Ticket #1845: suppress-warnings-on-x64_64.patch

File suppress-warnings-on-x64_64.patch, 173.0 KB (added by PHO, 4 years ago)
Line 
12 patches for repository http://darcs.haskell.org/ghc:
2
3Tue Sep 28 22:25:59 JST 2010  pho@cielonegro.org
4  * FIX #1845 (unconditional relative branch out of range)
5 
6  Don't use mmap on powerpc-apple-darwin as mmap doesn't support
7  reallocating but we need to allocate jump islands just after each
8  object images. Otherwise relative branches to jump islands can fail
9  due to 24-bits displacement overflow.
10
11Sun Oct 17 19:06:09 JST 2010  pho@cielonegro.org
12  * rts/Linker.c (ocAllocateSymbolExtras):
13     Suppress false warnings on x86_64 platforms.
14     It's OK to allocate jump islands separately for x86_64 because
15     PC-relative jumps/accesses are only limited to 32 bits.
16-----BEGIN PGP SIGNED MESSAGE-----
17Hash: SHA1
18
19
20New patches:
21
22[FIX #1845 (unconditional relative branch out of range)
23pho@cielonegro.org**20100928132559
24 Ignore-this: a3d9726969704a52c4a4d61f43c37162
25 
26 Don't use mmap on powerpc-apple-darwin as mmap doesn't support
27 reallocating but we need to allocate jump islands just after each
28 object images. Otherwise relative branches to jump islands can fail
29 due to 24-bits displacement overflow.
30] {
31hunk ./rts/Linker.c 72
32 #include <sys/wait.h>
33 #endif
34 
35- -#if defined(linux_HOST_OS) || defined(freebsd_HOST_OS) || defined(dragonfly_HOST_OS) || defined(netbsd_HOST_OS) || defined(openbsd_HOST_OS) || defined(darwin_HOST_OS)
36+#if defined(linux_HOST_OS    ) || defined(freebsd_HOST_OS) || \
37+    defined(dragonfly_HOST_OS) || defined(netbsd_HOST_OS ) || \
38+    defined(openbsd_HOST_OS  ) || \
39+    ( defined(darwin_HOST_OS ) && !defined(powerpc_HOST_ARCH) )
40+/* Don't use mmap on powerpc-apple-darwin as mmap doesn't support
41+ * reallocating but we need to allocate jump islands just after each
42+ * object images. Otherwise relative branches to jump islands can fail
43+ * due to 24-bits displacement overflow.
44+ */
45 #define USE_MMAP
46 #include <fcntl.h>
47 #include <sys/mman.h>
48hunk ./rts/Linker.c 1682
49    size_t fileSize;
50    int isObject;
51    char tmp[12];
52+#if !defined(USE_MMAP) && defined(darwin_HOST_OS)
53+   int misalignment;
54+#endif
55 
56    IF_DEBUG(linker, debugBelch("loadArchive `%s'\n", path));
57 
58hunk ./rts/Linker.c 1764
59        if (isObject) {
60            char *archiveMemberName;
61 
62+#if defined(USE_MMAP)
63            /* We can't mmap from the archive directly, as object
64               files need to be 8-byte aligned but files in .ar
65               archives are 2-byte aligned, and if we malloc the
66hunk ./rts/Linker.c 1772
67               mmap some anonymous memory and use that. We could
68               do better here. */
69            image = mmapForLinker(imageSize, MAP_ANONYMOUS, -1);
70+#elif defined(darwin_HOST_OS)
71+           // See loadObj()
72+           misalignment = machoGetMisalignment(f);
73+           image = stgMallocBytes(imageSize + misalignment, "loadArchive(file)");
74+           image += misalignment;
75+#else
76+           image = stgMallocBytes(imageSize, "loadArchive(file)");
77+#endif
78            n = fread ( image, 1, imageSize, f );
79            if (n != imageSize)
80hunk ./rts/Linker.c 1782
81- -               barf("loadObj: error whilst reading `%s'", path);
82+               barf("loadArchive: error whilst reading `%s'", path);
83 
84            archiveMemberName = stgMallocBytes(strlen(path) + fileNameSize + 3, "loadArchive(file)");
85            sprintf(archiveMemberName, "%s(%.*s)", path, (int)fileNameSize, file);
86hunk ./rts/Linker.c 1790
87            oc = mkOc(path, image, imageSize, archiveMemberName
88 #ifndef USE_MMAP
89 #ifdef darwin_HOST_OS
90- -                    , 0
91+                    , misalignment
92 #endif
93 #endif
94                     );
95hunk ./rts/Linker.c 1851
96    int fd;
97 #else
98    FILE *f;
99+#  if defined(darwin_HOST_OS)
100+   int misalignment;
101+#  endif
102 #endif
103hunk ./rts/Linker.c 1855
104+
105    IF_DEBUG(linker, debugBelch("loadObj %s\n", path));
106 
107    initLinker();
108hunk ./rts/Linker.c 1928
109     // We calculate the correct alignment from the header before
110     // reading the file, and then we misalign image on purpose so
111     // that the actual sections end up aligned again.
