20100530171043
3324 Ignore-this: ccc72caea2313be05cbac59bb54c0603
3325 If it fails, it still exits successfully.
3326]
3327[Add darwin to the list of OSes for which we use mmap
3328Ian Lynagh <igloo@earth.li>**20100529145016
3329 Ignore-this: a86d12a3334aaaafc86f7af9dbb0a7ae
3330 Patch from Barney Stratford
3331]
3332[Simplify the CPP logic in rts/Linker.c
3333Ian Lynagh <igloo@earth.li>**20100529144929
3334 Ignore-this: 1288f5b752cc1ab8b1c90cfd0ecfdf68
3335]
3336[Fix validate on OS X
3337Ian Lynagh <igloo@earth.li>**20100529154726]
3338[OS X x86_64 fix from Barney Stratford
3339Ian Lynagh <igloo@earth.li>**20100529122440]
3340[OS X 64 installer fixes from Barney Stratford
3341Ian Lynagh <igloo@earth.li>**20100528234935]
3342[fix warning
3343Simon Marlow <marlowsd@gmail.com>**20100525155812
3344 Ignore-this: f34eee3fe3d89579fd8d381c91ced750
3345]
3346[Fix doc bugs (#4071)
3347Simon Marlow <marlowsd@gmail.com>**20100525155728
3348 Ignore-this: aa25be196de567de360075022a1942f7
3349]
3350[Make sparks into weak pointers (#2185)
3351Simon Marlow <marlowsd@gmail.com>**20100525150435
3352 Ignore-this: feea0bb5006007b82c932bc3006124d7
3353 The new strategies library (parallel-2.0+, preferably 2.2+) is now
3354 required for parallel programming, otherwise parallelism will be lost.
3355]
3356[If you say 'make' or 'make stage=2' here, pretend we're in the ghc dir
3357Simon Marlow <marlowsd@gmail.com>**20100525085301
3358 Ignore-this: 78b740337aa460915c812cbbcdae5321
3359]
3360[Another attempt to get these #defines right
3361Simon Marlow <marlowsd@gmail.com>**20100525154313
3362 Ignore-this: 460ca0c47d81cd25eae6542114f67899
3363 Apparently on Solaris it is an error to omit _ISOC99_SOURCE when using
3364 _POSIX_C_SOURCE==200112L.
3365]
3366[Add configure flags for the location of GMP includes/library; fixes #4022
3367Ian Lynagh <igloo@earth.li>**20100525221616
3368 Ignore-this: fc3060caf995d07274ec975eeefbdf3e
3369]
3370[Refactor pretty printing of TyThings to fix Trac #4015
3371simonpj@microsoft.com**20100525153126
3372 Ignore-this: 8f15053b7554f62caa84201d2e4976d2
3373]
3374[When haddocking, we need the dependencies to have been built
3375Ian Lynagh <igloo@earth.li>**20100525145830
3376 as haddock loads the .hi files with the GHC API.
3377]
3378[Fix profiling output; spotted by jlouis
3379Ian Lynagh <igloo@earth.li>**20100525111217
3380 We were outputing the number of words allocated in a column titled "bytes".
3381]
3382[Improve printing of TyThings; fixes Trac #4087
3383simonpj@microsoft.com**20100525114045
3384 Ignore-this: da2a757a533454bba80b9b77cc5a771
3385]
3386[Spelling in comments
3387simonpj@microsoft.com**20100525114001
3388 Ignore-this: 270f3da655e526cf04e27db7a01e29c0
3389]
3390[Refactor (again) the handling of default methods
3391simonpj@microsoft.com**20100525113910
3392 Ignore-this: 6686f6cdb878d57abf6b49fec64fcbb1
3393 
3394 This patch fixes Trac #4056, by
3395 
3396  a) tidying up the treatment of default method names
3397  b) removing the 'module' argument to newTopSrcBinder
3398 
3399 The details aren't that interesting, but the result
3400 is much tidier. The original bug was a 'nameModule' panic,
3401 caused by trying to find the module of a top-level name.
3402 But TH quotes generate Internal top-level names that don't
3403 have a module, and that is generally a good thing. 
3404 
3405 Fixing that in turn led to the default-method refactoring,
3406 which also makes the Name for a default method be handled
3407 in the same way as other derived names, generated in BuildTyCl
3408 via a call newImplicitBinder.  Hurrah.
3409]
3410[Don't do SpecConstr on NOINLINE things (Trac #4064)
3411simonpj@microsoft.com**20100525112807
3412 Ignore-this: 452be0a2cef0042fb67275c2827b5f72
3413 
3414 Since the RULE from specialising gets the same Activation as
3415 the inlining for the Id itself there's no point in specialising
3416 a NOINLINE thing, because the rule will be permanently switched
3417 off.
3418 
3419 See Note [Transfer activation] in SpecConstr
3420 and Note [Auto-specialisation and RULES] in Specialise.
3421]
3422[Change our #defines to work on FreeBSD too
3423Simon Marlow <marlowsd@gmail.com>**20100524105828
3424 Ignore-this: b23ede46211e67859206c0ec57d6a86f
3425 With glibc, things like _POSIX_C_SOURCE and _ISOC99_SOURCE are
3426 additive, but on FreeBSD they are mutually exclusive.  However, it
3427 turns out we only need to define _POSIX_C_SOURCE and _XOPEN_SOURCE to
3428 get all the C99 stuff we need too, so there's no need for any #ifdefs.
3429 
3430 Submitted by: Gabor PALI <pgj@FreeBSD.org>
3431]
3432[Add a missing UNTAG_CLOSURE, causing bus errors on Sparc
3433Simon Marlow <marlowsd@gmail.com>**20100524105547
3434 Ignore-this: a590b5391d6f05d50c8c088456c3c166
3435 We just about got away with this on x86 which isn't
3436 alignment-sensitive.  The result of the memory load is compared
3437 against a few different values, but there is a fallback case that
3438 happened to be the right thing when the pointer was tagged.  A good
3439 bug to find, nonetheless.
3440]
3441[Add wiki links
3442Simon Marlow <marlowsd@gmail.com>**20100520095953
3443 Ignore-this: c22f126cde166e6207922b2eb51d29e3
3444]
3445[the 'stage=0' trick to disable all compiler builds stopped working; fix it
3446Simon Marlow <marlowsd@gmail.com>**20100520104455
3447 Ignore-this: bb6fae9056471612c8dbf06916188c33
3448]
3449[Comments and formatting only
3450benl@ouroborus.net**20100524014021
3451 Ignore-this: 64579c38154728b632e358bec751cc0b
3452]
3453[Core prettyprinter fixes. Patch from Tim Chevalier. Fixes #4085
3454Ian Lynagh <igloo@earth.li>**20100522225048]
3455[Correct install-name for dynamic Darwin rts
3456pho@cielonegro.org**20100508151155
3457 Ignore-this: 6d31716c8c113dcb46e9cb925c4201df
3458]
3459[Fix the RTS debug_p build
3460Ian Lynagh <igloo@earth.li>**20100522163127]
3461[Unset $CFLAGS for "GNU non-executable stack" configure test; fixes #3889
3462Ian Lynagh <igloo@earth.li>**20100521165005
3463 With gcc 4.4 we get
3464     Error: can't resolve `.note.GNU-stack' {.note.GNU-stack section} - `.Ltext0' {.text section}
3465 when running gcc with the -g flag. To work around this we unset
3466 CFLAGS when running the test.
3467]
3468[Don't run "set -o igncr" before configuring libffi
3469Ian Lynagh <igloo@earth.li>**20100520162918
3470 Ignore-this: 489fa94df23f2adf4ff63c8ede2c0794
3471 It used to make the build work on cygwin, but now it breaks it instead:
3472     config.status: creating include/Makefile
3473     gawk: ./confLqjohp/subs.awk:1: BEGIN {\r
3474     gawk: ./confLqjohp/subs.awk:1: ^ backslash not last character on line
3475     config.status: error: could not create include/Makefile
3476     make[2]: *** [libffi/stamp.ffi.configure-shared] Error 1
3477     make[1]: *** [all] Error 2
3478]
3479[Stop passing -Wl,-macosx_version_min to gcc
3480Ian Lynagh <igloo@earth.li>**20100520154003
3481 Fixes a build failure on OS X 10.6. When linking
3482     rts/dist/build/libHSrts-ghc6.13.20100519.dylib
3483 we got
3484     ld: symbol dyld_stub_binding_helper not defined (usually in crt1.o/dylib1.o/bundle1.o)
3485     collect2: ld returned 1 exit status
3486]
3487[Fix build on FreeBSD; patch from Gabor PALI
3488Ian Lynagh <igloo@earth.li>**20100519140552]
3489[Fix package shadowing order (#4072)
3490Simon Marlow <marlowsd@gmail.com>**20100519104617
3491 Ignore-this: 26ea5e4bb5dff18618b807a54c7d6ebb
3492 
3493 Later packages are supposed to shadow earlier ones in the stack,
3494 unless the ordering is overriden with -package-id flags.
