Ticket #3165: fixRegression.patch

File fixRegression.patch, 211.7 KB (added by mnislaih, 3 years ago)
Line 
11 patch for repository http://darcs.haskell.org/ghc:
2
3Fri Dec  3 20:23:46 GMT 2010  pepeiborra@gmail.com
4  * Fix the behaviour of :history for ticks surrounding top level functions
5
6New patches:
7
8[Fix the behaviour of :history for ticks surrounding top level functions
9pepeiborra@gmail.com**20101203202346
10 Ignore-this: 8059d4859c52c0c9a235b937cb8cde1d
11] hunk ./compiler/deSugar/Coverage.lhs 694
12 allocATickBox boxLabel pos fvs | isGoodSrcSpan' pos =
13   sameFileName pos
14     (return Nothing) $ TM $ \ env st ->
15-  let me = (pos, declPath env, map (nameOccName.idName) ids, boxLabel)
16+  let mydecl_path
17+        | null (declPath env), TopLevelBox x <- boxLabel = x
18+        | otherwise = declPath env
19+      me = (pos, mydecl_path, map (nameOccName.idName) ids, boxLabel)
20       c = tickBoxCount st
21       mes = mixEntries st
22       ids = occEnvElts fvs
23
24Context:
25
26[FIX #1845 (unconditional relative branch out of range)
27pho@cielonegro.org**20101130143014
28 Ignore-this: df234bd8ad937104c455656fe3c33732
29 
30 Don't use mmap on powerpc-apple-darwin as mmap doesn't support
31 reallocating but we need to allocate jump islands just after each
32 object images. Otherwise relative branches to jump islands can fail
33 due to 24-bits displacement overflow.
34]
35[rts/Linker.c (loadArchive):
36pho@cielonegro.org**20101130142700
37 Ignore-this: bc84f9369ce5c2d289440701b7a3a2ab
38 
39 This routine should be aware of Mach-O misalignment of malloc'ed memory regions.
40]
41[rts/Linker.c (machoGetMisalignment):
42pho@cielonegro.org**20101130123355
43 Ignore-this: 75425600049efd587e9873578e26392f
44 
45 Use fseek(3) instead of rewind(3) to move the file position indicator back to the initial position. Otherwise we can't use this function in loadArchive().
46]
47[rts/Linker.c (ocFlushInstructionCache):
48pho@cielonegro.org**20101130121425
49 Ignore-this: 1e2c207e4b1d17387617ec5d645204b7
50 
51 I found this function causes a segfault when ocAllocateSymbolExtras() has allocated a separate memory region for jump islands.
52]
53[Remove NewQualifiedOperators
54Ian Lynagh <igloo@earth.li>**20101201181117
55 The extension was rejected by Haskell', and deprecated in 7.0.
56]
57[fix ref to utils/ext-core, which moved to Hackage (extcore package)
58Simon Marlow <marlowsd@gmail.com>**20101201092147
59 Ignore-this: 272a7daaa335ef60bcc645db70b4d68b
60]
61[fix floating-point/FFI section: fenv is C99, not POSIX
62Simon Marlow <marlowsd@gmail.com>**20101201092119
63 Ignore-this: ce8b3edd428e4f77691dd739b5b4ae73
64]
65[Fixed some 'unused vars' warnings
66keller@cse.unsw.edu.au**20101130013425
67 Ignore-this: 35790d443faa23b87e4ba442e62376a3
68]
69[vectScalarLam handles int, float, and double now
70keller@cse.unsw.edu.au**20101129231043
71 Ignore-this: 6d67bdc8dd8577184040e791e6f3d0
72]
73[Handling of lets, letrec and case when checking if a lambda expr needs to be vectorised
74keller@cse.unsw.edu.au**20101115051225
75 Ignore-this: 1db6ed63d7b3f6d093e019322b407ff7
76]
77[Document the behaviour of fenv.h functions with GHC (#4391)
78Simon Marlow <marlowsd@gmail.com>**20101126125336
79 Ignore-this: bc4eab49428d567505a28add6fed90f1
80]
81[Remove the no-ghci-lib warning in ghc-pkg
82Ian Lynagh <igloo@earth.li>**20101127235805
83 GHCi libs are no longer necessary, as we can use the .a or .so versions
84 instead.
85]
86[Add GNU-variant support to the .a parser, and other improvements/tidyups
87Ian Lynagh <igloo@earth.li>**20101127223945]
88[Re-indent only
89Ian Lynagh <igloo@earth.li>**20101127191646]
90[Improve linker debugging for archive files
91Ian Lynagh <igloo@earth.li>**20101127190907]
92[Always enable the archive-loading code
93Ian Lynagh <igloo@earth.li>**20101127173000
94 If the GHCi .o lib doesn't exist, load the .a instead
95]
96[Inherit the ForceSpecConstr flag in non-recursive nested bindings
97Roman Leshchinskiy <rl@cse.unsw.edu.au>**20101127125025
98 Ignore-this: 401391eae25cefcb4afaba2e357decc1
99 
100 This makes sure that join points are fully specialised in loops which are
101 marked as ForceSpecConstr.
102]
103[Document -ddump-rule-firings and -ddump-rule-rewrites
104Roman Leshchinskiy <rl@cse.unsw.edu.au>**20101127123528
105 Ignore-this: beade2efe0cd767c0ce9d4f45a3380ba
106]
107[New flag -dddump-rule-rewrites
108Roman Leshchinskiy <rl@cse.unsw.edu.au>**20101127122022
109 Ignore-this: c0ef5b8a199fbd1ef020258d2cde85a3
110 
111 Now, -ddump-rule-firings only shows the names of the rules that fired (it would
112 show "before" and "after" with -dverbose-core2core previously) and
113 -ddump-rule-rewrites always shows the "before" and "after" bits, even without
114 -dverbose-core2core.
115]
116[Acutally, wild-card variables *can* have occurrences
117simonpj@microsoft.com**20101126162409
118 Ignore-this: 544bffed75eeccef03a1097f98524eea
119 
120 This patch removes the Lint test, and comments why
121]
122[Tidy up the handling of wild-card binders, and make Lint check it
123simonpj@microsoft.com**20101126133210
124 Ignore-this: 9e0be9f7867d53046ee5b0e478a0f433
125 
126 See Note [WildCard binders] in SimplEnv.  Spotted by Roman.
127]
128[Substitution should just substitute, not optimise
129simonpj@microsoft.com**20101125172356
130 Ignore-this: 657628d9b6796ceb5f915c43d56e4a06
131 
132 This was causing Trac #4524, by optimising
133      (e |> co)  to   e
134 on the LHS of a rule.  Result, the template variable
135 'co' wasn't bound any more.
136 
137 Now that substition doesn't optimise, it seems sensible to call
138 simpleOptExpr rather than substExpr when substituting in the
139 RHS of rules.  Not a big deal either way.
140]
141[Make SpecConstr "look through" identity coercions
142simonpj@microsoft.com**20101125172138
143 Ignore-this: c1cc585ed890a7702c33987e971e0af6
144]
145[Comment only
146simonpj@microsoft.com**20101125172011
147 Ignore-this: 3c7be8791badd00dcca9610ebb8981d1
148]
149[White space only
150simonpj@microsoft.com**20101101080748
151 Ignore-this: f7133fc6b22ae263c6672543a8534a6f
152]
153[Keep a maximum of 6 spare worker threads per Capability (#4262)
154Simon Marlow <marlowsd@gmail.com>**20101125135729
155 Ignore-this: a020786569656bf2f3a1717b65d463bd
156]
157[Unicide OtherNumber category should be allowed in identifiers (#4373)
158Simon Marlow <marlowsd@gmail.com>**20101115095444
159 Ignore-this: e331b6ddb17550163ee91bd283348800
160]
161[vectoriser: fix warning
162Ben Lippmeier <benl@ouroborus.net>**20101126044036
163 Ignore-this: e1a66bb405bf2f3f56b42c3b13fd4bf3
164]
165[vectoriser: fix warning
166Ben Lippmeier <benl@ouroborus.net>**20101126042950
167 Ignore-this: df8dd25bcfb3946c2974b13953a2f2c7
168]
169[vectoriser: take class directly from the instance tycon
170Ben Lippmeier <benl@ouroborus.net>**20101126042900
171 Ignore-this: 626a416717a5a059f39e53f4ec95fc66
172]
173[vectoriser: comments only
174Ben Lippmeier <benl@ouroborus.net>**20101125073201
175 Ignore-this: 8846ea8895307083bd1ebbc5d7fb1c5
176]
177[vectoriser: follow changes in mkClass
178Ben Lippmeier <benl@ouroborus.net>**20101125062349
179 Ignore-this: d5018cc022686d4272e126ca9a12283a
180]
181[vectoriser: tracing wibbles
182Ben Lippmeier <benl@ouroborus.net>**20101125062332
183 Ignore-this: c2024d8f03bc03bee2851f4f1c139fd5
184]
185[mkDFunUnfolding wants the type of the dfun to be a PredTy
186benl@ouroborus.net**20100914062939
187 Ignore-this: 7aa6e6b140746184cf00355b50c83b66
188]
189[vectoriser: fix conflicts
190Ben Lippmeier <benl@ouroborus.net>**20101125060904
191 Ignore-this: cc3decab1affada8629ca3818b76b3bf
192]
193[Comments and formatting only
194benl@ouroborus.net**20100914062903
195 Ignore-this: b0fc25f0952cafd56cc25353936327d4
196]
197[Comments and formatting to type environment vectoriser
198benl@ouroborus.net**20100909080405
199 Ignore-this: ab8549d53f845c9d82ed9a525fda3906
200]
201[Don't mix implicit and explicit layout
202Ian Lynagh <igloo@earth.li>**20101124231514]
203[Whitespace only
204Ian Lynagh <igloo@earth.li>**20101124230655]
205[Separate NondecreasingIndentation out into its own extension
206Ian Lynagh <igloo@earth.li>**20101124220507]
207[Add another GHC layout rule relaxation to RelaxedLayout
208Ian Lynagh <igloo@earth.li>**20101124205957]
209[Remove an unused build system variable: GhcDir
210Ian Lynagh <igloo@earth.li>**20101124140455]
211[Remove unused build system variable: GhcHasEditline
212Ian Lynagh <igloo@earth.li>**20101124140415]
213[Remove unused variables from the build system: HBC, NHC, MKDEPENDHS
214Ian Lynagh <igloo@earth.li>**20101124140052]
215[Remove references to Haskell 98
216Ian Lynagh <igloo@earth.li>**20101123233536
217 They are no longer right, as we have Haskell' generating new Haskell
218 standards.
219]
220[Tweak a configure test
221Ian Lynagh <igloo@earth.li>**20101123170621]
222[Add a configure test for the visibility hidden attribute
223Ian Lynagh <igloo@earth.li>**20101123170541]
224[sanity: fix places where we weren't filling fresh memory with 0xaa
225Simon Marlow <marlowsd@gmail.com>**20101029092843
226 Ignore-this: 2cb18f7f5afcaf33371aeffce67e218f
227]
228[Just some alpha renaming
229Ian Lynagh <igloo@earth.li>**20101121144455
230 Ignore-this: d5e807c5470840efc199e29f7d50804c
231]
232[For bindists, build ghc-pwd with stage 1
233Ian Lynagh <igloo@earth.li>**20101121183520
234 Ignore-this: a3b5c8b78c81ec1b6d5fbf23da346ff5
235 rather then the bootstrapping compiler. This fixes problems where the
236 bootstrapping compiler dynamically links against libraries not on the
237 target machine.
238]
239[Makefile tweak
240Ian Lynagh <igloo@earth.li>**20101121183342
241 Ignore-this: cd55a2819c1a5fd36da1bc7a75d2ded1
242]
243[Fix a makefile include ordering sanity check
244Ian Lynagh <igloo@earth.li>**20101121174916
245 Ignore-this: d0bdd41c4b618944d04ecb4f54fdd0f1
246]
247[Add an extension for GHC's layout-rule relaxations
248Ian Lynagh <igloo@earth.li>**20101120215340
249 Still TODO: Add the other relaxation (#1060) and update the alternative
250 layout rule to use the extension.
251]
252[Tweak the bindist configure.ac.in
253Ian Lynagh <igloo@earth.li>**20101120173735]
254[configure.ac tweaks
255Ian Lynagh <igloo@earth.li>**20101120170245]
256[When testing the bindist, tell it where gcc is
257Ian Lynagh <igloo@earth.li>**20101120155920
258 The location isn't baked into the bindist, as it may differ from
259 machine to machine.
260]
261[Comments only
262simonpj@microsoft.com**20101119100153
263 Ignore-this: 7abd5d965ea805770449d6f8dadbb921
264]
265[ForceSpecConstr now forces specialisation even for arguments which aren't scrutinised
266Roman Leshchinskiy <rl@cse.unsw.edu.au>**20101118212839
267 Ignore-this: db45721d29a694e53746f8b76513efa4
268]
269[Move the superclass generation to the canonicaliser
270simonpj@microsoft.com**20101118120533
271 Ignore-this: 5e0e525402a240b709f2b8104c1682b2
272 
273 Doing superclass generation in the canonicaliser (rather than
274 TcInteract) uses less code, and is generally more efficient.
275 
276 See Note [Adding superclasses] in TcCanonical.
277 
278 Fixes Trac #4497.
279]
280[Fix the generation of in-scope variables for IfaceLint check
281simonpj@microsoft.com**20101118090057
282 Ignore-this: bbcdba61ddf89d07fe69ca99c2017e3f
283]
284[Comments only
285simonpj@microsoft.com**20101118090034
286 Ignore-this: fa2936d35a0f7be4e4535ea9e2b7bf7b
287]
288[Omit bogus test for -XDeriveFunctor
289simonpj@microsoft.com**20101118090028
290 Ignore-this: a534243011809ebbb788b910961601c5
291 
292 It was duplicated in the case of 'deriving( Functor )'
293 and wrong for 'deriving( Foldable )'
294]
295[Improve error message on advice from a user
296simonpj@microsoft.com**20101118085306
297 Ignore-this: bd4f3858ff24e602e985288f27d536f3
298 
299 See Trac #4499
300]
301[TAG 2010-11-18
302Ian Lynagh <igloo@earth.li>**20101118011554
303 Ignore-this: ccadbe7fadd1148d2ee3caa8c8821ec5
304]
305[Fix the type sanity test in genprimopcode's Syntax.hs
306Ian Lynagh <igloo@earth.li>**20101117163031
307 We assume this is what it's supposed to be checking. Certainly the old
308 test (t2 == t2) can't be right.
309 Spotted by Andres Loeh.
310]
311[Build random with the stage2 compiler
312Ian Lynagh <igloo@earth.li>**20101117155259
313 Ignore-this: 7e376e4df479ff9d5d45604c99faafed
314]
315[remove -XNoMonomorphismRestriction
316Simon Marlow <marlowsd@gmail.com>**20101117124926
317 Ignore-this: b5ee3730f25e188e5920c42e169134fc
318 This was apparently needed at some point during the new typechecker
319 development, but does not seem to be required now.
320]
321[add some {-# LANGUAGE BangPatterns #-} to mollify GHC
322Simon Marlow <marlowsd@gmail.com>**20101117113705
323 Ignore-this: 136e7ff19b009eadbf2e0e6662c37d0f
324]
325[Eventlog: Put correct size for startup event
326scpmw@leeds.ac.uk**20101105151655
327 Ignore-this: 8b6eb4fa2137c8dfe50c5917e3a609a7
328]
329[Document SPECIALISE for imported functions
330simonpj@microsoft.com**20101117111559
331 Ignore-this: 5c9e83d15b85fe5a8639321e00e5dcaa
332 
333 This is a really useful new facility, but I'd forgotten to document it.
334 Pls merge to 7.0 branch
335]
336[Fix Trac #4498: bang-pattern bindings are monomorphic
337simonpj@microsoft.com**20101117101058
338 Ignore-this: 2a739aeca590b4dd1907078ba80133ff
339 
340 This patch forces bang patterns to be monomorphic,
341 and documents this fact.
342]
343[Comments only
344simonpj@microsoft.com**20101117100839
345 Ignore-this: 792d932904fef3fd4fe80fa8d97649db
346]
347[Fix Trac #4501: a transposition error in DynFlags
348simonpj@microsoft.com**20101117100832
349 Ignore-this: b81eca419581a7cec773556514915814
350 
351 Push to STABLE
352]
353[Refactoring of the way that inlinings and rules are activated
354simonpj@microsoft.com**20101116173719
355 Ignore-this: d195c39a646e1fac3804fb044644d226
356 
357 Principally, the SimplifierMode now carries several (currently
358 four) flags in *all* phases, not just the "Gentle" phase.
359 This makes things simpler and more uniform.
360 
361 As usual I did more refactoring than I had intended.
362 
363 This stuff should go into 7.0.2 in due course, once
364 we've checked it solves the DPH performance problems.
365]
366[Some infrastruture for lambda-lifting
367simonpj@microsoft.com**20101116173500
368 Ignore-this: bb0b7db06898b9fa731602107febbf7
369 
370 This stuff should have no effect but it sets things
371 up so that we can try floating out lambdas of n value
372 arguments.
373 
374 The new (secret) flag is -ffloatt-lam-args=n.
375 
376 This is *not* working yet, but it's got tangled up with
377 other stuff I want to commit, and it does no harm.
378]
379[Occurrence analyser takes account of the phase when handing RULES
380simonpj@microsoft.com**20101116173312
381 Ignore-this: 50558fb8ec8fe1d0d50db46a7153b077
382 
383 See Note [Finding rule RHS free vars]
384 
385 This should make Roman happy.
386]
387[Add -fwarn-lazy-unlifted-bindings to the list of flags
388simonpj@microsoft.com**20101116172211
389 Ignore-this: 4f150f347bb74027adacb64545f6b757
390]
391[Warn a bit less often about unlifted bindings.
392simonpj@microsoft.com**20101116171843
393 Ignore-this: 4b1caf1659695931bd17015699542026
394 
395 Warn when
396    (a) a pattern bindings binds unlifted values
397    (b) it has no top-level bang
398    (c) the RHS has a *lifted* type
399 
400 Clause (c) is new, argued for by Simon M
401 
402 Eg     x# = 4# + 4#      -- No warning
403        (# a,b #) = blah  -- No warning
404        I# x = blah       -- Warning
405]
406[Libraries bytestirng, time, binary need -fno-warn-identities
407simonpj@microsoft.com**20101116171610
408 Ignore-this: 17a6e0074510089ba03cfe4c15044b2d
409 
410 They all have redundant uses of fromIntegral, but are no under
411 GHC HQ control.
412]
413[Improve documentation for -fwarn-incomplete-patterns
414simonpj@microsoft.com**20101116171527
415 Ignore-this: d8386202cc322207436db0c5b185dab
416]
417[Add warning for probable identities (fromIntegral and friends)
418simonpj@microsoft.com**20101116171510
419 Ignore-this: 75d0912e716daa5727bb369c4c2d9354
420 
421 See Trac #4488.  The basic idea is to check for
422 
423     fun :: ty -> ty
424 
425 where fun is one of
426   toIntegerName     toRationalName
427   fromIntegralName  realToFracName
428 
429 There's a (documented) flag to control it -fwarn-identities.
430 Currently -Wall switches it on.
431]
432[Remove unncessary fromIntegral calls
433simonpj@microsoft.com**20101116171247
434 Ignore-this: 62a3dee6edb7e42c24b745ab96838eb8
435]
436[Fix -fwarn-missing-import-lists (fix Trac #4489)
437simonpj@microsoft.com**20101115232142
438 Ignore-this: 656b3a76540a488a7111ba7c9ec8ebc4
439]
440[Deriving Typeable changes
441simonpj@microsoft.com**20101115231146
442 Ignore-this: ce91c3f9cb19770115654b5d45f7dfb4
443 
444 * Fix a bug that led to a crash with
445     data family T a
446     deriving Functor T
447 
448 * Allow deriving Typeable for data families
449     data family T a
450     deriving Typeable1 T
451 
452 * Some refactoring and tidying
453]
454[Fix bug #3165 (:history throws irrefutable pattern failed)
455pepeiborra@gmail.com**20101115223623
456 Ignore-this: 73edf56e502b4d0385bc044133b27946
457 
458 I ran across this bug and took the time to fix it, closing
459 a long time due TODO in InteractiveEval.hs
460 
461 Instead of looking around to find the enclosing declaration
462 of a tick, this patch makes use of the information already collected during the
463 coverage desugaring phase
464]
465[Ensure that instance overlap errors are report properly
466simonpj@microsoft.com**20101115142805
467 Ignore-this: 2fca29a95bdc69a4c783cbcc663a10f7
468 
469 This (annoyingly) requires us to re-flatten the class predicate.
470 See Note [Flattening in error message generation]
471]
472[Ensure that unification variables alloc'd during solving are untouchable
473simonpj@microsoft.com**20101115121540
474 Ignore-this: 4cdb38180488e605489ce5d018998089
475 
476 This fixes Trac #4494.  See Note [Extra TcsTv untouchables] in TcSimplify.
477]
478[Fix paths used for doc publishing in the nightly builds
479Ian Lynagh <igloo@earth.li>**20101114205527]
480[Remove -Ofile docs
481Ian Lynagh <igloo@earth.li>**20101114173332
482 They said:
483     (NOTE: not supported since GHC 4.x.  Please ask if you're interested
484     in this.)
485]
486[Update to docbook 4.5; fixes trac #4447
487Ian Lynagh <igloo@earth.li>**20101114155023]
488[Add a build system dependency; fixes #4357
489Ian Lynagh <igloo@earth.li>**20101114140311]
490[Build system tweak: Inline DQ now it's the same on all platforms
491Ian Lynagh <igloo@earth.li>**20101114134636]
492[Fix up the ifdefs in Task.c
493Ian Lynagh <igloo@earth.li>**20101113125314]
494[Use standard task ID print style (hexadecimal).
495Edward Z. Yang <ezyang@mit.edu>**20101111000912
496 Ignore-this: 932c7eb57ab9ae47db12c28800555ecd
497]
498[Remove use of (now-gone) VarPatOut
499simonpj@microsoft.com**20101112230701
500 Ignore-this: 539600aa23278dc68da3eef8387f852
501]
502[Remove the now-unused constructor VarPatOut
503simonpj@microsoft.com**20101105131636
504 Ignore-this: a26cc84eabb9890fc8c375f9c8c5bd9d
505]
506[Track change in isInlinePragma
507simonpj@microsoft.com**20101105131545
508 Ignore-this: 256f4940744d84086c0f1c99a9ab8778
509 
510 In TcBinds we want to use isAnyInlinePragma, to get
511 both INLINE and INLINABLE.  I don't know why this isn't
512 leading to failures for others!  The (bogus) error I got,
513 triggered by this bug was:
514 
515   libraries\haskeline\System\Console\Haskeline\Key.hs:23:1:
516     You cannot SPECIALISE `M.findWithDefault'
517       because its definition has no INLINE/INLINABLE pragma
518]
519[More modules that need LANGUAGE BangPatterns
520simonpj@microsoft.com**20101112174439
521 Ignore-this: 2cc3d2a24b09eae4ddc98afeda35f011
522]
523[Add LANGUAGE BangPatterns to modules that use them
524simonpj@microsoft.com**20101112172049
525 Ignore-this: a9970f684f200fcc820ebb032f084bea
526]
527[Warn about top-level bangs (Trac #4477)
528simonpj@microsoft.com**20101112160801
529 Ignore-this: e9f6c85503c8b19780e5730ddde1e8f0
530 
531 A simple refactoring in the parser
532]
533[Slurp author spellings file from base and add more authors
534Ben Lippmeier <benl@ouroborus.net>**20101112134641
535 Ignore-this: 4bd29f0759fc8a117145c73db0763eeb
536]
537[A (final) re-engineering of the new typechecker
538simonpj@microsoft.com**20101112130011
539 Ignore-this: 23cf1f193a670ca199a05fa5fb940491
540 
541 Regression testing and user feedback for GHC 7.0 taught
542 us a lot.  This patch fixes numerous small bugs, and some
543 major ones (eg Trac #4484, #4492), and improves type
544 error messages.
545 
546 The main changes are:
547 
548 * Entirely remove the "skolem equivalance class" stuff;
549   a very useful simplification
550 
551 * Instead, when flattening "wanted" constraints we generate
552   unification variables (not flatten-skolems) for the
553   flattened type function application
554 
555 * We then need a fixup pass at the end, TcSimplify.solveCTyFunEqs,
556   which resolves any residual equalities of form
557       F xi ~ alpha
558 
559 * When we come across a definite failure (e.g. Int ~ [a]),
560   we now defer reporting the error until the end, in case we
561   learn more about 'a'.  That is particularly important for
562   occurs-check errors.  These are called "frozen" type errors.
563 
564 * Other improvements in error message generation.
565 
566 * Better tracing messages
567]
568[Allow the old [$foo| ... |] syntax for quasi-quotes
569simonpj@microsoft.com**20101112083052
570 Ignore-this: 868e0e07fc6bbc9772bcba54e8ca79d
571 
572 This is just a backward-compatibility thing, to be removed
573 eventually.
574]
575[Fix multi-line string (minor glitch in stage-1 compiler)
576simonpj@microsoft.com**20101111134503
577 Ignore-this: 5360d617003f0e97f91cfaef7f3c45df
578]
579[Add upstream URLs for vector and primitive
580Roman Leshchinskiy <rl@cse.unsw.edu.au>**20101110210258
581 Ignore-this: 6d995644fc2b2c6c408efd64366454da
582]
583[Obey the -rtsopts flag when making shared libraries; part of #4464
584Ian Lynagh <igloo@earth.li>**20101110145449]
585[Avoid scary message from tar if dph isn't in the tree
586Ian Lynagh <igloo@earth.li>**20101103213354
587 Ignore-this: b7010c5551be36e4149867cdb363bef0
588]
589[Whitespace only
590Ian Lynagh <igloo@earth.li>**20101103213318
591 Ignore-this: b6e90d57be101e8beae83516d9ccbe35
592]
593[Remove no-longer-necessary withFlattenedDynflags
594Ian Lynagh <igloo@earth.li>**20101103212718]
595[Use liftIO rather than io
596Ian Lynagh <igloo@earth.li>**20101103212216]
597[Fix a #if test
598Ian Lynagh <igloo@earth.li>**20101102192446
599 showMultiLineString is only in >= 7.1.
600]
601[Fix building genprimopcode with GHC 7.0, and remove haskell98 imports
602Ian Lynagh <igloo@earth.li>**20101102185831]
603[On Windows, when returning memory to the OS, we try to release it
604Ian Lynagh <igloo@earth.li>**20101101161802
605 Ignore-this: 899e1ede2f532caf9fe6d5f8581492c2
606 as well as decommiting it.
607]
608[Whitespace only, in rts/win32/OSMem.c
609Ian Lynagh <igloo@earth.li>**20101029204158]
610[Improve error messages
611simonpj@microsoft.com**20101101164630
612 Ignore-this: 17a9c2ff6b3f6dfdc80f826052302f86
613 
614 In particular, instead of
615    Cannot match 'a' with 'b'
616 we get
617    Could not deduce (a~b) from context (F a ~ b)
618 or whatever
619]
620[Uncomment and fix an OS X linker debug print
621Ian Lynagh <igloo@earth.li>**20101029135805
622 Ignore-this: a2e84c0f10cc9ee4fe1af8451a46114f
623]
624[fix bugs in tryTakeMVar/tryPutMVar
625Simon Marlow <marlowsd@gmail.com>**20101029090423
626 Ignore-this: f9909d1065042e4123225e638ca680b6
627 I'm surprised that these haven't caused any problems (or maybe they have?)
628]
629[Remove duplicate PrimopWrappers generation
630Simon Marlow <marlowsd@gmail.com>**20101028101624
631 Ignore-this: d739fc94db726025f229018b03e7fbc9
632 We were generating PrimopWrappers.hs twice in two different places,
633 one for the normal build and one for Haddock (I think).  Unfortunately
634 one of these was causing extra rebuilding, because the file dropped
635 into the dist-install/build/autogen directory, where it was picked up
636 by the automatic HS_SRCS detection in subsequent make runs.  This was
637 exposed by a recent change to HS_SRCS:
638 
639   Wed Oct 20 11:47:59 BST 2010  Ian Lynagh <igloo@earth.li>
640     * Look for sources in Cabal's autogen directory too
641 
642 This should fix the stage3 failures in the build bots.
643]
644[Comment out bogus debug line that broke build on OSX
645benl@ouroborus.net**20101029074405
646 Ignore-this: 63afeda706de08a63a70f0baffcb9b1f
647]
648[Fix import warning
649benl@ouroborus.net**20101029071617
650 Ignore-this: c3fdf649695be62e4a611d84423501f7
651]
652[Cleanup comments and formatting only
653benl@ouroborus.net**20101029065837
654 Ignore-this: 393131d43ae57e4c1e7ac0dce734c452
655]
656[Nicer error message for #3782
657benl@ouroborus.net**20101029063320
658 Ignore-this: fc746cad57410123a29f37f61f13dd3c
659 It now says:
660 
661 ghc-stage2: sorry! (this is work in progress)
662   (GHC version 7.1.20101028 for i386-apple-darwin):
663        Vectorise.Builtins.indexBuiltin
664     
665     DPH builtin function 'sumTyCon' of size '11' is not yet implemented.
666     This function does not appear in your source program, but it is needed
667     to compile your code in the backend. This is a known, current limitation
668     of DPH. If you want it to to work you should send mail to cvs-ghc@haskell.org
669     and ask what you can do to help (it might involve some GHC hacking).
670 
671 
672 I added 'pprSorry' that behaves like 'pprPanic' except it say sorry! instead
673 of panic!, and doesn't ask the user to report a bug.
674]
675[Replace some exit(n) calls with stg_exit(n); fixes trac #4445
676Ian Lynagh <igloo@earth.li>**20101028134854
677 Also changed exitcode of -1 to 1 in hpc.
678]
679[Remove -ds-multi-tyvar static flag
680simonpj@microsoft.com**20101028142548
681 Ignore-this: 6e69569e3d5966d0c42785b619237816
682 
683 It was always secret, and has now vanished as silently as it came
684]
685[Deprecate -fmethod-sharing, which no longer does anything
686simonpj@microsoft.com**20101028142429
687 Ignore-this: 717621bd06fbe97e564b3766517781a2
688 
689 We can remove it altogether in GHC 7.2
690]
691[Do w/w *recursive* things even if they are small
692simonpj@microsoft.com**20101028142150
693 Ignore-this: cd93dcfb62a9a41e29e1eb6d81097631
694 
695 A wibble to my previous change in WorkWrap, fixes simplrun010
696]
697[Two signficant changes to the simplifier
698simonpj@microsoft.com**20101027193729
699 Ignore-this: 9b35e8ad975ba1cebbba28028f1c7f43
700 
701 1. Do eta-expansion at let-bindings, not lambdas.
702    I have wanted to do this for a long time.
703    See Note [Eta-expanding at let bindings] in SimplUtils
704 
705 2. Simplify the rather subtle way in which InlineRules (the
706    template captured by an INLINE pragma) was simplified.
707    Now, these templates are always simplified in "gentle"
708    mode only, and only INLINE things inline inside them.
709 
710    See Note Note [Gentle mode], Note [Inlining in gentle mode]
711    and Note [RULEs enabled in SimplGently] in SimplUtils
712]
713[Yet another go at CoreArity
714simonpj@microsoft.com**20101027185630
715 Ignore-this: 400f04c4c7b6999567410081c672b081
716 
717 Amazingly, there were still Wrong Things in the arity analysis,
718 exposed by my fiddling with eta expansion.
719 
720 I simplified the code, clarified the comments, added more examples,
721 and tidied it all up.  I hope it's better this time.
722]
723[Buglet in tcIface, now that nested binders can have pragmas
724simonpj@microsoft.com**20101027184235
725 Ignore-this: de2db50370c35b8ae92ec2574d806b33
726 
727 This fix ties the knot for recursive groups properly
728]
729[Add strictness for raiseIO# primop
730simonpj@microsoft.com**20101027184128
731 Ignore-this: a4e862bbf08bdb837c6f3c851b185543
732]
733[Don't worker-wrapper INLINABLE things
734simonpj@microsoft.com**20101027184107
735 Ignore-this: 301a4b14ee8d71e97a83c8a976564a03
736 
737 See Note [Don't w/w INLINABLE things] in WorkWrap
738 
739 This fixes a bug that Milan found.
740]
741[Refactoring and tidyup of HscMain and related things (also fix #1666)
742Simon Marlow <marlowsd@gmail.com>**20101027121132
743 Ignore-this: af7a9a9c1e1f993be689ccd57b38dd74
744 While trying to fix #1666 (-Werror aborts too early) I decided to some
745 tidyup in GHC/DriverPipeline/HscMain.
746 
747  - The GhcMonad overloading is gone from DriverPipeline and HscMain
748    now.  GhcMonad is now defined in a module of its own, and only
749    used in the top-level GHC layer.  DriverPipeline and HscMain
750    use the plain IO monad and take HscEnv as an argument.
751 
752  - WarnLogMonad is gone.  printExceptionAndWarnings is now called
753    printException (the old name is deprecated).  Session no longer
754    contains warnings.
755 
756  - HscMain has its own little monad that collects warnings, and also
757    plumbs HscEnv around.  The idea here is that warnings are collected
758    while we're in HscMain, but on exit from HscMain (any function) we
759    check for warnings and either print them (via log_action, so IDEs
760    can still override the printing), or turn them into an error if
761    -Werror is on.
762 
763  - GhcApiCallbacks is gone, along with GHC.loadWithLogger.  Thomas
764    Schilling told me he wasn't using these, and I don't see a good
765    reason to have them.