112- -   misalignment = machoGetMisalignment(f);
113- -   image = stgMallocBytes(fileSize + misalignment, "loadObj(image)");
114- -   image += misalignment;
115- -#  else
116- -   image = stgMallocBytes(fileSize, "loadObj(image)");
117- -#  endif
118+    misalignment = machoGetMisalignment(f);
119+    image = stgMallocBytes(fileSize + misalignment, "loadObj(image)");
120+    image += misalignment;
121+#   else
122+    image = stgMallocBytes(fileSize, "loadObj(image)");
123+#   endif
124 
125    {
126        int n;
127hunk ./rts/Linker.c 2214
128      */
129     if( m > n ) // we need to allocate more pages
130     {
131+        errorBelch("%s: WARNING: Allocating jump islands separately from "
132+                   "the object image itself. This may interfere with "
133+                   "relative branches to them.",
134+                   OC_INFORMATIVE_FILENAME(oc));
135+
136         oc->symbol_extras = mmapForLinker(sizeof(SymbolExtra) * count,
137                                           MAP_ANONYMOUS, -1);
138     }
139hunk ./rts/Linker.c 5181
140     struct mach_header header;
141     int misalignment;
142     
143- -    fread(&header, sizeof(header), 1, f);
144- -    rewind(f);
145+    {
146+        int n = fread(&header, sizeof(header), 1, f);
147+        if (n != 1) {
148+            barf("machoGetMisalignment: can't read the Mach-O header");
149+        }
150+    }
151+    fseek(f, -sizeof(header), SEEK_CUR);
152 
153 #if x86_64_HOST_ARCH || powerpc64_HOST_ARCH
154     if(header.magic != MH_MAGIC_64) {
155hunk ./rts/Linker.c 5191
156- -        errorBelch("Bad magic. Expected: %08x, got: %08x.\n",
157- -                   MH_MAGIC_64, header->magic);
158- -        return 0;
159+        barf("Bad magic. Expected: %08x, got: %08x.",
160+             MH_MAGIC_64, header.magic);
161     }
162 #else
163     if(header.magic != MH_MAGIC) {
164hunk ./rts/Linker.c 5196
165- -        errorBelch("Bad magic. Expected: %08x, got: %08x.\n",
166- -                   MH_MAGIC, header->magic);
167- -        return 0;
168+        barf("Bad magic. Expected: %08x, got: %08x.",
169+             MH_MAGIC, header.magic);
170     }
171 #endif
172 
173}
174[rts/Linker.c (ocAllocateSymbolExtras):