3495 Unfortunately an earlier fix for something else had sorted the list of
3496 packages so that it was in lexicographic order by installedPackageId,
3497 and sadly our test (cabal/shadow) didn't pick this up because the
3498 lexicographic ordering happened to work for the test.  I've now fixed
3499 the test so it tries both orderings.
3500]
3501[Set more env variables when configuring libffi
3502Ian Lynagh <igloo@earth.li>**20100518185014
3503 We now tell it where to find ld, nm and ar
3504]
3505[Set the location of ar to be the in-tree ar on Windows
3506Ian Lynagh <igloo@earth.li>**20100518181556]
3507[Change another / to </> to avoid building paths containing \/
3508Ian Lynagh <igloo@earth.li>**20100518172015
3509 This will hopefully fix #2889.
3510]
3511[Fix #4074 (I hope).
3512Simon Marlow <marlowsd@gmail.com>**20100518113214
3513 Ignore-this: 73cd70f5bc6f5add5247b61985c03fc1
3514 
3515 1. allow multiple threads to call startTimer()/stopTimer() pairs
3516 2. disable the timer around fork() in forkProcess()
3517 
3518 A corresponding change to the process package is required.
3519]
3520[we don't have a gcc-lib in LIB_DIR any more
3521Simon Marlow <marlowsd@gmail.com>**20100401102351
3522 Ignore-this: f41acd2d8f8e6763aa8bd57a0b44a7e4
3523]
3524[In validate, use gmake if available; based on a patch from Gabor PALI
3525Ian Lynagh <igloo@earth.li>**20100517200654]
3526[Remove duplicate "./configure --help" output; fixes #4075
3527Ian Lynagh <igloo@earth.li>**20100516141206]
3528[Update various 'sh boot's to 'perl boot'
3529Ian Lynagh <igloo@earth.li>**20100516122609
3530 Spotted by Marco Túlio Gontijo e Silva
3531]
3532[add missing initialisation for eventBufMutex
3533Simon Marlow <marlowsd@gmail.com>**20100514094943
3534 Ignore-this: 7f75594a8cb54fbec5aebd46bb959f45
3535]
3536[Undo part of #4003 patch
3537Simon Marlow <marlowsd@gmail.com>**20100513142017
3538 Ignore-this: cb65db86a38a7e5ccee9f779e489d104
3539 We still need the workaround for when compiling HEAD with 6.12.2
3540 
3541]
3542[Fix makefile loop (#4050)
3543pho@cielonegro.org**20100507140707
3544 Ignore-this: 3a1cb13d0600977e74d17ac26cbef83d
3545 
3546 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.
3547]
3548[fix !TABLES_NEXT_TO_CODE
3549Simon Marlow <marlowsd@gmail.com>**20100510151934
3550 Ignore-this: fccb859b114bef1c3122c98e60af51
3551]
3552[looksLikeModuleName: allow apostrophe in module names (#4051)
3553Simon Marlow <marlowsd@gmail.com>**20100510094741
3554 Ignore-this: df9348f3ba90608bec57257b47672985
3555]
3556[add the proper library dependencies for GhcProfiled=YES
3557Simon Marlow <marlowsd@gmail.com>**20100506122118
3558 Ignore-this: 6236993aa308ab5b5e1e5ea5f65982a
3559]
3560[Fix Trac #4003: fix the knot-tying in checkHiBootIface
3561simonpj@microsoft.com**20100511075026
3562 Ignore-this: a9ce2a318386fdc8782848df84592002
3563 
3564 I had incorrectly "optimised" checkHiBootIface so that it forgot
3565 to update the "knot-tied" type environment.
3566 
3567 This patch fixes the HEAD
3568]
3569[Re-engineer the derived Ord instance generation code (fix Trac #4019)
3570simonpj@microsoft.com**20100510133333
3571 Ignore-this: 8fe46e4dad27fbee211a7928acf372c2
3572   
3573 As well as fixing #4019, I rejigged the way that Ord instances are
3574 generated, which should make them faster in general.  See the
3575 Note [Generating Ord instances].
3576 
3577 I tried to measure the performance difference from this change, but
3578 the #4019 fix only removes one conditional branch per iteration, and
3579 I couldn't measure a consistent improvement.  But still, tihs is
3580 better than before.
3581]
3582[Make arity of INLINE things consistent
3583simonpj@microsoft.com**20100510133005
3584 Ignore-this: 15e7abf803d1dcb3f4ca760d2d939d0d
3585 
3586 We eta-expand things with INLINE pragmas;
3587 see Note [Eta-expanding INLINE things].
3588 
3589 But I eta-expanded it the wrong amount when the function
3590 was overloaded.  Ooops.
3591]
3592[Compacting GC fix, we forgot to thread the new bq field of StgTSO.
3593Simon Marlow <marlowsd@gmail.com>**20100510082325
3594 Ignore-this: a079c8446e2ad53efff6fd95d0f3ac80
3595]
3596[Add version constraints for the boot packages; fixes trac #3852
3597Ian Lynagh <igloo@earth.li>**20100509175051
3598 When using the bootstrapping compiler, we now explicitly constrain
3599 the version of boot packages (Cabal, extensible-exceptions, etc) to the
3600 in-tree version, so that the build system is less fragile should the
3601 user have a newer version installed for the bootstrapping compiler.
3602]
3603[Don't include inter-package dependencies when compiling with stage 0; #4031
3604Ian Lynagh <igloo@earth.li>**20100509130511
3605 This fixes a problem when building with GHC 6.12 on Windows, where
3606 dependencies on stage 0 (bootstrapping compiler) packages have absolute
3607 paths c:/ghc/..., and make gets confused by the colon.
3608]
3609[Add a ghc.mk for bindisttest/
3610Ian Lynagh <igloo@earth.li>**20100508223911]
3611[Move some make variables around so they are available when cleaning
3612Ian Lynagh <igloo@earth.li>**20100508212405]
3613[Optimise checkremove a bit
3614Ian Lynagh <igloo@earth.li>**20100508202006]
3615[Improve the bindisttest Makefile
3616Ian Lynagh <igloo@earth.li>**20100508195450]
3617[Add tools to test that cleaning works properly
3618Ian Lynagh <igloo@earth.li>**20100508194105]
3619[Tweak the ghc-pkg finding code
3620Ian Lynagh <igloo@earth.li>**20100508125815
3621 It now understand the ghc-stage[123] names we use in-tree, and it won't
3622 go looking for any old ghc-pkg if it can't find the one that matches
3623 ghc.