766 
767  - there's a new pure API to the parser (suggestion from Neil Mitchell):
768       parser :: String
769              -> DynFlags
770              -> FilePath
771              -> Either ErrorMessages (WarningMessages,
772                                       Located (HsModule RdrName))
773]
774[follow changes in the GHC API
775Simon Marlow <marlowsd@gmail.com>**20101027111003
776 Ignore-this: ccf2c92944b0c9e191c7a25833347507
777]
778[tweaks
779Simon Marlow <marlowsd@gmail.com>**20101013184322
780 Ignore-this: 4861806ebfb190fcfe5bc2a07fef8b1a
781]
782[comment
783Simon Marlow <marlowsd@gmail.com>**20101013184300
784 Ignore-this: 7c9493c3f9a9420b18244cad3ceeb814
785]
786[Warn (not error) for SPECIALISE pragmas without INLINE
787simonpj@microsoft.com**20101027162922
788 Ignore-this: 412a6613556904752ea2b995b39d9edf
789 
790 See Note [SPECIALISE pragmas] in TcBinds.  Fixes Trac #444.
791]
792[fix stage1_libs pseudo-target
793Simon Marlow <marlowsd@gmail.com>**20101026122906
794 Ignore-this: 5ce50e0a6daebdbffbb4edc09f7c85c7
795]
796[Fix #4424.  Allow `-hide-all-packages` to be set dynamically.
797Thomas Schilling <nominolo@googlemail.com>**20101021170432
798 Ignore-this: 59fa27d6c74252a89a14290fa57fa093
799]
800[Fix whitespace and layout in PrelRules.
801Michal Terepeta <michal.terepeta@gmail.com>**20101024174134
802 Ignore-this: ade82debc07bac72079c7264623abade
803]
804[Fix for #4318 (Linker failure on OS X 10.6)
805gwright@antiope.com**20101026135153
806 Ignore-this: 15a0e05d9f029af63bd599762fd56a54
807 
808 This patch fixes two bugs in the Mach-O linker and adds debugging statements
809 to the same. The bugs:
810 
811 1. The test for symbol->n_value == 0 is removed and replaced by a test of the
812 flag field.  Checking the n_value field was just wrong; the value of a
813 symbol should only be examined when allocating space for a common block,
814 in which case the n_value field gives the size of the block.  This bug
815 led to an infrequently occuring linker crash.
816 
817 I believe the behavior of the linker now agrees with the intent of the
818 sketchy Apple documentation.
819 
820 2. Jump islands were being filled with garbage instead of the the location
821 of the referenced symbol. This caused relocations of type X86_64_RELOC_GOT and
822 X86_64_RELOC_GOT_LOAD to eventually lead to crashes.  The fix is simply to
823 look up the symbol.
824 
825 Enough debug statements have been added to follow the operation of the Mach-O
826 linker while it loads modules.  They are not yet as informative and well
827 organized as for ELF.  Improving the debug statements will require some
828 reorganization of the code -- the Mach-O linker seems basically sound, but
829 is crying out for some refactoring and commenting.
830     
831]
832[Use removeDups to find top-level duplicate names
833Ian Lynagh <igloo@earth.li>**20101026211546
834 findDupsEq is quadratic, whereas removeDups is n log n.
835 Fixes T1969 regression.
836]
837[Remove unnecessary imports
838Ian Lynagh <igloo@earth.li>**20101026191906
839 Ignore-this: 399a59e416337d6fa22bfd060a8ca0e9
840]
841[Add NoFoo form to the extensions list that dummy-ghc generates
842Ian Lynagh <igloo@earth.li>**20101026145445]
843[Fix a long-standing bug the float-out pass
844simonpj@microsoft.com**20101026111547
845 Ignore-this: 304a6ec24a49eea430e14a353a26001a
846 
847 We were failing to float out a binding that could be floated,
848 because of a confusion in the Lam case of floatExpr.
849 
850 In investigating this I also discoverd that there is really
851 no point at all in giving a different level to variables in
852 a binding group, so I've now given them all the same (in
853 SetLevels.lvlLamBndrs
854 
855 The overall difference is quite minor in a nofib run:
856 
857         Program           Size    Allocs   Runtime   Elapsed
858 -------------------------------------------------------------
859             Min          +0.0%     -8.5%    -28.4%    -28.7%
860             Max          +0.0%     +0.7%     -0.7%     -1.1%
861  Geometric Mean          +0.0%     -0.0%    -11.6%    -11.8%
862 
863 I don't trust those runtimes, but smaller is good!  The 8.5%
864 improvement in allocation in fulsom, and seems real.  The
865 0.7% allocation increase only happens in programs with
866 very small allocation.  I tracked one down to a call of this form
867 
868   GHC.IO.Handle.Internals.mkDuplexHandle5
869     = \ args -> GHC.IO.Handle.Internals.openTextEncoding1
870                   mb_codec ha_type
871                   (\mb_encoder mb_decoder -> blah)
872 
873 With the new floater the argument of openTextEncoding1 becomes
874 
875      (let lvl = .. in \mb_encoder mb_decoder -> blah)
876 
877 And rightly so.  However in fact this argument is a continuation
878 and hence is called once, so the floating is fruitless.
879 
880 Roll on one-shot-function analysis (which I know how to do
881 but fail to get to!).
882]
883[Comments only
884simonpj@microsoft.com**20100831085001
885 Ignore-this: 7b317432fb34721460080f5677a5bc9f
886]
887[Fix initialisation of strictness in the demand analyser
888simonpj@microsoft.com**20101026081757
889 Ignore-this: d89b117caa95b51b6c24584ac03bedf6
890 
891 Previously, the demand analyser assumed that every binder
892 starts off with no strictness info.  But now that we are
893 preserving strictness on nesting bindings in interface files,
894 that assumption is no longer correct, because an inlined function
895 might have a nested binding with strictness set.
896 
897 So we need to know when we are in the initial sweep, so that we can
898 set the strictness to 'bottom'.
899 
900 See Note [Initialising strictness]
901]
902[Don't look through SCC in exprIsConApp_maybe
903simonpj@microsoft.com**20101026080754
904 Ignore-this: f75b6bbb0ff5fc044934f91f6d8cd93a
905 
906 By not taking account of SCC notes we were inadvertently
907 discarding some, which led to mis-counted entries with -O
908 
909 Should fix (part of) Trac #4414
910]
911[Make isIrrefutableHsPat say True for existentials; fixes Trac #4439
912simonpj@microsoft.com**20101026070405
913 Ignore-this: 5acc2be8b24b70dacdc58b284ee0923f
914 
915 At last. A bug that is really easy to fix.
916]
917[Serialise nested unfoldings across module boundaries
918simonpj@microsoft.com**20101025152817
919 Ignore-this: 1520586f152501d4acb084ebf9cd3136
920 
921 As Roman reported in #4428, nested let-bindings weren't
922 being recorded with their unfoldings.  Needless to say,
923 fixing this had more knock-on effects than I expected.
924]
925[Do not (ever) use substExprSC in the simplifier
926simonpj@microsoft.com**20101025152622
927 Ignore-this: d710a5c06495d4aa4b4b5408ade4dcde
928 
929 "Short-cut" substitution means "do nothing if the substitution
930 is empty". We *never* want do to that in the simplifier because
931 even though the substitution is empty, the in-scope set has
932 useful information:
933 
934  * We get up-to-date unfoldings; and that in turn may
935    reduce the number of iterations of the simplifier
936 
937  * We avoid space leaks, because failing to substitute may
938    hang on to old Ids from a previous iteration
939 
940 (This is what was causing the late inlining of foo in
941 Trac #4428.)
942]
943[Use new showMultiLineString to fix Trac #4436
944simonpj@microsoft.com**20101025151236
945 Ignore-this: 423715c38054018a54c7c5d433e0aba1
946 
947 There is an accompanying patch for libraries
948     base
949     template-haskell
950]
951[For a non-recursive let, make sure we extend the value environment
952simonpj@microsoft.com**20101025143108
953 Ignore-this: 6e72cf2aa0d400dc0484715a62f04d88
954 
955 This has been wrong for a long time.  Fixes Trac #4431.
956]
957[instance Outputable IntMap
958simonpj@microsoft.com**20101025140704
959 Ignore-this: 2674f9b03e752f781d3669cc14402b14
960]
961[Make the ffi.h->ffitarget.h dep a proper dep, rather than an order-only dep
962Ian Lynagh <igloo@earth.li>**20101024134052]
963[libffi: missing dependency on ffitarget.h
964pho@cielonegro.org**20101017113321
965 Ignore-this: 3e7f0c802e39184596219e3dcfd06fe9
966 dist-install/build/ffi.h should have a dependency on ffitarget.h as *_stub.c requires it during the stage-2 build.
967]
968[Remove the need to explicitly flatten the dynflags
969Ian Lynagh <igloo@earth.li>**20101023210442]
970[Follow GHC.Bool/GHC.Types merge
971Ian Lynagh <igloo@earth.li>**20101023165422
972 Ignore-this: fcdb2115e7f8c3bfabb8e91cb028aa2a
973]
974[Don't build packages that need TH if we are making a profiling compiler
975Ian Lynagh <igloo@earth.li>**20101023145600
976 You can't use TH with a profiling compiler.
977]
978[OS X uninstaller: Pass create-links its argument
979Ian Lynagh <igloo@earth.li>**20101023124400]
980[Tweak the OS X installer uninstaller script
981Ian Lynagh <igloo@earth.li>**20101023000951]
982[Improve the OS X installer
983Ian Lynagh <igloo@earth.li>**20101022223104
984 Major changes are:
985 * Version number now includes full GHC version and arch
986 * Uninstaller copes with multiple versions
987]
988[Optimised the representation of Inert Sets to use Maps to get to the relevant inert constraint faster.
989dimitris@microsoft.com**20101022125840]
990[Add rebindable syntax for if-then-else
991simonpj@microsoft.com**20101022143400
992 Ignore-this: 3a1979aef8864a6bae4bfd9be9a543ec
993 
994 There are two main changes
995 
996  * New LANGUAGE option RebindableSyntax, which implies NoImplicitPrelude
997 
998  * if-the-else becomes rebindable, with function name "ifThenElse"
999    (but case expressions are unaffected)
1000 
1001 Thanks to Sam Anklesaria for doing most of the work here
1002]
1003[Lint should check for duplicate top-level bindings with same qualified name
1004simonpj@microsoft.com**20101022072405
1005 Ignore-this: 41db7e1fbbd1863c11bdf9c01cf75e07
1006 
1007 This would have produced a more civilised error for Trac #4396
1008]
1009[Fix unused import warning on OS X
1010Ian Lynagh <igloo@earth.li>**20101022134024]
1011[Switch more uniqFromSupply+splitUniqSupply's to takeUniqFromSupply
1012Ian Lynagh <igloo@earth.li>**20101021142824
1013 Ignore-this: 27991ada81de2d3ef7f4757ac6cc8d11
1014]
1015[Whitespace only
1016Ian Lynagh <igloo@earth.li>**20101021135551]
1017[Whitespace only
1018Ian Lynagh <igloo@earth.li>**20101021134404]
1019[Whitespace only
1020Ian Lynagh <igloo@earth.li>**20101021134133]
1021[Windows installer improvements from Claus
1022Ian Lynagh <igloo@earth.li>**20101021122942
1023 - add link to inno setup docs, so readers don't have to guess what
1024   ghc.iss might be
1025 
1026 - add Task section, and associated most Registry actions with separate
1027   (sub)tasks, so that file associations, default action, and PATH
1028   setting can be optional
1029 
1030 - copy license file into doc directory
1031 
1032 - install icon file again, so that DefaultIcon is no longer a dangling
1033   pointer (#4352)
1034 
1035 - only delete ghc_haskell key if empty (in case there were other tools
1036   using it)
1037 
1038 - add versioned GHCi to right-click menu (to allow for multiple tool installs)
1039]
1040[Fix Trac #4396, by localising pattern binders in the desugarer
1041simonpj@microsoft.com**20101021170324
1042 Ignore-this: d787a5a4c93c93757ca6c0979db9e0b3
1043 
1044 See Note [Localise pattern binders]
1045]
1046[White space only
1047simonpj@microsoft.com**20101021170232
1048 Ignore-this: 72260210a6c31dd61888b32780b0dc9f
1049]
1050[Add an assertion
1051simonpj@microsoft.com**20101021170222
1052 Ignore-this: 11fbd842db26b18519067ff9e6c63faa
1053]
1054[Use takeUniqFromSupply in ByteCodeGen
1055Ian Lynagh <igloo@earth.li>**20101021121454]
1056[Fix some whitespace
1057Ian Lynagh <igloo@earth.li>**20101021121059]
1058[Use takeUniqFromSupply in emitProcWithConvention
1059Ian Lynagh <igloo@earth.li>**20101021120853
1060 We were using the supply's unique, and then passing the same supply to
1061 initUs_, which sounds like a bug waiting to happen.
1062]
1063[Use takeUniqFromSupply in IfaceEnv
1064Ian Lynagh <igloo@earth.li>**20101021120304
1065 This is a little nicer than having to explicitly split supplies and
1066 throw half of them away.
1067]
1068[Define takeUniqFromSupply
1069Ian Lynagh <igloo@earth.li>**20101021115822]
1070[Remove some extraneous whitespace
1071Ian Lynagh <igloo@earth.li>**20101021115600]
1072[Add a comment on why some seq's are done
1073Ian Lynagh <igloo@earth.li>**20101021113404
1074 Ignore-this: 6c8e59eeb2f8733f7ff7686a28463728
1075]
1076[Tidy up RuntimeUnkSkols a bit more
1077simonpj@microsoft.com**20101021100640
1078 Ignore-this: eb3f5617eff99512a2e72d748c7ad99
1079]
1080[Fix haddock markup
1081simonpj@microsoft.com**20101021085420
1082 Ignore-this: a2705c5aaea64648770eb06ffbcf1e8f
1083]
1084[Improve rule checking, to fix panic Trac #4398
1085simonpj@microsoft.com**20101021085402
1086 Ignore-this: 7f1f8496f336b4e52374fc99110db4a8
1087 
1088 Lots of comments with decomposeRuleLhs
1089]
1090[Improve the simple expression optimiser so it does simple beta reduction
1091simonpj@microsoft.com**20101021085030
1092 Ignore-this: 7a8f99b59d01389ffada547dd29005f9
1093]
1094[Template Haskell: add view patterns (Trac #2399)
1095Reiner Pope <reiner.pope@gmail.com>**20101010131720
1096 Ignore-this: 710855eb7e9cdef96478f2f4da9cfbf4
1097]
1098[Tidy-up sweep, following the Great Skolemisation Simplification
1099simonpj@microsoft.com**20101021070837
1100 Ignore-this: ec821a200d54fb7c9d1135e2c4ce581
1101]
1102[Fix haddock markup
1103Ian Lynagh <igloo@earth.li>**20101020151952
1104 Ignore-this: 3d7bdf6a84a6264ae4cea215c415d821
1105]
1106[Tweak the haddock rules; no functional change
1107Ian Lynagh <igloo@earth.li>**20101020151636
1108 Ignore-this: bee8a936b3ebd33832fdd58161ba725
1109]
1110[Don't seq unfoldings
1111Ian Lynagh <igloo@earth.li>**20101020143710
1112 Ignore-this: 972ae43afdca1b7c6713c6ee5af14da4
1113 We generate intermediate unfoldings which are just thrown away, so
1114 evaluating them is a waste of time.
1115]
1116[Avoid hanging on to old unfoldings; fixes #4367 (compiler space regression)
1117Ian Lynagh <igloo@earth.li>**20101020131539]
1118[Comments and layout only
1119simonpj@microsoft.com**20101020132438
1120 Ignore-this: ae58e1d0dbb2d7d9ce02fe765cd3591f
1121]
1122[Look for sources in Cabal's autogen directory too
1123Ian Lynagh <igloo@earth.li>**20101020104759
1124 Ignore-this: 709b8b491ed89c9d979b93dbaa6e7069
1125]
1126[Follow Cabal change: Use usedExtensions rather than extensions
1127Ian Lynagh <igloo@earth.li>**20101020104739
1128 Ignore-this: 18f7fe6729e573c0b75746b5f891bb71
1129]
1130[fix markup
1131Simon Marlow <marlowsd@gmail.com>**20101020123116
1132 Ignore-this: 51372bccc13905f45610b106cacce31c
1133]
1134[remove xref to hasktags
1135Simon Marlow <marlowsd@gmail.com>**20101020123106
1136 Ignore-this: 1211f44dca795532adc729d08fde30b8
1137]
1138[Update the documentation on using DLL's from Windows, fixing several errors, in particular those relating to bug 3605
1139Neil Mitchell **20101010100709
1140 Ignore-this: 13b44b074b123c21f1b2257c9b1d0585
1141]
1142[(1) More lenient kind checking, (2) Fixed orientation problems and avoiding double unifications, (3) Comments
1143dimitris@microsoft.com**20101020115526]
1144[Midstream changes to deal with spontaneous solving and flatten skolem equivalence classes
1145dimitris@microsoft.com**20101019171514]
1146[hasktags was dropped in GHC 6.12, remove it from the docs
1147Simon Marlow <marlowsd@gmail.com>**20101018084401
1148 Ignore-this: e17307814d9c91f1197c07e0ea421f88
1149]
1150[Refactor, plus fix Trac #4418
1151simonpj@microsoft.com**20101020090946
1152 Ignore-this: 401774da6fbd69d567a27a58be3dd9b8
1153 
1154 We weren't doing fundeps for derived superclasses
1155]
1156[Add a comment, connecting the seq to the test (#4367) that shows its usefulness
1157simonpj@microsoft.com**20101020075941
1158 Ignore-this: 2dae7e226168a25e5753467c08e32414
1159]
1160[Define setIdUnfoldingLazily, and use it in Vectorise
1161Ian Lynagh <igloo@earth.li>**20101019201537
1162 Fixes a loop in the compiler, when running the dph tests
1163]
1164[Evaluate the results in coreToStgApp
1165Ian Lynagh <igloo@earth.li>**20101019170532
1166 Ignore-this: 3b3439d35ac21e664b5d24a070d6e8ca
1167]
1168[Retab CoreToStg, and remove trailing whitespace
1169Ian Lynagh <igloo@earth.li>**20101019155530]
1170[seq the unfolding in setUnfoldingInfo
1171Ian Lynagh <igloo@earth.li>**20101019154552
1172 Ignore-this: a18677dfdc1f1dddc82e89b3267d2bb1
1173 Contrary to the comment, for the module in #4367 at least, it is a big
1174 improvement. Without it we get a huge spike of drag.
1175]
1176[Comments only
1177simonpj@microsoft.com**20101019153531
1178 Ignore-this: 159849457d44f7036f22004075ac1377
1179]
1180[Fix debugger
1181simonpj@microsoft.com**20101019153522
1182 Ignore-this: ea1fb8253a1ac45bdc58d4f1ea9aa0f9
1183 
1184 A bit yukky; see Note [Runtime skolems] in TcErrors.
1185 But it works, and the debugger just is yukky in places.
1186]
1187[Fix IPRun by fixing the inferred quantification mechanism
1188simonpj@microsoft.com**20101019090220
1189 Ignore-this: e7dc8c35979136237fa9a572b70f298e
1190]
1191[Recover after an error in an implication constraint
1192simonpj@microsoft.com**20101019090202
1193 Ignore-this: fbbbdc2c3fc0bac0662d6b40e5d57d47
1194]
1195[Reject programs with equality superclasses for now
1196simonpj@microsoft.com**20101019090140
1197 Ignore-this: 3e3eb8c876fa95290f0278e800fce5bc
1198]
1199[Layout and tiny refactoring only
1200simonpj@microsoft.com**20101019090124
1201 Ignore-this: ee21565fde8cd45514d83d0ca9af3253
1202]
1203[Clean up the debugger code
1204simonpj@microsoft.com**20101019090100
1205 Ignore-this: 8677a9084d787f45542419a58e5d8866
1206 
1207 In particular there is much less fiddly skolemisation now
1208 Things are not *quite* right (break001 and 006 still fail),
1209 but they are *much* better than before.
1210]
1211[Add new VarEnv functions minusVarEnv, intersectsVarEnv, unionInScope
1212simonpj@microsoft.com**20101019085609
1213 Ignore-this: a3a8e23fe522079e3eb9c34ea8032602
1214]
1215[Major pass through type checker:(1) prioritizing equalities, (2) improved Derived mechanism, (3) bugfixes
1216dimitris@microsoft.com**20101018151510]
1217[(1) Caching FD improvements for efficiency, (2) preventing cascading deriveds from entering the inert, (3) Fixing bugs in the creation of FlexiTcS variables
1218dimitris@microsoft.com**20101015164421]
1219[Midstream changes for performance improvement related to superclasses and functional dependencies.
1220dimitris@microsoft.com**20101014143811]
1221[Minor
1222dimitris@microsoft.com**20101012080951]
1223[Commentary changes
1224dimitris@microsoft.com**20101011142235]
1225[Kind checking bugfix (#4356) and preventing wanteds from rewriting wanteds
1226dimitris@microsoft.com**20101008173700]
1227[Fix a retainer profiling segfault
1228Ian Lynagh <igloo@earth.li>**20101019132727
1229 Ignore-this: 19f9e2a6dc8a64b30f3769f4101f82db
1230 The bitmap type wasn't big enough to hold large bitmaps on 64 bit
1231 platforms. Profiling GHC was segfaulting when retainStack was handling a
1232 size 33 bitmap.
1233]
1234[Fix -auto-all: Add SCCs to IDs which have a monotype too
1235Ian Lynagh <igloo@earth.li>**20101018153957]
1236[Define SpecConstrAnnotation in GHC.Exts, and import it from there
1237simonpj@microsoft.com**20101018135746
1238 Ignore-this: a6ade815584a1d03be55a6c417de0ab0
1239 
1240 Reason: avoid having to link the entire ghc package in modules
1241 that use compile-time annotations:
1242 
1243      import GHC.Exts( SpecConstrAnnotation )
1244      {-# ANN type T ForceSpecConstr #-}
1245 
1246 It's a kind of bug that the package exporting SpecConstrAnnotation
1247 is linked even though it is only needed at compile time, but putting
1248 the data type declaration in GHC.Exts is a simple way to sidestep
1249 the problem
1250 
1251 See See Note [SpecConstrAnnotation] in SpecConstr
1252]
1253[Fix warnings in AsmCodeGen
1254David Terei <davidterei@gmail.com>**20101007143559
1255 Ignore-this: 1c34b3a39968a5ca123862726f36a8ee
1256]
1257[LLVM: Fix compilation of writebarrier, #4308
1258David Terei <davidterei@gmail.com>**20101004153843
1259 Ignore-this: df256e78b90e3fc263a88d4f0af6ea96
1260]
1261[Change how the OS X installer's create-links finds the versin number
1262Ian Lynagh <igloo@earth.li>**20101017122352
1263 Ignore-this: 543d3f40c2ba5daf8b2b4ea22eb4bfca
1264 It now gets created by configure, rather than trying to work out the
1265 version number at runtime.
1266]
1267[Add more quoting to distrib/MacOS/installer-scripts/create-links
1268Ian Lynagh <igloo@earth.li>**20101017112648
1269 Ignore-this: 5119ea2efc925d568d96c616af550c7c
1270]
1271[change os x installer to allow multiple installed versions
1272Evan Laforge <qdunkan@gmail.com>**20100929234538
1273 Ignore-this: 4f70551293f53a52c39c5962d8a5bd5b
1274 This puts the ghc version into the package name so they are considered separate
1275 packages.
1276]
1277[Only put the boot packages in the haddock contents/index
1278Ian Lynagh <igloo@earth.li>**20101016180031
1279 Ignore-this: 3dc276dc734fec207def24a1d10eb369
1280 We don't install dph etc, so don't put them in the doc index.
1281]
1282[Correct the regexp used to search for extra packages
1283Ian Lynagh <igloo@earth.li>**20101016123421
1284 Ignore-this: 98e3f8fda1c5678730a00ec1ac25fde3
1285 We weren't ignoring comment lines
1286]
1287[New member "archiveMemberName" for struct _ObjectCode
1288pho@cielonegro.org**20100927224145
1289 Ignore-this: 628bc605e7dc4f0c4856c6f7ad23d9ee
1290 
1291 struct _ObjectCode should be able to retain the name of archive members.
1292 Though currently the only use of those names are for debugging outputs.
1293]
1294[Add a -fghci-sandbox flag so that we can en/disable the ghci sandbox
1295Ian Lynagh <igloo@earth.li>**20101015172746
1296 Ignore-this: 52bc5729437a93a925d3586d9803623c
1297 It's on by default (which matches the previous behaviour).
1298 
1299 Motivation:
1300 GLUT on OS X needs to run on the main thread. If you
1301 try to use it from another thread then you just get a
1302 white rectangle rendered. For this, or anything else
1303 with such restrictions, you can turn the GHCi sandbox off
1304 and things will be run in the main thread.
1305]
1306[Fix boot; it was failing if darcs-all or validate were missing
1307Ian Lynagh <igloo@earth.li>**20101015164549
1308 Ignore-this: 14b3e98836647aff345e35bca2102f93
1309 (which is the case in sdists)
1310]
1311[Comments and layout
1312simonpj@microsoft.com**20101015131924
1313 Ignore-this: 126fbdb629a08c1380c7a1f5cd967d97
1314]
1315[Make (Located a) an instance of Eq, Ord
1316simonpj@microsoft.com**20101015131857
1317 Ignore-this: 5da50f8dab06fcbc23ce149cd5080062
1318 
1319 Fulfils Trac #4369
1320]
1321[Give user-defined rules precedence over built-in rules
1322simonpj@microsoft.com**20101015131814
1323 Ignore-this: 76fbf36cb1a09e31c10c68c06b98937e
1324 
1325 This fixes Trac #4397.  See comments with 'isMoreSpecific'.
1326]
1327[Fix Trac #4401: meta-tyvars allocated by the constraint solver are always touchable
1328simonpj@microsoft.com**20101015130818
1329 Ignore-this: 7fcdaee825d9cadcf69c8c9b8967a446
1330 
1331   See Note [Touchable meta type variables] in TcSMonad
1332]
1333[Remove GHC.extendGlobalRdrScope, GHC.extendGlobalTypeScope
1334simonpj@microsoft.com**20101013091107
1335 Ignore-this: df65a43d261353ad53811b25507e913e
1336 
1337 These functions were added by
1338 
1339    Tue Apr 18 03:36:06 BST 2006  Lemmih <lemmih@gmail.com>
1340    * Make the initial rdr and type scope available in the ghc-api
1341 
1342 The are extremely dubious, because they extend the Rdr and Type
1343 env for every compilation.  The right thing to do is to use
1344 the InteractiveContext for temporary extensions.
1345 
1346 So far as we know, no one uses them.  And if they are being used
1347 it's probably a mistake.  So we're backing them out.
1348]
1349[InlinePrag needs an arity only for INLINE, not INLINABLE
1350Simon Marlow <marlowsd@gmail.com>**20101015094925
1351 Ignore-this: 3b338f0b2b49fa714b195067e0026ef7
1352 This doesn't fix anything (we think), but it's morally correct.
1353]
1354[Fix #4346 (INLINABLE pragma not behaving consistently)
1355Simon Marlow <marlowsd@gmail.com>**20101015094836
1356 Ignore-this: 9a9d8ad42cfdf7f619adfac284bae021
1357 Debugged thanks to lots of help from Simon PJ: we weren't updating the
1358 UnfoldingGuidance when the unfolding changed.
1359 Also, a bit of refactoring and additinoal comments.
1360]
1361[Have boot check that we have the dph packages when validating
1362Ian Lynagh <igloo@earth.li>**20101014140556
1363 Ignore-this: 61e365127933f4dbfb62be66115455bd
1364]
1365[Add more documentation for interruptible foreign calls
1366Simon Marlow <marlowsd@gmail.com>**20101014084253
1367 Ignore-this: 28ec0ddd958926a08ab816b6975da344
1368]
1369[minor refactoring
1370Simon Marlow <marlowsd@gmail.com>**20100926105819
1371 Ignore-this: 70d38e0a31a096c94dad5f346b0d91f6
1372]
1373[Fix for interruptible FFI handling
1374Simon Marlow <marlowsd@gmail.com>**20100925193442
1375 Ignore-this: a63ba3c60f577002a1d32b30bb45090c
1376 Set tso->why_blocked before calling maybePerformBlockedException(), so
1377 that throwToSingleThreaded() doesn't try to unblock the current thread
1378 (it is already unblocked).
1379]
1380[interruptible FFI: more robust handling of the exception case in the interpreter
1381Simon Marlow <marlowsd@gmail.com>**20100925193317
1382 Ignore-this: 7f9e67835a3bd096154db2dcf533ec66
1383]
1384[Don't interrupt when task blocks exceptions, don't immediately start exception.
1385Edward Z. Yang <ezyang@mit.edu>**20100925033026
1386 Ignore-this: 6c0669995a2b66abf29d68b3711cb78e
1387]
1388[Interruptible FFI calls with pthread_kill and CancelSynchronousIO. v4
1389Edward Z. Yang <ezyang@mit.edu>**20100919002905
1390 Ignore-this: 43c260f90eeb9c03413f6e749e23101d
1391 
1392 This is patch that adds support for interruptible FFI calls in the form
1393 of a new foreign import keyword 'interruptible', which can be used
1394 instead of 'safe' or 'unsafe'.  Interruptible FFI calls act like safe
1395 FFI calls, except that the worker thread they run on may be interrupted.
1396 
1397 Internally, it replaces BlockedOnCCall_NoUnblockEx with
1398 BlockedOnCCall_Interruptible, and changes the behavior of the RTS
1399 to not modify the TSO_ flags on the event of an FFI call from
1400 a thread that was interruptible.  It also modifies the bytecode
1401 format for foreign call, adding an extra Word16 to indicate
1402 interruptibility.
1403 
1404 The semantics of interruption vary from platform to platform, but the
1405 intent is that any blocking system calls are aborted with an error code.
1406 This is most useful for making function calls to system library
1407 functions that support interrupting.  There is no support for pre-Vista
1408 Windows.
1409 
1410 There is a partner testsuite patch which adds several tests for this
1411 functionality.
1412]
1413[Remove ghc-pkg's dependency on haskell98
1414Ian Lynagh <igloo@earth.li>**20101013194356
1415 Ignore-this: 273fdede3f21e682faa4dc6e61f4e7aa
1416]
1417[Build haskell98 and haskell2010 with stage2
1418Ian Lynagh <igloo@earth.li>**20101013182759
1419 Stops us accidentally depending on them
1420]
1421[Fix warning: Remove unused import
1422Ian Lynagh <igloo@earth.li>**20101013141224
1423 Ignore-this: ae8891440811f4676d9bb8e721ad22ca
1424]
1425[Fix warnings
1426benl@ouroborus.net**20101013040335
1427 Ignore-this: 4e5875827d2840de863000cdb35afb0e
1428]
1429[RegAlloc: Track slot liveness over jumps in spill cleaner
1430benl@ouroborus.net**20101013015414
1431 Ignore-this: ccd4a148908b7fbdc6ea76acf527c16b
1432]
1433[Bump Cabal dep
1434Ian Lynagh <igloo@earth.li>**20101012154528
1435 Ignore-this: da8539b957b74c7da91efff8ac25872
1436]
1437[Remove __HASKELL1__, __HASKELL98__, __CONCURRENT_HASKELL__
1438Ian Lynagh <igloo@earth.li>**20101012134700
1439 Ignore-this: d0db531be58537c52802de1c62694eb1
1440 We used to define these CPP symbols, but nothing on hackage uses them
1441 and the first 2 are no longer correct (as we support multiple Haskell
1442 versions).
1443]
1444[Follow Cabal changes: Cabal no longer has a docbook userguide
1445Ian Lynagh <igloo@earth.li>**20101012130538
1446 Ignore-this: bf0b30e465c13f82725d72aec145e939
1447 For now we don't build the Cabal userguide, but we should add markdown
1448 support so that we can do so.
1449]
1450[Fix build on Windows: ghc-pkg/Main.hs needs ForeignFunctionInterface
1451Ian Lynagh <igloo@earth.li>**20101012112111]
1452[Remove unnecessary import
1453Ian Lynagh <igloo@earth.li>**20101010222231
1454 Ignore-this: 6c7d5cee72837e2af43c2807e10ad602
1455]
1456[Make "./validate --slow" run the full testsuite
1457Ian Lynagh <igloo@earth.li>**20101007004327
1458 Ignore-this: 2dee8262c19fd5d5034a186203e20d0f
1459]
1460[Fix build following haskell98 and -fglasgow-exts changes
1461Ian Lynagh <igloo@earth.li>**20101006160656
1462 Ignore-this: c56bdb0f01da897b3de75bcab1e2887c
1463]
1464[Don't automatically link the haskell98 package
1465Ian Lynagh <igloo@earth.li>**20101006130235
1466 The default language is now Haskell2010, so this was a little odd.
1467 Also, --make is now on by default, so this was largely irrelevant.
1468]
1469[Deprecate -fglasgow-exts
1470Ian Lynagh <igloo@earth.li>**20101006124413
1471 Ignore-this: 70be67898a633ab31d76f6fca557d55
1472]
1473[Remove Opt_GADTs and Opt_TypeFamilies from -fglasgow-exts
1474Ian Lynagh <igloo@earth.li>**20101006122000
1475 Ignore-this: 7153ec880b2b5f9d332ae2f57b97afcc
1476 This means most code doesn't get caught by monomorphic local bindings.
1477]
1478[Fix Trac #4360: omitted case in combineCtLoc
1479simonpj@microsoft.com**20101008135747
1480 Ignore-this: 834636a97af6469862a822809253db41
1481]
1482[Beautiful new approach to the skolem-escape check and untouchable
1483simonpj@microsoft.com**20101008133751
1484 Ignore-this: 33517a772cfdfbf4aa4678609f3dcd71
1485 
1486 Instead of keeping a *set* of untouchable variables in each
1487 implication contraints, we keep a *range* of uniques for the
1488 *touchable* variables of an implication.  This are precisely
1489 the ones we would call the "existentials" if we were French.
1490 
1491 It turns out that the code is more efficient, and vastly easier
1492 to get right, than the set-based approach.
1493 
1494 Fixes Trac #4355 among others
1495]
1496[Do less simplification when doing let-generalisation
1497simonpj@microsoft.com**20101008133542
1498 Ignore-this: 71366d0de37f10ffba2edc9f3927ddbe
1499 
1500 This fixes Trac #4361.  In a rather delicate way, but
1501 no more delicate than before.  A more remoseless typechecker
1502 would reject #4361 altogether.