175pho@cielonegro.org**20101017100609
176 Ignore-this: c75c11ad4b6502282ac3d67c25dbdeca
177    Suppress false warnings on x86_64 platforms.
178    It's OK to allocate jump islands separately for x86_64 because
179    PC-relative jumps/accesses are only limited to 32 bits.
180] {
181hunk ./rts/Linker.c 2214
182      */
183     if( m > n ) // we need to allocate more pages
184     {
185+#if !defined(x86_64_HOST_ARCH)
186         errorBelch("%s: WARNING: Allocating jump islands separately from "
187                    "the object image itself. This may interfere with "
188                    "relative branches to them.",
189hunk ./rts/Linker.c 2219
190                    OC_INFORMATIVE_FILENAME(oc));
191- -
192+#endif
193         oc->symbol_extras = mmapForLinker(sizeof(SymbolExtra) * count,
194                                           MAP_ANONYMOUS, -1);
195     }
196}
197
198Context:
199
200[Only put the boot packages in the haddock contents/index
201Ian Lynagh <igloo@earth.li>**20101016180031
202 Ignore-this: 3dc276dc734fec207def24a1d10eb369
203 We don't install dph etc, so don't put them in the doc index.
204]
205[Correct the regexp used to search for extra packages
206Ian Lynagh <igloo@earth.li>**20101016123421
207 Ignore-this: 98e3f8fda1c5678730a00ec1ac25fde3
208 We weren't ignoring comment lines
209]
210[Add a -fghci-sandbox flag so that we can en/disable the ghci sandbox
211Ian Lynagh <igloo@earth.li>**20101015172746
212 Ignore-this: 52bc5729437a93a925d3586d9803623c
213 It's on by default (which matches the previous behaviour).
214 
215 Motivation:
216 GLUT on OS X needs to run on the main thread. If you
217 try to use it from another thread then you just get a
218 white rectangle rendered. For this, or anything else
219 with such restrictions, you can turn the GHCi sandbox off
220 and things will be run in the main thread.
221]
222[Fix boot; it was failing if darcs-all or validate were missing
223Ian Lynagh <igloo@earth.li>**20101015164549
224 Ignore-this: 14b3e98836647aff345e35bca2102f93
225 (which is the case in sdists)
226]
227[Comments and layout
228simonpj@microsoft.com**20101015131924
229 Ignore-this: 126fbdb629a08c1380c7a1f5cd967d97
230]
231[Make (Located a) an instance of Eq, Ord
232simonpj@microsoft.com**20101015131857
233 Ignore-this: 5da50f8dab06fcbc23ce149cd5080062
234 
235 Fulfils Trac #4369
236]
237[Give user-defined rules precedence over built-in rules
238simonpj@microsoft.com**20101015131814
239 Ignore-this: 76fbf36cb1a09e31c10c68c06b98937e
240 
241 This fixes Trac #4397.  See comments with 'isMoreSpecific'.
242]
243[Fix Trac #4401: meta-tyvars allocated by the constraint solver are always touchable
244simonpj@microsoft.com**20101015130818
245 Ignore-this: 7fcdaee825d9cadcf69c8c9b8967a446
246 
247   See Note [Touchable meta type variables] in TcSMonad
248]
249[Remove GHC.extendGlobalRdrScope, GHC.extendGlobalTypeScope
250simonpj@microsoft.com**20101013091107
251 Ignore-this: df65a43d261353ad53811b25507e913e
252 
253 These functions were added by
254 
255    Tue Apr 18 03:36:06 BST 2006  Lemmih <lemmih@gmail.com>
256    * Make the initial rdr and type scope available in the ghc-api
257 
258 The are extremely dubious, because they extend the Rdr and Type
259 env for every compilation.  The right thing to do is to use
260 the InteractiveContext for temporary extensions.
261 
262 So far as we know, no one uses them.  And if they are being used
263 it's probably a mistake.  So we're backing them out.
264]
265[InlinePrag needs an arity only for INLINE, not INLINABLE
266Simon Marlow <marlowsd@gmail.com>**20101015094925
267 Ignore-this: 3b338f0b2b49fa714b195067e0026ef7
268 This doesn't fix anything (we think), but it's morally correct.
269]
270[Fix #4346 (INLINABLE pragma not behaving consistently)