3624]
3625[Add a way to show what cleaning would be done, without actually doing it
3626Ian Lynagh <igloo@earth.li>**20100508122438]
3627[Tidy up the "rm" flags in the build system
3628Ian Lynagh <igloo@earth.li>**20100508115745]
3629[Fix crash in nested callbacks (#4038)
3630Simon Marlow <marlowsd@gmail.com>**20100507093222
3631 Ignore-this: cade85e361534ce711865a4820276388
3632 Broken by "Split part of the Task struct into a separate struct
3633 InCall".
3634]
3635[Add $(GhcDynamic) knob, set to YES to get stage2 linked with -dynamic
3636Simon Marlow <marlowsd@gmail.com>**20100428205241
3637 Ignore-this: 1db8bccf92099785ecac39aebd27c92d
3638 Default currently NO.
3639 
3640 Validate passed with GhcDynamic=YES on x86/Linux here.
3641 
3642 The compiler is currently slower on x86 when linked -dynamic,
3643 because the GC inner loop has been adversely affected by -fPIC, I'm
3644 looking into how to fix it.
3645]
3646[omit "dyn" from the way appended to the __stginit label
3647Simon Marlow <marlowsd@gmail.com>**20100428204914
3648 Ignore-this: 14183f3defa9f2bde68fda6729b740bc
3649 When GHCi is linked dynamically, we still want to be able to load
3650 non-dynamic object files.
3651]
3652[improvements to findPtr(), a neat hack for browsing the heap in gdb
3653Simon Marlow <marlowsd@gmail.com>**20100506115427
3654 Ignore-this: ac57785bb3e13b97a5945f753f068738
3655]
3656[Fix +RTS -G1
3657Simon Marlow <marlowsd@gmail.com>**20100506110739
3658 Ignore-this: 86a5de39a94d3331a4ee1213f82be497
3659]
3660[Enable the "redundant specialise pragmas" warning; fixes trac #3855
3661Ian Lynagh <igloo@earth.li>**20100506175351]
3662[Find the correct external ids when there's a wrapper
3663simonpj@microsoft.com**20100506164135
3664 Ignore-this: 636266407b174b05b2b8646cc73062c0
3665 
3666 We were failing to externalise the wrapper id for a function
3667 that had one.
3668]
3669[Add a comment about pattern coercions
3670simonpj@microsoft.com**20100506164027
3671 Ignore-this: 17428089f3df439f65d892e23e8ed61a
3672]
3673[Comments only
3674simonpj@microsoft.com**20100506163829
3675 Ignore-this: 169167b6463873ab173cc5750c5be469
3676]
3677[Make a missing name in mkUsageInfo into a panic
3678simonpj@microsoft.com**20100506163813
3679 Ignore-this: b82ff1b8bf89f74f146db7cb5cc4c4d7
3680 
3681 We really want to know about this!
3682]
3683[Refactoring of hsXxxBinders
3684simonpj@microsoft.com**20100506163737
3685 Ignore-this: 97c6667625262b160f9746f7bea1c980
3686 
3687 This patch moves various functions that extract the binders
3688 from a HsTyClDecl, HsForeignDecl etc into HsUtils, and gives
3689 them consistent names.
3690]
3691[Fix Trac #3966: warn about useless UNPACK pragmas
3692simonpj@microsoft.com**20100506163337
3693 Ignore-this: 5beb24b686eda6113b614dfac8490df1
3694 
3695 Warning about useless UNPACK pragmas wasn't as easy as I thought.
3696 I did quite a bit of refactoring, which improved the code by refining
3697 the types somewhat.  In particular notice that in DataCon, we have
3698 
3699     dcStrictMarks   :: [HsBang]
3700     dcRepStrictness :: [StrictnessMarks]
3701 
3702 The former relates to the *source-code* annotation, the latter to
3703 GHC's representation choice.
3704]
3705[Make tcg_dus behave more sanely; fixes a mkUsageInfo panic
3706simonpj@microsoft.com**20100506162719
3707 Ignore-this: d000bca15b0e127e297378ded1bfb81b
3708 
3709 The tcg_dus field used to contain *uses* of type and class decls,
3710 but not *defs*.  That was inconsistent, and it really went wrong
3711 for Template Haskell bracket.  What happened was that
3712  foo = [d| data A = A
3713                   f :: A -> A
3714                   f x = x |]
3715 would find a "use" of A when processing the top level of the module,
3716 which in turn led to a mkUsageInfo panic in MkIface.  The cause was
3717 the fact that the tcg_dus for the nested quote didn't have defs for
3718 A.
3719]
3720[Add a HsExplicitFlag to SpliceDecl, to improve Trac #4042
3721simonpj@microsoft.com**20100506161523
3722 Ignore-this: e4e563bac2fd831cc9e94612f5b4fa9d
3723 
3724 The issue here is that
3725 
3726     g :: A -> A
3727     f
3728     data A = A
3729 
3730 is treated as if you'd written $(f); that is the call of
3731 f is a top-level Template Haskell splice.  This patch
3732 makes sure that we *first* check the -XTemplateHaskellFlag
3733 and bleat about a parse error if it's off.  Othewise we
3734 get strange seeing "A is out of scope" errors.
3735]
3736[Change an assert to a warn
3737simonpj@microsoft.com**20100506161111
3738 Ignore-this: 739a4fb4c7940376b0f2c8ad52a1966c
3739 
3740 This is in the constraint simplifier which I'm about
3741 to rewrite, so I'm hoping the assert isn't fatal!
3742]
3743[Tidy up debug print a little
3744simonpj@microsoft.com**20100506161027
3745 Ignore-this: bd5492878e06bee1cddcbb3fc4df66d8
3746]
3747[Remove useless UNPACK pragmas
3748simonpj@microsoft.com**20100506161012
3749 Ignore-this: 3e5ab1a7cf58107034412a798bc214e5
3750]
3751[Add WARNM2 macro, plus some refactoring
3752simonpj@microsoft.com**20100506160808
3753 Ignore-this: 2ab4f1f0b5d94be683036e77aec09255
3754]
3755[Use -Wwarn for the binary package, becuase it has redundant UNPACK pragmas
3756simonpj@microsoft.com**20100506160750
3757 Ignore-this: cf0d3a11473e28bfce9602e716e69a5f
3758]
3759[Fix Trac #3966: warn about unused UNPACK pragmas
3760simonpj@microsoft.com**20100409201812
3761 Ignore-this: c96412596b39c918b5fb9b3c39ce2119
3762]
3763[Fix Trac #3953: fail earlier when using a bogus quasiquoter
3764simonpj@microsoft.com**20100409201748
3765 Ignore-this: ef48e39aa932caed538643985234f043
3766]
3767[Fix Trac #3965: tighten conditions when deriving Data
3768simonpj@microsoft.com**20100409184420
3769 Ignore-this: 96f7d7d2da11565d26b465d7d0497ac9
3770 
3771 It's tricky to set up the context for a Data instance.  I got it wrong
3772 once, and fixed it -- hence the "extra_constraints" in
3773 TcDeriv.inferConstraints. 
3774 
3775 But it still wasn't right!  The tricky bit is that dataCast1 is only
3776 generated when T :: *->*, and dataCast2 when T :: *->*->*. (See
3777 the code in TcGenDeriv for dataCastX.
3778]
3779[Fix Trac #3964: view patterns in DsArrows
3780simonpj@microsoft.com**20100409165557
3781 Ignore-this: d823c182831d5e2e592e995b16180e2f
3782 
3783 Just a missing case; I've eliminated the catch-all so
3784 that we get a warning next time we extend HsPat
3785]
3786[Fix Trac #3955: renamer and type variables
3787simonpj@microsoft.com**20100409163710
3788 Ignore-this: bd5ec64d76c0f583bf5f224792bf294c
3789 
3790 The renamer wasn't computing the free variables of a type declaration
3791 properly.  This patch refactors a bit, and makes it more robust,
3792 fixing #3955 and several other closely-related bugs.  (We were
3793 omitting some free variables and that could just possibly lead to a
3794 usage-version tracking error.
3795]
3796[Layout only
3797simonpj@microsoft.com**20100409163506
3798 Ignore-this: 1f14990b5aa0b9821b84452fb34e9f41
3799]
3800[Give a better deprecated message for INCLUDE pragmas; fixes #3933
3801Ian Lynagh <igloo@earth.li>**20100506130910
3802 We now have a DeprecatedFullText constructor, so we can override the
3803 "-#include is deprecated: " part of the warning.