1503 
1504 See Note [Avoid unecessary constraint simplification]
1505]
1506[Suppress ambiguity errors if any other errors occur
1507simonpj@microsoft.com**20101008111318
1508 Ignore-this: 40f014265c1ab15fe172baaf76c23c87
1509]
1510[Fix Trac #4361: be more discerning when inferring types
1511simonpj@microsoft.com**20101008111227
1512 Ignore-this: 33656f9a151f494b26b6318b5fcffef
1513 
1514 Note [Avoid unecessary constraint simplification] in TcSimplify
1515]
1516[Float out partial applications
1517Simon Marlow <marlowsd@gmail.com>**20101008092709
1518 Ignore-this: 2dc9d10597c19d0598ef2fc3cf74156d
1519 
1520 This fixes at least one case of performance regression in 7.0, and
1521 is nice win on nofib:
1522 
1523         Program           Size    Allocs   Runtime   Elapsed
1524             Min          +0.3%    -63.0%    -38.5%    -38.7%
1525             Max          +1.2%     +0.2%     +0.9%     +0.9%
1526  Geometric Mean          +0.6%     -3.0%     -6.4%     -6.6%
1527]
1528[Suppress knock-on typechecker errors
1529simonpj@microsoft.com**20101008094348
1530 Ignore-this: 8d125926286a7614fa1ce998e3b26d04
1531 
1532 The error cascade caused puzzling errors in T4093b, and
1533 suppressing some seems like a good plan.  Very few test
1534 outputs change.
1535]
1536[Some refactoring and simplification in TcInteract.occurCheck
1537simonpj@microsoft.com**20101007163500
1538 Ignore-this: d43d09370ab27b8796062e2e98ce7e9
1539]
1540[Comments only
1541simonpj@microsoft.com**20101007130301
1542 Ignore-this: ab46592edd3d24786bbce42c50feb4fd
1543]
1544[Implement auto-specialisation of imported Ids
1545simonpj@microsoft.com**20101007111051
1546 Ignore-this: 45257ff6e9597e4fa4de10b0657e27d6
1547 
1548 This big-ish patch arranges that if an Id 'f' is
1549   * Type-class overloaded
1550        f :: Ord a => [a] -> [a]
1551   * Defined with an INLINABLE pragma
1552        {-# INLINABLE f #-}
1553   * Exported from its defining module 'D'
1554 
1555 then in any module 'U' that imports D
1556 
1557 1. Any call of 'f' at a fixed type will generate
1558    (a) a specialised version of f in U
1559    (b) a RULE that rewrites unspecialised calls to the
1560        specialised on
1561 
1562   e.g. if the call is (f Int dOrdInt xs) then the
1563   specialiser will generate
1564      $sfInt :: [Int] -> [Int]
1565      $sfInt = <code for f, imported from D, specialised>
1566      {-# RULE forall d.  f Int d = $sfInt #-}
1567 
1568 2. In addition, you can give an explicit {-# SPECIALISE -#}
1569    pragma for the imported Id
1570      {-# SPECIALISE f :: [Bool] -> [Bool] #-}
1571    This too generates a local specialised definition,
1572    and the corresponding RULE
1573 
1574 The new RULES are exported from module 'U', so that any module
1575 importing U will see the specialised versions of 'f', and will
1576 not re-specialise them.
1577 
1578 There's a flag -fwarn-auto-orphan that warns you if the auto-generated
1579 RULES are orphan rules. It's not in -Wall, mainly to avoid lots of
1580 error messages with existing packages.
1581 
1582 Main implementation changes
1583 
1584  - A new flag on a CoreRule to say if it was auto-generated.
1585    This is persisted across interface files, so there's a small
1586    change in interface file format.
1587 
1588  - Quite a bit of fiddling with plumbing, to get the
1589    {-# SPECIALISE #-} pragmas for imported Ids.  In particular, a
1590    new field tgc_imp_specs in TcGblEnv, to keep the specialise
1591    pragmas for imported Ids between the typechecker and the desugarer.
1592 
1593  - Some new code (although surprisingly little) in Specialise,
1594    to deal with calls of imported Ids
1595]
1596[Make NameEnv back into type NameEnv a = UniqFM a
1597simonpj@microsoft.com**20101007104638
1598 Ignore-this: 24857c013461788be354520e84f4c286
1599 
1600 I don't think the type distinction of declaring NameEnv with a newtype
1601 (as it was) is really useful to us. Moreover, VarEnv is a UniqFM, and
1602 I do sometimes want to build an envt with Ids and look up with Names.
1603 
1604 This may not be the last word on the subject.
1605]
1606[Improve the rule-matcher
1607simonpj@microsoft.com**20101007103700
1608 Ignore-this: 9de96237dc4b73a43326bd568e34b53b
1609 
1610 Previously it was rejecting the match
1611 
1612   Template: forall s t. map s t
1613   Actual:   map Int t
1614 
1615 which should obviously be fine.  It turns out that this kind of match
1616 comes up when specialising.  By freshening that t we could avoid the
1617 difficulty, but morally the (forall t) binds t and the rule should
1618 be alpha-equivalent regardless of the forall'd variables.
1619 
1620 This patch makes it so, and incidentally makes matching a little
1621 more efficient.  See Note [Eta expansion] in VarEnv.
1622]
1623[Fix Trac #4345: simplifier bug
1624simonpj@microsoft.com**20101007102720
1625 Ignore-this: 261c1c9f094df344ce34de814f8b60c5
1626 
1627 This is another long-standing bug, in which there was a possibility
1628 that a loop-breaker could lose its loop-breaker-hood OccInfo,
1629 and then the simplifer re-simplified the expression. Result, either
1630 non-termination or, in the case of #4345, an unbound identifier.
1631 
1632 The fix is very simple, in Id.transferPolyIdInfo.
1633 See Note [transferPolyIdInfo].
1634]
1635[Avoid redundant simplification
1636simonpj@microsoft.com**20101007095935
1637 Ignore-this: 61bd1a2c508260f558866e6a88c29fa3
1638 
1639 When adding specialisation for imported Ids, I noticed that the
1640 Glorious Simplifier was repeatedly (and fruitlessly) simplifying the
1641 same term.  It turned out to be easy to fix this, because I already
1642 had a flag in the ApplyTo and Select constructors of SimplUtils.SimplCont.
1643 
1644 See Note [Avoid redundant simplification]
1645]
1646[Make the occurrence analyser deal correctly with RULES for imported Ids
1647simonpj@microsoft.com**20101007094100
1648 Ignore-this: 335b1cad013524e42b31e88c0a7a00f6
1649 
1650 This patch fixes a long-standing lurking bug, but it surfaced when I
1651 was adding specialisation for imported Ids.
1652 
1653 See Note [ImpRuleUsage], which explains the issue.   The solution
1654 seems more complicated than the problem really deserves, but I
1655 could not think of a simpler way, so I just bit the bullet and
1656 wrote the code.  Improvements welcome.
1657]
1658[Make warning-free
1659simonpj@microsoft.com**20101007092007
1660 Ignore-this: 4bae0c470a8a1f96d21990d1f3cf1f93
1661]
1662[This is just white-space and layout
1663simonpj@microsoft.com**20101007091618
1664 Ignore-this: 759c0335df70fce32558e967f140803a
1665 
1666 (At least, I don't think there is anything else.)
1667]
1668[Fix an ASSERT failure in FamInstEnv
1669simonpj@microsoft.com**20101007091327
1670 Ignore-this: a8c08ccb7ec2bc65864a674b5441539
1671 
1672 I added a lot of comments too, to explain the preconditions;
1673 esp Note [FamInstEnv]
1674]
1675[Fix a looping bug in the new occur-check code
1676simonpj@microsoft.com**20101007084104
1677 Ignore-this: a02a2deafb9ec986ef1565f4596049ed
1678]
1679[Fix test T4235 with -O
1680simonpj@microsoft.com**20101006155223
1681 Ignore-this: f0fa0fe2f0c493e362d528d71f7a64e1
1682 
1683 The tag2Enum rule wasn't doing the right thing for
1684 enumerations with a phantom type parameter, like
1685    data T a = A | B
1686]
1687[Make warning-free
1688simonpj@microsoft.com**20101006155033
1689 Ignore-this: 221a3c95a6079c6ecc0468996a38b048
1690]
1691[Major bugfixing pass through the type checker
1692dimitris@microsoft.com**20101006152854]
1693[Typechecker performance fixes and flatten skolem bugfixing
1694dimitris@microsoft.com**20101004130200
1695 Ignore-this: 86721ba3f09479c146a0710796b43459
1696]
1697[Performance bug fixes
1698dimitris@microsoft.com**20100923143918]
1699[Fix Trac #4371: matching of view patterns
1700simonpj@microsoft.com**20101006115316
1701 Ignore-this: 494b28b91f1e6392b2f1521cda0e83b1
1702]
1703[Remove unused NoMatchContext construtor
1704simonpj@microsoft.com**20101006115251
1705 Ignore-this: 8985ff1dac51fb652bd65657a630a792
1706]
1707[Refactoring: mainly rename ic_env_tvs to ic_untch
1708simonpj@microsoft.com**20101006102830
1709 Ignore-this: 32999403a3f447e14b59cec7896027ff
1710 
1711 Plus remember to zonk the free_tvs in TcUnify.newImplication
1712]
1713[remove unnecessary/broken definition of mask_
1714Simon Marlow <marlowsd@gmail.com>**20101002195118
1715 Ignore-this: 4cdc9c95d40e01cdfe2d0ac411476603
1716]
1717[-fwarn-tabs: add "Warning" to the message
1718Simon Marlow <marlowsd@gmail.com>**20101002195100
1719 Ignore-this: 589a36daa3426ab51f2fb140e38df6c
1720]
1721[give a better error message in the non-threaded RTS for out-of-range FDs
1722Simon Marlow <marlowsd@gmail.com>**20100929212916
1723 Ignore-this: e94c9f390b8f79d24895a80f9d16c8d9
1724 
1725 # ./aw
1726 aw: file descriptor 1027 out of range for select (0--1024).
1727 Recompile with -threaded to work around this.
1728]
1729[Fix a very rare crash in GHCi
1730Simon Marlow <marlowsd@gmail.com>**20101005144735
1731 Ignore-this: dad1cd08934bae2ba47e72c0c000acfa
1732 When a BCO with a zero-length bitmap was right at the edge of
1733 allocated memory, we were reading a word of non-existent memory.
1734 
1735 This showed up as a segfault in T789(ghci) for me, but the crash was
1736 extremely sensitive and went away with most changes.
1737 
1738 Also, optimised scavenge_large_bitmap a bit while I was in there.
1739]
1740[Using 'stdcall' when it is not supported is only a warning now (#3336)
1741Simon Marlow <marlowsd@gmail.com>**20100924152445
1742 Ignore-this: 66c5903a600a47485a7583535bb38455
1743]
1744[remove unnecessary stg_noForceIO (#3508)
1745Simon Marlow <marlowsd@gmail.com>**20100924150202
1746 Ignore-this: dec52de9cd9da7dcedae12b20691aba9
1747]
1748[Replace an outputStr with putStrLn calls; fixes #4332
1749Ian Lynagh <igloo@earth.li>**20101003125707
1750 Ignore-this: eb4b5f60d9d9d3f7dc203869927b28ba
1751]
1752[make test and fulltest targets in the main Makefile; fixes #4297
1753Ian Lynagh <igloo@earth.li>**20100930224741
1754 You can now run "make test" in the root, and the fast testsuite will be
1755 run with cleaning enabled. It will also put the summary in
1756 testsuite_summary.txt.
1757]
1758[Don't show the loaded packages in ":show packages"; fixes #4300
1759Ian Lynagh <igloo@earth.li>**20100930210128
1760 It's never worked properly, and the information is in ":show linker".
1761]
1762[Handle EXTRA_LIBRARIES when building programs
1763Ian Lynagh <igloo@earth.li>**20100930192552
1764 Ignore-this: 401a26e18d25dcaee010b13eaed8f011
1765 And set hp2ps's EXTRA_LIBRARIES. Based on a patch from Sergei Trofimovich.
1766]
1767[Fix the doc directory on Windows
1768Ian Lynagh <igloo@earth.li>**20100929133328]
1769[Remove an unused import on Windows
1770Ian Lynagh <igloo@earth.li>**20100929000024
1771 Ignore-this: 2899e0e5a47122e637fb5c8aa0df52ab
1772]
1773[Use showCommandForUser when showing tracing commands
1774Ian Lynagh <igloo@earth.li>**20100928235844
1775 Ignore-this: 8a4a9c9f8a8029e708c4297b096b6ef1
1776]
1777[Fix hsc2hs docs: 'gcc' is now the default compiler, not 'ghc'; fixes #4341
1778Ian Lynagh <igloo@earth.li>**20100928201938]
1779[Use an empty signal handler for SIGPIPE instead of SIG_IGN
1780Simon Marlow <marlowsd@gmail.com>**20100925193548
1781 Ignore-this: b4dc5de32740a7c5fd8fe4b3bfb1300f
1782 
1783 This is so that the SIGPIPE handler gets reset to the default
1784 automatically on exec().
1785]
1786[Fix the TH deps
1787Ian Lynagh <igloo@earth.li>**20100925210029
1788 Ignore-this: 32b832301a3625d4ba70f84c5c4f94d2
1789]
1790[Check inplace doesn't exist before we try to create it
1791Ian Lynagh <igloo@earth.li>**20100924191858
1792 This fixes rerunning configure in a tree which already has an inplace
1793 directory. Edward Z Yang ran into this; I guess whether it actually
1794 fails depends on details of your installation, or we'd have run into
1795 it sooner.
1796]
1797[Fix an egregious bug: INLINE pragmas on monomorphic Ids were being ignored
1798simonpj@microsoft.com**20100924155815
1799 Ignore-this: 38c6eec6710a92df7662a55fc5132c15
1800 
1801 I had do to some refactoring to make this work nicely
1802 but now it does. I can't think how this escaped our
1803 attention for so long!
1804]
1805[Eta expand only lambdas that bind a non-dictionary Id
1806simonpj@microsoft.com**20100924155707
1807 Ignore-this: 7cc265eaf6c0bb3fa12eb311d92594ac
1808 
1809 See Note [When to eta expand]. The idea is that dictionary
1810 lambdas are invisible to the user, so we shouldn't eta
1811 expand them.
1812]
1813[Add a comment
1814simonpj@microsoft.com**20100924155620
1815 Ignore-this: de210a1afdd40328824803e1d77b4d7f
1816]
1817[Add a debug print
1818simonpj@microsoft.com**20100924155614
1819 Ignore-this: 1a58b6d297fc77d6ded8eec7ea9f895d
1820]
1821[Just moving comments around
1822simonpj@microsoft.com**20100924155600
1823 Ignore-this: 96635b8e8c9d88b50d82938568152ef8
1824]
1825[use putStrLn instead of Haskeline's outputStrLn
1826Simon Marlow <marlowsd@gmail.com>**20100924133154
1827 Ignore-this: 7581ae11714a9a52e78ba098c3c216b3
1828 use of the latter caused problems for Claus Reinke's macros that
1829 redirect stdout.
1830]
1831[Change "OPTIONS" to "OPTIONS_GHC" in error messages; fixes #4327
1832Ian Lynagh <igloo@earth.li>**20100924120423
1833 Ignore-this: 1697c83a5c346db640c0a2e22c69ff55
1834]
1835[Add deps for TH uses in vector
1836Ian Lynagh <igloo@earth.li>**20100923220244
1837 Ignore-this: 54c3386b1c268821fcdd34b84bc8c6a4
1838]
1839[Bump Cabal dep
1840Ian Lynagh <igloo@earth.li>**20100923143241]
1841[Update Cabal's version number
1842Ian Lynagh <igloo@earth.li>**20100923141719]
1843[Build primitive with stage2
1844Ian Lynagh <igloo@earth.li>**20100923140525
1845 Ignore-this: 110a819b78a57629a7edf1d4facdc191
1846]
1847[Fix the Windows __chkstk build error (missing Linker symbol)
1848Simon Marlow <marlowsd@gmail.com>**20100924113837
1849 Ignore-this: 48f0907bb1bd5eaa0730b94a6bd94ea
1850]
1851[emit a helpful error message for missing DPH packages
1852Simon Marlow <marlowsd@gmail.com>**20100923141957
1853 Ignore-this: 55ff2ee90c94524e023e014243bfe5df
1854]
1855[Fix computation of installed packages
1856simonpj@microsoft.com**20100924084737
1857 Ignore-this: a597d2fa8be5135ba8ead6d2624b3d71
1858 
1859 This is a follow-on to Simon's patch yesterday, developed
1860 with him.  It cleans up the computation of how packages
1861 are installed, and installs the right ones.
1862]
1863[Fix braino in WwLib/Literal patch
1864simonpj@microsoft.com**20100924070914
1865 Ignore-this: f6eb3a42e10f8aa7920de541cdfe76d8
1866]
1867[For now, switch off incomplete-pattern warnings in containers
1868simonpj@microsoft.com**20100923130117
1869 Ignore-this: 7ffa58567f7a33aafe256492999da325
1870 
1871 Put it back on when my patch is applied to the containers repo.
1872 (the one that removes two refuable lambdas)
1873]
1874[Make -funfolding-dict-threshold work properly
1875simonpj@microsoft.com**20100923130032
1876 Ignore-this: 417788f5b09d1d624f6b6371852c80c7
1877 
1878 and increase its default value. This makes overloaded functions
1879 a bit keener to inline.  Which fixes Trac #4321
1880]
1881[Impredicative types is no longer deprecated
1882simonpj@microsoft.com**20100923125910
1883 Ignore-this: 2bbaeb38b5e8424551677c0add627683
1884]
1885[Do not make FunctionalDependencies force MonoLocalBinds
1886simonpj@microsoft.com**20100923125900
1887 Ignore-this: f4ae1fd07c87ec14f60bdfe3863ba7a9
1888]
1889[move CHECKED settings to the right place
1890Simon Marlow <marlowsd@gmail.com>**20100923123558
1891 Ignore-this: e00a0eb5855463cc9b953670b3bbf211
1892]
1893[turn off -Werror for primitive and vector
1894Simon Marlow <marlowsd@gmail.com>**20100923122055
1895 Ignore-this: 54d7b80f3f893385e1c3ef431e2a8a7b
1896]
1897[Add primitive and vector packages for DPH support
1898Simon Marlow <marlowsd@gmail.com>**20100923104542
1899 Ignore-this: c070d015385b0a0797394132dcbb7670
1900 DPH is now using the public vector package instead of its internal
1901 version.
1902 
1903 vector and primitive are not "boot" packages; they aren't required to
1904 build GHC, but they are required to validate (because we include DPH
1905 when validating).
1906 
1907 If you say './darcs-all get --no-dph' then you don't get DPH, vector,
1908 or primitive.
1909]
1910[Refactoring and tidy up in the build system
1911Simon Marlow <marlowsd@gmail.com>**20100923095642
1912 Ignore-this: f7bf3a1fd160149d89b26f464b064fb1
1913 
1914 Instead of the ghc-stage and ghc-stage2-package files in a package, we
1915 now have a list of these in ghc.mk.  There are other similar lists (of
1916 boot-packages and non-installable packages), so this is not too bad,
1917 and is simpler.
1918 
1919 While poking around in the top-level ghc.mk file I spotted various
1920 opportunities to clean up and re-order some of the cruft that has
1921 accumulated over time.
1922]
1923[Allow absent State# RealWorld arguments
1924simonpj@microsoft.com**20100923111356
1925 Ignore-this: c2d57633dec0293ebe6723ea3a4bb5df
1926]
1927[Add notSCCNote, and use it
1928simonpj@microsoft.com**20100923105949
1929 Ignore-this: c8cc758656558a7f366bf784d75f0304
1930 
1931 The point here is that SCCs get in the way of eta
1932 expansion and we must treat them uniformly.
1933]
1934[Remove use of lambda with a refutable pattern
1935simonpj@microsoft.com**20100923105901
1936 Ignore-this: d7d48b94e5744717a838591a1cc79cf0
1937]
1938[Avoid ASSERT black hole
1939simonpj@microsoft.com**20100923105820
1940 Ignore-this: 5419d450871be22c8781ac3f0f40d76a
1941 
1942 When this ASSERT tripped in CoreToStg it tried to print out
1943 too much, which tripped the asssertion again.  Result: an
1944 infinite loop with no output at all.  Hard to debug!
1945]
1946[Rejig the absent-arg stuff for unlifted types
1947simonpj@microsoft.com**20100923105732
1948 Ignore-this: 69daa35816b948b0c4d259c73a5e928e
1949 
1950 This is what was giving the "absent entered" messages
1951 See Note [Absent errors] in WwLib.  We now return a
1952 suitable literal for absent values of unlifted type.
1953]
1954[Remove -fwarn-simple-patterns, and make -fwarn-incomplete-patterns include lambdas
1955simonpj@microsoft.com**20100922133934
1956 Ignore-this: e851a2fb0377e10c28c506f0bf14cc85
1957 
1958 This makes
1959      \(x:xs) -> e
1960 want when you have -fwarn-incomplete-patterns, which is consistent.
1961]
1962[Get rid of non-exhaustive lambda
1963simonpj@microsoft.com**20100922133801
1964 Ignore-this: 748b2d5b43b02b6591b81abe7c105cd6
1965]
1966[Fix an ASSERT failure with profiling
1967simonpj@microsoft.com**20100922133741
1968 Ignore-this: 170b2e94d6ee8cc7444cc4bb515328a0
1969 
1970 The problem arose with this kind of thing
1971 
1972    x = (,) (scc "blah" Nothing)
1973 
1974 Then 'x' is marked NoCafRefs by CoreTidy, becuase it has
1975 arity 1, and doesn't mention any caffy things.
1976 
1977 That in turns means that CorePrep must not float out the
1978 sat binding to give
1979 
1980   sat = scc "blah" Nothing
1981   x = (,) sat
1982 
1983 Rather we must generate
1984 
1985   x = \eta. let sat = scc "blah" Nothing
1986             in (,) sat eta
1987 
1988 URGH! This Caf stuff is such a mess.
1989]
1990[Remove an out of date paragraph from the user guide; fixes #4331
1991Ian Lynagh <igloo@earth.li>**20100922225239]
1992[Fix bindisttest when GhcProfiled = YES
1993Ian Lynagh <igloo@earth.li>**20100921222634
1994 Ignore-this: 47c620fd6bec745e3eb699d9f53441d8
1995]
1996[Fixes for when HADDOCK_DOCS=NO
1997Ian Lynagh <igloo@earth.li>**20100921213916
1998 Ignore-this: e0e069555c6db9d01a8ac70ba4dde591
1999]
2000[Bump version to 7.1
2001Ian Lynagh <igloo@earth.li>**20100921195935
2002 Ignore-this: 4563987e6885d5ef55995ec0fa0d5ae8
2003]
2004[Don't use -march=i686 on powerpc-apple-darwin
2005Ian Lynagh <igloo@earth.li>**20100921193721
2006 Thorikil ran into this when doing a PPC OS X build. We now also don't
2007 use -m32 on PPC/OSX, but I don't think it should be necessary. We can
2008 add it back if it does turn out to be.
2009]
2010[add a simple trace facility to the build system
2011Simon Marlow <marlowsd@gmail.com>**20100921134729
2012 Ignore-this: d23ea2d62a648d0711b4b07d98e1b79f
2013 
2014 saying
2015 
2016   make TRACE=1
2017 
2018 prints most of the macro calls and their arguments.  It's easy to
2019 trace new macros; see rules/trace.mk.
2020]
2021[fix building with extra packages (packages were added to BUILD_DIRS twice)
2022Simon Marlow <marlowsd@gmail.com>**20100921100153
2023 Ignore-this: 4b425dff9777871ad5ba3e05e1d14483
2024 Also add some comments about what extra-packages is doing
2025]
2026[add extra packages to $(EXTRA_PACKAGES), so we avoid installing them by default
2027Simon Marlow <marlowsd@gmail.com>**20100920144307
2028 Ignore-this: 3395825d911a8bf7ba8385518d8b517b
2029]
2030[Fix indexing error in archive loader
2031Ian Lynagh <igloo@earth.li>**20100921121642]
2032[Add some -Dl belches
2033Ian Lynagh <igloo@earth.li>**20100921121624]
2034[Add casts to fix warnings
2035Ian Lynagh <igloo@earth.li>**20100921121714]
2036[Add support for BSD-variant large filenames in .a archives
2037Ian Lynagh <igloo@earth.li>**20100921000451]
2038[Tell Cabal that we're not building GHCi libs if UseArchivesForGhci=YES
2039Ian Lynagh <igloo@earth.li>**20100920230449]
2040["UseArchivesForGhci = YES" on darwin
2041Ian Lynagh <igloo@earth.li>**20100920211538]
2042[Add a dependency that my OS X build has recently started tripping up over
2043Ian Lynagh <igloo@earth.li>**20100920210239]
2044[Add "Use archives for ghci" to --info output
2045Ian Lynagh <igloo@earth.li>**20100920210523]
2046[Implement archive loading for ghci
2047Ian Lynagh <igloo@earth.li>**20100920201620]
2048[Tweak gen_contents_index now dph may not be there
2049Ian Lynagh <igloo@earth.li>**20100920201513]
2050[Filter out the FFI library when loading package in ghci
2051Ian Lynagh <igloo@earth.li>**20100920181032
2052 The FFI GHCi import lib isn't needed as
2053 compiler/ghci/Linker.lhs + rts/Linker.c link the
2054 interpreted references to FFI to the compiled FFI.
2055 We therefore filter it out so that we don't get
2056 duplicate symbol errors.
2057]
2058[Loosen the conditions for -XUndecidableInstances; fixes Trac #4200
2059simonpj@microsoft.com**20100919162623
2060 Ignore-this: 2f4323e278b1ce9250549727ffd0aa1b
2061]
2062[Further improvements in error messages
2063simonpj@microsoft.com**20100919153355
2064 Ignore-this: b6fa0b11ae893df1a3ca68f78e427fa
2065]
2066[Add a flag -fwarn-missing-local-sigs, and improve -fwarn-mising-signatures
2067simonpj@microsoft.com**20100919153327
2068 Ignore-this: fda8dfca450054ea692be0ee30b01885
2069 
2070 The new flag prints out a warning if you have a local,
2071 polymorphic binding that lacks a type signature. It's meant
2072 to help with the transition to the new typechecker, which
2073 discourages local let-generalisation.
2074 
2075 At the same time I moved the missing-signature code to TcHsSyn,
2076 where it takes place as part of zonking.  That way the
2077 types are reported after all typechecking is complete,
2078 thereby fixing Trac #3696.  (It's even more important for
2079 local bindings, which is why I made the change.)
2080]
2081[Include the "stupid theta" in the type of $con2tag
2082simonpj@microsoft.com**20100919152201
2083 Ignore-this: d95fae78a0e66f48bbd5862573a11f4d
2084]
2085[Add a release note about the typechecker
2086Ian Lynagh <igloo@earth.li>**20100919132927]
2087[Enable shared libs on OpenBSD
2088Matthias Kilian <kili@outback.escape.de>**20100918205040
2089 Ignore-this: 729dd7ac0bba5d758f43bc31b541896
2090]
2091[Add separate functions for querying DynFlag and ExtensionFlag options
2092Ian Lynagh <igloo@earth.li>**20100918163815
2093 and remove the temporary DOpt class workaround.
2094]
2095[Fix mkUserGuidePart deps
2096Ian Lynagh <igloo@earth.li>**20100918145904
2097 We need to directly depend on the stage1 libs. The stage1 compiler lib
2098 doesn't depend on them.
2099]
2100[Fix build on cygwin: Normalise slashes in .depend files to be /
2101Ian Lynagh <igloo@earth.li>**20100918132328
2102 Ignore-this: 664f5ef4a41a4461eb34fe2ca7f2729a
2103]
2104[extra packages info is now read from packages file
2105Ian Lynagh <igloo@earth.li>**20100917224409
2106 rather than being repeated in the build system
2107]
2108[Tweak darcs-all
2109Ian Lynagh <igloo@earth.li>**20100917194435]
2110[Bump dependencies
2111Ian Lynagh <igloo@earth.li>**20100917183609]
2112[Library release notes for 7.0.1
2113Ian Lynagh <igloo@earth.li>**20100917174850]
2114[Fix overriding of implicit parameters in the solver
2115simonpj@microsoft.com**20100917140403
2116 Ignore-this: af76732309c7e2ca6b04f49327e9c14b
2117]
2118[Minor type printing amomaly
2119simonpj@microsoft.com**20100917140204
2120 Ignore-this: c90cb2e51421b4543a827e096051772e
2121]
2122[Spaces only
2123simonpj@microsoft.com**20100917140156
2124 Ignore-this: 7e34479502f7cb87d762355e40cbd012
2125]
2126[Minor refactoring
2127simonpj@microsoft.com**20100917140150
2128 Ignore-this: 6c0648b949b91b7e2f23c136b124faf2
2129]
2130[Add types of implicit parameters as untouchable
2131simonpj@microsoft.com**20100917140138
2132 Ignore-this: ba80740a557a9ba062dc7756e2320d17
2133 
2134 This is a tricky point:
2135    see Note [Implicit parameter untouchables]
2136]
2137[Better pretty printing of implicit parameters
2138simonpj@microsoft.com**20100917140054
2139 Ignore-this: 867dd67818a5bd687b2b6a1b59e15775
2140]
2141[Yet more error message improvement
2142simonpj@microsoft.com**20100917121206
2143 Ignore-this: 647fe8129d1d39d81e8249debd8df94e
2144]
2145[More error message wibbles
2146simonpj@microsoft.com**20100917094721
2147 Ignore-this: 8ec2f150b96b26af2e9ab7ac2b371fc7
2148]
2149[More error refactoring
2150simonpj@microsoft.com**20100917092834
2151 Ignore-this: 2d570ac0b9cc11305ddd33d093d11324
2152]
2153[Refactor type errors a bit
2154simonpj@microsoft.com**20100917080726
2155 Ignore-this: 33da4549373f585064e2ee22b50ad6ac
2156 
2157 Improves kind error messages in paticular
2158]
2159[Fix a very subtle shadowing bug in optCoercion
2160simonpj@microsoft.com**20100916170452
2161 Ignore-this: 9041cfb3c93e27a5e644e57815313aae
2162 
2163 See Note [Subtle shadowing in coercions]
2164 
2165 This is what was going wrong in Trac 4160.
2166]
2167[Fix bad error in tyVarsOfType
2168simonpj@microsoft.com**20100916170348
2169 Ignore-this: 67c8ce96a668cf6e3a38b82c893bcd81
2170 
2171 We weren't gathering the type variables free in the kind
2172 of a coercion binder!
2173]
2174[More assertions
2175simonpj@microsoft.com**20100916170310
2176 Ignore-this: 7fdcb53c99d791621a3d7e01ef454404
2177]
2178[Add more location info in CoreLint
2179simonpj@microsoft.com**20100916170229
2180 Ignore-this: 6558bab544b4f30189e0430668db87c3
2181]
2182[Print coercion variables as such (debugging change only)
2183simonpj@microsoft.com**20100916165944
2184 Ignore-this: c6d2001c1d8279a2288cb63bc339577d
2185]
2186[Remove pprTrace
2187simonpj@microsoft.com**20100915225935
2188 Ignore-this: 28185bbfa9732386f3c0f3eb4781a637
2189]
2190[Remove dead code dealing with type refinement
2191simonpj@microsoft.com**20100915223230
2192 Ignore-this: 62824b5c2ec1077c7642163352559621
2193]
2194[Use mkAppTy
2195simonpj@microsoft.com**20100915223205
2196 Ignore-this: e79e087b6a49219e9088846a1253a153
2197 
2198 Using AppTy in CoreLint was giving a bogus Lint failure
2199]
2200[Comments only
2201simonpj@microsoft.com**20100915221253
2202 Ignore-this: 3a45ea614188ccbb4a462de5cac96eda
2203]
2204[Extend eta reduction to work with casted arguments
2205simonpj@microsoft.com**20100915221229
2206 Ignore-this: 24b103dcdf70331211507af929789f86
2207 
2208 See Trac #4201, and
2209 Note [Eta reduction with casted arguments]
2210 
2211 Thanks to Louis Wasserman for suggesting this, and
2212 implementing an early version of the patch
2213]
2214[Allow "INLINEABLE" as a synonym
2215simonpj@microsoft.com**20100915154249
2216 Ignore-this: f41f80cb769e9acd5b463b170df698d0
2217]
2218[Documentation for INLINABLE
2219simonpj@microsoft.com**20100915154235
2220 Ignore-this: f942c02bcadc0d2d2f05b9369f93e280
2221]
2222[Implement TH reification of instances (Trac #1835)
2223simonpj@microsoft.com**20100915151242
2224 Ignore-this: 97dfa83db7da8f6cbd1b96801a57f8c5
2225 
2226 Accompanying patch for template-haskell package is reqd
2227]
2228[errno corresponding to ERROR_NO_DATA should be EPIPE (non-threaded RTS)
2229Simon Marlow <marlowsd@gmail.com>**20100915141809
2230 Ignore-this: 709c7280fbaa762e7071fb8796e8c01e
2231]
2232[Windows: use a thread-local variable for myTask()
2233Simon Marlow <marlowsd@gmail.com>**20100915120627
2234 Ignore-this: 13ffa4f19ebd319fe672af53af8d0b9a
2235 Which entailed fixing an incorrect #ifdef in Task.c
2236]
2237[Fix typo
2238Ian Lynagh <igloo@earth.li>**20100915140814]
2239[Add quotes in error message
2240simonpj@microsoft.com**20100915144724
2241 Ignore-this: c5158047c0aa41947a79e4c8edbe54f4
2242]
2243[Fix isDefaultInlinePragma
2244simonpj@microsoft.com**20100915144710
2245 Ignore-this: c9addf6bf811b23dc12603cf8521aa6c
2246]
2247[Implement INLINABLE pragma
2248simonpj@microsoft.com**20100915124442
2249 Ignore-this: 80a4ab2c2d65b27868dc9b2e954d6c6f
2250 
2251 Implements Trac #4299.  Documentation to come.
2252]
2253[Less voluminous error when derived code doesn't typecheck
2254simonpj@microsoft.com**20100915072301
2255 Ignore-this: eca7871dcc50c1070a0b530711adea27
2256]
2257[Improve pretty-printing of family instances
2258simonpj@microsoft.com**20100915123219
2259 Ignore-this: 25ec6bcc7e8a7f7c303b38ca201db90e
2260 
2261 Fixed Trac #4246
2262]
2263[Fix Trac #4240: -ddump-minimal-imports
2264simonpj@microsoft.com**20100915121937
2265 Ignore-this: ab85057cb829a42ea44a92f7b4af24a3
2266 
2267 See Note [Partial export] for the details.
2268 I also fixed one egregious bug that was just
2269 waiting to bite: we were using loadSysInterface
2270 instead of loadSrcInterface.