271Simon Marlow <marlowsd@gmail.com>**20101015094836
272 Ignore-this: 9a9d8ad42cfdf7f619adfac284bae021
273 Debugged thanks to lots of help from Simon PJ: we weren't updating the
274 UnfoldingGuidance when the unfolding changed.
275 Also, a bit of refactoring and additinoal comments.
276]
277[Have boot check that we have the dph packages when validating
278Ian Lynagh <igloo@earth.li>**20101014140556
279 Ignore-this: 61e365127933f4dbfb62be66115455bd
280]
281[Add more documentation for interruptible foreign calls
282Simon Marlow <marlowsd@gmail.com>**20101014084253
283 Ignore-this: 28ec0ddd958926a08ab816b6975da344
284]
285[minor refactoring
286Simon Marlow <marlowsd@gmail.com>**20100926105819
287 Ignore-this: 70d38e0a31a096c94dad5f346b0d91f6
288]
289[Fix for interruptible FFI handling
290Simon Marlow <marlowsd@gmail.com>**20100925193442
291 Ignore-this: a63ba3c60f577002a1d32b30bb45090c
292 Set tso->why_blocked before calling maybePerformBlockedException(), so
293 that throwToSingleThreaded() doesn't try to unblock the current thread
294 (it is already unblocked).
295]
296[interruptible FFI: more robust handling of the exception case in the interpreter
297Simon Marlow <marlowsd@gmail.com>**20100925193317
298 Ignore-this: 7f9e67835a3bd096154db2dcf533ec66
299]
300[Don't interrupt when task blocks exceptions, don't immediately start exception.
301Edward Z. Yang <ezyang@mit.edu>**20100925033026
302 Ignore-this: 6c0669995a2b66abf29d68b3711cb78e
303]
304[Interruptible FFI calls with pthread_kill and CancelSynchronousIO. v4
305Edward Z. Yang <ezyang@mit.edu>**20100919002905
306 Ignore-this: 43c260f90eeb9c03413f6e749e23101d
307 
308 This is patch that adds support for interruptible FFI calls in the form
309 of a new foreign import keyword 'interruptible', which can be used
310 instead of 'safe' or 'unsafe'.  Interruptible FFI calls act like safe
311 FFI calls, except that the worker thread they run on may be interrupted.
312 
313 Internally, it replaces BlockedOnCCall_NoUnblockEx with
314 BlockedOnCCall_Interruptible, and changes the behavior of the RTS
315 to not modify the TSO_ flags on the event of an FFI call from
316 a thread that was interruptible.  It also modifies the bytecode
317 format for foreign call, adding an extra Word16 to indicate
318 interruptibility.
319 
320 The semantics of interruption vary from platform to platform, but the
321 intent is that any blocking system calls are aborted with an error code.
322 This is most useful for making function calls to system library
323 functions that support interrupting.  There is no support for pre-Vista
324 Windows.
325 
326 There is a partner testsuite patch which adds several tests for this
327 functionality.
328]
329[Remove ghc-pkg's dependency on haskell98
330Ian Lynagh <igloo@earth.li>**20101013194356
331 Ignore-this: 273fdede3f21e682faa4dc6e61f4e7aa
332]
333[Build haskell98 and haskell2010 with stage2
334Ian Lynagh <igloo@earth.li>**20101013182759
335 Stops us accidentally depending on them
336]
337[Fix warning: Remove unused import
338Ian Lynagh <igloo@earth.li>**20101013141224
339 Ignore-this: ae8891440811f4676d9bb8e721ad22ca
340]
341[Fix warnings
342benl@ouroborus.net**20101013040335
343 Ignore-this: 4e5875827d2840de863000cdb35afb0e
344]
345[RegAlloc: Track slot liveness over jumps in spill cleaner
346benl@ouroborus.net**20101013015414
347 Ignore-this: ccd4a148908b7fbdc6ea76acf527c16b
348]
349[Bump Cabal dep
350Ian Lynagh <igloo@earth.li>**20101012154528
351 Ignore-this: da8539b957b74c7da91efff8ac25872
352]
353[Remove __HASKELL1__, __HASKELL98__, __CONCURRENT_HASKELL__
354Ian Lynagh <igloo@earth.li>**20101012134700
355 Ignore-this: d0db531be58537c52802de1c62694eb1
356 We used to define these CPP symbols, but nothing on hackage uses them
357 and the first 2 are no longer correct (as we support multiple Haskell
358 versions).
359]
360[Follow Cabal changes: Cabal no longer has a docbook userguide
361Ian Lynagh <igloo@earth.li>**20101012130538
362 Ignore-this: bf0b30e465c13f82725d72aec145e939
363 For now we don't build the Cabal userguide, but we should add markdown
364 support so that we can do so.