3804]
3805[De-haddock a comment that confuses haddock
3806Ian Lynagh <igloo@earth.li>**20100506123607]
3807[Fix comment to not confuse haddock
3808Ian Lynagh <igloo@earth.li>**20100506113642]
3809[Detect EOF when trying to parse a string in hp2ps
3810Ian Lynagh <igloo@earth.li>**20100506000830]
3811[Make the demand analyser sdd demands for strict constructors
3812simonpj@microsoft.com**20100505200936
3813 Ignore-this: eb32632adbc354eb7a5cf884c263e0d3
3814 
3815 This opportunity was spotted by Roman, and is documented in
3816 Note [Add demands for strict constructors] in DmdAnal.
3817]
3818[Fix interaction of exprIsCheap and the lone-variable inlining check
3819simonpj@microsoft.com**20100505200723
3820 Ignore-this: f3cb65085c5673a99153d5d7b6559ab1
3821 
3822 See Note [Interaction of exprIsCheap and lone variables] in CoreUnfold
3823 
3824 This buglet meant that a nullary definition with an INLINE pragma
3825 counter-intuitively didn't get inlined at all.  Roman identified
3826 the bug.
3827]
3828[Matching cases in SpecConstr and Rules
3829simonpj@microsoft.com**20100505200543
3830 Ignore-this: f5c28c780fbf8badce84c6fdc9aa1779
3831 
3832 This patch has zero effect.  It includes comments,
3833 a bit of refactoring, and a tiny bit of commment-out
3834 code go implement the "matching cases" idea below.
3835 
3836 In the end I've left it disabled because while I think
3837 it does no harm I don't think it'll do any good either.
3838 But I didn't want to lose the idea totally. There's
3839 a thread called "Storable and constant memory" on
3840 the libraries@haskell.org list (Apr 2010) about it.
3841 
3842 Note [Matching cases]
3843 ~~~~~~~~~~~~~~~~~~~~~
3844 {- NOTE: This idea is currently disabled.  It really only works if
3845          the primops involved are OkForSpeculation, and, since
3846         they have side effects readIntOfAddr and touch are not.
3847         Maybe we'll get back to this later .  -}
3848   
3849 Consider
3850    f (case readIntOffAddr# p# i# realWorld# of { (# s#, n# #) ->
3851       case touch# fp s# of { _ ->
3852       I# n# } } )
3853 This happened in a tight loop generated by stream fusion that
3854 Roman encountered.  We'd like to treat this just like the let
3855 case, because the primops concerned are ok-for-speculation.
3856 That is, we'd like to behave as if it had been
3857    case readIntOffAddr# p# i# realWorld# of { (# s#, n# #) ->
3858    case touch# fp s# of { _ ->
3859    f (I# n# } } )
3860]
3861[Comments only
3862simonpj@microsoft.com**20100504163629
3863 Ignore-this: 3be12df04714aa820bce706b5dc8a9cb
3864]
3865[Comments only
3866simonpj@microsoft.com**20100504163529
3867 Ignore-this: 791e2fd39c7d880ce1dc80ebdf3a5398
3868]
3869[Comments only
3870simonpj@microsoft.com**20100504163457
3871 Ignore-this: f19e9ffeb3d65770b1595bca5f97a59d
3872]
3873[Comments only (about type families)
3874simonpj@microsoft.com**20100417145032
3875 Ignore-this: dd39425ef2155d52dbf55a4d5fd97cb8
3876]
3877[Fix hp2ps when the .hp file has large string literals
3878Ian Lynagh <igloo@earth.li>**20100505191921]
3879[In build system, call package-config after including package data
3880Ian Lynagh <igloo@earth.li>**20100504225035
3881 Otherwise the $1_$2_HC_OPTS variable gets clobbered.
3882]
3883[runghc: flush stdout/stderr on an exception (#3890)
3884Simon Marlow <marlowsd@gmail.com>**20100505133848
3885 Ignore-this: 224c1898cec64cb1c94e0d7033e7590e
3886]
3887[Remove the Unicode alternative for ".." (#3894)
3888Simon Marlow <marlowsd@gmail.com>**20100505121202
3889 Ignore-this: 2452cd67281667106f9169747b6d784f
3890]
3891[tidyup; no functional changes
3892Simon Marlow <marlowsd@gmail.com>**20100505115015
3893 Ignore-this: d0787e5cdeef1dee628682fa0a46019
3894]
3895[Make the running_finalizers flag task-local
3896Simon Marlow <marlowsd@gmail.com>**20100505114947
3897 Ignore-this: 345925d00f1dca203941b3c5d84c90e1
3898 Fixes a bug reported by Lennart Augustsson, whereby we could get an
3899 incorrect error from the RTS about re-entry from a finalizer,
3900]
3901[add a MAYBE_GC() in killThread#, fixes throwto003(threaded2) looping
3902Simon Marlow <marlowsd@gmail.com>**20100505114746
3903 Ignore-this: efea04991d6feed04683a42232fc85da
3904]
3905[Allow filepath-1.2.*
3906Simon Marlow <marlowsd@gmail.com>**20100505101139
3907 Ignore-this: 1b5580cd9cd041ec48f40cd37603326a
3908]
3909[BlockedOnMsgThrowTo is possible in resurrectThreads (#4030)
3910Simon Marlow <marlowsd@gmail.com>**20100505094534
3911 Ignore-this: ac24a22f95ffeaf480187a1620fdddb2
3912]
3913[Don't raise a throwTo when the target is masking and BlockedOnBlackHole
3914Simon Marlow <marlowsd@gmail.com>**20100505094506
3915 Ignore-this: 302616931f61667030d77ddfbb02374e
3916]
3917[Fix build with GHC 6.10
3918Ian Lynagh <igloo@earth.li>**20100504180302
3919 In GHC 6.10, intersectionWith is (a -> b -> a) instead of (a -> b -> c),
3920 so we need to jump through some hoops to get the more general type.
3921]
3922[The libffi patches are no longer needed
3923Ian Lynagh <igloo@earth.li>**20100504171603]
3924[Use the in-tree windres; fixes trac #4032
3925Ian Lynagh <igloo@earth.li>**20100504170941]
3926[Print unfoldings on lambda-bound variables
3927Simon PJ <simonpj@microsoft.com>**20100503181822
3928 Ignore-this: 2fd5a7502cc6273d96258e0914f0f8cd
3929 
3930 ...in the unusual case where they have one;
3931 see Note [Case binders and join points] in Simplify.lhs
3932]
3933[Replace FiniteMap and UniqFM with counterparts from containers.
3934Milan Straka <fox@ucw.cz>**20100503171315
3935 Ignore-this: a021972239163dbf728284b19928cebb
3936 
3937 The original interfaces are kept. There is small performance improvement:
3938 - when compiling for five nofib, we get following speedups:
3939     Average                -----           -2.5%
3940     Average                -----           -0.6%
3941     Average                -----           -0.5%
3942     Average                -----           -5.5%
3943     Average                -----          -10.3%
3944 - when compiling HPC ten times, we get:
3945     switches                          oldmaps   newmaps
3946     -O -fasm                          117.402s  116.081s (98.87%)
3947     -O -fasm -fregs-graph             119.993s  118.735s (98.95%)
3948     -O -fasm -fregs-iterative         120.191s  118.607s (98.68%)
3949]
3950[Make the demand analyser take account of lambda-bound unfoldings
3951Simon PJ <simonpj@microsoft.com>**20100503151630
3952 Ignore-this: 2ee8e27d4df2debfc79e6b8a17c32bc1
3953 
3954 This is a long-standing lurking bug. See Note [Lamba-bound unfoldings]
3955 in DmdAnal.