2271]
2272[Comments only
2273simonpj@microsoft.com**20100915105707
2274 Ignore-this: ab3a5f16f8260b7b8570e748bf97998a
2275]
2276[implement setThreadAffinity on Windows (#1741)
2277Simon Marlow <marlowsd@gmail.com>**20100914155844
2278 Ignore-this: a14c7b4ef812007042342d0a25478f0b
2279]
2280[COFF: cope with new debug sections in gcc 4.x (fixes ghciprog004)
2281Simon Marlow <marlowsd@gmail.com>**20100914153026
2282 Ignore-this: f340e40a2b0390836bc61bba144a04ed
2283 Also updated the object file parser to properly handle the overflow
2284 case for section names longer than 8 chars.
2285]
2286[eliminate clutter from make output
2287Simon Marlow <marlowsd@gmail.com>**20100915105712
2288 Ignore-this: bfa4480dd239dda2a02ac391b6a9219c
2289]
2290[rts_isProfiled should be a visible API (fixes T2615(dyn))
2291Simon Marlow <marlowsd@gmail.com>**20100915083941
2292 Ignore-this: b8ac09bb9d1a929bf45c6122f8485561
2293]
2294[Fix the "lost due to fragmentation" calculation
2295Simon Marlow <marlowsd@gmail.com>**20100914145945
2296 Ignore-this: cdffcc9f3061c3a33da5171be111fc43
2297 It was counting the space used by block descriptors as "lost"
2298]
2299[fix +RTS -S output: use peak_mblocks_allocated, now that mblocks can be freed
2300Simon Marlow <marlowsd@gmail.com>**20100914135030
2301 Ignore-this: 65d21e5f86d3ab6ab4d6c255180b6968
2302]
2303[Fix egregious bug in deeplyInstantiate
2304simonpj@microsoft.com**20100915070325
2305 Ignore-this: 22ede973038877af2673339aaf5de6cf
2306 
2307 This resulted in an infinite loop in applyTypeToArgs, in syb
2308]
2309[Improve HsSyn pretty printing
2310simonpj@microsoft.com**20100915070255
2311 Ignore-this: 7c8e2d86a482453c7e69e22bc31cb03f
2312]
2313[Remove (most of) the FiniteMap wrapper
2314Ian Lynagh <igloo@earth.li>**20100914201703
2315 We still have
2316     insertList, insertListWith, deleteList
2317 which aren't in Data.Map, and
2318     foldRightWithKey
2319 which works around the fold(r)WithKey addition and deprecation.
2320]
2321[Improve ASSERT
2322simonpj@microsoft.com**20100914113900
2323 Ignore-this: dbc0363be5924f543316e77f7d18dd77
2324]
2325[Comment on what an "enumeration" type is
2326simonpj@microsoft.com**20100914113850
2327 Ignore-this: c09c8591e3140f305d55fbf945adbf95
2328]
2329[Make absent-arg wrappers work for unlifted types (fix Trac #4306)
2330simonpj@microsoft.com**20100914113827
2331 Ignore-this: 1945e56779329e8b79780403710aba98
2332 
2333 Previously we were simply passing arguments of unlifted
2334 type to a wrapper, even if they were absent, which was
2335 stupid.
2336 
2337 See Note [Absent error Id] in WwLib.
2338]
2339[Comments only
2340simonpj@microsoft.com**20100914113641
2341 Ignore-this: 3191ce856c9b5d9700cedc9b149b8097
2342]
2343[Move error-ids to MkCore (from PrelRules)
2344simonpj@microsoft.com**20100914113635
2345 Ignore-this: c3d820db62ba6139dd7c96bf97e51bb5
2346 
2347 and adjust imports accordingly
2348]
2349[More wibbles to deriving error messages
2350simonpj@microsoft.com**20100914113523
2351 Ignore-this: bd2df662644961138fa209aec843a2aa
2352]
2353[Fix getThreadCPUTime()
2354Simon Marlow <marlowsd@gmail.com>**20100913153838
2355 Ignore-this: 950e048a5724086534b74c609c7d5ed
2356 ever since the patch "Check with sysconf _POSIX_THREAD_CPUTIME", it
2357 has been returning incorrect results, because the sysconf variable to
2358 check should have been _SC_THREAD_CPUTIME, not _POSIX_THREAD_CPUTIME.
2359]
2360[filter out the gcc-lib directory from the rts package's library-dirs
2361Simon Marlow <marlowsd@gmail.com>**20100913101259
2362 Ignore-this: 46dc1dccbfee8a65f9243e125eee117f
2363 fixes problems when building with GHC 6.10 on Windows
2364]
2365[Don't include GC time in heap profiles (#4225)
2366Simon Marlow <marlowsd@gmail.com>**20100913133852
2367 Ignore-this: 68ac48b004b311384b5996c6b33ba5cc
2368]
2369[Use clock_gettime (if available) to measure the process CPU time
2370Simon Marlow <marlowsd@gmail.com>**20100913133818
2371 Ignore-this: 8c9300df9b929bfc1db4713c9b6065b3
2372 This is much more accurate than getrusage, which was giving misleading
2373 results when trying to time very quick operations like a minor GC.
2374]
2375[make stg_arg_bitmaps public, and available via the GHCi linker (#3672)
2376Simon Marlow <marlowsd@gmail.com>**20100913105235
2377 Ignore-this: e18efd0bd77c521e5530fb59e93b5a42
2378]
2379[fix typo
2380Simon Marlow <marlowsd@gmail.com>**20100913105100
2381 Ignore-this: 6049eea21208864203b2d79db2edd143
2382]
2383[Update release notes and docs with LLVM info.
2384David Terei <davidterei@gmail.com>**20100914072135
2385 Ignore-this: 5b3d0e5c9d5da98ed6ae9c2e8e1f6f30
2386]
2387[Remove defaultExtensionFlags
2388Ian Lynagh <igloo@earth.li>**20100913165949
2389 The default should do into languageExtensions instead
2390]
2391[Improve crash message
2392simonpj@microsoft.com**20100913170407
2393 Ignore-this: 5c26a9979f18be8cd12cea823c9f4b5a
2394]
2395[Fix Trac #4302, plus a little refactoring
2396simonpj@microsoft.com**20100913170355
2397 Ignore-this: cf6886b587aa0e8d723362183625d946
2398]
2399[Fix build with 6.10
2400Ian Lynagh <igloo@earth.li>**20100913160048]
2401[Haddock fixes
2402simonpj@microsoft.com**20100913120510
2403 Ignore-this: f3157d6969f10d4cbd593000a477138b
2404]
2405[Remove two old junk files
2406simonpj@microsoft.com**20100913103426
2407 Ignore-this: ed7af5ef1b9592178909a8d876345302
2408]
2409[Super-monster patch implementing the new typechecker -- at last
2410simonpj@microsoft.com**20100913095048
2411 Ignore-this: 14d14a1e4d7a414f5ae8d9d89d1c6a4b
2412 
2413 This major patch implements the new OutsideIn constraint solving
2414 algorithm in the typecheker, following our JFP paper "Modular type
2415 inference with local assumptions". 
2416 
2417 Done with major help from Dimitrios Vytiniotis and Brent Yorgey.
2418 
2419]
2420[Fix simplifier statistics
2421simonpj@microsoft.com**20100909085441
2422 Ignore-this: 48e383655aafc912dea15c4d94382863
2423]
2424[Trace output
2425simonpj@microsoft.com**20100908170056
2426 Ignore-this: 4b67fa4b310fbf0a16b852686d2d3294
2427]
2428[Better debug output
2429simonpj@microsoft.com**20100908170047
2430 Ignore-this: 410cef00616dda7c0c162f65216e8ca3
2431]
2432[Add Outputable instance for OccEncl
2433simonpj@microsoft.com**20100908150510
2434 Ignore-this: 6362ef9028287d84f070eaf8963c1bfc
2435]
2436[Better simplifier counting
2437simonpj@microsoft.com**20100907214840
2438 Ignore-this: 9d4722703f8f47447e86a28c8c50e0ea
2439]
2440[Put liftStringName into the known-key names
2441simonpj@microsoft.com**20100906112415
2442 Ignore-this: 287064d14ff484da1a6dea6924bc9235
2443]
2444[Deprecate NoRelaxedPolyRec
2445simonpj@microsoft.com**20100903234519
2446 Ignore-this: 607217e77f6bc1b91bf57dfd8dd2b967
2447]
2448[Buglet in Core Lint
2449simonpj@microsoft.com**20100903234457
2450 Ignore-this: 277535d51b396d3b4b0265a0939c2d4
2451]
2452[Give seqId the right type
2453simonpj@microsoft.com**20100903093556
2454 Ignore-this: d1fc7a73dea160614a8d4ddc930f99cd
2455]
2456[Remove dead code
2457simonpj@microsoft.com**20100903093548
2458 Ignore-this: 92cc3f7651445aa349ee7f114d3ec758
2459]
2460[Comments and layout
2461simonpj@microsoft.com**20100903093502
2462 Ignore-this: 9987d1409e654992c1cb1be35cb87728
2463]
2464[Remove checkFreeness; no longer needed
2465simonpj@microsoft.com**20100902233227
2466 Ignore-this: c96a12ac9794290aa30402317d88c095
2467]
2468[Assert
2469simonpj@microsoft.com**20100902073642
2470 Ignore-this: 4be1ab2f6096665ae5ec7fdd1f025a67
2471]
2472[Add aserts
2473simonpj@microsoft.com**20100902073211
2474 Ignore-this: e1409441217fd070c5a7f9ee4cca99ab
2475]
2476[Wibbles
2477simonpj@microsoft.com**20100831113540
2478 Ignore-this: 903811ab493a7b560a62eb86fcf3ee25
2479]
2480[Wibble to allow phantom types in Enum
2481simonpj@microsoft.com**20100825112711
2482 Ignore-this: fdef1c50d92b4a3d46bbe4cbfd8a83ea
2483]
2484[Add HsCoreTy to HsType
2485simonpj@microsoft.com**20100824141845
2486 Ignore-this: 4ca742b099f9cc90af3167f1012dbba6
2487 
2488 The main thing here is to allow us to provide type
2489 signatures for 'deriving' bindings without pain.
2490]
2491[Comments
2492simonpj@microsoft.com**20100823223654
2493 Ignore-this: dd412a55839430c436902d8699d6900b
2494]
2495[Wibbles to error message
2496simonpj@microsoft.com**20100823163308
2497 Ignore-this: 4d6cd8e613762dca8135c2e3b09264ec
2498]
2499[Correct type signatures
2500simonpj@microsoft.com**20100823153045
2501 Ignore-this: 42942309221a443258246098f9c0a13b
2502]
2503[Add missing signatures
2504simonpj@microsoft.com**20100823112413
2505 Ignore-this: 8ee1ce40456306de469938c02df4fed5
2506]
2507[Add type signatures in "deriving" bindings
2508simonpj@microsoft.com**20100820234230
2509 Ignore-this: 4726b28968cf65ec16cb65b7e0e7303e
2510]
2511[Minor
2512dimitris@microsoft.com**20100820131021]
2513[Be a bit less aggressive in mark-many inside a cast
2514simonpj@microsoft.com**20100819104804
2515 Ignore-this: 3fd48fe7647ec7a58c2032cd86ca4d4f
2516]
2517[Wibble
2518simonpj@microsoft.com**20100818185738
2519 Ignore-this: d5c939311377c0d0c7244aa339193315
2520]
2521[Pretty printing change
2522simonpj@microsoft.com**20100818065436
2523 Ignore-this: 4f7e70976dbe52f95effb3e634dfef5d
2524]
2525[Remember to zonk FlatSkols!
2526simonpj@microsoft.com**20100811143555
2527 Ignore-this: 84f7f9dbda97f561a918c69308ddef9a
2528]
2529[De-polymorphise
2530simonpj@microsoft.com**20100730151217
2531 Ignore-this: a9304487b983e517a9083fd697f77576
2532]
2533[Work around missing type signature in Happy
2534simonpj@microsoft.com**20100730122405
2535 Ignore-this: 7f241a655d93c5ad7763a7ffe8db0c7a
2536 
2537 Happy generates
2538 
2539       notHappyAtAll = error "Blah"
2540 
2541 without a type signature, and currently the new
2542 typechecker doesn't generalise it.  This patch
2543 says "no monomorphism restriction" which makes it
2544 generalise again.
2545 
2546 Better would be to add a type sig to Happy's template
2547]
2548[Add two local type signatures
2549simonpj@microsoft.com**20100729152611
2550 Ignore-this: afa99bcc515469aa0990d44d8c18a9e6
2551]
2552[Second test from Simon's laptop
2553simonpj@microsoft.com**20100729091703
2554 Ignore-this: 4dc64cadae314a5a1b05cc5326918a83
2555]
2556[Test commit from Simon's laptop
2557simonpj@microsoft.com**20100729091344
2558 Ignore-this: 109eff835cc19e9f93799d12f09b0ba7
2559]
2560[Add OutsideIn flag
2561simonpj@microsoft.com**20100728075525
2562 Ignore-this: 69c2f5c3a15fa653f6da80598aa8d74d
2563]
2564[Layout only
2565simonpj@microsoft.com**20100727141539
2566 Ignore-this: 1a58a8fe80ba8bced18ae81a2efb9495
2567]
2568[Improvement to SimplUtils.mkLam
2569simonpj@microsoft.com**20100727131659
2570 Ignore-this: 739beaefa79baa7e0ebeb5b2b6d1ea91
2571]
2572[Give the correct kind to unsafeCoerce#
2573simonpj@microsoft.com**20100727131538
2574 Ignore-this: 6b787de3b398c6d7a61fa04fccd15fd6
2575]
2576[Suppress warnings about recursive INLINE in output of desugarer
2577simonpj@microsoft.com**20100727094549
2578 Ignore-this: a361f7238c0fcba526d46326722c42e
2579]
2580[Rename CorePrep.tryEtaReduce to tryEtaReducePrep
2581simonpj@microsoft.com**20100726231253
2582 Ignore-this: 4375ddace205745244ba224ae012252
2583 
2584 This avoids the name clash with the similar but
2585 not identical CoreUtils.tryEtaReduce
2586]
2587[Add a trace message
2588simonpj@microsoft.com**20100719211111
2589 Ignore-this: b5daebe46e50c8cf28cc693f84bbf099
2590]
2591[Don't use RelaxedPolyRec in the compiler; it's built in now
2592simonpj@microsoft.com**20100719170441
2593 Ignore-this: a2e4489cdf63478e46282a421ee7aec3
2594]
2595[Remove duplicated #defines for FreeBSD
2596Matthias Kilian <kili@outback.escape.de>**20100912181518
2597 Ignore-this: d16214fef8635c7c9ef4edec4e8e7896
2598]
2599[Don't fail with absolute silence
2600Matthias Kilian <kili@outback.escape.de>**20100912150506
2601 Ignore-this: 479e2321f39b263fa2d9f80491e5e9f7
2602]
2603[Add a release note: "-dynload wrapper" removed
2604Ian Lynagh <igloo@earth.li>**20100911195809]
2605[put back the conversion of warn-lazy-unlifted-bindings into an error until 7.2
2606Ian Lynagh <igloo@earth.li>**20100911193434
2607 I think we'll currently still have too many people with old versions of
2608 alex/happy to want to make this an error now.
2609]
2610[6.14 -> 7.0
2611Ian Lynagh <igloo@earth.li>**20100911192837]
2612[Add a couple more release notes
2613Ian Lynagh <igloo@earth.li>**20100911162059]
2614[Document -dsuppress-module-prefixes
2615Ian Lynagh <igloo@earth.li>**20100911162005]
2616[Enable -fregs-graph with -O2; fixes #2790
2617Ian Lynagh <igloo@earth.li>**20100910191301]
2618[Remove unused code
2619Ian Lynagh <igloo@earth.li>**20100909170207]
2620[Fix warnings
2621Ian Lynagh <igloo@earth.li>**20100909154348]
2622[Fix warnings
2623Ian Lynagh <igloo@earth.li>**20100909150957]
2624[Remove context completion
2625lykahb@gmail.com**20100901160153
2626 Ignore-this: dc61b259dcb7063f0c76f56788b5d2af
2627 Now completion suggests to remove only modules added to context before.
2628]
2629[avoid Foreign.unsafePerformIO
2630Ross Paterson <ross@soi.city.ac.uk>**20100909125531
2631 Ignore-this: 5cabeae4cffec8fc17ef7c0cabbea22a
2632]
2633[updates to the release notes
2634Simon Marlow <marlowsd@gmail.com>**20100909111450
2635 Ignore-this: a4d25ad8815c305b7e0f21fd4f6ee37b
2636]
2637[newAlignedPinnedByteArray#: avoid allocating an extra word sometimes
2638Simon Marlow <marlowsd@gmail.com>**20100909110805
2639 Ignore-this: 996a3c0460068ab2835b4920905b3e75
2640]
2641[Finish breaking up vectoriser utils
2642benl@ouroborus.net**20100909061311
2643 Ignore-this: 217fe1d58a3e8bb13200bcb81353a416
2644]
2645[Move VectType module to Vectorise tree
2646benl@ouroborus.net**20100909042451
2647 Ignore-this: 5af8cf394d4835911259ca3ffb6774c5
2648]
2649[Sort all the PADict/PData/PRDict/PRepr stuff into their own modules
2650benl@ouroborus.net**20100909035147
2651 Ignore-this: 53436329773347cad793adbd83e90a9e
2652]
2653[Break out Repr and PADict stuff for vectorisation of ADTs to their own modules
2654benl@ouroborus.net**20100909025759
2655 Ignore-this: d2b7d2f79332eda13416449742f7cf1c
2656]
2657[Break out conversion functions to own module
2658benl@ouroborus.net**20100909023332
2659 Ignore-this: 613f2666b6ca7f2f8876fcc1e4a59593
2660]
2661[Comments and formatting only
2662benl@ouroborus.net**20100909022117
2663 Ignore-this: c8e30139d730669e5db44f0ef491a588
2664]
2665[Remove "-dynload wrapper"; fixes trac #4275
2666Ian Lynagh <igloo@earth.li>**20100908213251]
2667[Don't set visibility on Windows
2668Ian Lynagh <igloo@earth.li>**20100905122442
2669 With gcc 4.5.0-1, using visibility hidden gives:
2670     error: visibility attribute not supported in this configuration; ignored
2671]
2672[Fix warnings on Windows
2673Ian Lynagh <igloo@earth.li>**20100905111201
2674 Ignore-this: c5cce63bb1e0c7a27271bed78d25fbc5
2675]
2676[Fix gcc wrapper for new mingw binaries
2677Ian Lynagh <igloo@earth.li>**20100905001807
2678 Ignore-this: f6acc8c911055ffce632bac138ccc939
2679]
2680[Don't pass our gcc options to stage0 ghc's gcc; they may not be suitable
2681Ian Lynagh <igloo@earth.li>**20100905103129]
2682[Update intree-mingw creation
2683Ian Lynagh <igloo@earth.li>**20100904225559]
2684[Update commands to build in-tree mingw
2685Ian Lynagh <igloo@earth.li>**20100904215112]
2686[Break out hoisting utils into their own module
2687benl@ouroborus.net**20100908074102
2688 Ignore-this: e3ba4ed0252a2def1ed88a9e14c58fea
2689]
2690[Break out closure utils into own module
2691benl@ouroborus.net**20100908072040
2692 Ignore-this: 216172b046ff101cf31a1753667a5383
2693]
2694[Move VectVar module to Vectorise tree
2695benl@ouroborus.net**20100908065904
2696 Ignore-this: 1fba5333d29927dba4275381e1a7f315
2697]
2698[Break out vectorisation of expressions into own module
2699benl@ouroborus.net**20100908065128
2700 Ignore-this: 6a952b80fb024b5291f166477eb1976
2701]
2702[Break out TyCon classifier into own module
2703benl@ouroborus.net**20100908063111
2704 Ignore-this: da754c4ef6960b4e152ea1bf8c04ab6f
2705]
2706[Break out vectorisation of TyConDecls into own module
2707benl@ouroborus.net**20100908052004
2708 Ignore-this: c0ab4fb2a05ca396efe348b384db1ebf
2709]
2710[Break out type vectorisation into own module
2711benl@ouroborus.net**20100907110311
2712 Ignore-this: 67bd70a21d16468daf68dd3ec1ff7d62
2713]
2714[Tidy up the ArchHasAdjustorSupport definition
2715Ian Lynagh <igloo@earth.li>**20100904144234]
2716[ppc: switch handling of 'foreign import wrapper' (FIW) to libffi
2717Sergei Trofimovich <slyfox@community.haskell.org>**20100829192859
2718 Ignore-this: 662ea926681ebea0759e2a04a38e82b7
2719 
2720 Joseph Jezak reported darcs-2.4.4 SIGSEGV in interactive mode in ghc-6.12.3.
2721 So I've concluded ppc also has rotten native adjustor. I don't have hardware
2722 to verify the patch (ticket #3516 should help to test it), but I think it will
2723 help (as similar patch helped for ia64 and ppc64).
2724]
2725[Binary no longer has unusable UNPACK pragmas, so no need to turn of -Werror
2726Ian Lynagh <igloo@earth.li>**20100904133339]
2727[Don't haddock packages that we aren't going to install
2728Ian Lynagh <igloo@earth.li>**20100903231921]
2729[Give haddock per-package source entity paths; fixes #3810
2730Ian Lynagh <igloo@earth.li>**20100903221335]
2731[update for containers-0.4
2732Simon Marlow <marlowsd@gmail.com>**20100903105131
2733 Ignore-this: 556eac0e4926c9b8af6b66d7b069302c
2734]
2735[Fix for nursery resizing: the first block's back pointer should be NULL
2736Simon Marlow <marlowsd@gmail.com>**20100827102818
2737 Ignore-this: fb68938e3f1e291e3c9e5e8047f9dcd2
2738 I'm not sure if this could lead to a crash or not, but it upsets +RTS -DS
2739 Might be related to #4265
2740]
2741[Add some -no-user-package-conf flags
2742Ian Lynagh <igloo@earth.li>**20100902224726
2743 Stops user-installed packages breaking the build
2744]
2745[Fix warnings: Remove unused imports
2746Ian Lynagh <igloo@earth.li>**20100902204342]
2747[Finish breaking up VectBuiltIn and VectMonad, and add comments
2748benl@ouroborus.net**20100831100724
2749 Ignore-this: 65604f3d22d03433abc12f10be40050d
2750]
2751[Fix warnings
2752benl@ouroborus.net**20100830083746
2753 Ignore-this: 2a0e000985f694582a6f9a9261ff2739
2754]
2755[Break up vectoriser builtins module
2756benl@ouroborus.net**20100830070900
2757 Ignore-this: b86bd36a7875abdcf16763902ba2e637
2758]
2759[Move VectCore to Vectorise tree
2760benl@ouroborus.net**20100830053415
2761 Ignore-this: d5763ca6424285b39a58c7792f4a84a1
2762]
2763[Split out vectoriser environments into own module
2764benl@ouroborus.net**20100830050252
2765 Ignore-this: 5319111c74831394d2c29b9aedf5a766
2766]
2767[Comments and formatting to vectoriser, and split out varish stuff into own module
2768benl@ouroborus.net**20100830042722
2769 Ignore-this: d3f0c98ed8124dd0fca9a2ccea3e15fd
2770]
2771[Fix warnings
2772benl@ouroborus.net**20100830040340
2773 Ignore-this: d6cfad803ad4617e7fdaa62e4a895282
2774]
2775[Fix warning about multiply exported name
2776benl@ouroborus.net**20100830035243
2777 Ignore-this: 27ce2c1d22d9f99929d16a426343044e
2778]
2779[Vectorisation of method types
2780benl@ouroborus.net**20100830032941
2781 Ignore-this: 75614571d5c246a4906edb3b39ab1e0b
2782]
2783[Comments and formatting to vectoriser
2784benl@ouroborus.net**20100830032516
2785 Ignore-this: fe665b77108501c7960d858be3290761
2786]
2787[Implement -dsuppress-module-prefixes
2788benl@ouroborus.net**20100830032428
2789 Ignore-this: 2bb8bad9c60ef9044132bba118010687
2790]
2791[Whitespace only
2792benl@ouroborus.net**20100527045629
2793 Ignore-this: 4c160dfa77727e659817b6af9c84684a
2794]
2795[Disambiguate a function name
2796Ian Lynagh <igloo@earth.li>**20100828225827]
2797[users_guide.xml is now generated
2798Ian Lynagh <igloo@earth.li>**20100828225751]
2799[Pass more -pgm flags in the ghc wrapper; fixes #3863
2800Ian Lynagh <igloo@earth.li>**20100827204537]
2801[Add a new-IO manager release note
2802Ian Lynagh <igloo@earth.li>**20100827171616]
2803[Merge a duplicate release note
2804Ian Lynagh <igloo@earth.li>**20100827171511]
2805[Typo, spotted by Johan Tibell
2806Ian Lynagh <igloo@earth.li>**20100827153914]
2807[First pass at 6.14.1 release notes
2808Ian Lynagh <igloo@earth.li>**20100826220811]
2809[Fix typo
2810Ian Lynagh <igloo@earth.li>**20100824201330]
2811[FIX BUILD: add rts_isProfiled to the symbol table
2812Simon Marlow <marlowsd@gmail.com>**20100826094319
2813 Ignore-this: 9536ddb0a94721c8dec03a2a981cfa83
2814]
2815[Fix the DPH package cleaning/profiled mess even more (the build was broken)
2816Simon Marlow <marlowsd@gmail.com>**20100826084436
2817 Ignore-this: 49d7e4db2fb53b856c213c74c8969d82
2818]
2819[Remove the debugging memory allocator - valgrind does a better job
2820Simon Marlow <marlowsd@gmail.com>**20100824113537
2821 Ignore-this: a3731a83dc18b0fd0de49452e695a7ca
2822 
2823 I got fed up with the constant bogus output from the debugging memory
2824 allocator in RtsUtils.c.  One problem is that we allocate memory in
2825 constructors that then isn't tracked, because the debugging allocator
2826 hasn't been initialised yet.
2827 
2828 The bigger problem is that for a given piece of leaking memory it's
2829 impossible to find out where it was allocated; however Valgrind gives
2830 output like this:
2831 
2832 ==6967== 8 bytes in 1 blocks are still reachable in loss record 1 of 7
2833 ==6967==    at 0x4C284A8: malloc (vg_replace_malloc.c:236)
2834 ==6967==    by 0x4C28522: realloc (vg_replace_malloc.c:525)
2835 ==6967==    by 0x6745E9: stgReallocBytes (RtsUtils.c:213)
2836 ==6967==    by 0x68D812: setHeapAlloced (MBlock.c:91)
2837 ==6967==    by 0x68D8E2: markHeapAlloced (MBlock.c:116)
2838 ==6967==    by 0x68DB56: getMBlocks (MBlock.c:240)
2839 ==6967==    by 0x684F55: alloc_mega_group (BlockAlloc.c:305)
2840 ==6967==    by 0x6850C8: allocGroup (BlockAlloc.c:358)
2841 ==6967==    by 0x69484F: allocNursery (Storage.c:390)
2842 ==6967==    by 0x694ABD: allocNurseries (Storage.c:436)
2843 ==6967==    by 0x6944F2: initStorage (Storage.c:217)
2844 ==6967==    by 0x673E3C: hs_init (RtsStartup.c:160)
2845 
2846 which tells us exactly what the leaking bit of memory is.  So I don't
2847 think we need our own debugging allocator.
2848]
2849[free the entries in the thread label table on exit
2850Simon Marlow <marlowsd@gmail.com>**20100824112606
2851 Ignore-this: c9d577c06548cda80791e590e40d35b3
2852]
2853[Panic in the right way
2854simonpj@microsoft.com**20100825091614
2855 Ignore-this: e6ea4f6dfd2aea088828ea7a945ddd5f
2856]
2857[Fix the DPH/profiled make thing (again)
2858simonpj@microsoft.com**20100825091602
2859 Ignore-this: bc58fa48034ac40cf7be4170958ea29e
2860]
2861[Don't test for gcc flags before we've found gcc
2862Ian Lynagh <igloo@earth.li>**20100824131401]
2863[Change how the dblatex/lndir problem is worked around
2864Ian Lynagh <igloo@earth.li>**20100824130938
2865 Hack: dblatex normalises the name of the main input file using
2866 os.path.realpath, which means that if we're in a linked build tree,
2867 it find the real source files rather than the symlinks in our link
2868 tree. This is fine for the static sources, but it means it can't
2869 find the generated sources.
2870 
2871 We therefore also generate the main input file, so that it really
2872 is in the link tree, and thus dblatex can find everything.
2873]
2874[Clean the generated userguide sources
2875Ian Lynagh <igloo@earth.li>**20100824105827
2876 Ignore-this: 39b4f9702c688c053ed3273b20969597
2877]
2878[DPH should not even be built if GhcProfiled
2879simonpj@microsoft.com**20100823133439
2880 Ignore-this: 62acbf83de5b70ff6d27ab38ce9218ae
2881 
2882 It's not just when cleaning!
2883]
2884[The templateHaskellOk check should only run in stage2
2885simonpj@microsoft.com**20100823133353
2886 Ignore-this: f6dc9292923a1ca201953c5f58c0af3c
2887 
2888 Because rtsIsProfiled is only available in stage2
2889]
2890[Add a couple of missing tests for EAGER_BLACKHOLE
2891Simon Marlow <marlowsd@gmail.com>**20100823104654
2892 Ignore-this: 70c981b86370b0c7564b29b057650897
2893 This was leading to looping and excessive allocation, when the
2894 computation should have just blocked on the black hole. 
2895 
2896 Reported by Christian Höner zu Siederdissen <choener@tbi.univie.ac.at>
2897 on glasgow-haskell-users.
2898]
2899[Don't check for swept blocks in -DS.
2900Marco Túlio Gontijo e Silva <marcot@marcot.eti.br>**20100718225526
2901 Ignore-this: ad5dcf3c247bc19fbef5122c1142f3b2
2902 
2903 The checkHeap function assumed the allocated part of the block contained only
2904 alive objects and slops.  This was not true for blocks that are collected using
2905 mark sweep.  The code in this patch skip the test for this kind of blocks.
2906]
2907[Fix "darcs get"
2908Ian Lynagh <igloo@earth.li>**20100822183542]
2909[Add "darcs-all upstreampull"
2910Ian Lynagh <igloo@earth.li>**20100822163419
2911 This pulls from the upstream repos, for those packages which have
2912 upstreams
2913]
2914[Generate the bit in the user guide where we say what -fglasgow-exts does
2915Ian Lynagh <igloo@earth.li>**20100822155514
2916 Stops the docs going out of sync with the code.
2917]
2918[Factor out the packages file parsing in darcs-all
2919Ian Lynagh <igloo@earth.li>**20100822154813]
2920[Document --supported-extensions
2921Ian Lynagh <igloo@earth.li>**20100822134530]
2922[fix extraction of command stack of arguments of arrow "forms" (fixes #4236)
2923Ross Paterson <ross@soi.city.ac.uk>**20100822090022
2924 Ignore-this: a93db04ec4f20540642a19cdc67d1666
2925 
2926 The command stack was being extracted (by unscramble) with the outermost
2927 type first, contrary to the comment on the function.
2928]
2929[minor fix to comment
2930Ross Paterson <ross@soi.city.ac.uk>**20100822085838
2931 Ignore-this: 8d203ba2600eaf4cf21b043dcfa96cdc
2932]
2933[Add the RTS library path to the library search path
2934Ian Lynagh <igloo@earth.li>**20100820155523
2935 In case the RTS is being explicitly linked in. For #3807.
2936]
2937[Remove some duplication of C flags
2938Ian Lynagh <igloo@earth.li>**20100819233743
2939 We now use the CONF_CC_OPTS_STAGEn C flags in machdepCCOpts, rather than
2940 repeating them there.
2941]
2942[Set -fno-stack-protector in CONF_CC_OPTS_STAGE* rathre than extra-gcc-opts
2943Ian Lynagh <igloo@earth.li>**20100819233031
2944 The latter is only used when compiling .hc files, whereas we need it for
2945 .c files too.
2946]
2947[Give clearer errors for bad input in the packages file; suggested by pejo
2948Ian Lynagh <igloo@earth.li>**20100819232420]
2949[Set -march=i686 on OS X x86 in the configure variables
2950Ian Lynagh <igloo@earth.li>**20100819230939
2951 We used to set it only in machdepCCOpts, so this is more consistent
2952]
2953[Give each stage its own Config.hs
2954Ian Lynagh <igloo@earth.li>**20100819224709
2955 This also means the file is generated in a dist directory, not a
2956 source directory.
2957]
2958[Fix cleaning when GhcProfiled = YES
2959Ian Lynagh <igloo@earth.li>**20100819131346
2960 We need to include the DPH cleaning rules, even though we don't build DPH
2961 when GhcProfiled = YES.
2962]
2963[stgReallocBytes(DEBUG): don't fail when the ptr passed in is NULL
2964Simon Marlow <marlowsd@gmail.com>**20100817150836
2965 Ignore-this: 4b5063e65e01399f64a33f0d0555ff38
2966]
2967[Use make-command in rules/bindist.mk
2968Ian Lynagh <igloo@earth.li>**20100818191243
2969 Rather than it having its own specialised version
2970]
2971[Use make-command when installing packages
2972Ian Lynagh <igloo@earth.li>**20100818190600]
2973[Add _DATA_FILES to package-data.mk files
2974Ian Lynagh <igloo@earth.li>**20100818185801]
2975[Add a "make-command" utility Makefile function
2976Ian Lynagh <igloo@earth.li>**20100818183055]
2977[LLVM: Nicer format for lack of shared lib warning
2978David Terei <davidterei@gmail.com>**20100817145207
2979 Ignore-this: 753d45762601d87761614937a1bb6716
2980]
2981[fix FP_CHECK_ALIGNMENT for autoconf 2.66 (fixes #4252)
2982Ross Paterson <ross@soi.city.ac.uk>**20100816142442
2983 Ignore-this: cd784b8888d32b3b2cc2cc0969ec40f
2984 
2985 Recent versions of AS_LITERAL_IF don't like *'s.  Fix from
2986 
2987 http://blog.gmane.org/gmane.comp.sysutils.autoconf.general/month=20100701
2988]
2989[Refactor the command-line argument parsing (again)
2990simonpj@microsoft.com**20100816074453
2991 Ignore-this: 26dc9e37a88660a887a2e316ed7a9803
2992 
2993 This change allows the client of CmdLineParser a bit more flexibility,
2994 by giving him an arbitrary computation (not just a deprecation
2995 message) for each flag. 