365]
366[Fix build on Windows: ghc-pkg/Main.hs needs ForeignFunctionInterface
367Ian Lynagh <igloo@earth.li>**20101012112111]
368[Remove unnecessary import
369Ian Lynagh <igloo@earth.li>**20101010222231
370 Ignore-this: 6c7d5cee72837e2af43c2807e10ad602
371]
372[Make "./validate --slow" run the full testsuite
373Ian Lynagh <igloo@earth.li>**20101007004327
374 Ignore-this: 2dee8262c19fd5d5034a186203e20d0f
375]
376[Fix build following haskell98 and -fglasgow-exts changes
377Ian Lynagh <igloo@earth.li>**20101006160656
378 Ignore-this: c56bdb0f01da897b3de75bcab1e2887c
379]
380[Don't automatically link the haskell98 package
381Ian Lynagh <igloo@earth.li>**20101006130235
382 The default language is now Haskell2010, so this was a little odd.
383 Also, --make is now on by default, so this was largely irrelevant.
384]
385[Deprecate -fglasgow-exts
386Ian Lynagh <igloo@earth.li>**20101006124413
387 Ignore-this: 70be67898a633ab31d76f6fca557d55
388]
389[Remove Opt_GADTs and Opt_TypeFamilies from -fglasgow-exts
390Ian Lynagh <igloo@earth.li>**20101006122000
391 Ignore-this: 7153ec880b2b5f9d332ae2f57b97afcc
392 This means most code doesn't get caught by monomorphic local bindings.
393]
394[Fix Trac #4360: omitted case in combineCtLoc
395simonpj@microsoft.com**20101008135747
396 Ignore-this: 834636a97af6469862a822809253db41
397]
398[Beautiful new approach to the skolem-escape check and untouchable
399simonpj@microsoft.com**20101008133751
400 Ignore-this: 33517a772cfdfbf4aa4678609f3dcd71
401 
402 Instead of keeping a *set* of untouchable variables in each
403 implication contraints, we keep a *range* of uniques for the
404 *touchable* variables of an implication.  This are precisely
405 the ones we would call the "existentials" if we were French.
406 
407 It turns out that the code is more efficient, and vastly easier
408 to get right, than the set-based approach.
409 
410 Fixes Trac #4355 among others
411]
412[Do less simplification when doing let-generalisation
413simonpj@microsoft.com**20101008133542
414 Ignore-this: 71366d0de37f10ffba2edc9f3927ddbe
415 
416 This fixes Trac #4361.  In a rather delicate way, but
417 no more delicate than before.  A more remoseless typechecker
418 would reject #4361 altogether.
419 
420 See Note [Avoid unecessary constraint simplification]
421]
422[Suppress ambiguity errors if any other errors occur
423simonpj@microsoft.com**20101008111318
424 Ignore-this: 40f014265c1ab15fe172baaf76c23c87
425]
426[Fix Trac #4361: be more discerning when inferring types
427simonpj@microsoft.com**20101008111227
428 Ignore-this: 33656f9a151f494b26b6318b5fcffef
429 
430 Note [Avoid unecessary constraint simplification] in TcSimplify
431]
432[Float out partial applications
433Simon Marlow <marlowsd@gmail.com>**20101008092709
434 Ignore-this: 2dc9d10597c19d0598ef2fc3cf74156d
435 
436 This fixes at least one case of performance regression in 7.0, and
437 is nice win on nofib:
438 
439         Program           Size    Allocs   Runtime   Elapsed
440             Min          +0.3%    -63.0%    -38.5%    -38.7%
441             Max          +1.2%     +0.2%     +0.9%     +0.9%
442  Geometric Mean          +0.6%     -3.0%     -6.4%     -6.6%
443]
444[Suppress knock-on typechecker errors
445simonpj@microsoft.com**20101008094348
446 Ignore-this: 8d125926286a7614fa1ce998e3b26d04
447 
448 The error cascade caused puzzling errors in T4093b, and
449 suppressing some seems like a good plan.  Very few test
450 outputs change.