3956 
3957 I'm still not really happy with this lambda-bound-unfolding stuff.
3958]
3959[Fix dynamic libs on OS X, and enable them by default
3960Ian Lynagh <igloo@earth.li>**20100503150302]
3961[Switch back to using bytestring from the darcs repo; partially fixes #3855
3962Ian Lynagh <igloo@earth.li>**20100502113458]
3963[Fix some cpp warnings when building on FreeBSD; patch from Gabor PALI
3964Ian Lynagh <igloo@earth.li>**20100428150700]
3965[Fix "make 2"
3966Ian Lynagh <igloo@earth.li>**20100427162212
3967 The new Makefile logic was enabling the stage 1 rules when stage=2,
3968 so "make 2" was rebuilding stage 1.
3969]
3970[Inplace programs depend on their shell wrappers
3971Ian Lynagh <igloo@earth.li>**20100427160038]
3972[--make is now the default (#3515), and -fno-code works with --make (#3783)
3973Simon Marlow <marlowsd@gmail.com>**20100427122851
3974 Ignore-this: 33330474fa4703f32bf9997462b4bf3c
3975 If the command line contains any Haskell source files, then we behave
3976 as if --make had been given.
3977 
3978 The meaning of the -c flag has changed (back): -c now selects one-shot
3979 compilation, but stops before linking.  However, to retain backwards
3980 compatibility, -c is still allowed with --make, and means the same as
3981 --make -no-link.  The -no-link flag has been un-deprecated.
3982 
3983 -fno-code is now allowed with --make (#3783); the fact that it was
3984 disabled before was largely accidental, it seems.  We also had some
3985 regressions in this area: it seems that -fno-code was causing a .hc
3986 file to be emitted in certain cases.  I've tidied up the code, there
3987 was no need for -fno-code to be a "mode" flag, as far as I can tell.
3988 
3989 -fno-code does not emit interface files, nor does it do recompilation
3990 checking, as suggested in #3783.  This would make Haddock emit
3991 interface files, for example, and I'm fairly sure we don't want to do
3992 that.  Compiling with -fno-code is pretty quick anyway, perhaps we can
3993 get away without recompilation checking.
3994]
3995[remove duplicate docs for -e in --help output (#4010)
3996Simon Marlow <marlowsd@gmail.com>**20100426140642
3997 Ignore-this: 187ff893ba8ffa0ec127867a7590e38d
3998]
3999[workaround for #4003, fixes HEAD build with 6.12.2
4000Simon Marlow <marlowsd@gmail.com>**20100426103428
4001 Ignore-this: c4bc445dc8052d4e6efef3f1daf63562
4002]
4003[Make sure all the clean rules are always included
4004Ian Lynagh <igloo@earth.li>**20100424181823
4005 In particular, this fixes a problem where stage3 bits weren't being cleaned
4006]
4007[Correct the name of the amd64/FreeBSD platform in PlatformSupportsSharedLibs
4008Ian Lynagh <igloo@earth.li>**20100424132830
4009 We weren't getting sharedlibs on amd64/FreeBSD because of this
4010]
4011[Include DPH docs in bindists
4012Ian Lynagh <igloo@earth.li>**20100424123101]
4013[reinstate eta-expansion during SimplGently, to fix inlining of sequence_
4014Simon Marlow <marlowsd@gmail.com>**20100423124853
4015 Ignore-this: 4fa0fd5bafe0d6b58fc81076f50d5f8d
4016]
4017[fix 64-bit value for W_SHIFT, which thankfully appears to be not used
4018Simon Marlow <marlowsd@gmail.com>**20100422213605
4019 Ignore-this: 525c062d2456c224ec8d0e083edd3b55
4020]
4021[Add missing constant folding and optimisation for unsigned division
4022Simon Marlow <marlowsd@gmail.com>**20100422213443
4023 Ignore-this: fb10d1cda0852fab0cbcb47247498fb3
4024 Noticed by Denys Rtveliashvili <rtvd@mac.com>, see #4004
4025]
4026[Fix the GHC API link in the main doc index.html
4027Ian Lynagh <igloo@earth.li>**20100422213226]
4028[Give the right exit code in darcs-all
4029Ian Lynagh <igloo@earth.li>**20100421171339
4030 Our END block was calling system, which alters $?. So now we save and
4031 restore it.
4032]
4033[Use StgWord64 instead of ullong
4034Ian Lynagh <igloo@earth.li>**20100421162336
4035 This patch also fixes ullong_format_string (renamed to showStgWord64)
4036 so that it works with values outside the 32bit range (trac #3979), and
4037 simplifies the without-commas case.
4038]
4039[Implement try10Times in Makefile
4040Ian Lynagh <igloo@earth.li>**20100420165909
4041 Avoid using seq, as FreeBSD has jot instead.
4042]
4043[Fix crash in non-threaded RTS on Windows
4044Simon Marlow <marlowsd@gmail.com>**20100420122125
4045 Ignore-this: 28b0255a914a8955dce02d89a7dfaca
4046 The tso->block_info field is now overwritten by pushOnRunQueue(), but
4047 stg_block_async_info was assuming that it still held a pointer to the
4048 StgAsyncIOResult.  We must therefore save this value somewhere safe
4049 before putting the TSO on the run queue.
4050]
4051[Expand the scope of the event_buf_mutex to cover io_manager_event
4052Simon Marlow <marlowsd@gmail.com>**20100420122026
4053 Ignore-this: 185a6d84f7d4a35997f10803f6dacef1
4054 I once saw a failure that I think was due to a race on
4055 io_manager_event, this should fix it.
4056]
4057[Flags -auto and -auto-all operate only on functions not marked INLINE.
4058Milan Straka <fox@ucw.cz>**20100331191050
4059 Ignore-this: 3b63580cfcb3c33d62ad697c36d94d05
4060]
4061[Spelling correction for LANGUAGE pragmas
4062Max Bolingbroke <batterseapower@hotmail.com>**20100413192825
4063 Ignore-this: 311b51ba8d43f6c7fd32f48db9a88dee
4064]
4065[Update the user guide so it talks about the newer "do rec" notation everywhere
4066Ian Lynagh <igloo@earth.li>**20100416205416
4067 Some of the problems highlighted in trac #3968.
4068]
4069[Fix typo
4070Ian Lynagh <igloo@earth.li>**20100416205412]
4071[Fix Trac #3950: unifying types of different kinds
4072simonpj@microsoft.com**20100412151845
4073 Ignore-this: d145b9de5ced136ef2c39f3ea4a04f4a
4074 
4075 I was assuming that the unifer only unified types of the
4076 same kind, but now we can "defer" unsolved constraints that
4077 invariant no longer holds.  Or at least is's more complicated
4078 to ensure. 
4079 
4080 This patch takes the path of not assuming the invariant, which
4081 is simpler and more robust.  See
4082 Note [Mismatched type lists and application decomposition]
4083]
4084[Fix Trac #3943: incorrect unused-variable warning
4085simonpj@microsoft.com**20100412151630
4086 Ignore-this: 52459f2b8b02c3cb120abe674dc9a060
4087 
4088 In fixing this I did the usual little bit of refactoring
4089]
4090[Convert boot and boot-pkgs to perl
4091Ian Lynagh <igloo@earth.li>**20100415143919
4092 This stops us having to worry about sh/sed/... portability.
4093]
4094[Use $(MAKE), not make, when recursively calling make
4095Ian Lynagh <igloo@earth.li>**20100415121453]
4096[Remove the ghc_ge_609 makefile variables
4097Ian Lynagh <igloo@earth.li>**20100412235658
4098 They are now guaranteed to be YES
4099]
4100[Increase the minimum version number required to 6.10 in configure.ac
4101Ian Lynagh <igloo@earth.li>**20100412235313]
4102[The bootstrapping compiler is now required to be > 609
4103Ian Lynagh <igloo@earth.li>**20100409161046]
4104[Handle IND_STATIC in isRetainer
4105Ian Lynagh <igloo@earth.li>**20100409104207
4106 IND_STATIC used to be an error, but at the moment it can happen
4107 as isAlive doesn't look through IND_STATIC as it ignores static
4108 closures. See trac #3956 for a program that hit this error.