2996 
2997 There are several clients, so there are lots of boilerplate changes.
2998 
2999 Immediate motivation: if RTS is not profiled, we want to make
3000 Template Haskell illegal.  That wasn't with the old setup.
3001]
3002[Add upstream repo to the packages file
3003Ian Lynagh <igloo@earth.li>**20100815154741]
3004[Make the "tag" column of the packages file always present
3005Ian Lynagh <igloo@earth.li>**20100815151657
3006 It makes the parsing simpler if we always have the same number of columns
3007]
3008[Disable object splitting on OSX; works around #4013
3009Ian Lynagh <igloo@earth.li>**20100815134759]
3010[Return memory to the OS; trac #698
3011Ian Lynagh <igloo@earth.li>**20100813170402]
3012[Reduce the xargs -s value we use on Windows
3013Ian Lynagh <igloo@earth.li>**20100812223721
3014 With 30000 I was getting:
3015     xargs: value for -s option should be < 28153
3016]
3017[LLVM: Enable shared lib support on Linux x64
3018David Terei <davidterei@gmail.com>**20100813191534
3019 Ignore-this: 642ed37af38e5f17d419bf4f09332671
3020]
3021[Re-do the arity calculation mechanism again (fix Trac #3959)
3022simonpj@microsoft.com**20100813161151
3023 Ignore-this: d4a2aa48150b503b20c25351a79decfb
3024 
3025 After rumination, yet again, on the subject of arity calculation,
3026 I have redone what an ArityType is (it's purely internal to the
3027 CoreArity module), and documented it better.  The result should
3028 fix #3959, and I hope the related #3961, #3983.
3029 
3030 There is lots of new documentation: in particular
3031  * Note [ArityType] 
3032    describes the new datatype for arity info
3033 
3034  * Note [State hack and bottoming functions]
3035    says how bottoming functions are dealt with, particularly
3036    covering catch# and Trac #3959
3037 
3038 I also found I had to be careful not to eta-expand single-method
3039 class constructors; see Note [Newtype classes and eta expansion].
3040 I think this part could be done better, but it works ok.
3041]
3042[Comments only
3043simonpj@microsoft.com**20100813161019
3044 Ignore-this: baf68300d8bc630bf0b7ab27647b33a0
3045]
3046[Modify FloatOut to fix Trac #4237
3047simonpj@microsoft.com**20100813163120
3048 Ignore-this: ffc8d00d4b7f0a8a785fcef312900413
3049 
3050 The problem was that a strict binding was getting floated
3051 out into a letrec. This only happened when profiling was
3052 on.  It exposed a fragility in the floating strategy.  This
3053 patch makes it more robust.  See
3054       Note [Avoiding unnecessary floating]
3055]
3056[Fix egregious bug in SetLevels.notWorthFloating
3057simonpj@microsoft.com**20100813161429
3058 Ignore-this: d22865f48d417e6a6b732de3dfba378f
3059 
3060 This bug just led to stupid code, which would
3061 later be optimised away, but better not to generate
3062 stupid code in the first place.
3063]
3064[Delete GhcLibProfiled
3065simonpj@microsoft.com**20100813140152
3066 Ignore-this: 2e1a3f677308be726bd022f45e2fd856
3067 
3068 Simon M and I looked at this, and we think GhcLibProfiled is
3069 (a) not needed (b) confusing.
3070 
3071 Ian should review.
3072 
3073 Really, if GhcProfiled is on we should also
3074 check that 'p' is in the GhcLibWays
3075]
3076[Do not build DPH when GhcProfiled (fixes #4172)
3077simonpj@microsoft.com**20100813140021
3078 Ignore-this: 9e20181643b456e841f845ae0cab0a9a
3079 
3080 Reason: DPH uses Template Haskell and TH doesn't work
3081 in a profiled compiler
3082]
3083[Fix Trac #4220
3084simonpj@microsoft.com**20100812131319
3085 Ignore-this: 33141cfd81627592150a9e5973411ff8
3086 
3087 For deriving Functor, Foldable, Traversable with empty
3088 data cons I just generate a null equation
3089    f _ = error "urk"
3090 
3091 There are probably more lurking (eg Enum) but this will do for now.
3092]
3093[Improve the Specialiser, fixing Trac #4203
3094simonpj@microsoft.com**20100812131133
3095 Ignore-this: 482afbf75165e24a80527a6e52080c07
3096 
3097 Simply fixing #4203 is a tiny fix: in case alterantives we should
3098 do dumpUDs *including* the case binder. 
3099 
3100 But I realised that we can do better and wasted far too much time
3101 implementing the idea.  It's described in
3102     Note [Floating dictionaries out of cases]
3103]
3104[Comments
3105simonpj@microsoft.com**20100812101456
3106 Ignore-this: 6362fe887d25688c12ef2c3cf5554ce4
3107]
3108[Comments only
3109simonpj@microsoft.com**20100812101439
3110 Ignore-this: 7ed2f5fc08811cbe9958c2309a9ed1fa
3111]
3112[Fix bug in linting of shadowed case-alternative binders
3113simonpj@microsoft.com**20100812101413
3114 Ignore-this: 9212a5e2c03421749f5935b3944ecf53
3115]
3116[Comments and spacing only
3117simonpj@microsoft.com**20100812101347
3118 Ignore-this: ed59a7dae7decb24470709dc1c118dbb
3119]
3120[Add more info to more parse error messages (#3811)
3121Ian Lynagh <igloo@earth.li>**20100809233108]
3122[Run finalizers *after* updating the stable pointer table (#4221)
3123Simon Marlow <marlowsd@gmail.com>**20100810133739
3124 Ignore-this: b0462f80dd64eac71e599d8a9f6dd665
3125 Silly bug really, we were running the C finalizers while the StablePtr
3126 table was still in a partially-updated state during GC, but finalizers
3127 are allowed to call freeStablePtr() (via hs_free_fun_ptr(), for
3128 example), and chaos ensues.
3129]
3130[Do the dependency-omitting for 'make 1' in a slightly different way
3131Simon Marlow <marlowsd@gmail.com>**20100810093446
3132 Ignore-this: af15edd3a1492cbd93111316b57e02e4
3133 
3134 I encountered a couple of things that broke after Ian's previous
3135 patch: one was my nightly build scripts that use 'make stage=2' at the
3136 top level, and the other is 'make fast' in libraries/base, which uses
3137 'stage=0' to avoid building any compilers.
3138 
3139 So my version of this patch is more direct: it just turns off the
3140 appropriate dependencies using a variable set by 'make 1', 'make 2',
3141 etc.
3142]
3143[Integrate new I/O manager, with signal support
3144Johan Tibell <johan.tibell@gmail.com>**20100724102355
3145 Ignore-this: eb092857a2a1b0ca966649caffe7ac2b
3146]
3147[Add DoAndIfThenElse support
3148Ian Lynagh <igloo@earth.li>**20100808194625]
3149[Make another parse error more informative
3150Ian Lynagh <igloo@earth.li>**20100808193340]
3151[Make a parse error say what it is failing to parse; part of #3811
3152Ian Lynagh <igloo@earth.li>**20100808155732]
3153[Send ghc progress output to stdout; fixes #3636
3154Ian Lynagh <igloo@earth.li>**20100808142542]
3155[Fix the HsColour test in the build system
3156Ian Lynagh <igloo@earth.li>**20100805155319
3157 Ignore-this: ba2752b04801a253e891b31e1914485d
3158]
3159[Fix the -lm configure test; fixes #4155
3160Ian Lynagh <igloo@earth.li>**20100805142508
3161 Ignore-this: 358b8b1074d2d22fb8d362ea6d8b80d6
3162]
3163[Don't restrict filenames in line pragmas to printable characters; fixes #4207
3164Ian Lynagh <igloo@earth.li>**20100805135011
3165 Ignore-this: e3d32312127165e40e6eaa919193d60b
3166 "printable" is ASCII-only, whereas in other locales we can get things like
3167 # 1 "<línea-de-orden>"
3168]
3169[Ensure extension flags are flattened in the Cmm phase
3170Ian Lynagh <igloo@earth.li>**20100805133614
3171 If we start with a .cmmcpp file then they don't get flattened in
3172 the CmmCpp phase, as we don't run that phase.
3173]
3174[Add "cmmcpp" as a Haskellish source suffix
3175Ian Lynagh <igloo@earth.li>**20100805132555]
3176[On amd64/OSX we don't need to be given memory in the first 31bits
3177Ian Lynagh <igloo@earth.li>**20100805120600
3178 Ignore-this: 42eb64e25ad4b66ae022884305e0297b
3179 as PIC is always on
3180]
3181[NCG: Don't worry about trying to re-freeze missing coalescences
3182benl@ouroborus.net**20100702053319
3183 Ignore-this: ea05cbee19b6c5c410db41292cbb64b0
3184]
3185[Make -rtsopts more flexible
3186Ian Lynagh <igloo@earth.li>**20100805011137
3187 The default is a new "some" state, which allows only known-safe flags
3188 that we want on by default. Currently this is only "--info".
3189]
3190[Test for (fd < 0) before trying to FD_SET it
3191Ian Lynagh <igloo@earth.li>**20100804173636]
3192[Remove "On by default" comments in DynFlags
3193Ian Lynagh <igloo@earth.li>**20100802110803
3194 Ignore-this: 2a51055277b5ce9f0e98e1438b212027
3195 These make less sense now we support multiple languges. The
3196 "languageExtensions" function gives the defaults.
3197]
3198[Fix build: Add newline to end of ghc-pkg/Main.hs
3199Ian Lynagh <igloo@earth.li>**20100801183206]
3200[Add a versions haddock binary for Windows
3201Ian Lynagh <igloo@earth.li>**20100801180917]
3202[ghc-pkg: don't fail, if a file is already removed
3203ich@christoph-bauer.net**20100725162606
3204 Ignore-this: 5501d6812c31f4da525c7fb24f6dcaed
3205]
3206[Remove push-all from file list in boot script (push-all no longer exists)
3207Ian Lynagh <igloo@earth.li>**20100801121841
3208 Ignore-this: eec130f06610d8728a57626682860a1a
3209]
3210[Add error checking to boot-pkgs script
3211Ian Lynagh <igloo@earth.li>**20100801121432
3212 Ignore-this: 8afd6663db443c774bad45d75bbfe950
3213]
3214[Add more error checking to the boot script
3215Ian Lynagh <igloo@earth.li>**20100801113628]
3216[Remove libHSrtsmain.a before creating it
3217Ian Lynagh <igloo@earth.li>**20100801005432
3218 Otherwise it isn't updated properly if rts/Main.c changes
3219]
3220[Expose the functions haddock needs even when haddock is disabled; #3558
3221Ian Lynagh <igloo@earth.li>**20100731115506]
3222[Always haddock by default
3223Ian Lynagh <igloo@earth.li>**20100730235001
3224 Revert this patch:
3225     Matthias Kilian <kili@outback.escape.de>**20090920181319
3226     Don't build haddock if HADDOC_DOCS = NO, and disable HADDOC_DOCS
3227         if GhcWithInterpreter = NO
3228     Haddock uses TcRnDriver.tcRnGetInfo, which is only available if
3229     GHCI is built. Set HADDOC_DOCS to NO if GhcWithInterpreter is NO,
3230     and disable the haddock build if HADDOC_DOCS = NO.
3231]
3232[Add a debugTrace for the phases that we run
3233Ian Lynagh <igloo@earth.li>**20100729201503]
3234[* Add StringPrimL as a constructor for Template Haskell (Trac #4168)
3235simonpj@microsoft.com**20100730131922
3236 Ignore-this: 520d0a0a14b499b299e8b2be8d148ff0
3237   
3238 There are already constructors for IntPrim, FloatPrim etc,
3239 so this makes it more uniform.
3240   
3241 There's a corresponding patch for the TH library
3242]
3243[Add thread affinity support for FreeBSD
3244Gabor Pali <pgj@FreeBSD.org>**20100720001409
3245 Ignore-this: 6c117b8219bfb45445089e82ee470410
3246 - Implement missing functions for setting thread affinity and getting real
3247   number of processors.
3248 - It is available starting from 7.1-RELEASE, which includes a native support
3249   for managing CPU sets.
3250 - Add __BSD_VISIBLE, since it is required for certain types to be visible in
3251   addition to POSIX & C99.
3252]
3253[Disable symbol visibility pragmas for FreeBSD
3254Ian Lynagh <igloo@earth.li>**20100729012507
3255 Do not use GCC pragmas for controlling visibility, because it causes
3256 "undefined reference" errors at link-time.  The true reasons are
3257 unknown, however FreeBSD 8.x includes GCC 4.2.1 in the base system,
3258 which might be buggy.
3259]
3260[Fix numeric escape sequences parsing
3261Anton Nikishaev <anton.nik@gmail.com>**20100721194208
3262 Ignore-this: dd71935b1866b5624f7975c45ad519a1
3263 This fixes trac bug #1344
3264]
3265[Explicitly give the right path to perl when making the OS X installer; #4183
3266Ian Lynagh <igloo@earth.li>**20100728163030]
3267[Set -fno-stack-protector in extra-gcc-opts; fixes #4206
3268Ian Lynagh <igloo@earth.li>**20100728161957
3269 We were using it only when building the RTS, and only on certain
3270 platforms. However, some versions of OS X need the flag, while others
3271 don't support it, so we now test for it properly.
3272]
3273[Make PersistentLinkerState fields strict; fixes #4208
3274Ian Lynagh <igloo@earth.li>**20100727201911
3275 Ignore-this: fc5cfba48cd16624f6bb15a7a03a3b4
3276 We modify fields a lot, so we retain the old value if they aren't forced.
3277]
3278[Don't rebuild dependency files unnecessarily when doing "make 1" etc
3279Ian Lynagh <igloo@earth.li>**20100726211512
3280 Ignore-this: d91a729e5113aa964cc67768e92e57ef
3281]
3282[LLVM: If user specifies optlo, don't use '-O' levels
3283David Terei <davidterei@gmail.com>**20100726105650
3284 Ignore-this: e05e103b09d1de937540ffad7983f88e
3285]
3286[Flatten flags for ghci's :show
3287Ian Lynagh <igloo@earth.li>**20100725135320]
3288[Add support for Haskell98 and Haskell2010 "languages"
3289Ian Lynagh <igloo@earth.li>**20100724230121]
3290[Rename "language" varibles etc to "extension", and add --supported-extensions
3291Ian Lynagh <igloo@earth.li>**20100724223624]
3292[Separate language option handling into 2 phases
3293Ian Lynagh <igloo@earth.li>**20100724212013
3294 We now first collect the option instructions (from the commandline,
3295 from pragmas in source files, etc), and then later flatten them into
3296 the list of enabled options. This will enable us to use different
3297 standards (H98, H2010, etc) as a base upon which to apply the
3298 instructions, when we don't know what the base will be when we start
3299 collecting instructions.
3300]
3301[Separate the language flags from the other DynFlag's
3302Ian Lynagh <igloo@earth.li>**20100724133103
3303 Ignore-this: 47bb8d42e621e47016b66c7472bd6cb5
3304]
3305[Set stage-specific CC/LD opts in the bindist configure.ac
3306Ian Lynagh <igloo@earth.li>**20100724113748
3307 Ignore-this: f06926d185a35ddd05490ca4a257e992
3308]
3309[Use different CC/LD options for different stages
3310Ian Lynagh <igloo@earth.li>**20100723223059]
3311[Add some error belchs to the linker, when we find bad magic numbers
3312Ian Lynagh <igloo@earth.li>**20100723200822]
3313[Add some more linker debugging prints
3314Ian Lynagh <igloo@earth.li>**20100723180237]
3315[When (un)loading an object fails, say which object in teh panic
3316Ian Lynagh <igloo@earth.li>**20100723162649]
3317[Add a release note: GHCi import syntax
3318Ian Lynagh <igloo@earth.li>**20100721193647]
3319[Deprecate NewQualifiedOperators extension (rejected by H')
3320Ian Lynagh <igloo@earth.li>**20100719150909
3321 Ignore-this: 6e7e3bedc5360c5975f73497b3e6cba5
3322]
3323[LLVM: Allow optlc and optlo to override default params for these systools
3324David Terei <davidterei@gmail.com>**20100722181631
3325 Ignore-this: e60af7941996f7170fb3bfb02a002082
3326]
3327[LLVM: Code and speed improvement to dominateAllocs pass.
3328David Terei <davidterei@gmail.com>**20100721143654
3329 Ignore-this: 9fb7058c8a2afc005521298c7b8d0036
3330]
3331[Comments only
3332simonpj@microsoft.com**20100721144257
3333 Ignore-this: b3091ddcd1df271eb85fe90978ab7adc
3334]
3335[Fix inlining for default methods
3336simonpj@microsoft.com**20100721144248
3337 Ignore-this: 61a11a8f741f775000c6318aae4b3191
3338 
3339 This was discombobulated by a patch a week ago;
3340 now fixed, quite straightforwardly.  See
3341 Note [Default methods and instances]
3342]
3343[Allow reification of existentials and GADTs
3344simonpj@microsoft.com**20100721090437
3345 Ignore-this: 20f1ccd336cc25aff4d4d67a9ac2211a
3346 
3347 It turns out that TH.Syntax is rich enough to express even GADTs,
3348 provided we express them in equality-predicate form.  So for
3349 example
3350 
3351   data T a where
3352      MkT1 :: a -> T [a]
3353      MkT2 :: T Int
3354 
3355 will appear in TH syntax like this
3356 
3357   data T a = forall b. (a ~ [b]) => MkT1 b
3358            | (a ~ Int) => MkT2
3359 
3360 While I was at it I also improved the reification of types,
3361 so that we use TH.TupleT and TH.ListT when we can.
3362]
3363[add numSparks# primop (#4167)
3364Simon Marlow <marlowsd@gmail.com>**20100720153746
3365 Ignore-this: f3f925e7de28f3f895213aefbdbe0b0f
3366]
3367[LLVM: Decrease max opt level used under OSX to avoid bug
3368David Terei <davidterei@gmail.com>**20100720160938
3369 Ignore-this: 34b0b3550f00b27b00ad92f8232745e5
3370 
3371 Currently, many programs compiled with GHC at -O2 and LLVM
3372 set to -O3 will segfault (only under OSX). Until this issue
3373 is fixed I have simply 'solved' the segfault by lowering
3374 the max opt level for LLVM used to -O2 under OSX.
3375 
3376 All these recent changes to OSX should mean its finally as
3377 stable as Linux and Windows.
3378]
3379[LLVM: Fix OSX to work again with TNTC disabled.
3380David Terei <davidterei@gmail.com>**20100720160845
3381 Ignore-this: 8dc98139cfa536b2a64aa364d040b581
3382]
3383[LLVM: Fix printing of local vars so LLVM works with -fnew-codegen
3384David Terei <davidterei@gmail.com>**20100720160302
3385 Ignore-this: d883c433dfaed67921a8c5360e1f9f6a
3386]
3387[Use a separate mutex to protect all_tasks, avoiding a lock-order-reversal
3388Simon Marlow <marlowsd@gmail.com>**20100716150832
3389 Ignore-this: ffbdb4ee502e0f724d57acb9bfbe9d92
3390 In GHC 6.12.x I found a rare deadlock caused by this
3391 lock-order-reversal:
3392 
3393 AQ cap->lock
3394   startWorkerTask
3395     newTask
3396       AQ sched_mutex
3397 
3398 scheduleCheckBlackHoles
3399   AQ sched_mutex
3400    unblockOne_
3401     wakeupThreadOnCapabilty
3402       AQ cap->lock
3403 
3404 so sched_mutex and cap->lock are taken in a different order in two
3405 places.
3406 
3407 This doesn't happen in the HEAD because we don't have
3408 scheduleCheckBlackHoles, but I thought it would be prudent to make
3409 this less likely to happen in the future by using a different mutex in
3410 newTask.  We can clearly see that the all_tasks mutex cannot be
3411 involved in a deadlock, becasue we never call anything else while
3412 holding it.
3413]
3414['make fast' in a package does not build any compilers
3415Simon Marlow <marlowsd@gmail.com>**20100715125904
3416 Ignore-this: f27e70faf3944831dad16e89a4e273da
3417]
3418[LLVM: Fix up botched last commit
3419David Terei <davidterei@gmail.com>**20100719104823
3420 Ignore-this: a32e0f6a38cb9e02527eb8ca69b3eb59
3421]
3422[LLVM: Fix warning introduce in last commit.
3423David Terei <davidterei@gmail.com>**20100719103411
3424 Ignore-this: e9c92a9402aff50d60ab26e6ad441bfc
3425]
3426[LLVM: Use mangler to fix up stack alignment issues on OSX
3427David Terei <davidterei@gmail.com>**20100718231000
3428 Ignore-this: 9f6e8cb855269cb3a5ac1a23480d0e71
3429]
3430[Fix #4195 (isGadtSyntaxTyCon returns opposite result)
3431illissius@gmail.com**20100715134134
3432 Ignore-this: a90403f893030432b5c15d743647f350
3433]
3434[Update to time 1.2.0.3
3435Ian Lynagh <igloo@earth.li>**20100717181810
3436 Ignore-this: 1ccb4801a73f399e6718ce556543ede1
3437]
3438[Reorder RTS --info output
3439Ian Lynagh <igloo@earth.li>**20100717162356]
3440[Fix unreg prof build: Define CCS_SYSTEM in stg/MiscClosures.h
3441Ian Lynagh <igloo@earth.li>**20100717142832
3442 Ignore-this: 9675f3f51b6dac40483155344e7f45b6
3443]
3444[Make mkDerivedConstants as a stage 1 program
3445Ian Lynagh <igloo@earth.li>**20100717000827
3446 Ignore-this: 5357403461b209b8606f1d33defb51cf
3447 This way it gets the defines for the right platform when cross-compiling
3448]
3449[Don't generate Haskell dependencies if we don't have any Haskell sources
3450Ian Lynagh <igloo@earth.li>**20100717000800
3451 Ignore-this: 454abd0358f535b7e789327125c9206c
3452]
3453[Link programs that have no Haskell objects with gcc rather than ghc
3454Ian Lynagh <igloo@earth.li>**20100716235303
3455 Ignore-this: f65588b69675edea616cc434e769b0a4
3456]
3457[Use gcc to build C programs for stages >= 1
3458Ian Lynagh <igloo@earth.li>**20100716223703
3459 Ignore-this: 9f843a4e17285cda582117504707f9e7
3460]
3461[Add platform info to "ghc --info" output
3462Ian Lynagh <igloo@earth.li>**20100716141953]
3463[Tidy up Config.hs generation
3464Ian Lynagh <igloo@earth.li>**20100716140630]
3465[Fix HC porting test in makefiles
3466Ian Lynagh <igloo@earth.li>**20100716010808
3467 Ignore-this: 6052c1dd022a6108ab2236a299ee1d84
3468 Now that we are trying to support cross compilation, we can't use
3469     "$(TARGETPLATFORM)" != "$(HOSTPLATFORM)"
3470 as a test for HC-porting.
3471]
3472[Change a BUILD var to a HOST var
3473Ian Lynagh <igloo@earth.li>**20100716002558]
3474[Remove an unnecessary #include
3475Ian Lynagh <igloo@earth.li>**20100715233930
3476 Ignore-this: dcede249de6be7e3c9305c9279c2ca07
3477]
3478[Split up some make commands, so that errors aren't overlooked
3479Ian Lynagh <igloo@earth.li>**20100715152237
3480 Ignore-this: fb69b0a25d9ca71dae5e75d38db675cd
3481 When we ask make to run "a | b", if a fails then the pipeline might
3482 still exit successfuly.
3483]
3484[Remove an unnecessary #include
3485Ian Lynagh <igloo@earth.li>**20100715143000
3486 Ignore-this: 4e098cac5dda2dd595ca0a0f5121853c
3487]
3488[Simplify some more CPP __GLASGOW_HASKELL__ tests
3489Ian Lynagh <igloo@earth.li>**20100715142500]
3490[Remove some code only used with GHC 6.11.*
3491Ian Lynagh <igloo@earth.li>**20100715141720]
3492[__GLASGOW_HASKELL__ >= 609 is now always true
3493Ian Lynagh <igloo@earth.li>**20100715141544]
3494[Correct the values in ghc_boot_platform.h
3495Ian Lynagh <igloo@earth.li>**20100714223717
3496 Ignore-this: 4c99116f7ac73fadbd6d16807f57a693
3497]
3498[Change some TARGET checks to HOST checks
3499Ian Lynagh <igloo@earth.li>**20100714184715]
3500[LLVM: Add inline assembly to binding.
3501David Terei <davidterei@gmail.com>**20100714152530
3502 Ignore-this: 72a7b5460c128ed511e8901e5889fe2b
3503]
3504[LLVM: Fix mistype in last commit which broke TNTC under win/linux.
3505David Terei <davidterei@gmail.com>**20100714153339
3506 Ignore-this: 302d7957e3dded80368ebade5312ab35
3507]
3508[Remove unnecessary #include
3509Ian Lynagh <igloo@earth.li>**20100713153704
3510 Ignore-this: c37d3127b1dc68f59270c07173994c28
3511]
3512[Change some TARGET tests to HOST tests in the RTS
3513Ian Lynagh <igloo@earth.li>**20100713141034
3514 Which was being used seemed to be random
3515]
3516[LLVM: Add in new LLVM mangler for implementing TNTC on OSX
3517David Terei <davidterei@gmail.com>**20100713183243
3518 Ignore-this: 394fb74d7f9657d8b454bd0148d24bf7
3519]
3520[Refactor where an error message is generated
3521simonpj@microsoft.com**20100713115733
3522 Ignore-this: f94467856238586fcbbe48537141cf78
3523]
3524[Comments only
3525simonpj@microsoft.com**20100713115703
3526 Ignore-this: 5815442c4e69b9ec331b34242a596253
3527]
3528[Comments on data type families
3529simonpj@microsoft.com**20100713115640
3530 Ignore-this: 90a333bb7f7d64a49fb7dd180d893f6b
3531]
3532[Fix Trac #T4136: take care with nullary symbol constructors
3533simonpj@microsoft.com**20100707135945
3534 Ignore-this: 2a717a24fefcd593ea41c23dad351db0
3535 
3536 When a nullary constructor is a symbol eg (:=:) we need
3537 to take care.  Annoying.
3538]
3539[Fix Trac #4127 (and hence #4173)
3540simonpj@microsoft.com**20100707123125
3541 Ignore-this: 98bb6d0f7182b59f8c93596c61f9785d
3542 
3543 The change involves a little refactoring, so that the default
3544 method Ids are brought into scope earlier, before the value
3545 declarations are compiled.  (Since a value decl may contain
3546 an instance decl in a quote.)
3547 
3548 See Note [Default method Ids and Template Haskell] in
3549 TcTyClsDcls.
3550]
3551[Fix second bug in Trac #4127
3552simonpj@microsoft.com**20100701140124
3553 Ignore-this: c8d1cc27364fe9ee5a52acb1ecb5cdd9
3554 
3555 This bug concerned the awkward shadowing we do for
3556 Template Haskell declaration brackets.  Lots of
3557 comments in
3558 
3559   Note [Top-level Names in Template Haskell decl quotes]
3560]
3561[ia64: switch handling of 'foreign import wrapper' (FIW) to libffi
3562Sergei Trofimovich <slyfox@community.haskell.org>**20100709213922
3563 Ignore-this: fd07687e0089aebabf62de85d2be693
3564 
3565 I tried to build darcs-2.4.4 with ghc-6.12.3 and got coredumps when darcs is used
3566 in interactive mode. I tried test from ticket #3516 and found out FIW code is broken.
3567 Instead of fixing it I just switched to libffi. Result built successfully, passed
3568 'foreign import wrapper' test from ticket #3516 and builds working darcs.
3569]
3570[* storage manager: preserve upper address bits on 64bit machines (thanks to zygoloid)
3571Sergei Trofimovich <slyfox@community.haskell.org>**20100709115917
3572 Ignore-this: 9f1958a19992091ddc2761c389ade940
3573 
3574 Patch does not touch amd64 as it's address lengts is 48 bits at most, so amd64 is unaffected.
3575 
3576 the issue: during ia64 ghc bootstrap (both 6.10.4 and 6.12.3) I
3577 got the failure on stage2 phase:
3578     "inplace/bin/ghc-stage2"   -H32m -O -H64m -O0 -w ...
3579     ghc-stage2: internal error: evacuate: strange closure type 15
3580         (GHC version 6.12.3 for ia64_unknown_linux)
3581         Please report this as a GHC bug:  http://www.haskell.org/ghc/reportabug
3582     make[1]: *** [libraries/dph/dph-base/dist-install/build/Data/Array/Parallel/Base/Hyperstrict.o] Aborted
3583 
3584 gdb backtrace (break on 'barf'):
3585 Breakpoint 1 at 0x400000000469ec31: file rts/RtsMessages.c, line 39.
3586 (gdb) run -B/var/tmp/portage/dev-lang/ghc-6.12.3/work/ghc-6.12.3/inplace/bin --info
3587 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
3588 [Thread debugging using libthread_db enabled]
3589 
3590 Breakpoint 1, barf (s=0x40000000047915b0 "evacuate: strange closure type %d") at rts/RtsMessages.c:39
3591 39        va_start(ap,s);
3592 (gdb) bt
3593 #0  barf (s=0x40000000047915b0 "evacuate: strange closure type %d") at rts/RtsMessages.c:39
3594 #1  0x400000000474a1e0 in evacuate (p=0x6000000000147958) at rts/sm/Evac.c:756
3595 #2  0x40000000046d68c0 in scavenge_srt (srt=0x6000000000147958, srt_bitmap=7) at rts/sm/Scav.c:348
3596 ...
3597 
3598 > 16:52:53 < zygoloid> slyfox: i'm no ghc expert but it looks like HEAP_ALLOCED_GC(q)
3599 >                      is returning true for a FUN_STATIC closure
3600 > 17:18:43 < zygoloid> try: p HEAP_ALLOCED_miss((unsigned long)(*p) >> 20, *p)
3601 > 17:19:12 < slyfox> (gdb) p HEAP_ALLOCED_miss((unsigned long)(*p) >> 20, *p)
3602 > 17:19:12 < slyfox> $1 = 0
3603 > 17:19:40 < zygoloid> i /think/ that means the mblock_cache is broken
3604 > 17:22:45 < zygoloid> i can't help further. however i am suspicious that you seem to have pointers with similar-looking low 33
3605 >                      bits and different high 4 bits, and it looks like such pointers get put into the same bucket in
3606 >                      mblock_cache.
3607 ...
3608 > 17:36:16 < zygoloid> slyfox: try changing the definition of MbcCacheLine to StgWord64, see if that helps
3609 > 17:36:31 < zygoloid> that's in includes/rts/storage/MBlock.h
3610 And it helped!
3611]
3612[Fixing link failure of compiler on ia64 ('-Wl,' prefixed value passed directly to ld)
3613Sergei Trofimovich <slyfox@community.haskell.org>**20100708180943
3614 Ignore-this: ced99785e1f870ee97e5bec658e2504f
3615 
3616     /usr/bin/ld -Wl,--relax -r -o dist-stage1/build/HSghc-6.10.4.o \
3617                                   dist-stage1/build/BasicTypes.o dist-stage1/build/DataCon.o ...
3618     /usr/bin/ld: unrecognized option '-Wl,--relax'
3619 
3620 If we just drop '-Wl,' part it will not help as '-r' and '--relax' are incompatible.
3621 
3622 Looks like '-Wl,--relax' was skipped by earlier binutils' ld as unknown option.
3623 Removing ia64 specific path.
3624]
3625[LLVM: Allow getelementptr to use LlvmVar for indexes.
3626David Terei <davidterei@gmail.com>**20100712152529
3627 Ignore-this: 9e158d9b89a86bca8abf11d082328278
3628]
3629[Move all the warning workarounds to one place
3630Ian Lynagh <igloo@earth.li>**20100710161723]
3631[xhtml is now warning-free
3632Ian Lynagh <igloo@earth.li>**20100710144635]
3633[Move a bit of build system code
3634Ian Lynagh <igloo@earth.li>**20100709224534]
3635[adapt to the new async exceptions API
3636Simon Marlow <marlowsd@gmail.com>**20100709125238
3637 Ignore-this: 55d845e40b9daed3575c1479d8dda1d5
3638]
3639[quiet some new spewage
3640Simon Marlow <marlowsd@gmail.com>**20100709091521
3641 Ignore-this: de7f91976bbc9789e6fd7091f05c25c0
3642]
3643[New asynchronous exception control API (ghc parts)
3644Simon Marlow <marlowsd@gmail.com>**20100708144851
3645 Ignore-this: 56320c5fc61ae3602d586609387aae22
3646 
3647 As discussed on the libraries/haskell-cafe mailing lists
3648   http://www.haskell.org/pipermail/libraries/2010-April/013420.html
3649 
3650 This is a replacement for block/unblock in the asychronous exceptions
3651 API to fix a problem whereby a function could unblock asynchronous
3652 exceptions even if called within a blocked context.
3653 
3654 The new terminology is "mask" rather than "block" (to avoid confusion
3655 due to overloaded meanings of the latter).