4109]
4110[Add Data and Typeable instances to HsSyn
4111David Waern <david.waern@gmail.com>**20100330011020
4112 Ignore-this: c3f2717207b15539fea267c36b686e6a
4113 
4114 The instances (and deriving declarations) have been taken from the ghc-syb
4115 package.
4116]
4117[Fix for derefing ThreadRelocated TSOs in MVar operations
4118Simon Marlow <marlowsd@gmail.com>**20100407092824
4119 Ignore-this: 94dd7c68a6094eda667e2375921a8b78
4120]
4121[sanity check fix
4122Simon Marlow <marlowsd@gmail.com>**20100407092746
4123 Ignore-this: 9c18cd5f5393e5049015ca52e62a1269
4124]
4125[get the reg liveness right in the putMVar# heap check
4126Simon Marlow <marlowsd@gmail.com>**20100407092724
4127 Ignore-this: b1ba07a59ecfae00e9a1f8391741abc
4128]
4129[initialise the headers of MSG_BLACKHOLE objects properly
4130Simon Marlow <marlowsd@gmail.com>**20100407081712
4131 Ignore-this: 183dcd0ca6a395d08db2be12b02bdd79
4132]
4133[initialise the headers of MVAR_TSO_QUEUE objects properly
4134Simon Marlow <marlowsd@gmail.com>**20100407081514
4135 Ignore-this: 4b4a2f30cf2fb69ca4128c41744687bb
4136]
4137[undo debugging code
4138Simon Marlow <marlowsd@gmail.com>**20100406142740
4139 Ignore-this: 323c2248f817b6717c19180482fc4b00
4140]
4141[putMVar#: fix reg liveness in the heap check
4142Simon Marlow <marlowsd@gmail.com>**20100406135832
4143 Ignore-this: cddd2c7807ac7612c9b2c4c0d384d284
4144]
4145[account for the new BLACKHOLEs in the GHCi debugger
4146Simon Marlow <marlowsd@gmail.com>**20100406133406
4147 Ignore-this: 4d4aeb4bbada3f50dc1fb0123f565e8f
4148]
4149[don't forget to deRefTSO() in tryWakeupThread()
4150Simon Marlow <marlowsd@gmail.com>**20100406130411
4151 Ignore-this: 171d57c4f8653835dec0b69f9be9881c
4152]
4153[Fix bug in popRunQueue
4154Simon Marlow <marlowsd@gmail.com>**20100406091453
4155 Ignore-this: 9d3cec8f18f5c5cbd51751797386eb6f
4156]
4157[fix bug in migrateThread()
4158Simon Marlow <marlowsd@gmail.com>**20100401105840
4159 Ignore-this: 299bcf0d1ea0f8865f3e845eb93d2ad3
4160]
4161[Remove the IND_OLDGEN and IND_OLDGEN_PERM closure types
4162Simon Marlow <marlowsd@gmail.com>**20100401093519
4163 Ignore-this: 95f2480c8a45139835eaf5610217780b
4164 These are no longer used: once upon a time they used to have different
4165 layout from IND and IND_PERM respectively, but that is no longer the
4166 case since we changed the remembered set to be an array of addresses
4167 instead of a linked list of closures.
4168]
4169[Change the representation of the MVar blocked queue
4170Simon Marlow <marlowsd@gmail.com>**20100401091605
4171 Ignore-this: 20a35bfabacef2674df362905d7834fa
4172 
4173 The list of threads blocked on an MVar is now represented as a list of
4174 separately allocated objects rather than being linked through the TSOs
4175 themselves.  This lets us remove a TSO from the list in O(1) time
4176 rather than O(n) time, by marking the list object.  Removing this
4177 linear component fixes some pathalogical performance cases where many
4178 threads were blocked on an MVar and became unreachable simultaneously
4179 (nofib/smp/threads007), or when sending an asynchronous exception to a
4180 TSO in a long list of thread blocked on an MVar.
4181 
4182 MVar performance has actually improved by a few percent as a result of
4183 this change, slightly to my surprise.
4184 
4185 This is the final cleanup in the sequence, which let me remove the old
4186 way of waking up threads (unblockOne(), MSG_WAKEUP) in favour of the
4187 new way (tryWakeupThread and MSG_TRY_WAKEUP, which is idempotent).  It
4188 is now the case that only the Capability that owns a TSO may modify
4189 its state (well, almost), and this simplifies various things.  More of
4190 the RTS is based on message-passing between Capabilities now.
4191]
4192[eliminate some duplication with a bit of CPP
4193Simon Marlow <marlowsd@gmail.com>**20100330154355
4194 Ignore-this: 838f7d341f096ca14c86ab9c81193e36
4195]
4196[Make ioManagerDie() idempotent
4197Simon Marlow <marlowsd@gmail.com>**20100401100705
4198 Ignore-this: a5996b43cdb2e2d72e6e971d7ea925fb
4199 Avoids screeds of "event buffer overflowed; event dropped" in
4200 conc059(threaded1).
4201]
4202[Move a thread to the front of the run queue when another thread blocks on it
4203Simon Marlow <marlowsd@gmail.com>**20100329144521
4204 Ignore-this: c518ff0d41154680edc811d891826a29
4205 This fixes #3838, and was made possible by the new BLACKHOLE
4206 infrastructure.  To allow reording of the run queue I had to make it
4207 doubly-linked, which entails some extra trickiness with regard to
4208 GC write barriers and suchlike.
4209]
4210[remove non-existent MUT_CONS symbols
4211Simon Marlow <marlowsd@gmail.com>**20100330152600
4212 Ignore-this: 885628257a9d03f2ece2a754d993014a
4213]
4214[change throwTo to use tryWakeupThread rather than unblockOne
4215Simon Marlow <marlowsd@gmail.com>**20100329144613
4216 Ignore-this: 10ad4965e6c940db71253f1c72218bbb
4217]
4218[tiny GC optimisation
4219Simon Marlow <marlowsd@gmail.com>**20100329144551
4220 Ignore-this: 9e095b9b73fff0aae726f9937846ba92
4221]
4222[New implementation of BLACKHOLEs
4223Simon Marlow <marlowsd@gmail.com>**20100329144456
4224 Ignore-this: 96cd26793b4e6ab9ddd0d59aae5c2f1d
4225 
4226 This replaces the global blackhole_queue with a clever scheme that
4227 enables us to queue up blocked threads on the closure that they are
4228 blocked on, while still avoiding atomic instructions in the common
4229 case.
4230 
4231 Advantages:
4232 
4233  - gets rid of a locked global data structure and some tricky GC code
4234    (replacing it with some per-thread data structures and different
4235    tricky GC code :)
4236 
4237  - wakeups are more prompt: parallel/concurrent performance should
4238    benefit.  I haven't seen anything dramatic in the parallel
4239    benchmarks so far, but a couple of threading benchmarks do improve
4240    a bit.
4241 
4242  - waking up a thread blocked on a blackhole is now O(1) (e.g. if
4243    it is the target of throwTo).
4244 
4245  - less sharing and better separation of Capabilities: communication
4246    is done with messages, the data structures are strictly owned by a
4247    Capability and cannot be modified except by sending messages.
4248 
4249  - this change will utlimately enable us to do more intelligent
4250    scheduling when threads block on each other.  This is what started
4251    off the whole thing, but it isn't done yet (#3838).
4252 
4253 I'll be documenting all this on the wiki in due course.
4254 
4255]
4256[Fix warnings (allow pushOnRunQueue() to not be inlined)
4257Simon Marlow <marlowsd@gmail.com>**20100401114559
4258 Ignore-this: f40bfbfad70a5165a946d11371605b7d
4259]
4260[remove out of date comment
4261Simon Marlow <marlowsd@gmail.com>**20100401105853
4262 Ignore-this: 26af88dd418ee0bcda7223b3b7e4e8d2
4263]
4264[tidy up spacing in stderr traces
4265Simon Marlow <marlowsd@gmail.com>**20100326163122
4266 Ignore-this: 16558b0433a274be217d4bf39aa4946
4267]
4268[Fix an assertion that was not safe when running in parallel
4269Simon Marlow <marlowsd@gmail.com>**20100325143656
4270 Ignore-this: cad08fb8900eb3a475547af0189fcc47
4271]
4272[Never jump directly to a thunk's entry code, even if it is single-entry
4273Simon Marlow <marlowsd@gmail.com>**20100325114847
4274 Ignore-this: 938da172c06a97762ef605c8fccfedf1
4275 I don't think this fixes any bugs as we don't have single-entry thunks
4276 at the moment, but it could cause problems for parallel execution if
4277 we ever did re-introduce update avoidance.
4278]
4279[Rename forgotten -dverbose-simpl to -dverbose-core2core in the docs.
4280Milan Straka <fox@ucw.cz>**20100331153626
4281 Ignore-this: 2da58477fb96e1cfb80f37dddd7c422c
4282]
4283[Add -pa and -V to the documentation of time profiling options.
4284Milan Straka <fox@ucw.cz>**20100329191121
4285 Ignore-this: be74d216481ec5a19e5f40f85e6e3d65
4286]
4287[Keep gcc 4.5 happy
4288Simon Marlow <marlowsd@gmail.com>**20100330120425
4289 Ignore-this: 7811878cc2bd1ce9cfbb5bf102fe3454
4290]
4291[Fix warning compiling Linker.c for PPC Mac
4292naur@post11.tele.dk**20100403182355
4293 Ignore-this: e2d2448770c9714ce17dd6cf3e297063
4294 The warning message eliminated is:
4295 > rts/Linker.c:4756:0:
4296 >      warning: nested extern declaration of 'symbolsWithoutUnderscore'
4297]
4298[Fix error compiling AsmCodeGen.lhs for PPC Mac (mkRtsCodeLabel)
4299naur@post11.tele.dk**20100403181656
4300 Ignore-this: deb7524ea7852a15a2ac0849c8c82f74
4301 The error messages eliminated are:
4302 > compiler/nativeGen/AsmCodeGen.lhs:875:31:
4303 >     Not in scope: `mkRtsCodeLabel'
4304 > compiler/nativeGen/AsmCodeGen.lhs:879:31:
4305 >     Not in scope: `mkRtsCodeLabel'
4306 > compiler/nativeGen/AsmCodeGen.lhs:883:31:
4307 >     Not in scope: `mkRtsCodeLabel'
4308]
4309[Fix error compiling AsmCodeGen.lhs for PPC Mac (DestBlockId)
4310naur@post11.tele.dk**20100403180643
4311 Ignore-this: 71e833e94ed8371b2ffabc2cf80bf585
4312 The error message eliminated is:
4313 > compiler/nativeGen/AsmCodeGen.lhs:637:16:
4314 >     Not in scope: data constructor `DestBlockId'
4315]
4316[Fix boot-pkgs's sed usage to work with Solaris's sed
4317Ian Lynagh <igloo@earth.li>**20100401153441]
4318[Pass "-i org.haskell.GHC" to packagemaker when building the OS X installer
4319Ian Lynagh <igloo@earth.li>**20100331144707
4320 This seems to fix this failure:
4321 [...]
4322 ** BUILD SUCCEEDED **
4323 rm -f -f GHC-system.pmdoc/*-contents.xml
4324 /Developer/usr/bin/packagemaker -v --doc GHC-system.pmdoc\
4325              -o /Users/ian/to_release/ghc-6.12.1.20100330/GHC-6.12.1.20100330-i386.pkg
4326 2010-03-31 15:08:15.695 packagemaker[13909:807] Setting to : 0 (null)
4327 2010-03-31 15:08:15.709 packagemaker[13909:807] Setting to : 0 org.haskell.glasgowHaskellCompiler.ghc.pkg
4328 2010-03-31 15:08:15.739 packagemaker[13909:807] relocate: (null) 0
4329 2010-03-31 15:08:15.740 packagemaker[13909:807] *** Terminating app due to uncaught exception 'NSInvalidArgumentException', reason: '*** -[NSXMLDocument initWithXMLString:options:error:]: nil argument'
4330 2010-03-31 15:08:15.741 packagemaker[13909:807] Stack: (
4331     2511962091,
4332     2447007291,
4333     2511961547,
4334     2511961610,
4335     2432803204,
4336     453371,
4337     447720,
4338     436209,
4339     435510,
4340     9986,
4341     9918
4342 )
4343 make[1]: *** [framework-pkg] Trace/BPT trap
4344 make: *** [framework-pkg] Error 2
4345]
4346[Use machdepCCOpts when compiling the file to toggle -(no-)rtsopts
4347Ian Lynagh <igloo@earth.li>**20100331161302
4348 Should fix toggling on OS X "Snow Leopard". Diagnosed by Roman Leshchinskiy.
4349]
4350[Avoid a non-portable use of tar reported by Roman Leshchinskiy
4351Ian Lynagh <igloo@earth.li>**20100330145802]
4352[Don't install EXTRA_PACKAGES by default
4353Simon Marlow <marlowsd@gmail.com>**20100330142714
4354 Ignore-this: d4cc8f87a6de8d9d1d6dc9b77130b3
4355]
4356[fix a non-portable printf format
4357Simon Marlow <marlowsd@gmail.com>**20100330134437
4358 Ignore-this: d41c23c54ec29654cb2049de1e588570
4359]
4360[avoid single quote in #error
4361Simon Marlow <marlowsd@gmail.com>**20100330120346
4362 Ignore-this: 663f39e7a27fead2f648fbf22d345bb4
4363]
4364[use FMT_Word64 instead of locally-defined version
4365Simon Marlow <marlowsd@gmail.com>**20100330114650
4366 Ignore-this: 82697b8095dffb3a8e196c687006ece0
4367]
4368[remove old/unused DotnetSupport and GhcLibsWithUnix
4369Simon Marlow <marlowsd@gmail.com>**20100330123732
4370 Ignore-this: c68814868b3671abdc369105bbeafe6c
4371]
4372[fix return type cast in f.i.wrapper when using libffi (#3516)
4373Simon Marlow <marlowsd@gmail.com>**20100329154220
4374 Ignore-this: f898eb8c9ae2ca2009e539735b92c438
4375 
4376 Original fix submitted by
4377   Sergei Trofimovich <slyfox@community.haskell.org>
4378 modified by me:
4379  - exclude 64-bit types
4380  - compare uniques, not strings
4381  - #include "ffi.h" is conditional
4382]
4383[libffi: install 'ffitarget.h' header as sole 'ffi.h' is unusable
4384Simon Marlow <marlowsd@gmail.com>**20100329135734
4385 Ignore-this: f9b555ea289d8df1aa22cb6faa219a39
4386 Submitted by: Sergei Trofimovich <slyfox@community.haskell.org>
4387 Re-recorded against HEAD.
4388]
4389[avoid a fork deadlock (see comments)
4390Simon Marlow <marlowsd@gmail.com>**20100329132329
4391 Ignore-this: 3377f88b83bb3b21e42d7fc5f0d866f
4392]
4393[tidy up the end of the all_tasks list after forking
4394Simon Marlow <marlowsd@gmail.com>**20100329132253
4395 Ignore-this: 819d679875be5f344e816210274d1c29
4396]
4397[Add a 'setKeepCAFs' external function (#3900)
4398Simon Marlow <marlowsd@gmail.com>**20100329110036
4399 Ignore-this: ec532a18cad4259a09847b0b9ae2e1d2
4400]
4401[Explicitly check whether ar supports the @file syntax
4402Ian Lynagh <igloo@earth.li>**20100329123325
4403 rather than assuming that all GNU ar's do.
4404 Apparently OpenBSD's older version doesn't.
4405]
4406[Fix the format specifier for Int64/Word64 on Windows
4407Ian Lynagh <igloo@earth.li>**20100327182126
4408 mingw doesn't understand %llu/%lld - it treats them as 32-bit rather
4409 than 64-bit. We use %I64u/%I64d instead.