3656 
3657 In GHC, we changed the names of some primops:
3658 
3659   blockAsyncExceptions#   -> maskAsyncExceptions#
3660   unblockAsyncExceptions# -> unmaskAsyncExceptions#
3661   asyncExceptionsBlocked# -> getMaskingState#
3662 
3663 and added one new primop:
3664 
3665   maskUninterruptible#
3666 
3667 See the accompanying patch to libraries/base for the API changes.
3668]
3669[remove outdated comment
3670Simon Marlow <marlowsd@gmail.com>**20100708100840
3671 Ignore-this: afb2e9f6fe1f1acda51b0cbdf2637176
3672]
3673[remove 'mode: xml' emacs settings (#2208)
3674Simon Marlow <marlowsd@gmail.com>**20100708100817
3675 Ignore-this: 3a8d997fb90e01ca88dc47fb95feeba0
3676]
3677[typo in comment
3678Simon Marlow <marlowsd@gmail.com>**20100616111359
3679 Ignore-this: d3ef9288d6d6b9ab3bacbe09e0d9801c
3680]
3681[Win32 getProcessElapsedTime: use a higher-resolution time source
3682Simon Marlow <marlowsd@gmail.com>**20100708093223
3683 Ignore-this: 821989d4ff7ff2bff40cee71a881521c
3684 QueryPerformanceCounter() on Windows gives much better resolution than
3685 GetSystemTimeAsFileTime().
3686]
3687[alpha: switch handling of 'foreign import wrapper' (FIW) to libffi
3688Sergei Trofimovich <slyfox@community.haskell.org>**20100708065318
3689 Ignore-this: ddee15876737a6aa7f6dabc8ff79ce0d
3690 
3691 I tried to build ghc-6.12.3 and found out FIW part of code
3692 does not compile anymore. It uses absent functions under #ifdef.
3693 Instead of fixing it I just switched to libffi. Result built successfully
3694 and passed 'foreign import wrapper' test I wrote for trac ticket #3516.
3695 
3696 I didn't try to build -HEAD yet, but this patch only removes code, so
3697 it should not make -HEAD worse.
3698]
3699[Reorder the CPP flags so -optP can override the platform defines
3700Ian Lynagh <igloo@earth.li>**20100708203523]
3701[Add docs for DatatypeContexts extension
3702Ian Lynagh <igloo@earth.li>**20100707230907
3703 Ignore-this: 8158f03b35a2d7442a75fe85d6f1b1c7
3704]
3705[Make datatype contexts an extension (on by default) (DatatypeContexts)
3706Ian Lynagh <igloo@earth.li>**20100707212529
3707 Ignore-this: 6885ff510a0060610eeeba65122caef5
3708]
3709[LLVM: Fix various typos in comments
3710David Terei <davidterei@gmail.com>**20100707220448
3711 Ignore-this: 1ba3e722f150492da2f9d485c5795e80
3712]
3713[Handle haddock headers when looking for LANGUAGE/OPTIONS_GHC pragmas
3714Ian Lynagh <igloo@earth.li>**20100707120423
3715 Ignore-this: a75aa67690284a6cee3e62c943d4fd01
3716]
3717[Make pragState call mkPState, rather than duplicating everything
3718Ian Lynagh <igloo@earth.li>**20100706173007
3719 Ignore-this: 61fe24b99dbe7a42efff1a9dd703a75c
3720 This also means that extsBitmap gets set, whereas is was just being set
3721 to 0 before.
3722]
3723[LLVM: Add alias type defenitions to LlvmModule.
3724David Terei <davidterei@gmail.com>**20100707142053
3725 Ignore-this: eee6ad5385563ccf08e664d2634a03f2
3726]
3727[LLVM: Use packed structure type instead of structure type
3728David Terei <davidterei@gmail.com>**20100707120320
3729 Ignore-this: a06e0359d182291b81cae56993ca385e
3730 
3731 The regular structure type adds padding to conform to the platform ABI,
3732 which causes problems with structures storing doubles under windows since
3733 we don't conform to the platform ABI there. So we use packed structures
3734 instead now that don't do any padding.
3735]
3736[Make mkPState and pragState take their arguments in the same order
3737Ian Lynagh <igloo@earth.li>**20100706172611]
3738[Remove an out-of-date comment
3739Ian Lynagh <igloo@earth.li>**20100706172217
3740 Ignore-this: 710ebd7d2dc01c1b0f1e58a5b6f85701
3741]
3742[LLVM: Stop llvm saving stg caller-save regs across C calls
3743David Terei <davidterei@gmail.com>**20100705162629
3744 Ignore-this: 28b4877b31b9358e682e38fc54b88658
3745 
3746 This is already handled by the Cmm code generator so LLVM is simply
3747 duplicating work. LLVM also doesn't know which ones are actually live
3748 so saves them all which causes a fair performance overhead for C calls
3749 on x64. We stop llvm saving them across the call by storing undef to
3750 them just before the call.
3751]
3752[LLVM: Add in literal undefined value to binding
3753David Terei <davidterei@gmail.com>**20100705161544
3754 Ignore-this: 95d8361b11584aaeec44c30e76916470
3755]
3756[LLVM: Add a literal NULL value to binding
3757David Terei <davidterei@gmail.com>**20100705161308
3758 Ignore-this: 9507b4b12c1157498704a9d1e5860f12
3759 
3760 Patch from Erik de Castro Lopo <erikd@mega-nerd.com>.
3761]
3762[refactor import declaration support (#2362)
3763Simon Marlow <marlowsd@gmail.com>**20100705104557
3764 Ignore-this: ee034ac377078a7e92bfada1907c86a0
3765]
3766[Disable dynamic linking optimisations on OS X
3767Simon Marlow <marlowsd@gmail.com>**20100705103014
3768 Ignore-this: b04420d3705c51112797758d17b2e40c
3769 To improve performance of the RTS when dynamically linked on x86, I
3770 previously disabled -fPIC for certain critical modules (the GC, and a
3771 few others).  However, build reports suggest that the dynamic linker
3772 on OS X doesn't like this, so I'm disabling this optimsation on that
3773 platform.
3774]
3775[trac #2362 (full import syntax in ghci)
3776amsay@amsay.net**20100625032632
3777 Ignore-this: a9d0859d84956beb74e27b797431bf9c
3778 'import' syntax is seperate from ':module' syntax
3779]
3780[Simplify ghc-pkg's Cabal dependencies
3781Ian Lynagh <igloo@earth.li>**20100704184155
3782 We no longer support building with a compiler that doesn't come with
3783 base 4.
3784]
3785[Use Cabal to configure the dist-install ghc-pkg; fixes trac #4156
3786Ian Lynagh <igloo@earth.li>**20100704132612]
3787[Remove dead code (standalone deriving flag no longer used in parser)
3788Ian Lynagh <igloo@earth.li>**20100701162058]
3789[LLVM: Use the inbounds keyword for getelementptr instructions.
3790David Terei <davidterei@gmail.com>**20100702160511
3791 Ignore-this: 3708e658a4c82b78b1402393f4405541
3792]
3793[threadPaused: fix pointer arithmetic
3794Simon Marlow <marlowsd@gmail.com>**20100701085046
3795 Ignore-this: b78210e5d978f18ffd235f1c78a55a23
3796 Noticed by Henrique Ferreiro <hferreiro@udc.es>, thanks!
3797]
3798[LLVM: Change more operations to use getelementptr
3799David Terei <davidterei@gmail.com>**20100701161856
3800 Ignore-this: fb24eb124e203f50680c6fec3ff9fe7d
3801]
3802[Add the haskell2010 package
3803Simon Marlow <marlowsd@gmail.com>**20100630125532
3804 Ignore-this: e9b011313f283a8ff2fcda7d029a01f
3805]
3806[LLVM: Use getelementptr instruction for a lot of situations
3807David Terei <davidterei@gmail.com>**20100630181157
3808 Ignore-this: 34d314dd8dffad9bdcffdc525261a49d
3809 
3810 LLVM supports creating pointers in two ways, firstly through
3811 pointer arithmetic (by casting between pointers and ints)
3812 and secondly using the getelementptr instruction. The second way
3813 is preferable as it gives LLVM more information to work with.
3814 
3815 This patch changes a lot of pointer related code from the first
3816 method to the getelementptr method.
3817]
3818[remove out of date comments; point to the wiki
3819Simon Marlow <marlowsd@gmail.com>**20100625100313
3820 Ignore-this: 95f363a373534b9471b1818102ec592d
3821]
3822[NCG: allocatableRegs is only giving us 8 SSE regs to allocate to
3823benl@ouroborus.net**20100629054321
3824 Ignore-this: b3e0fa0b4ce988a0258dc12261989ee0
3825]
3826[LLVM: Use intrinsic functions for pow, sqrt, sin, cos
3827David Terei <davidterei@gmail.com>**20100628182949
3828 Ignore-this: 98a0befaca3fe2b36d710d8ff9f062c4
3829 
3830 Instead of calling the C library for these Cmm functions
3831 we use intrinsic functions provided by llvm. LLVM will
3832 then either create a compile time constant if possible, or
3833 use a cpu instruction or as a last resort call the C
3834 library.
3835]
3836[LLVM: Fix test '2047' under linux-x64
3837David Terei <davidterei@gmail.com>**20100628165256
3838 Ignore-this: 41735d4f431a430db636621650ccd71e
3839]
3840[LLVM: Fix test 'ffi005' under linux-x64
3841David Terei <davidterei@gmail.com>**20100628155355
3842 Ignore-this: 841f3142c63cc898ac4c3f89698a837e
3843]
3844[LLVM: Update to use new fp ops introduced in 2.7
3845David Terei <davidterei@gmail.com>**20100628144037
3846 Ignore-this: 5dd2e5964e3c039d297ed586841e706b
3847]
3848[Add noalias and nocapture attributes to pointer stg registers
3849David Terei <davidterei@gmail.com>**20100628115120
3850 Ignore-this: 492a1e723cb3a62498d240d7de92dd7
3851 
3852 At the moment this gives a very slight performance boost of around 1 - 2%.
3853 Future changes to the generated code though so that pointers are kept as
3854 pointers more often instead of being cast to integer types straight away
3855 should hopefully improve the benefit this brings.
3856 
3857]
3858[during shutdown, only free the heap if we waited for foreign calls to exit
3859Simon Marlow <marlowsd@gmail.com>**20100628090536
3860 Ignore-this: d545384a4f641d701455d08ef1217479
3861]
3862[Fix typo in -ddump-pass's document.
3863shelarcy <shelarcy@gmail.com>**20100620070759
3864 Ignore-this: f4f1ddb53f147949e948147d89190c37
3865]
3866[Add #undefs for posix source symbols when including papi.h
3867dmp@rice.edu**20100624163514
3868 Ignore-this: 8a1cba21b880d12a75a75f7e96882053
3869 
3870 Validation fails when validating with PAPI support (i.e. GhcRtsWithPapi  = YES
3871 in validate.mk).  The problem is that the posix symbols are defined by a header
3872 included from papi.h. Compilation then fails because these symbols are
3873 redefined in PosixSource.h.
3874 
3875 This patch adds an undefine for the posix symbols after including papi.h and
3876 before including PosixSource.h. The #undefines are localized to Papi.c since
3877 that is the only case where they are getting defined twice.
3878]
3879[Use machdepCCOpts in runPhase_MoveBinary; fixes trac #3952
3880Ian Lynagh <igloo@earth.li>**20100625220953]
3881[LLVM: Fix bug with calling tail with empty list
3882David Terei <davidterei@gmail.com>**20100625115729
3883 Ignore-this: 46b4b32c8d92372a2d49794a96fe1613
3884]
3885[Fix warnings
3886benl@ouroborus.net**20100624091339
3887 Ignore-this: 5ba4bbd6abb9c9d1fb8c5d21ab73f218
3888]
3889[NCG: Comments and formatting only
3890benl@ouroborus.net**20100624083121
3891 Ignore-this: 86002e72c30d06bcc876d8c49f4caa5a
3892]
3893[NCG: Do the actual reversing of SCCs
3894benl@ouroborus.net**20100624082717
3895 Ignore-this: 12d2027ea118e751fbb48b27126150ef
3896]
3897[NCG: Fix dumping of graphs in regalloc stats for graph allocator
3898benl@ouroborus.net**20100624082625
3899 Ignore-this: 2b971bc9e0318099a9afb0e0db135730
3900]
3901[NCG: Reverse SCCs after each round in the graph allocator
3902benl@ouroborus.net**20100624082437
3903 Ignore-this: f0152e4039d6f16f7b5a99b286538116
3904]
3905[NCG: Don't actually complain on unreachable code blocks
3906benl@ouroborus.net**20100624081445
3907 Ignore-this: e7335ae6120917cb858c38c7c6da8e24
3908]
3909[NCG: Do explicit check for precondition of computeLiveness
3910benl@ouroborus.net**20100624080747
3911 Ignore-this: e7053c4e5e4c3c746b5ebf016913424a
3912 
3913  computeLiveness requires the SCCs of blocks to be in reverse dependent
3914  order, and if they're not it was silently giving bad liveness info,
3915  yielding a bad allocation.
3916 
3917  Now it complains, loudly.
3918]
3919[NCG: Fix off-by-one error in realRegSqueeze
3920benl@ouroborus.net**20100623095813
3921 Ignore-this: ab0698686d4c250da8e207f734f8252d
3922]
3923[NCG: Handle stripping of liveness info from procs with no blocks (like stg_split_marker)
3924benl@ouroborus.net**20100623091209
3925 Ignore-this: c0319b6cc62ec713afe4eb03790406e3
3926]
3927[NCG: Emit a warning on unreachable code block instead of panicing
3928benl@ouroborus.net**20100623085002
3929 Ignore-this: d20314b79e3c31e764ed4cd97290c696
3930]
3931[NCG: Remember to keep the entry block first when erasing liveness info
3932Ben.Lippmeier@anu.edu.au**20090917104429
3933 Ignore-this: 1b0c1df19d622858d50ffb6a01f2cef0
3934]
3935[NCG: Refactor representation of code with liveness info
3936Ben.Lippmeier@anu.edu.au**20090917090730
3937 Ignore-this: 2aebb3b02ebd92e547c5abad9feb0f0d
3938 
3939  * I've pushed the SPILL and RELOAD instrs down into the
3940    LiveInstr type to make them easier to work with.
3941 
3942  * When the graph allocator does a spill cycle it now just
3943    re-annotates the LiveCmmTops instead of converting them
3944    to NatCmmTops and back.
3945 
3946  * This saves working out the SCCS again, and avoids rewriting
3947    the SPILL and RELOAD meta instructions into real machine
3948    instructions.
3949]
3950[NCG: Add sanity checking to linear allocator
3951Ben.Lippmeier@anu.edu.au**20090917090335
3952 Ignore-this: 5a442be8b5087d04bc8b58dffa9ea080
3953 If there are are unreachable basic blocks in the native code then the
3954 linear allocator might loop. Detect this case and panic instead.
3955]
3956[NCG: Refactor LiveCmmTop to hold a list of SCCs instead of abusing ListGraph
3957Ben.Lippmeier@anu.edu.au**20090917060332
3958 Ignore-this: 3fec8d69ed0f760e53a202f873d5d9cb
3959]
3960[NCG: Allow the liveness map in a LiveInfo to be Nothing
3961Ben.Lippmeier@anu.edu.au**20090917043937
3962 Ignore-this: 5f82422d54d1b0ffc0589eb7e82fb7a4
3963]
3964[NCG: Also show the result of applying coalesings with -ddump-asm-regalloc-stages
3965Ben.Lippmeier.anu.edu.au**20090917034427
3966 Ignore-this: 76bd6d5ca43adb2167cb25832cbaa80b
3967]
3968[Fix panic when running "ghc -H"; trac #3364
3969Ian Lynagh <igloo@earth.li>**20100624234011
3970 The problem is that showing SDoc's looks at the static flags global
3971 variables, but those are panics while we are parsing the static flags.
3972 We work around this by explicitly using a fixed prettyprinter style.
3973]
3974[Allow for stg registers to have pointer type in llvm BE.
3975David Terei <davidterei@gmail.com>**20100621175839
3976 Ignore-this: fc09b1a8314aef0bde945c77af1124fb
3977 
3978 Before all the stg registers were simply a bit type or
3979 floating point type but now they can be declared to have
3980 a pointer type to one of these. This will allow various
3981 optimisations in the future in llvm since the type is
3982 more accurate.
3983]
3984[Add support for parameter attributes to the llvm BE binding
3985David Terei <davidterei@gmail.com>**20100624111744
3986 Ignore-this: 77f3c0c7bf8f81c4a154dc835ae7bcba
3987 
3988 These allow annotations of the code produced by the backend
3989 which should bring some perforamnce gains. At the moment
3990 the attributes aren't being used though.
3991]
3992[Cast some more nats to StgWord to be on the safe side
3993Simon Marlow <marlowsd@gmail.com>**20100624105700
3994 Ignore-this: e6176683856f9872fdeb2358bb065bb8
3995 And add a comment about the dangers of int overflow
3996]
3997[comments only
3998Simon Marlow <marlowsd@gmail.com>**20100624105105
3999 Ignore-this: fc8f762f4c3a5ffca2f8da2bc63ac2a4
4000]
4001[Fix an arithmetic overflow bug causing crashes with multi-GB heaps
4002Simon Marlow <marlowsd@gmail.com>**20100624104654
4003 Ignore-this: 67210755aa098740ff5230347be0fd5d
4004]
4005[Add support for collecting PAPI native events
4006dmp@rice.edu**20100622195953
4007 Ignore-this: 7269f9c4dfb2912a024eb632200fcd1
4008 
4009 This patch extends the PAPI support in the RTS to allow collection of native
4010 events. PAPI can collect data for native events that are exposed by the
4011 hardware beyond the PAPI present events. The native events supported on your
4012 hardware can found by using the papi_native_avail tool.
4013 
4014 The RTS already allows users to specify PAPI preset events from the command
4015 line. This patch extends that support to allow users to specify native events.
4016 The changes needed are:
4017 
4018 1) New option (#) for the RTS PAPI flag for native events. For example, to
4019    collect the native event 0x40000000, use ./a.out +RTS -a#0x40000000 -sstderr
4020 
4021 2) Update the PAPI_FLAGS struct to store whether the user specified event is a
4022    papi preset or a native event
4023 
4024 3) Update init_countable_events function to add the native events after parsing
4025    the event code and decoding the name using PAPI_event_code_to_name
4026 
4027]
4028[Don't warn about unused bindings with parents in .hs-boot files; trac #3449
4029Ian Lynagh <igloo@earth.li>**20100624110351]
4030[fix the home_imps filter to allow for 'import "this" <module>'
4031Simon Marlow <marlowsd@gmail.com>**20100621125535
4032 Ignore-this: da4e605b0513afc32a4e7caa921a2c76
4033 In the PackageImports extension, import "this" means "import from the
4034 current package".
4035]
4036[Use the standard C wrapper code for the ghc-$version.exe wrapper
4037Ian Lynagh <igloo@earth.li>**20100622202859
4038 Ignore-this: 60cd3e6db3afb63e6ba9e2db3b033580
4039]
4040[Don't rely on "-packagefoo" working; use "-package foo" instead
4041Ian Lynagh <igloo@earth.li>**20100622202547]
4042[Remove unnecessary C #includes
4043Ian Lynagh <igloo@earth.li>**20100622172919]
4044[Make the ghci.exe wrapper call the right ghc.exe
4045Ian Lynagh <igloo@earth.li>**20100622172247]
4046[More updates to datalayout description in llvm BE
4047David Terei <davidterei@gmail.com>**20100622165339
4048 Ignore-this: b0c604fe7673b0aa7c7064694d574437
4049]
4050[Remove LlvmAs phase as the llvm opt tool now handles this phase
4051David Terei <davidterei@gmail.com>**20100622144044
4052 Ignore-this: b9fd8f959702b6af014e2fa654bede3
4053 
4054 This phase originally invoked the llvm-as tool that turns a textual
4055 llvm assembly file into a bit code file for the rest of llvm to deal
4056 with. Now the llvm opt tool can do this itself, so we don't need to
4057 use llvm-as anymore.
4058]
4059[Update datalayout info in llvm BE
4060David Terei <davidterei@gmail.com>**20100622123457
4061 Ignore-this: 89b043d211225dcd819f30549afe1840
4062]
4063[Fix handling of float literals in llvm BE
4064David Terei <davidterei@gmail.com>**20100622121642
4065 Ignore-this: a3b5f382ad4b5a426ad4b581664506fa
4066]
4067[Declare some top level globals to be constant when appropriate
4068David Terei <davidterei@gmail.com>**20100621174954
4069 Ignore-this: 44832f65550d4f995d11c01cc1affef5
4070 
4071 This involved removing the old constant handling mechanism
4072 which was fairly hard to use. Now being constant or not is
4073 simply a property of a global variable instead of a separate
4074 type.
4075]
4076[Reduce the number of passes over the cmm in llvm BE
4077David Terei <davidterei@gmail.com>**20100621125220
4078 Ignore-this: cb2f4e54e8d0f982d5087fbeee35c73c
4079]
4080[Fix negate op not working for -0 in llvm backend
4081David Terei <davidterei@gmail.com>**20100621123606
4082 Ignore-this: c5d76e5cffa781fed074137851b1347f
4083]
4084[ROLLBACK: picCCOpts: -dynamic should not entail -optc-fPIC
4085Simon Marlow <marlowsd@gmail.com>**20100621100409
4086 Ignore-this: f2fac7df33d3919199befc59bd455414
4087 and add a comment to explain why it was wrong.  This fixes the dyn
4088 test failures that sprang up recently.
4089]
4090[Check files are really created in libffi
4091Ian Lynagh <igloo@earth.li>**20100620163724
4092 when we think that the libffi build creates them, so they just depend
4093 on the libffi build stamp.
4094]
4095[Improve the missing-import-list warning
4096Ian Lynagh <igloo@earth.li>**20100620124320
4097 Ignore-this: 551e5fdf2dfb56b49d249e0cebaa6115
4098]
4099[Tweak missing-import-list warning
4100Ian Lynagh <igloo@earth.li>**20100620122622
4101 Ignore-this: 360cdf59ae13d66ded181129325506c4
4102]
4103[trac #1789 (warnings for missing import lists)
4104amsay@amsay.net**20100618150649
4105 Ignore-this: b0b0b1e048fbca0817c1e6fade1153fa
4106]
4107[Refix docs for sizeofByteArray#/sizeofMutableByteArray# (#3800)
4108Ian Lynagh <igloo@earth.li>**20100620103749]
4109[Remove some old commented out code
4110Ian Lynagh <igloo@earth.li>**20100620000459]
4111[SET_ARR_HDR's last argument is now a number of bytes, rather than words
4112Ian Lynagh <igloo@earth.li>**20100619235214
4113 This avoids unnecessary work and potential loss of information
4114]
4115[Replace an (incorrect) bytes-to-words calculation with ROUNDUP_BYTES_TO_WDS
4116Ian Lynagh <igloo@earth.li>**20100619234310]
4117[FIX #38000 Store StgArrWords payload size in bytes
4118Antoine Latter <aslatter@gmail.com>**20100101183346
4119 Ignore-this: 7bf3ab4fc080c46311fc10b179361bb6
4120]
4121[Add win32 datalayout support to llvm backend
4122David Terei <davidterei@gmail.com>**20100618131733
4123 Ignore-this: 4b7bffaa8ef38c628ab852c1a6c1c009
4124]
4125[Remove unused 'ddump-opt-llvm' flag
4126David Terei <davidterei@gmail.com>**20100618101237
4127 Ignore-this: f78467496d986897e49d82646ee2907e
4128]
4129[generate "movl lbl(%reg1), %reg2" instructions, better codegen for -fPIC
4130Simon Marlow <marlowsd@gmail.com>**20100618082258
4131 Ignore-this: a25567ebff9f575303ddc8f2deafebbf
4132]
4133[joinToTargets: fix a case of panic "handleComponent cyclic"
4134Simon Marlow <marlowsd@gmail.com>**20100618082147
4135 Ignore-this: 765baeefbb5a41724004acd92405cecc
4136]
4137[comment typo
4138Simon Marlow <marlowsd@gmail.com>**20100618082102
4139 Ignore-this: e495610b7dd5ec30b02938638b56cb7
4140]
4141[Add support of TNTC to llvm backend
4142David Terei <davidterei@gmail.com>**20100618093205
4143 Ignore-this: 2c27d21668374a5b0d5e844882c69439
4144 
4145 We do this through a gnu as feature called subsections,
4146 where you can put data/code into a numbered subsection
4147 and those subsections will be joined together in descending
4148 order by gas at compile time.
4149]
4150[Don't automatically insert a -fvia-C flag in an unregisterised compiler
4151Ian Lynagh <igloo@earth.li>**20100617190901
4152 Ignore-this: eb25a9a338fade9e17c153da7c5f27e9
4153 The default object mode is already HscC, so it's unnecessary, and
4154 -fvia-C generates a deprecated flag warning now.
4155]
4156[In PosixSource.h, conditionally define things based on platform
4157Ian Lynagh <igloo@earth.li>**20100617174912
4158 This may not be ideal, but it should get GHC building on all platforms
4159 again.
4160]
4161[disable -fPIC for the GC for performance reasons
4162Simon Marlow <marlowsd@gmail.com>**20100617140025
4163 Ignore-this: c7c152bbff71ef7891eaee8ff39fc281
4164 see comment for details
4165]
4166[picCCOpts: -dynamic should not entail -optc-fPIC
4167Simon Marlow <marlowsd@gmail.com>**20100617115259
4168 Ignore-this: d71e42bd56e4bd107d2c431b801855e5
4169]
4170[Make getAllocations() visible
4171Simon Marlow <marlowsd@gmail.com>**20100617113259
4172 Ignore-this: 1b7fb38a01358c0acbe8987df07d23f2
4173]
4174[Fix the symbol visibility pragmas
4175Simon Marlow <marlowsd@gmail.com>**20100617105758
4176 Ignore-this: 76552500865473a1dbebbc1cc2def9f0
4177]
4178[pick up changes to $(GhcStage1HcOpts) without re-configuring the ghc package
4179Simon Marlow <marlowsd@gmail.com>**20100616124718
4180 Ignore-this: afb56d5560c813051285607fefb15493
4181]
4182[Fix bindisttest Makefile
4183Ian Lynagh <igloo@earth.li>**20100616205611
4184 Ignore-this: 39cd352152422f378572fc3859c5a377
4185]
4186[Remove some more unused make variables
4187Ian Lynagh <igloo@earth.li>**20100616180519]
4188[Convert some more variable names to FOO_CMD, for consistency
4189Ian Lynagh <igloo@earth.li>**20100616175916]
4190[Rename some variables from FOO to FOO_CMD
4191Ian Lynagh <igloo@earth.li>**20100616161108
4192 This fixes a problem with commands like gzip, where if $GZIP is exported
4193 in the environment, then when make runs a command it'll put the Makefile
4194 variable's value in the environment. But gzip treats $GZIP as arguments
4195 for itself, so when we run gzip it thinks we're giving it "gzip" as an
4196 argument.
4197]
4198[Make the "show" target work anywhere in the build tree
4199Ian Lynagh <igloo@earth.li>**20100616122910
4200 Ignore-this: 299d40cbe16112accd9f14e56fa12158
4201]
4202[Change ghc-pwd's license to a string Cabal recognises
4203Ian Lynagh <igloo@earth.li>**20100615204015
4204 Ignore-this: c935b6ad7f605aab0168997a90b40fc6
4205]
4206[fix warning
4207Simon Marlow <marlowsd@gmail.com>**20100604205933
4208 Ignore-this: 2aaa4ed6a8b9ae1e39adc4696aaf14a3
4209]
4210[--install-signal-handles=no does not affect the timer signal (#1908)
4211Simon Marlow <marlowsd@gmail.com>**20100527214627
4212 Ignore-this: b0c51f1abdb159dc360662485095a11a
4213]
4214[Small optimisation: allocate nursery blocks contiguously
4215Simon Marlow <marlowsd@gmail.com>**20100509194928
4216 Ignore-this: e650e99e9ea9493d2efb245d565beef4
4217 This lets automatic prefetching work better, for a tiny performance boost
4218]
4219[fix -fforce-recomp setting: module is PrimOp, not PrimOps
4220Simon Marlow <marlowsd@gmail.com>**20100507084507
4221 Ignore-this: f76e0d9b643682ec0e8fb7d91afdea68
4222]
4223[it should be an error to use relative directories (#4134)
4224Simon Marlow <marlowsd@gmail.com>**20100615151740
4225 Ignore-this: 2068021701832e018ca41b22877921d5
4226]
4227[missing include-dirs or library-dirs is only a warning now (#4104)
4228Simon Marlow <marlowsd@gmail.com>**20100615151702
4229 Ignore-this: e3114123cef147bbd28ccb64581a1afb
4230]
4231[fix #3822: desugaring case command in arrow notation
4232Ross Paterson <ross@soi.city.ac.uk>**20100615225110
4233 Ignore-this: 477d6c460b4174b94b4cd113fa5b9d19
4234 
4235 Get the set of free variables from the generated case expression:
4236 includes variables in the guards and decls that were missed before,
4237 and is also a bit simpler.
4238]
4239[Deprecate the -fvia-C flag; trac #3232
4240Ian Lynagh <igloo@earth.li>**20100615151836
4241 Ignore-this: c2452b2648bf7e44546465c1b964fce
4242]
4243[Avoid using the new ~~ perl operator in the mangler
4244Ian Lynagh <igloo@earth.li>**20100615151236
4245 Ignore-this: 709a7ba4e514b1596841b3ba7e5c6cc
4246]
4247[stmAddInvariantToCheck: add missing init of invariant->lock (#4057)
4248Simon Marlow <marlowsd@gmail.com>**20100615123643
4249 Ignore-this: 3b132547fa934cecf71a846db2a5f70e
4250]
4251[Add new LLVM code generator to GHC. (Version 2)
4252David Terei <davidterei@gmail.com>**20100615094714
4253 Ignore-this: 4dd2fe5854b64a3f0339d484fd5c238
4254 
4255 This was done as part of an honours thesis at UNSW, the paper describing the
4256 work and results can be found at:
4257 
4258 http://www.cse.unsw.edu.au/~pls/thesis/davidt-thesis.pdf
4259 
4260 A Homepage for the backend can be found at:
4261 
4262 http://hackage.haskell.org/trac/ghc/wiki/Commentary/Compiler/Backends/LLVM
4263 
4264 Quick summary of performance is that for the 'nofib' benchmark suite, runtimes
4265 are within 5% slower than the NCG and generally better than the C code
4266 generator.  For some code though, such as the DPH projects benchmark, the LLVM
4267 code generator outperforms the NCG and C code generator by about a 25%
4268 reduction in run times.
4269 
4270]
4271[Fix Trac #4127: build GlobalRdrEnv in GHCi correctly
4272simonpj@microsoft.com**20100615070626
4273 Ignore-this: d907e3bfa7882878cea0af172aaf6e84
4274 
4275 GHCi was building its GlobalRdrEnv wrongly, so that the
4276 gre_par field was bogus.  That in turn fooled the renamer.
4277 The fix is easy: use the right function!  Namely, call
4278 RnNames.gresFromAvail rather than availsToNameSet.
4279]
4280[Comments, and improvement to pretty-printing of HsGroup
4281simonpj@microsoft.com**20100615070409
4282 Ignore-this: ec8358f2485370b20226a97ec84e9024
4283]
4284[Don't reverse bindings in rnMethodBinds (fix Trac #4126)
4285simonpj@microsoft.com**20100614163935
4286 Ignore-this: a6ffbb5af6f51b142ed0aeae8ee5e3a9
4287]
4288[Fix Trac #4120: generate a proper coercion when unifying forall types
4289simonpj@microsoft.com**20100614134311
4290 Ignore-this: 601592bb505305f1954cbe730f168da4
4291 
4292 This was just a blatant omission, which hasn't come up before.
4293 Easily fixed, happily.
4294]
4295[Use mkFunTy to ensure that invariants are respected
4296simonpj@microsoft.com**20100614134159
4297 Ignore-this: 67dcada7a4e8d9927581cd77af71b6f
4298]
4299[Remove redundant debug code
4300simonpj@microsoft.com**20100601154151
4301 Ignore-this: e6ff11c04c631cf6aac73788cbcf02b5
4302]
4303[Fix Trac #4099: better error message for type functions
4304simonpj@microsoft.com**20100531140413
4305 Ignore-this: 3f53ca98cf770577818b9c0937482577
4306 
4307 Now we only want about "T is a type function and might not be
4308 injective" when matchin (T x) against (T y), which is the case
4309 that is really confusing.
4310]
4311[Gruesome fix in CorePrep to fix embarassing Trac #4121
4312simonpj@microsoft.com**20100614132726
4313 Ignore-this: fe82d15474afaac3e6133adfd7a7e055
4314 
4315 This is a long-lurking bug that has been flushed into
4316 the open by other arity-related changes.  There's a
4317 long comment
4318 
4319      Note [CafInfo and floating]
4320 
4321 to explain. 
4322 
4323 I really hate the contortions we have to do through to keep correct
4324 CafRef information on top-level binders.  The Right Thing, I believe,
4325 is to compute CAF and arity information later, and merge it into the
4326 interface-file information when the latter is generated.
4327 
4328 But for now, this hackily fixes the problem.
4329]
4330[Fix a bug in CorePrep that meant output invariants not satisfied
4331simonpj@microsoft.com**20100531150013
4332 Ignore-this: d34eb36d8877d3caf1cf2b20de426abd
4333 
4334 In cpePair I did things in the wrong order so that something that
4335 should have been a CprRhs wasn't.  Result: a crash in CoreToStg.
4336 Fix is easy, and I added more informative type signatures too.
4337]
4338[Robustify the treatement of DFunUnfolding
4339simonpj@microsoft.com**20100531145332
4340 Ignore-this: 8f5506ada4d89f6ab8ad1e8c3ffb09ba
4341 
4342 See Note [DFun unfoldings] in CoreSyn.  The issue here is that
4343 you can't tell how many dictionary arguments a DFun needs just
4344 from looking at the Arity of the DFun Id: if the dictionary is
4345 represented by a newtype the arity might include the dictionary
4346 and value arguments of the (single) method.
4347 
4348 So we need to record the number of arguments need by the DFun
4349 in the DFunUnfolding itself.  Details in
4350    Note [DFun unfoldings] in CoreSyn
4351]
4352[Fix spelling in comment
4353simonpj@microsoft.com**20100614132259
4354 Ignore-this: bbf0d55f2e5f10ef9c74592c12f9201c
4355]
4356[Update docs on view patterns
4357simonpj@microsoft.com**20100614074801
4358 Ignore-this: 8617b9078800d4942d71f142a5b6c831
4359]
4360[Fix printing of splices; part of #4124
4361Ian Lynagh <igloo@earth.li>**20100613154838
4362 Just putting parens around non-atomic expressions isn't sufficient
4363 for splices, as only the $x and $(e) forms are valid input.