4410]
4411[Fix the ghci startmenu item
4412Ian Lynagh <igloo@earth.li>**20100326235934
4413 I'm not sure what changed, but it now doesn't work for me without
4414 the "Start in" field being set.
4415]
4416[Fix paths to docs in "Start Menu" entries in Windows installer; fixes #3847
4417Ian Lynagh <igloo@earth.li>**20100326155917]
4418[Add a licence file for the Windows installer to use
4419Ian Lynagh <igloo@earth.li>**20100326155130]
4420[Add gcc-g++ to the inplace mingw installation; fixes #3893
4421Ian Lynagh <igloo@earth.li>**20100326154714]
4422[Add the licence file to the Windows installer. Fixes #3934
4423Ian Lynagh <igloo@earth.li>**20100326152449]
4424[Quote the paths to alex and happy in configure
4425Ian Lynagh <igloo@earth.li>**20100325143449
4426 Ignore-this: d6d6e1a250f88985bbeea760e63a79db
4427]
4428[Use </> rather than ++ "/"
4429Ian Lynagh <igloo@earth.li>**20100325133237
4430 This stops us generating paths like
4431     c:\foo\/ghc460_0/ghc460_0.o
4432 which windres doesn't understand.
4433]
4434[Append $(exeext) to utils/ghc-pkg_dist_PROG
4435Ian Lynagh <igloo@earth.li>**20100324233447
4436 Fixes bindist creation
4437]
4438[A sanity check
4439Simon Marlow <marlowsd@gmail.com>**20100325110500
4440 Ignore-this: 3b3b76d898c822456857e506b7531e65
4441]
4442[do_checks: do not set HpAlloc if the stack check fails
4443Simon Marlow <marlowsd@gmail.com>**20100325110328
4444 Ignore-this: 899ac8c29ca975d03952dbf4608d758
4445 
4446 This fixes a very rare heap corruption bug, whereby
4447 
4448  - a context switch is requested, which sets HpLim to zero
4449    (contextSwitchCapability(), called by the timer signal or
4450    another Capability).
4451 
4452  - simultaneously a stack check fails, in a code fragment that has
4453    both a stack and a heap check.
4454 
4455 The RTS then assumes that a heap-check failure has occurred and
4456 subtracts HpAlloc from Hp, although in fact it was a stack-check
4457 failure and retreating Hp will overwrite valid heap objects.  The bug
4458 is that HpAlloc should only be set when Hp has been incremented by the
4459 heap check.  See comments in rts/HeapStackCheck.cmm for more details.
4460 
4461 This bug is probably incredibly rare in practice, but I happened to be
4462 working on a test that triggers it reliably:
4463 concurrent/should_run/throwto001, compiled with -O -threaded, args 30
4464 300 +RTS -N2, run repeatedly in a loop.
4465]
4466[comments and formatting only
4467Simon Marlow <marlowsd@gmail.com>**20100325104617
4468 Ignore-this: c0a211e15b5953bb4a84771bcddd1d06
4469]
4470[Change how perl scripts get installed; partially fixes #3863
4471Ian Lynagh <igloo@earth.li>**20100324171422
4472 We now regenerate them when installing, which means the path for perl
4473 doesn't get baked in
4474]
4475[Pass the location of gcc in the ghc wrapper script; partially fixes #3863
4476Ian Lynagh <igloo@earth.li>**20100324171408
4477 This means we don't rely on baking a path to gcc into the executable
4478]
4479[Quote the ar path in configure
4480Ian Lynagh <igloo@earth.li>**20100324162043]
4481[Remove unused cUSER_WAY_NAMES cUSER_WAY_OPTS
4482Ian Lynagh <igloo@earth.li>**20100324145048]
4483[Remove unused cCONTEXT_DIFF
4484Ian Lynagh <igloo@earth.li>**20100324145013]
4485[Remove unused cEnableWin32DLLs
4486Ian Lynagh <igloo@earth.li>**20100324144841]
4487[Remove unused cGHC_CP
4488Ian Lynagh <igloo@earth.li>**20100324144656]
4489[Fix the build for non-GNU-ar
4490Ian Lynagh <igloo@earth.li>**20100324132907]
4491[Tweak the Makefile code for making .a libs; fixes trac #3642
4492Ian Lynagh <igloo@earth.li>**20100323221325
4493 The main change is that, rather than using "xargs ar" we now put
4494 all the filenames into a file, and do "ar @file". This means that
4495 ar adds all the files at once, which works around a problem where
4496 files with the same basename in a later invocation were overwriting
4497 the existing file in the .a archive.
4498]
4499[Enable shared libraries on Windows; fixes trac #3879
4500Ian Lynagh <igloo@earth.li>**20100320231414
4501 Ignore-this: c93b35ec5b7a7fa6ddb286d17a616216
4502]
4503[Add the external core PDF to the new build system
4504Ian Lynagh <igloo@earth.li>**20100321161909]
4505[Allow specifying $threads directly when validating
4506Ian Lynagh <igloo@earth.li>**20100321112835]
4507[Remove LazyUniqFM; fixes trac #3880
4508Ian Lynagh <igloo@earth.li>**20100320213837]
4509[UNDO: slight improvement to scavenging ...
4510Simon Marlow <marlowsd@gmail.com>**20100319153413
4511 Ignore-this: f0ab581c07361f7b57eae02dd6ec893c
4512 
4513 Accidnetally pushed this patch which, while it validates, isn't
4514 correct.
4515 
4516 rolling back:
4517 
4518 Fri Mar 19 11:21:27 GMT 2010  Simon Marlow <marlowsd@gmail.com>
4519   * slight improvement to scavenging of update frames when a collision has occurred
4520 
4521     M ./rts/sm/Scav.c -19 +15
4522]
4523[slight improvement to scavenging of update frames when a collision has occurred
4524Simon Marlow <marlowsd@gmail.com>**20100319112127
4525 Ignore-this: 6de2bb9614978975f17764a0f259d9bf
4526]
4527[Don't install the utf8-string package
4528Ian Lynagh <igloo@earth.li>**20100317212709]
4529[Don't use -Bsymbolic when linking the RTS
4530Ian Lynagh <igloo@earth.li>**20100316233357
4531 This makes the RTS hooks work when doing dynamic linking
4532]
4533[Fix Trac #3920: Template Haskell kinds
4534simonpj@microsoft.com**20100317123519
4535 Ignore-this: 426cac7920446e04f3cc30bd1d9f76e2
4536 
4537 Fix two places where we were doing foldl instead of foldr
4538 after decomposing a Kind.  Strange that the same bug appears
4539 in two quite different places!
4540]
4541[copy_tag_nolock(): fix write ordering and add a write_barrier()
4542Simon Marlow <marlowsd@gmail.com>**20100316143103
4543 Ignore-this: ab7ca42904f59a0381ca24f3eb38d314
4544 
4545 Fixes a rare crash in the parallel GC.
4546 
4547 If we copy a closure non-atomically during GC, as we do for all
4548 immutable values, then before writing the forwarding pointer we better
4549 make sure that the closure itself is visible to other threads that
4550 might follow the forwarding pointer.  I imagine this doesn't happen
4551 very often, but I just found one case of it: in scavenge_stack, the
4552 RET_FUN case, after evacuating ret_fun->fun we then follow it and look
4553 up the info pointer.
4554]
4555[Add sliceP mapping to vectoriser builtins
4556benl@ouroborus.net**20100316060517
4557 Ignore-this: 54c3cafff584006b6fbfd98124330aa3
4558]
4559[Comments only
4560benl@ouroborus.net**20100311064518
4561 Ignore-this: d7dc718cc437d62aa5b1b673059a9b22
4562]
4563[TAG 2010-03-16
4564Ian Lynagh <igloo@earth.li>**20100316005137
4565 Ignore-this: 234e3bc29e2f26cc59d7b03d780cc352
4566]
4567Patch bundle hash:
45686690575d3faeeabffbbfafa5d709a13d929eaccf