4364]
4365[In ghci, catch IO exceptions when calling canonicalizePath
4366Ian Lynagh <igloo@earth.li>**20100613134627
4367 We now get an exception if the path doesn't exist
4368]
4369[Whitespace only
4370Ian Lynagh <igloo@earth.li>**20100612213119]
4371[Whitespace only
4372Ian Lynagh <igloo@earth.li>**20100612165450]
4373[Update ghci example output in user guide; patch from YitzGale in #4111
4374Ian Lynagh <igloo@earth.li>**20100612162250]
4375[Fix #4131 missing UNTAG_CLOSURE in messageBlackHole()
4376benl@ouroborus.net**20100611044614]
4377[messageBlackHole: fix deadlock bug caused by a missing 'volatile'
4378Simon Marlow <marlowsd@gmail.com>**20100610080636
4379 Ignore-this: 3cda3054bb45408aa9bd2d794b69c938
4380]
4381[Pass --no-tmp-comp-dir to Haddock (see comment)
4382Simon Marlow <marlowsd@gmail.com>**20100604083214
4383 Ignore-this: bfa4d74038637bd149f4d878b4eb8a87
4384]
4385[Track changes to DPH libs
4386Roman Leshchinskiy <rl@cse.unsw.edu.au>**20100607052903
4387 Ignore-this: 4dbc3f8418af3e74b3fc4f9a9dfe7764
4388]
4389[Track changes to DPH libs
4390Roman Leshchinskiy <rl@cse.unsw.edu.au>**20100607012642
4391 Ignore-this: 5d4e498171a3c57ab02621bfaea82cff
4392]
4393[In ghc-pkg, send warnings to stderr
4394Ian Lynagh <igloo@earth.li>**20100606161726
4395 Ignore-this: 56927d13b5e1c1ce2752734f0f9b665b
4396]
4397[Re-add newlines to enable layout for multi-line input.
4398Ian Lynagh <igloo@earth.li>**20100602180737
4399 Patch from Adam Vogt <vogt.adam@gmail.com>
4400 Partial fix for #3984
4401]
4402[Don't use unnecessary parens when printing types (Fix Trac 4107)
4403simonpj@microsoft.com**20100604110143
4404 Ignore-this: a833714ab13013c4345b222f4e87db1d
4405 
4406    f :: Eq a => a -> a
4407 rather than
4408    f :: (Eq a) => a -> a
4409]
4410[Track DPH library changes
4411Roman Leshchinskiy <rl@cse.unsw.edu.au>**20100604005728
4412 Ignore-this: 32bc2fbea6ad975e89545d4c42fd7c30
4413]
4414[fix --source-entity option passed to Haddock: we needed to escape a #
4415Simon Marlow <marlowsd@gmail.com>**20100603125459
4416 Ignore-this: d52ae6188b510c482bcebb23f0e553ae
4417]
4418[__stg_EAGER_BLACKHOLE_INFO -> __stg_EAGER_BLACKHOLE_info (#4106)
4419Simon Marlow <marlowsd@gmail.com>**20100602091419
4420 Ignore-this: 293315ac8f86fd366b8d61992ecc7961
4421]
4422[Add xhtml package (a new dependency of Haddock; not installed/shipped)
4423Simon Marlow <marlowsd@gmail.com>**20100602090101
4424 Ignore-this: af0ac8b91abe98f7fdb624ea0a4dee20
4425]
4426[Use UserInterrupt rather than our own Interrupted exception (#4100)
4427Simon Marlow <marlowsd@gmail.com>**20100602082345
4428 Ignore-this: 1909acf2f452593138b9f85024711714
4429]
4430[Add the global package DB to ghc --info (#4103)
4431Simon Marlow <marlowsd@gmail.com>**20100602082233
4432 Ignore-this: fd5c0e207e70eb0f62606c45dc5b8124
4433]
4434[rts/sm/GC.c: resize_generations(): Remove unneeded check of number of generations.
4435Marco Túlio Gontijo e Silva <marcot@debian.org>**20100528115612
4436 Ignore-this: 6f1bea62917c01c7adac636146132c97
4437 
4438 This "if" is inside another "if" which checks for RtsFlags.GcFlags.generations
4439 > 1, so testing this again is redundant, assuming the number of generations
4440 won't change during program execution.
4441]
4442[rts/sm/BlockAlloc.c: Small comment correction.
4443Marco Túlio Gontijo e Silva <marcot@debian.org>**20100526205839
4444 Ignore-this: bd2fcd4597cc872d80b0e2eeb1c3998a
4445]
4446[rts/sm/GC.c: Annotate constants.
4447Marco Túlio Gontijo e Silva <marcot@debian.org>**20100526205707
4448 Ignore-this: f232edb89383564d759ed890a18f602f
4449]
4450[includes/rts/storage/GC.h: generation_: n_words: Improve comment.
4451Marco Túlio Gontijo e Silva <marcot@debian.org>**20100526204615
4452 Ignore-this: f5d5feefa8f7b552303978f1804fea23
4453]
4454[Add PPC_RELOC_LOCAL_SECTDIFF support; patch from PHO in #3654
4455Ian Lynagh <igloo@earth.li>**20100601204211
4456 Ignore-this: 51293b7041cdce3ce7619ef11cf7ceb
4457]
4458[powerpc-apple-darwin now supports shared libs
4459Ian Lynagh <igloo@earth.li>**20100601173325]
4460[PIC support for PowerPC
4461pho@cielonegro.org**20100508143900
4462 Ignore-this: 3673859a305398c4acae3f4d7c997615
4463 
4464 PPC.CodeGen.getRegister was not properly handling PicBaseReg.
4465 It seems working with this patch, but I'm not sure this change is correct.
4466]
4467[Vectoriser: only treat a function as scalar if it actually computes something
4468Roman Leshchinskiy <rl@cse.unsw.edu.au>**20100601045630
4469 Ignore-this: e5d99a6ddb62052e3520094a5af47552
4470]
4471[Add a release notes file for 6.14.1
4472Ian Lynagh <igloo@earth.li>**20100530171117
4473 Ignore-this: 1941e6d3d1f4051b69ca2f17a1cf84d6
4474]
4475[Check dblatex actually creates the files we tell it to
4476Ian Lynagh <igloo@earth.li>**20100530171043
4477 Ignore-this: ccc72caea2313be05cbac59bb54c0603
4478 If it fails, it still exits successfully.
4479]
4480[Add darwin to the list of OSes for which we use mmap
4481Ian Lynagh <igloo@earth.li>**20100529145016
4482 Ignore-this: a86d12a3334aaaafc86f7af9dbb0a7ae
4483 Patch from Barney Stratford
4484]
4485[Simplify the CPP logic in rts/Linker.c
4486Ian Lynagh <igloo@earth.li>**20100529144929
4487 Ignore-this: 1288f5b752cc1ab8b1c90cfd0ecfdf68
4488]
4489[Fix validate on OS X
4490Ian Lynagh <igloo@earth.li>**20100529154726]
4491[OS X x86_64 fix from Barney Stratford
4492Ian Lynagh <igloo@earth.li>**20100529122440]
4493[OS X 64 installer fixes from Barney Stratford
4494Ian Lynagh <igloo@earth.li>**20100528234935]
4495[fix warning
4496Simon Marlow <marlowsd@gmail.com>**20100525155812
4497 Ignore-this: f34eee3fe3d89579fd8d381c91ced750
4498]
4499[Fix doc bugs (#4071)
4500Simon Marlow <marlowsd@gmail.com>**20100525155728
4501 Ignore-this: aa25be196de567de360075022a1942f7
4502]
4503[Make sparks into weak pointers (#2185)
4504Simon Marlow <marlowsd@gmail.com>**20100525150435
4505 Ignore-this: feea0bb5006007b82c932bc3006124d7
4506 The new strategies library (parallel-2.0+, preferably 2.2+) is now
4507 required for parallel programming, otherwise parallelism will be lost.
4508]
4509[If you say 'make' or 'make stage=2' here, pretend we're in the ghc dir
4510Simon Marlow <marlowsd@gmail.com>**20100525085301
4511 Ignore-this: 78b740337aa460915c812cbbcdae5321
4512]
4513[Another attempt to get these #defines right
4514Simon Marlow <marlowsd@gmail.com>**20100525154313
4515 Ignore-this: 460ca0c47d81cd25eae6542114f67899
4516 Apparently on Solaris it is an error to omit _ISOC99_SOURCE when using
4517 _POSIX_C_SOURCE==200112L.
4518]
4519[Add configure flags for the location of GMP includes/library; fixes #4022
4520Ian Lynagh <igloo@earth.li>**20100525221616
4521 Ignore-this: fc3060caf995d07274ec975eeefbdf3e
4522]
4523[Refactor pretty printing of TyThings to fix Trac #4015
4524simonpj@microsoft.com**20100525153126
4525 Ignore-this: 8f15053b7554f62caa84201d2e4976d2
4526]
4527[When haddocking, we need the dependencies to have been built
4528Ian Lynagh <igloo@earth.li>**20100525145830
4529 as haddock loads the .hi files with the GHC API.
4530]
4531[Fix profiling output; spotted by jlouis
4532Ian Lynagh <igloo@earth.li>**20100525111217
4533 We were outputing the number of words allocated in a column titled "bytes".
4534]
4535[Improve printing of TyThings; fixes Trac #4087
4536simonpj@microsoft.com**20100525114045
4537 Ignore-this: da2a757a533454bba80b9b77cc5a771
4538]
4539[Spelling in comments
4540simonpj@microsoft.com**20100525114001
4541 Ignore-this: 270f3da655e526cf04e27db7a01e29c0
4542]
4543[Refactor (again) the handling of default methods
4544simonpj@microsoft.com**20100525113910
4545 Ignore-this: 6686f6cdb878d57abf6b49fec64fcbb1
4546 
4547 This patch fixes Trac #4056, by
4548 
4549  a) tidying up the treatment of default method names
4550  b) removing the 'module' argument to newTopSrcBinder
4551 
4552 The details aren't that interesting, but the result
4553 is much tidier. The original bug was a 'nameModule' panic,
4554 caused by trying to find the module of a top-level name.
4555 But TH quotes generate Internal top-level names that don't
4556 have a module, and that is generally a good thing. 
4557 
4558 Fixing that in turn led to the default-method refactoring,
4559 which also makes the Name for a default method be handled
4560 in the same way as other derived names, generated in BuildTyCl
4561 via a call newImplicitBinder.  Hurrah.
4562]
4563[Don't do SpecConstr on NOINLINE things (Trac #4064)
4564simonpj@microsoft.com**20100525112807
4565 Ignore-this: 452be0a2cef0042fb67275c2827b5f72
4566 
4567 Since the RULE from specialising gets the same Activation as
4568 the inlining for the Id itself there's no point in specialising
4569 a NOINLINE thing, because the rule will be permanently switched
4570 off.
4571 
4572 See Note [Transfer activation] in SpecConstr
4573 and Note [Auto-specialisation and RULES] in Specialise.
4574]
4575[Change our #defines to work on FreeBSD too
4576Simon Marlow <marlowsd@gmail.com>**20100524105828
4577 Ignore-this: b23ede46211e67859206c0ec57d6a86f
4578 With glibc, things like _POSIX_C_SOURCE and _ISOC99_SOURCE are
4579 additive, but on FreeBSD they are mutually exclusive.  However, it
4580 turns out we only need to define _POSIX_C_SOURCE and _XOPEN_SOURCE to
4581 get all the C99 stuff we need too, so there's no need for any #ifdefs.
4582 
4583 Submitted by: Gabor PALI <pgj@FreeBSD.org>
4584]
4585[Add a missing UNTAG_CLOSURE, causing bus errors on Sparc
4586Simon Marlow <marlowsd@gmail.com>**20100524105547
4587 Ignore-this: a590b5391d6f05d50c8c088456c3c166
4588 We just about got away with this on x86 which isn't
4589 alignment-sensitive.  The result of the memory load is compared
4590 against a few different values, but there is a fallback case that
4591 happened to be the right thing when the pointer was tagged.  A good
4592 bug to find, nonetheless.
4593]
4594[Add wiki links
4595Simon Marlow <marlowsd@gmail.com>**20100520095953
4596 Ignore-this: c22f126cde166e6207922b2eb51d29e3
4597]
4598[the 'stage=0' trick to disable all compiler builds stopped working; fix it
4599Simon Marlow <marlowsd@gmail.com>**20100520104455
4600 Ignore-this: bb6fae9056471612c8dbf06916188c33
4601]
4602[Comments and formatting only
4603benl@ouroborus.net**20100524014021
4604 Ignore-this: 64579c38154728b632e358bec751cc0b
4605]
4606[Core prettyprinter fixes. Patch from Tim Chevalier. Fixes #4085
4607Ian Lynagh <igloo@earth.li>**20100522225048]
4608[Correct install-name for dynamic Darwin rts
4609pho@cielonegro.org**20100508151155
4610 Ignore-this: 6d31716c8c113dcb46e9cb925c4201df
4611]
4612[Fix the RTS debug_p build
4613Ian Lynagh <igloo@earth.li>**20100522163127]
4614[Unset $CFLAGS for "GNU non-executable stack" configure test; fixes #3889
4615Ian Lynagh <igloo@earth.li>**20100521165005
4616 With gcc 4.4 we get
4617     Error: can't resolve `.note.GNU-stack' {.note.GNU-stack section} - `.Ltext0' {.text section}
4618 when running gcc with the -g flag. To work around this we unset
4619 CFLAGS when running the test.
4620]
4621[Don't run "set -o igncr" before configuring libffi
4622Ian Lynagh <igloo@earth.li>**20100520162918
4623 Ignore-this: 489fa94df23f2adf4ff63c8ede2c0794
4624 It used to make the build work on cygwin, but now it breaks it instead:
4625     config.status: creating include/Makefile
4626     gawk: ./confLqjohp/subs.awk:1: BEGIN {\r
4627     gawk: ./confLqjohp/subs.awk:1: ^ backslash not last character on line
4628     config.status: error: could not create include/Makefile
4629     make[2]: *** [libffi/stamp.ffi.configure-shared] Error 1
4630     make[1]: *** [all] Error 2
4631]
4632[Stop passing -Wl,-macosx_version_min to gcc
4633Ian Lynagh <igloo@earth.li>**20100520154003
4634 Fixes a build failure on OS X 10.6. When linking
4635     rts/dist/build/libHSrts-ghc6.13.20100519.dylib
4636 we got
4637     ld: symbol dyld_stub_binding_helper not defined (usually in crt1.o/dylib1.o/bundle1.o)
4638     collect2: ld returned 1 exit status
4639]
4640[Fix build on FreeBSD; patch from Gabor PALI
4641Ian Lynagh <igloo@earth.li>**20100519140552]
4642[Fix package shadowing order (#4072)
4643Simon Marlow <marlowsd@gmail.com>**20100519104617
4644 Ignore-this: 26ea5e4bb5dff18618b807a54c7d6ebb
4645 
4646 Later packages are supposed to shadow earlier ones in the stack,
4647 unless the ordering is overriden with -package-id flags.
4648 Unfortunately an earlier fix for something else had sorted the list of
4649 packages so that it was in lexicographic order by installedPackageId,
4650 and sadly our test (cabal/shadow) didn't pick this up because the
4651 lexicographic ordering happened to work for the test.  I've now fixed
4652 the test so it tries both orderings.
4653]
4654[Set more env variables when configuring libffi
4655Ian Lynagh <igloo@earth.li>**20100518185014
4656 We now tell it where to find ld, nm and ar
4657]
4658[Set the location of ar to be the in-tree ar on Windows
4659Ian Lynagh <igloo@earth.li>**20100518181556]
4660[Change another / to </> to avoid building paths containing \/
4661Ian Lynagh <igloo@earth.li>**20100518172015
4662 This will hopefully fix #2889.
4663]
4664[Fix #4074 (I hope).
4665Simon Marlow <marlowsd@gmail.com>**20100518113214
4666 Ignore-this: 73cd70f5bc6f5add5247b61985c03fc1
4667 
4668 1. allow multiple threads to call startTimer()/stopTimer() pairs
4669 2. disable the timer around fork() in forkProcess()
4670 
4671 A corresponding change to the process package is required.
4672]
4673[we don't have a gcc-lib in LIB_DIR any more
4674Simon Marlow <marlowsd@gmail.com>**20100401102351
4675 Ignore-this: f41acd2d8f8e6763aa8bd57a0b44a7e4
4676]
4677[In validate, use gmake if available; based on a patch from Gabor PALI
4678Ian Lynagh <igloo@earth.li>**20100517200654]
4679[Remove duplicate "./configure --help" output; fixes #4075
4680Ian Lynagh <igloo@earth.li>**20100516141206]
4681[Update various 'sh boot's to 'perl boot'
4682Ian Lynagh <igloo@earth.li>**20100516122609
4683 Spotted by Marco Túlio Gontijo e Silva
4684]
4685[add missing initialisation for eventBufMutex
4686Simon Marlow <marlowsd@gmail.com>**20100514094943
4687 Ignore-this: 7f75594a8cb54fbec5aebd46bb959f45
4688]
4689[Undo part of #4003 patch
4690Simon Marlow <marlowsd@gmail.com>**20100513142017
4691 Ignore-this: cb65db86a38a7e5ccee9f779e489d104
4692 We still need the workaround for when compiling HEAD with 6.12.2
4693 
4694]
4695[Fix makefile loop (#4050)
4696pho@cielonegro.org**20100507140707
4697 Ignore-this: 3a1cb13d0600977e74d17ac26cbef83d
4698 
4699 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.
4700]
4701[fix !TABLES_NEXT_TO_CODE
4702Simon Marlow <marlowsd@gmail.com>**20100510151934
4703 Ignore-this: fccb859b114bef1c3122c98e60af51
4704]
4705[looksLikeModuleName: allow apostrophe in module names (#4051)
4706Simon Marlow <marlowsd@gmail.com>**20100510094741
4707 Ignore-this: df9348f3ba90608bec57257b47672985
4708]
4709[add the proper library dependencies for GhcProfiled=YES
4710Simon Marlow <marlowsd@gmail.com>**20100506122118
4711 Ignore-this: 6236993aa308ab5b5e1e5ea5f65982a
4712]
4713[Fix Trac #4003: fix the knot-tying in checkHiBootIface
4714simonpj@microsoft.com**20100511075026
4715 Ignore-this: a9ce2a318386fdc8782848df84592002
4716 
4717 I had incorrectly "optimised" checkHiBootIface so that it forgot
4718 to update the "knot-tied" type environment.
4719 
4720 This patch fixes the HEAD
4721]
4722[Re-engineer the derived Ord instance generation code (fix Trac #4019)
4723simonpj@microsoft.com**20100510133333
4724 Ignore-this: 8fe46e4dad27fbee211a7928acf372c2
4725   
4726 As well as fixing #4019, I rejigged the way that Ord instances are
4727 generated, which should make them faster in general.  See the
4728 Note [Generating Ord instances].
4729 
4730 I tried to measure the performance difference from this change, but
4731 the #4019 fix only removes one conditional branch per iteration, and
4732 I couldn't measure a consistent improvement.  But still, tihs is
4733 better than before.
4734]
4735[Make arity of INLINE things consistent
4736simonpj@microsoft.com**20100510133005
4737 Ignore-this: 15e7abf803d1dcb3f4ca760d2d939d0d
4738 
4739 We eta-expand things with INLINE pragmas;
4740 see Note [Eta-expanding INLINE things].
4741 
4742 But I eta-expanded it the wrong amount when the function
4743 was overloaded.  Ooops.
4744]
4745[Compacting GC fix, we forgot to thread the new bq field of StgTSO.
4746Simon Marlow <marlowsd@gmail.com>**20100510082325
4747 Ignore-this: a079c8446e2ad53efff6fd95d0f3ac80
4748]
4749[Add version constraints for the boot packages; fixes trac #3852
4750Ian Lynagh <igloo@earth.li>**20100509175051
4751 When using the bootstrapping compiler, we now explicitly constrain
4752 the version of boot packages (Cabal, extensible-exceptions, etc) to the
4753 in-tree version, so that the build system is less fragile should the
4754 user have a newer version installed for the bootstrapping compiler.
4755]
4756[Don't include inter-package dependencies when compiling with stage 0; #4031
4757Ian Lynagh <igloo@earth.li>**20100509130511
4758 This fixes a problem when building with GHC 6.12 on Windows, where
4759 dependencies on stage 0 (bootstrapping compiler) packages have absolute
4760 paths c:/ghc/..., and make gets confused by the colon.
4761]
4762[Add a ghc.mk for bindisttest/
4763Ian Lynagh <igloo@earth.li>**20100508223911]
4764[Move some make variables around so they are available when cleaning
4765Ian Lynagh <igloo@earth.li>**20100508212405]
4766[Optimise checkremove a bit
4767Ian Lynagh <igloo@earth.li>**20100508202006]
4768[Improve the bindisttest Makefile
4769Ian Lynagh <igloo@earth.li>**20100508195450]
4770[Add tools to test that cleaning works properly
4771Ian Lynagh <igloo@earth.li>**20100508194105]
4772[Tweak the ghc-pkg finding code
4773Ian Lynagh <igloo@earth.li>**20100508125815
4774 It now understand the ghc-stage[123] names we use in-tree, and it won't
4775 go looking for any old ghc-pkg if it can't find the one that matches
4776 ghc.
4777]
4778[Add a way to show what cleaning would be done, without actually doing it
4779Ian Lynagh <igloo@earth.li>**20100508122438]
4780[Tidy up the "rm" flags in the build system
4781Ian Lynagh <igloo@earth.li>**20100508115745]
4782[Fix crash in nested callbacks (#4038)
4783Simon Marlow <marlowsd@gmail.com>**20100507093222
4784 Ignore-this: cade85e361534ce711865a4820276388
4785 Broken by "Split part of the Task struct into a separate struct
4786 InCall".
4787]
4788[Add $(GhcDynamic) knob, set to YES to get stage2 linked with -dynamic
4789Simon Marlow <marlowsd@gmail.com>**20100428205241
4790 Ignore-this: 1db8bccf92099785ecac39aebd27c92d
4791 Default currently NO.
4792 
4793 Validate passed with GhcDynamic=YES on x86/Linux here.
4794 
4795 The compiler is currently slower on x86 when linked -dynamic,
4796 because the GC inner loop has been adversely affected by -fPIC, I'm
4797 looking into how to fix it.
4798]
4799[omit "dyn" from the way appended to the __stginit label
4800Simon Marlow <marlowsd@gmail.com>**20100428204914
4801 Ignore-this: 14183f3defa9f2bde68fda6729b740bc
4802 When GHCi is linked dynamically, we still want to be able to load
4803 non-dynamic object files.
4804]
4805[improvements to findPtr(), a neat hack for browsing the heap in gdb
4806Simon Marlow <marlowsd@gmail.com>**20100506115427
4807 Ignore-this: ac57785bb3e13b97a5945f753f068738
4808]
4809[Fix +RTS -G1
4810Simon Marlow <marlowsd@gmail.com>**20100506110739
4811 Ignore-this: 86a5de39a94d3331a4ee1213f82be497
4812]
4813[Enable the "redundant specialise pragmas" warning; fixes trac #3855
4814Ian Lynagh <igloo@earth.li>**20100506175351]
4815[Find the correct external ids when there's a wrapper
4816simonpj@microsoft.com**20100506164135
4817 Ignore-this: 636266407b174b05b2b8646cc73062c0
4818 
4819 We were failing to externalise the wrapper id for a function
4820 that had one.
4821]
4822[Add a comment about pattern coercions
4823simonpj@microsoft.com**20100506164027
4824 Ignore-this: 17428089f3df439f65d892e23e8ed61a
4825]
4826[Comments only
4827simonpj@microsoft.com**20100506163829
4828 Ignore-this: 169167b6463873ab173cc5750c5be469
4829]
4830[Make a missing name in mkUsageInfo into a panic
4831simonpj@microsoft.com**20100506163813
4832 Ignore-this: b82ff1b8bf89f74f146db7cb5cc4c4d7
4833 
4834 We really want to know about this!
4835]
4836[Refactoring of hsXxxBinders
4837simonpj@microsoft.com**20100506163737
4838 Ignore-this: 97c6667625262b160f9746f7bea1c980
4839 
4840 This patch moves various functions that extract the binders
4841 from a HsTyClDecl, HsForeignDecl etc into HsUtils, and gives
4842 them consistent names.
4843]
4844[Fix Trac #3966: warn about useless UNPACK pragmas
4845simonpj@microsoft.com**20100506163337
4846 Ignore-this: 5beb24b686eda6113b614dfac8490df1
4847 
4848 Warning about useless UNPACK pragmas wasn't as easy as I thought.
4849 I did quite a bit of refactoring, which improved the code by refining
4850 the types somewhat.  In particular notice that in DataCon, we have
4851 
4852     dcStrictMarks   :: [HsBang]
4853     dcRepStrictness :: [StrictnessMarks]
4854 
4855 The former relates to the *source-code* annotation, the latter to
4856 GHC's representation choice.
4857]
4858[Make tcg_dus behave more sanely; fixes a mkUsageInfo panic
4859simonpj@microsoft.com**20100506162719
4860 Ignore-this: d000bca15b0e127e297378ded1bfb81b
4861 
4862 The tcg_dus field used to contain *uses* of type and class decls,
4863 but not *defs*.  That was inconsistent, and it really went wrong
4864 for Template Haskell bracket.  What happened was that
4865  foo = [d| data A = A
4866                   f :: A -> A
4867                   f x = x |]
4868 would find a "use" of A when processing the top level of the module,
4869 which in turn led to a mkUsageInfo panic in MkIface.  The cause was
4870 the fact that the tcg_dus for the nested quote didn't have defs for
4871 A.
4872]
4873[Add a HsExplicitFlag to SpliceDecl, to improve Trac #4042
4874simonpj@microsoft.com**20100506161523
4875 Ignore-this: e4e563bac2fd831cc9e94612f5b4fa9d
4876 
4877 The issue here is that
4878 
4879     g :: A -> A
4880     f
4881     data A = A
4882 
4883 is treated as if you'd written $(f); that is the call of
4884 f is a top-level Template Haskell splice.  This patch
4885 makes sure that we *first* check the -XTemplateHaskellFlag
4886 and bleat about a parse error if it's off.  Othewise we
4887 get strange seeing "A is out of scope" errors.
4888]
4889[Change an assert to a warn
4890simonpj@microsoft.com**20100506161111
4891 Ignore-this: 739a4fb4c7940376b0f2c8ad52a1966c
4892 
4893 This is in the constraint simplifier which I'm about
4894 to rewrite, so I'm hoping the assert isn't fatal!
4895]
4896[Tidy up debug print a little
4897simonpj@microsoft.com**20100506161027
4898 Ignore-this: bd5492878e06bee1cddcbb3fc4df66d8
4899]
4900[Remove useless UNPACK pragmas
4901simonpj@microsoft.com**20100506161012
4902 Ignore-this: 3e5ab1a7cf58107034412a798bc214e5
4903]
4904[Add WARNM2 macro, plus some refactoring
4905simonpj@microsoft.com**20100506160808
4906 Ignore-this: 2ab4f1f0b5d94be683036e77aec09255
4907]
4908[Use -Wwarn for the binary package, becuase it has redundant UNPACK pragmas
4909simonpj@microsoft.com**20100506160750
4910 Ignore-this: cf0d3a11473e28bfce9602e716e69a5f
4911]
4912[Fix Trac #3966: warn about unused UNPACK pragmas
4913simonpj@microsoft.com**20100409201812
4914 Ignore-this: c96412596b39c918b5fb9b3c39ce2119
4915]
4916[Fix Trac #3953: fail earlier when using a bogus quasiquoter
4917simonpj@microsoft.com**20100409201748
4918 Ignore-this: ef48e39aa932caed538643985234f043
4919]
4920[Fix Trac #3965: tighten conditions when deriving Data
4921simonpj@microsoft.com**20100409184420
4922 Ignore-this: 96f7d7d2da11565d26b465d7d0497ac9
4923 
4924 It's tricky to set up the context for a Data instance.  I got it wrong
4925 once, and fixed it -- hence the "extra_constraints" in
4926 TcDeriv.inferConstraints. 
4927 
4928 But it still wasn't right!  The tricky bit is that dataCast1 is only
4929 generated when T :: *->*, and dataCast2 when T :: *->*->*. (See
4930 the code in TcGenDeriv for dataCastX.
4931]
4932[Fix Trac #3964: view patterns in DsArrows
4933simonpj@microsoft.com**20100409165557
4934 Ignore-this: d823c182831d5e2e592e995b16180e2f
4935 
4936 Just a missing case; I've eliminated the catch-all so
4937 that we get a warning next time we extend HsPat
4938]
4939[Fix Trac #3955: renamer and type variables
4940simonpj@microsoft.com**20100409163710
4941 Ignore-this: bd5ec64d76c0f583bf5f224792bf294c
4942 
4943 The renamer wasn't computing the free variables of a type declaration
4944 properly.  This patch refactors a bit, and makes it more robust,
4945 fixing #3955 and several other closely-related bugs.  (We were
4946 omitting some free variables and that could just possibly lead to a
4947 usage-version tracking error.
4948]
4949[Layout only
4950simonpj@microsoft.com**20100409163506
4951 Ignore-this: 1f14990b5aa0b9821b84452fb34e9f41
4952]
4953[Give a better deprecated message for INCLUDE pragmas; fixes #3933
4954Ian Lynagh <igloo@earth.li>**20100506130910
4955 We now have a DeprecatedFullText constructor, so we can override the
4956 "-#include is deprecated: " part of the warning.
4957]
4958[De-haddock a comment that confuses haddock
4959Ian Lynagh <igloo@earth.li>**20100506123607]
4960[Fix comment to not confuse haddock
4961Ian Lynagh <igloo@earth.li>**20100506113642]
4962[Detect EOF when trying to parse a string in hp2ps
4963Ian Lynagh <igloo@earth.li>**20100506000830]
4964[Make the demand analyser sdd demands for strict constructors
4965simonpj@microsoft.com**20100505200936
4966 Ignore-this: eb32632adbc354eb7a5cf884c263e0d3
4967 
4968 This opportunity was spotted by Roman, and is documented in
4969 Note [Add demands for strict constructors] in DmdAnal.
4970]
4971[Fix interaction of exprIsCheap and the lone-variable inlining check
4972simonpj@microsoft.com**20100505200723
4973 Ignore-this: f3cb65085c5673a99153d5d7b6559ab1
4974 
4975 See Note [Interaction of exprIsCheap and lone variables] in CoreUnfold
4976 
4977 This buglet meant that a nullary definition with an INLINE pragma
4978 counter-intuitively didn't get inlined at all.  Roman identified
4979 the bug.
4980]
4981[Matching cases in SpecConstr and Rules
4982simonpj@microsoft.com**20100505200543
4983 Ignore-this: f5c28c780fbf8badce84c6fdc9aa1779
4984 
4985 This patch has zero effect.  It includes comments,
4986 a bit of refactoring, and a tiny bit of commment-out
4987 code go implement the "matching cases" idea below.
4988 
4989 In the end I've left it disabled because while I think
4990 it does no harm I don't think it'll do any good either.
4991 But I didn't want to lose the idea totally. There's
4992 a thread called "Storable and constant memory" on
4993 the libraries@haskell.org list (Apr 2010) about it.
4994 
4995 Note [Matching cases]
4996 ~~~~~~~~~~~~~~~~~~~~~
4997 {- NOTE: This idea is currently disabled.  It really only works if
4998          the primops involved are OkForSpeculation, and, since
4999         they have side effects readIntOfAddr and touch are not.
5000         Maybe we'll get back to this later .  -}
5001   
5002 Consider
5003    f (case readIntOffAddr# p# i# realWorld# of { (# s#, n# #) ->
5004       case touch# fp s# of { _ ->
5005       I# n# } } )
5006 This happened in a tight loop generated by stream fusion that
5007 Roman encountered.  We'd like to treat this just like the let
5008 case, because the primops concerned are ok-for-speculation.
5009 That is, we'd like to behave as if it had been
5010    case readIntOffAddr# p# i# realWorld# of { (# s#, n# #) ->
5011    case touch# fp s# of { _ ->
5012    f (I# n# } } )
5013]
5014[Comments only
5015simonpj@microsoft.com**20100504163629
5016 Ignore-this: 3be12df04714aa820bce706b5dc8a9cb
5017]
5018[Comments only
5019simonpj@microsoft.com**20100504163529
5020 Ignore-this: 791e2fd39c7d880ce1dc80ebdf3a5398
5021]
5022[Comments only
5023simonpj@microsoft.com**20100504163457
5024 Ignore-this: f19e9ffeb3d65770b1595bca5f97a59d
5025]
5026[Comments only (about type families)
5027simonpj@microsoft.com**20100417145032
5028 Ignore-this: dd39425ef2155d52dbf55a4d5fd97cb8
5029]
5030[Fix hp2ps when the .hp file has large string literals
5031Ian Lynagh <igloo@earth.li>**20100505191921]
5032[In build system, call package-config after including package data
5033Ian Lynagh <igloo@earth.li>**20100504225035
5034 Otherwise the $1_$2_HC_OPTS variable gets clobbered.
5035]
5036[runghc: flush stdout/stderr on an exception (#3890)
5037Simon Marlow <marlowsd@gmail.com>**20100505133848
5038 Ignore-this: 224c1898cec64cb1c94e0d7033e7590e
5039]
5040[Remove the Unicode alternative for ".." (#3894)
5041Simon Marlow <marlowsd@gmail.com>**20100505121202
5042 Ignore-this: 2452cd67281667106f9169747b6d784f
5043]
5044[tidyup; no functional changes
5045Simon Marlow <marlowsd@gmail.com>**20100505115015
5046 Ignore-this: d0787e5cdeef1dee628682fa0a46019
5047]
5048[Make the running_finalizers flag task-local
5049Simon Marlow <marlowsd@gmail.com>**20100505114947
5050 Ignore-this: 345925d00f1dca203941b3c5d84c90e1
5051 Fixes a bug reported by Lennart Augustsson, whereby we could get an
5052 incorrect error from the RTS about re-entry from a finalizer,
5053]
5054[add a MAYBE_GC() in killThread#, fixes throwto003(threaded2) looping
5055Simon Marlow <marlowsd@gmail.com>**20100505114746
5056 Ignore-this: efea04991d6feed04683a42232fc85da
5057]
5058[Allow filepath-1.2.*
5059Simon Marlow <marlowsd@gmail.com>**20100505101139
5060 Ignore-this: 1b5580cd9cd041ec48f40cd37603326a
5061]
5062[BlockedOnMsgThrowTo is possible in resurrectThreads (#4030)
5063Simon Marlow <marlowsd@gmail.com>**20100505094534
5064 Ignore-this: ac24a22f95ffeaf480187a1620fdddb2
5065]
5066[Don't raise a throwTo when the target is masking and BlockedOnBlackHole
5067Simon Marlow <marlowsd@gmail.com>**20100505094506
5068 Ignore-this: 302616931f61667030d77ddfbb02374e
5069]
5070[Fix build with GHC 6.10
5071Ian Lynagh <igloo@earth.li>**20100504180302
5072 In GHC 6.10, intersectionWith is (a -> b -> a) instead of (a -> b -> c),
5073 so we need to jump through some hoops to get the more general type.
5074]
5075[The libffi patches are no longer needed
5076Ian Lynagh <igloo@earth.li>**20100504171603]
5077[Use the in-tree windres; fixes trac #4032
5078Ian Lynagh <igloo@earth.li>**20100504170941]
5079[Print unfoldings on lambda-bound variables
5080Simon PJ <simonpj@microsoft.com>**20100503181822
5081 Ignore-this: 2fd5a7502cc6273d96258e0914f0f8cd
5082 
5083 ...in the unusual case where they have one;
5084 see Note [Case binders and join points] in Simplify.lhs
5085]
5086[Replace FiniteMap and UniqFM with counterparts from containers.
5087Milan Straka <fox@ucw.cz>**20100503171315
5088 Ignore-this: a021972239163dbf728284b19928cebb
5089 
5090 The original interfaces are kept. There is small performance improvement:
5091 - when compiling for five nofib, we get following speedups:
5092     Average                -----           -2.5%
5093     Average                -----           -0.6%
5094     Average                -----           -0.5%
5095     Average                -----           -5.5%
5096     Average                -----          -10.3%
5097 - when compiling HPC ten times, we get:
5098     switches                          oldmaps   newmaps
5099     -O -fasm                          117.402s  116.081s (98.87%)
5100     -O -fasm -fregs-graph             119.993s  118.735s (98.95%)
5101     -O -fasm -fregs-iterative         120.191s  118.607s (98.68%)
5102]
5103[Make the demand analyser take account of lambda-bound unfoldings
5104Simon PJ <simonpj@microsoft.com>**20100503151630
5105 Ignore-this: 2ee8e27d4df2debfc79e6b8a17c32bc1
5106 
5107 This is a long-standing lurking bug. See Note [Lamba-bound unfoldings]
5108 in DmdAnal.
5109 
5110 I'm still not really happy with this lambda-bound-unfolding stuff.
5111]
5112[Fix dynamic libs on OS X, and enable them by default
5113Ian Lynagh <igloo@earth.li>**20100503150302]
5114[Switch back to using bytestring from the darcs repo; partially fixes #3855
5115Ian Lynagh <igloo@earth.li>**20100502113458]
5116[Fix some cpp warnings when building on FreeBSD; patch from Gabor PALI
5117Ian Lynagh <igloo@earth.li>**20100428150700]
5118[Fix "make 2"
5119Ian Lynagh <igloo@earth.li>**20100427162212
5120 The new Makefile logic was enabling the stage 1 rules when stage=2,
5121 so "make 2" was rebuilding stage 1.
5122]
5123[Inplace programs depend on their shell wrappers
5124Ian Lynagh <igloo@earth.li>**20100427160038]
5125[--make is now the default (#3515), and -fno-code works with --make (#3783)
5126Simon Marlow <marlowsd@gmail.com>**20100427122851
5127 Ignore-this: 33330474fa4703f32bf9997462b4bf3c
5128 If the command line contains any Haskell source files, then we behave
5129 as if --make had been given.
5130 
5131 The meaning of the -c flag has changed (back): -c now selects one-shot
5132 compilation, but stops before linking.  However, to retain backwards
5133 compatibility, -c is still allowed with --make, and means the same as
5134 --make -no-link.  The -no-link flag has been un-deprecated.
5135 
5136 -fno-code is now allowed with --make (#3783); the fact that it was
5137 disabled before was largely accidental, it seems.  We also had some
5138 regressions in this area: it seems that -fno-code was causing a .hc
5139 file to be emitted in certain cases.  I've tidied up the code, there
5140 was no need for -fno-code to be a "mode" flag, as far as I can tell.
5141 
5142 -fno-code does not emit interface files, nor does it do recompilation
5143 checking, as suggested in #3783.  This would make Haddock emit
5144 interface files, for example, and I'm fairly sure we don't want to do
5145 that.  Compiling with -fno-code is pretty quick anyway, perhaps we can
5146 get away without recompilation checking.
5147]
5148[remove duplicate docs for -e in --help output (#4010)
5149Simon Marlow <marlowsd@gmail.com>**20100426140642
5150 Ignore-this: 187ff893ba8ffa0ec127867a7590e38d
5151]
5152[workaround for #4003, fixes HEAD build with 6.12.2
5153Simon Marlow <marlowsd@gmail.com>**20100426103428
5154 Ignore-this: c4bc445dc8052d4e6efef3f1daf63562
5155]
5156[Make sure all the clean rules are always included
5157Ian Lynagh <igloo@earth.li>**20100424181823
5158 In particular, this fixes a problem where stage3 bits weren't being cleaned
5159]
5160[Correct the name of the amd64/FreeBSD platform in PlatformSupportsSharedLibs
5161Ian Lynagh <igloo@earth.li>**20100424132830
5162 We weren't getting sharedlibs on amd64/FreeBSD because of this
5163]
5164[Include DPH docs in bindists
5165Ian Lynagh <igloo@earth.li>**20100424123101]
5166[reinstate eta-expansion during SimplGently, to fix inlining of sequence_
5167Simon Marlow <marlowsd@gmail.com>**20100423124853
5168 Ignore-this: 4fa0fd5bafe0d6b58fc81076f50d5f8d
5169]
5170[fix 64-bit value for W_SHIFT, which thankfully appears to be not used
5171Simon Marlow <marlowsd@gmail.com>**20100422213605
5172 Ignore-this: 525c062d2456c224ec8d0e083edd3b55
5173]
5174[Add missing constant folding and optimisation for unsigned division
5175Simon Marlow <marlowsd@gmail.com>**20100422213443
5176 Ignore-this: fb10d1cda0852fab0cbcb47247498fb3
5177 Noticed by Denys Rtveliashvili <rtvd@mac.com>, see #4004
5178]
5179[Fix the GHC API link in the main doc index.html
5180Ian Lynagh <igloo@earth.li>**20100422213226]
5181[Give the right exit code in darcs-all
5182Ian Lynagh <igloo@earth.li>**20100421171339
5183 Our END block was calling system, which alters $?. So now we save and
5184 restore it.
5185]
5186[Use StgWord64 instead of ullong
5187Ian Lynagh <igloo@earth.li>**20100421162336
5188 This patch also fixes ullong_format_string (renamed to showStgWord64)
5189 so that it works with values outside the 32bit range (trac #3979), and
5190 simplifies the without-commas case.
5191]
5192[Implement try10Times in Makefile
5193Ian Lynagh <igloo@earth.li>**20100420165909
5194 Avoid using seq, as FreeBSD has jot instead.
5195]
5196[Fix crash in non-threaded RTS on Windows
5197Simon Marlow <marlowsd@gmail.com>**20100420122125
5198 Ignore-this: 28b0255a914a8955dce02d89a7dfaca
5199 The tso->block_info field is now overwritten by pushOnRunQueue(), but
5200 stg_block_async_info was assuming that it still held a pointer to the
5201 StgAsyncIOResult.  We must therefore save this value somewhere safe
5202 before putting the TSO on the run queue.
5203]
5204[Expand the scope of the event_buf_mutex to cover io_manager_event
5205Simon Marlow <marlowsd@gmail.com>**20100420122026
5206 Ignore-this: 185a6d84f7d4a35997f10803f6dacef1
5207 I once saw a failure that I think was due to a race on
5208 io_manager_event, this should fix it.
5209]
5210[Flags -auto and -auto-all operate only on functions not marked INLINE.
5211Milan Straka <fox@ucw.cz>**20100331191050
5212 Ignore-this: 3b63580cfcb3c33d62ad697c36d94d05
5213]
5214[Spelling correction for LANGUAGE pragmas
5215Max Bolingbroke <batterseapower@hotmail.com>**20100413192825
5216 Ignore-this: 311b51ba8d43f6c7fd32f48db9a88dee
5217]
5218[Update the user guide so it talks about the newer "do rec" notation everywhere
5219Ian Lynagh <igloo@earth.li>**20100416205416
5220 Some of the problems highlighted in trac #3968.
5221]
5222[Fix typo
5223Ian Lynagh <igloo@earth.li>**20100416205412]
5224[Fix Trac #3950: unifying types of different kinds
5225simonpj@microsoft.com**20100412151845
5226 Ignore-this: d145b9de5ced136ef2c39f3ea4a04f4a
5227 
5228 I was assuming that the unifer only unified types of the
5229 same kind, but now we can "defer" unsolved constraints that
5230 invariant no longer holds.  Or at least is's more complicated
5231 to ensure. 
5232 
5233 This patch takes the path of not assuming the invariant, which
5234 is simpler and more robust.  See
5235 Note [Mismatched type lists and application decomposition]
5236]
5237[Fix Trac #3943: incorrect unused-variable warning
5238simonpj@microsoft.com**20100412151630
5239 Ignore-this: 52459f2b8b02c3cb120abe674dc9a060
5240 
5241 In fixing this I did the usual little bit of refactoring
5242]
5243[Convert boot and boot-pkgs to perl
5244Ian Lynagh <igloo@earth.li>**20100415143919
5245 This stops us having to worry about sh/sed/... portability.
5246]
5247[Use $(MAKE), not make, when recursively calling make
5248Ian Lynagh <igloo@earth.li>**20100415121453]
5249[Remove the ghc_ge_609 makefile variables
5250Ian Lynagh <igloo@earth.li>**20100412235658
5251 They are now guaranteed to be YES
5252]
5253[Increase the minimum version number required to 6.10 in configure.ac
5254Ian Lynagh <igloo@earth.li>**20100412235313]
5255[The bootstrapping compiler is now required to be > 609
5256Ian Lynagh <igloo@earth.li>**20100409161046]
5257[Handle IND_STATIC in isRetainer
5258Ian Lynagh <igloo@earth.li>**20100409104207
5259 IND_STATIC used to be an error, but at the moment it can happen
5260 as isAlive doesn't look through IND_STATIC as it ignores static
5261 closures. See trac #3956 for a program that hit this error.
5262]
5263[Add Data and Typeable instances to HsSyn
5264David Waern <david.waern@gmail.com>**20100330011020
5265 Ignore-this: c3f2717207b15539fea267c36b686e6a
5266 
5267 The instances (and deriving declarations) have been taken from the ghc-syb
5268 package.
5269]
5270[Fix for derefing ThreadRelocated TSOs in MVar operations
5271Simon Marlow <marlowsd@gmail.com>**20100407092824
5272 Ignore-this: 94dd7c68a6094eda667e2375921a8b78
5273]
5274[sanity check fix
5275Simon Marlow <marlowsd@gmail.com>**20100407092746
5276 Ignore-this: 9c18cd5f5393e5049015ca52e62a1269
5277]
5278[get the reg liveness right in the putMVar# heap check
5279Simon Marlow <marlowsd@gmail.com>**20100407092724
5280 Ignore-this: b1ba07a59ecfae00e9a1f8391741abc
5281]
5282[initialise the headers of MSG_BLACKHOLE objects properly
5283Simon Marlow <marlowsd@gmail.com>**20100407081712
5284 Ignore-this: 183dcd0ca6a395d08db2be12b02bdd79
5285]
5286[initialise the headers of MVAR_TSO_QUEUE objects properly
5287Simon Marlow <marlowsd@gmail.com>**20100407081514
5288 Ignore-this: 4b4a2f30cf2fb69ca4128c41744687bb
5289]
5290[undo debugging code
5291Simon Marlow <marlowsd@gmail.com>**20100406142740
5292 Ignore-this: 323c2248f817b6717c19180482fc4b00
5293]
5294[putMVar#: fix reg liveness in the heap check
5295Simon Marlow <marlowsd@gmail.com>**20100406135832
5296 Ignore-this: cddd2c7807ac7612c9b2c4c0d384d284
5297]
5298[account for the new BLACKHOLEs in the GHCi debugger
5299Simon Marlow <marlowsd@gmail.com>**20100406133406
5300 Ignore-this: 4d4aeb4bbada3f50dc1fb0123f565e8f
5301]
5302[don't forget to deRefTSO() in tryWakeupThread()
5303Simon Marlow <marlowsd@gmail.com>**20100406130411
5304 Ignore-this: 171d57c4f8653835dec0b69f9be9881c
5305]
5306[Fix bug in popRunQueue
5307Simon Marlow <marlowsd@gmail.com>**20100406091453
5308 Ignore-this: 9d3cec8f18f5c5cbd51751797386eb6f
5309]
5310[fix bug in migrateThread()
5311Simon Marlow <marlowsd@gmail.com>**20100401105840
5312 Ignore-this: 299bcf0d1ea0f8865f3e845eb93d2ad3
5313]
5314[Remove the IND_OLDGEN and IND_OLDGEN_PERM closure types
5315Simon Marlow <marlowsd@gmail.com>**20100401093519
5316 Ignore-this: 95f2480c8a45139835eaf5610217780b
5317 These are no longer used: once upon a time they used to have different
5318 layout from IND and IND_PERM respectively, but that is no longer the
5319 case since we changed the remembered set to be an array of addresses
5320 instead of a linked list of closures.
5321]
5322[Change the representation of the MVar blocked queue
5323Simon Marlow <marlowsd@gmail.com>**20100401091605
5324 Ignore-this: 20a35bfabacef2674df362905d7834fa
5325 
5326 The list of threads blocked on an MVar is now represented as a list of
5327 separately allocated objects rather than being linked through the TSOs
5328 themselves.  This lets us remove a TSO from the list in O(1) time
5329 rather than O(n) time, by marking the list object.  Removing this
5330 linear component fixes some pathalogical performance cases where many
5331 threads were blocked on an MVar and became unreachable simultaneously
5332 (nofib/smp/threads007), or when sending an asynchronous exception to a
5333 TSO in a long list of thread blocked on an MVar.
5334 
5335 MVar performance has actually improved by a few percent as a result of
5336 this change, slightly to my surprise.
5337 
5338 This is the final cleanup in the sequence, which let me remove the old
5339 way of waking up threads (unblockOne(), MSG_WAKEUP) in favour of the
5340 new way (tryWakeupThread and MSG_TRY_WAKEUP, which is idempotent).  It
5341 is now the case that only the Capability that owns a TSO may modify
5342 its state (well, almost), and this simplifies various things.  More of
5343 the RTS is based on message-passing between Capabilities now.
5344]
5345[eliminate some duplication with a bit of CPP
5346Simon Marlow <marlowsd@gmail.com>**20100330154355
5347 Ignore-this: 838f7d341f096ca14c86ab9c81193e36
5348]
5349[Make ioManagerDie() idempotent
5350Simon Marlow <marlowsd@gmail.com>**20100401100705
5351 Ignore-this: a5996b43cdb2e2d72e6e971d7ea925fb
5352 Avoids screeds of "event buffer overflowed; event dropped" in
5353 conc059(threaded1).
5354]
5355[Move a thread to the front of the run queue when another thread blocks on it
5356Simon Marlow <marlowsd@gmail.com>**20100329144521
5357 Ignore-this: c518ff0d41154680edc811d891826a29
5358 This fixes #3838, and was made possible by the new BLACKHOLE
5359 infrastructure.  To allow reording of the run queue I had to make it
5360 doubly-linked, which entails some extra trickiness with regard to
5361 GC write barriers and suchlike.
5362]
5363[remove non-existent MUT_CONS symbols
5364Simon Marlow <marlowsd@gmail.com>**20100330152600
5365 Ignore-this: 885628257a9d03f2ece2a754d993014a
5366]
5367[change throwTo to use tryWakeupThread rather than unblockOne
5368Simon Marlow <marlowsd@gmail.com>**20100329144613
5369 Ignore-this: 10ad4965e6c940db71253f1c72218bbb
5370]
5371[tiny GC optimisation
5372Simon Marlow <marlowsd@gmail.com>**20100329144551
5373 Ignore-this: 9e095b9b73fff0aae726f9937846ba92
5374]
5375[New implementation of BLACKHOLEs
5376Simon Marlow <marlowsd@gmail.com>**20100329144456
5377 Ignore-this: 96cd26793b4e6ab9ddd0d59aae5c2f1d
5378 
5379 This replaces the global blackhole_queue with a clever scheme that
5380 enables us to queue up blocked threads on the closure that they are
5381 blocked on, while still avoiding atomic instructions in the common
5382 case.
5383 
5384 Advantages:
5385 
5386  - gets rid of a locked global data structure and some tricky GC code
5387    (replacing it with some per-thread data structures and different
5388    tricky GC code :)
5389 
5390  - wakeups are more prompt: parallel/concurrent performance should
5391    benefit.  I haven't seen anything dramatic in the parallel
5392    benchmarks so far, but a couple of threading benchmarks do improve
5393    a bit.
5394 
5395  - waking up a thread blocked on a blackhole is now O(1) (e.g. if
5396    it is the target of throwTo).
5397 
5398  - less sharing and better separation of Capabilities: communication
5399    is done with messages, the data structures are strictly owned by a
5400    Capability and cannot be modified except by sending messages.
5401 
5402  - this change will utlimately enable us to do more intelligent
5403    scheduling when threads block on each other.  This is what started
5404    off the whole thing, but it isn't done yet (#3838).
5405 
5406 I'll be documenting all this on the wiki in due course.
5407 
5408]
5409[Fix warnings (allow pushOnRunQueue() to not be inlined)
5410Simon Marlow <marlowsd@gmail.com>**20100401114559
5411 Ignore-this: f40bfbfad70a5165a946d11371605b7d
5412]
5413[remove out of date comment
5414Simon Marlow <marlowsd@gmail.com>**20100401105853
5415 Ignore-this: 26af88dd418ee0bcda7223b3b7e4e8d2
5416]
5417[tidy up spacing in stderr traces
5418Simon Marlow <marlowsd@gmail.com>**20100326163122
5419 Ignore-this: 16558b0433a274be217d4bf39aa4946
5420]
5421[Fix an assertion that was not safe when running in parallel
5422Simon Marlow <marlowsd@gmail.com>**20100325143656
5423 Ignore-this: cad08fb8900eb3a475547af0189fcc47
5424]
5425[Never jump directly to a thunk's entry code, even if it is single-entry
5426Simon Marlow <marlowsd@gmail.com>**20100325114847
5427 Ignore-this: 938da172c06a97762ef605c8fccfedf1
5428 I don't think this fixes any bugs as we don't have single-entry thunks
5429 at the moment, but it could cause problems for parallel execution if
5430 we ever did re-introduce update avoidance.
5431]
5432[Rename forgotten -dverbose-simpl to -dverbose-core2core in the docs.
5433Milan Straka <fox@ucw.cz>**20100331153626
5434 Ignore-this: 2da58477fb96e1cfb80f37dddd7c422c
5435]
5436[Add -pa and -V to the documentation of time profiling options.
5437Milan Straka <fox@ucw.cz>**20100329191121
5438 Ignore-this: be74d216481ec5a19e5f40f85e6e3d65
5439]
5440[Keep gcc 4.5 happy
5441Simon Marlow <marlowsd@gmail.com>**20100330120425
5442 Ignore-this: 7811878cc2bd1ce9cfbb5bf102fe3454
5443]
5444[Fix warning compiling Linker.c for PPC Mac
5445naur@post11.tele.dk**20100403182355
5446 Ignore-this: e2d2448770c9714ce17dd6cf3e297063
5447 The warning message eliminated is:
5448 > rts/Linker.c:4756:0:
5449 >      warning: nested extern declaration of 'symbolsWithoutUnderscore'
5450]
5451[Fix error compiling AsmCodeGen.lhs for PPC Mac (mkRtsCodeLabel)
5452naur@post11.tele.dk**20100403181656
5453 Ignore-this: deb7524ea7852a15a2ac0849c8c82f74
5454 The error messages eliminated are:
5455 > compiler/nativeGen/AsmCodeGen.lhs:875:31:
5456 >     Not in scope: `mkRtsCodeLabel'
5457 > compiler/nativeGen/AsmCodeGen.lhs:879:31:
5458 >     Not in scope: `mkRtsCodeLabel'
5459 > compiler/nativeGen/AsmCodeGen.lhs:883:31:
5460 >     Not in scope: `mkRtsCodeLabel'
5461]
5462[Fix error compiling AsmCodeGen.lhs for PPC Mac (DestBlockId)
5463naur@post11.tele.dk**20100403180643
5464 Ignore-this: 71e833e94ed8371b2ffabc2cf80bf585
5465 The error message eliminated is:
5466 > compiler/nativeGen/AsmCodeGen.lhs:637:16:
5467 >     Not in scope: data constructor `DestBlockId'
5468]
5469[Fix boot-pkgs's sed usage to work with Solaris's sed
5470Ian Lynagh <igloo@earth.li>**20100401153441]
5471[Pass "-i org.haskell.GHC" to packagemaker when building the OS X installer
5472Ian Lynagh <igloo@earth.li>**20100331144707
5473 This seems to fix this failure:
5474 [...]
5475 ** BUILD SUCCEEDED **
5476 rm -f -f GHC-system.pmdoc/*-contents.xml
5477 /Developer/usr/bin/packagemaker -v --doc GHC-system.pmdoc\
5478              -o /Users/ian/to_release/ghc-6.12.1.20100330/GHC-6.12.1.20100330-i386.pkg
5479 2010-03-31 15:08:15.695 packagemaker[13909:807] Setting to : 0 (null)
5480 2010-03-31 15:08:15.709 packagemaker[13909:807] Setting to : 0 org.haskell.glasgowHaskellCompiler.ghc.pkg
5481 2010-03-31 15:08:15.739 packagemaker[13909:807] relocate: (null) 0
5482 2010-03-31 15:08:15.740 packagemaker[13909:807] *** Terminating app due to uncaught exception 'NSInvalidArgumentException', reason: '*** -[NSXMLDocument initWithXMLString:options:error:]: nil argument'
5483 2010-03-31 15:08:15.741 packagemaker[13909:807] Stack: (
5484     2511962091,
5485     2447007291,
5486     2511961547,
5487     2511961610,
5488     2432803204,
5489     453371,
5490     447720,
5491     436209,
5492     435510,
5493     9986,
5494     9918
5495 )
5496 make[1]: *** [framework-pkg] Trace/BPT trap
5497 make: *** [framework-pkg] Error 2
5498]
5499[Use machdepCCOpts when compiling the file to toggle -(no-)rtsopts
5500Ian Lynagh <igloo@earth.li>**20100331161302
5501 Should fix toggling on OS X "Snow Leopard". Diagnosed by Roman Leshchinskiy.
5502]
5503[Avoid a non-portable use of tar reported by Roman Leshchinskiy
5504Ian Lynagh <igloo@earth.li>**20100330145802]
5505[Don't install EXTRA_PACKAGES by default
5506Simon Marlow <marlowsd@gmail.com>**20100330142714
5507 Ignore-this: d4cc8f87a6de8d9d1d6dc9b77130b3
5508]
5509[fix a non-portable printf format
5510Simon Marlow <marlowsd@gmail.com>**20100330134437
5511 Ignore-this: d41c23c54ec29654cb2049de1e588570
5512]
5513[avoid single quote in #error
5514Simon Marlow <marlowsd@gmail.com>**20100330120346
5515 Ignore-this: 663f39e7a27fead2f648fbf22d345bb4
5516]
5517[use FMT_Word64 instead of locally-defined version
5518Simon Marlow <marlowsd@gmail.com>**20100330114650
5519 Ignore-this: 82697b8095dffb3a8e196c687006ece0
5520]
5521[remove old/unused DotnetSupport and GhcLibsWithUnix
5522Simon Marlow <marlowsd@gmail.com>**20100330123732
5523 Ignore-this: c68814868b3671abdc369105bbeafe6c
5524]
5525[fix return type cast in f.i.wrapper when using libffi (#3516)
5526Simon Marlow <marlowsd@gmail.com>**20100329154220
5527 Ignore-this: f898eb8c9ae2ca2009e539735b92c438
5528 
5529 Original fix submitted by
5530   Sergei Trofimovich <slyfox@community.haskell.org>
5531 modified by me:
5532  - exclude 64-bit types
5533  - compare uniques, not strings
5534  - #include "ffi.h" is conditional
5535]
5536[libffi: install 'ffitarget.h' header as sole 'ffi.h' is unusable
5537Simon Marlow <marlowsd@gmail.com>**20100329135734
5538 Ignore-this: f9b555ea289d8df1aa22cb6faa219a39
5539 Submitted by: Sergei Trofimovich <slyfox@community.haskell.org>
5540 Re-recorded against HEAD.
5541]
5542[avoid a fork deadlock (see comments)
5543Simon Marlow <marlowsd@gmail.com>**20100329132329
5544 Ignore-this: 3377f88b83bb3b21e42d7fc5f0d866f
5545]
5546[tidy up the end of the all_tasks list after forking
5547Simon Marlow <marlowsd@gmail.com>**20100329132253
5548 Ignore-this: 819d679875be5f344e816210274d1c29
5549]
5550[Add a 'setKeepCAFs' external function (#3900)
5551Simon Marlow <marlowsd@gmail.com>**20100329110036
5552 Ignore-this: ec532a18cad4259a09847b0b9ae2e1d2
5553]
5554[Explicitly check whether ar supports the @file syntax
5555Ian Lynagh <igloo@earth.li>**20100329123325
5556 rather than assuming that all GNU ar's do.
5557 Apparently OpenBSD's older version doesn't.
5558]
5559[Fix the format specifier for Int64/Word64 on Windows
5560Ian Lynagh <igloo@earth.li>**20100327182126
5561 mingw doesn't understand %llu/%lld - it treats them as 32-bit rather
5562 than 64-bit. We use %I64u/%I64d instead.
5563]
5564[Fix the ghci startmenu item
5565Ian Lynagh <igloo@earth.li>**20100326235934
5566 I'm not sure what changed, but it now doesn't work for me without
5567 the "Start in" field being set.
5568]
5569[Fix paths to docs in "Start Menu" entries in Windows installer; fixes #3847
5570Ian Lynagh <igloo@earth.li>**20100326155917]
5571[Add a licence file for the Windows installer to use
5572Ian Lynagh <igloo@earth.li>**20100326155130]
5573[Add gcc-g++ to the inplace mingw installation; fixes #3893
5574Ian Lynagh <igloo@earth.li>**20100326154714]
5575[Add the licence file to the Windows installer. Fixes #3934
5576Ian Lynagh <igloo@earth.li>**20100326152449]
5577[Quote the paths to alex and happy in configure
5578Ian Lynagh <igloo@earth.li>**20100325143449
5579 Ignore-this: d6d6e1a250f88985bbeea760e63a79db
5580]
5581[Use </> rather than ++ "/"
5582Ian Lynagh <igloo@earth.li>**20100325133237
5583 This stops us generating paths like
5584     c:\foo\/ghc460_0/ghc460_0.o
5585 which windres doesn't understand.
5586]
5587[Append $(exeext) to utils/ghc-pkg_dist_PROG
5588Ian Lynagh <igloo@earth.li>**20100324233447
5589 Fixes bindist creation
5590]
5591[A sanity check
5592Simon Marlow <marlowsd@gmail.com>**20100325110500
5593 Ignore-this: 3b3b76d898c822456857e506b7531e65
5594]
5595[do_checks: do not set HpAlloc if the stack check fails
5596Simon Marlow <marlowsd@gmail.com>**20100325110328
5597 Ignore-this: 899ac8c29ca975d03952dbf4608d758
5598 
5599 This fixes a very rare heap corruption bug, whereby
5600 
5601  - a context switch is requested, which sets HpLim to zero
5602    (contextSwitchCapability(), called by the timer signal or
5603    another Capability).
5604 
5605  - simultaneously a stack check fails, in a code fragment that has
5606    both a stack and a heap check.
5607 
5608 The RTS then assumes that a heap-check failure has occurred and
5609 subtracts HpAlloc from Hp, although in fact it was a stack-check
5610 failure and retreating Hp will overwrite valid heap objects.  The bug
5611 is that HpAlloc should only be set when Hp has been incremented by the
5612 heap check.  See comments in rts/HeapStackCheck.cmm for more details.
5613 
5614 This bug is probably incredibly rare in practice, but I happened to be
5615 working on a test that triggers it reliably:
5616 concurrent/should_run/throwto001, compiled with -O -threaded, args 30
5617 300 +RTS -N2, run repeatedly in a loop.
5618]
5619[comments and formatting only
5620Simon Marlow <marlowsd@gmail.com>**20100325104617
5621 Ignore-this: c0a211e15b5953bb4a84771bcddd1d06
5622]
5623[Change how perl scripts get installed; partially fixes #3863
5624Ian Lynagh <igloo@earth.li>**20100324171422
5625 We now regenerate them when installing, which means the path for perl
5626 doesn't get baked in
5627]
5628[Pass the location of gcc in the ghc wrapper script; partially fixes #3863
5629Ian Lynagh <igloo@earth.li>**20100324171408
5630 This means we don't rely on baking a path to gcc into the executable
5631]
5632[Quote the ar path in configure
5633Ian Lynagh <igloo@earth.li>**20100324162043]
5634[Remove unused cUSER_WAY_NAMES cUSER_WAY_OPTS
5635Ian Lynagh <igloo@earth.li>**20100324145048]
5636[Remove unused cCONTEXT_DIFF
5637Ian Lynagh <igloo@earth.li>**20100324145013]
5638[Remove unused cEnableWin32DLLs
5639Ian Lynagh <igloo@earth.li>**20100324144841]
5640[Remove unused cGHC_CP
5641Ian Lynagh <igloo@earth.li>**20100324144656]
5642[Fix the build for non-GNU-ar
5643Ian Lynagh <igloo@earth.li>**20100324132907]
5644[Tweak the Makefile code for making .a libs; fixes trac #3642
5645Ian Lynagh <igloo@earth.li>**20100323221325
5646 The main change is that, rather than using "xargs ar" we now put
5647 all the filenames into a file, and do "ar @file". This means that
5648 ar adds all the files at once, which works around a problem where
5649 files with the same basename in a later invocation were overwriting
5650 the existing file in the .a archive.
5651]
5652[Enable shared libraries on Windows; fixes trac #3879
5653Ian Lynagh <igloo@earth.li>**20100320231414
5654 Ignore-this: c93b35ec5b7a7fa6ddb286d17a616216
5655]
5656[Add the external core PDF to the new build system
5657Ian Lynagh <igloo@earth.li>**20100321161909]
5658[Allow specifying $threads directly when validating
5659Ian Lynagh <igloo@earth.li>**20100321112835]
5660[Remove LazyUniqFM; fixes trac #3880
5661Ian Lynagh <igloo@earth.li>**20100320213837]
5662[UNDO: slight improvement to scavenging ...
5663Simon Marlow <marlowsd@gmail.com>**20100319153413
5664 Ignore-this: f0ab581c07361f7b57eae02dd6ec893c
5665 
5666 Accidnetally pushed this patch which, while it validates, isn't
5667 correct.
5668 
5669 rolling back:
5670 
5671 Fri Mar 19 11:21:27 GMT 2010  Simon Marlow <marlowsd@gmail.com>
5672   * slight improvement to scavenging of update frames when a collision has occurred
5673 
5674     M ./rts/sm/Scav.c -19 +15
5675]
5676[slight improvement to scavenging of update frames when a collision has occurred
5677Simon Marlow <marlowsd@gmail.com>**20100319112127
5678 Ignore-this: 6de2bb9614978975f17764a0f259d9bf
5679]
5680[Don't install the utf8-string package
5681Ian Lynagh <igloo@earth.li>**20100317212709]
5682[Don't use -Bsymbolic when linking the RTS
5683Ian Lynagh <igloo@earth.li>**20100316233357
5684 This makes the RTS hooks work when doing dynamic linking
5685]
5686[Fix Trac #3920: Template Haskell kinds
5687simonpj@microsoft.com**20100317123519
5688 Ignore-this: 426cac7920446e04f3cc30bd1d9f76e2
5689 
5690 Fix two places where we were doing foldl instead of foldr
5691 after decomposing a Kind.  Strange that the same bug appears
5692 in two quite different places!
5693]
5694[copy_tag_nolock(): fix write ordering and add a write_barrier()
5695Simon Marlow <marlowsd@gmail.com>**20100316143103
5696 Ignore-this: ab7ca42904f59a0381ca24f3eb38d314
5697 
5698 Fixes a rare crash in the parallel GC.
5699 
5700 If we copy a closure non-atomically during GC, as we do for all
5701 immutable values, then before writing the forwarding pointer we better
5702 make sure that the closure itself is visible to other threads that
5703 might follow the forwarding pointer.  I imagine this doesn't happen
5704 very often, but I just found one case of it: in scavenge_stack, the
5705 RET_FUN case, after evacuating ret_fun->fun we then follow it and look
5706 up the info pointer.
5707]
5708[Add sliceP mapping to vectoriser builtins
5709benl@ouroborus.net**20100316060517
5710 Ignore-this: 54c3cafff584006b6fbfd98124330aa3
5711]
5712[Comments only
5713benl@ouroborus.net**20100311064518
5714 Ignore-this: d7dc718cc437d62aa5b1b673059a9b22
5715]
5716[TAG 2010-03-16
5717Ian Lynagh <igloo@earth.li>**20100316005137
5718 Ignore-this: 234e3bc29e2f26cc59d7b03d780cc352
5719]
5720Patch bundle hash:
5721700633c5569b1acebd02e49ad2f84d3c8f4e2995