451]
452[Some refactoring and simplification in TcInteract.occurCheck
453simonpj@microsoft.com**20101007163500
454 Ignore-this: d43d09370ab27b8796062e2e98ce7e9
455]
456[Comments only
457simonpj@microsoft.com**20101007130301
458 Ignore-this: ab46592edd3d24786bbce42c50feb4fd
459]
460[Implement auto-specialisation of imported Ids
461simonpj@microsoft.com**20101007111051
462 Ignore-this: 45257ff6e9597e4fa4de10b0657e27d6
463 
464 This big-ish patch arranges that if an Id 'f' is
465   * Type-class overloaded
466        f :: Ord a => [a] -> [a]
467   * Defined with an INLINABLE pragma
468        {-# INLINABLE f #-}
469   * Exported from its defining module 'D'
470 
471 then in any module 'U' that imports D
472 
473 1. Any call of 'f' at a fixed type will generate
474    (a) a specialised version of f in U
475    (b) a RULE that rewrites unspecialised calls to the
476        specialised on
477 
478   e.g. if the call is (f Int dOrdInt xs) then the
479   specialiser will generate
480      $sfInt :: [Int] -> [Int]
481      $sfInt = <code for f, imported from D, specialised>
482      {-# RULE forall d.  f Int d = $sfInt #-}
483 
484 2. In addition, you can give an explicit {-# SPECIALISE -#}
485    pragma for the imported Id
486      {-# SPECIALISE f :: [Bool] -> [Bool] #-}
487    This too generates a local specialised definition,
488    and the corresponding RULE
489 
490 The new RULES are exported from module 'U', so that any module
491 importing U will see the specialised versions of 'f', and will
492 not re-specialise them.
493 
494 There's a flag -fwarn-auto-orphan that warns you if the auto-generated
495 RULES are orphan rules. It's not in -Wall, mainly to avoid lots of
496 error messages with existing packages.
497 
498 Main implementation changes
499 
500  - A new flag on a CoreRule to say if it was auto-generated.
501    This is persisted across interface files, so there's a small
502    change in interface file format.
503 
504  - Quite a bit of fiddling with plumbing, to get the
505    {-# SPECIALISE #-} pragmas for imported Ids.  In particular, a
506    new field tgc_imp_specs in TcGblEnv, to keep the specialise
507    pragmas for imported Ids between the typechecker and the desugarer.
508 
509  - Some new code (although surprisingly little) in Specialise,
510    to deal with calls of imported Ids
511]
512[Make NameEnv back into type NameEnv a = UniqFM a
513simonpj@microsoft.com**20101007104638
514 Ignore-this: 24857c013461788be354520e84f4c286
515 
516 I don't think the type distinction of declaring NameEnv with a newtype
517 (as it was) is really useful to us. Moreover, VarEnv is a UniqFM, and
518 I do sometimes want to build an envt with Ids and look up with Names.
519 
520 This may not be the last word on the subject.
521]
522[Improve the rule-matcher
523simonpj@microsoft.com**20101007103700
524 Ignore-this: 9de96237dc4b73a43326bd568e34b53b
525 
526 Previously it was rejecting the match
527 
528   Template: forall s t. map s t
529   Actual:   map Int t
530 
531 which should obviously be fine.  It turns out that this kind of match
532 comes up when specialising.  By freshening that t we could avoid the
533 difficulty, but morally the (forall t) binds t and the rule should
534 be alpha-equivalent regardless of the forall'd variables.
535 
536 This patch makes it so, and incidentally makes matching a little
537 more efficient.  See Note [Eta expansion] in VarEnv.
538]
539[Fix Trac #4345: simplifier bug
540simonpj@microsoft.com**20101007102720
541 Ignore-this: 261c1c9f094df344ce34de814f8b60c5
542 
543 This is another long-standing bug, in which there was a possibility
544 that a loop-breaker could lose its loop-breaker-hood OccInfo,
545 and then the simplifer re-simplified the expression. Result, either
546 non-termination or, in the case of #4345, an unbound identifier.
547 
548 The fix is very simple, in Id.transferPolyIdInfo.
549 See Note [transferPolyIdInfo].
550]
551[Avoid redundant simplification
552simonpj@microsoft.com**20101007095935
553 Ignore-this: 61bd1a2c508260f558866e6a88c29fa3
554 
555 When adding specialisation for imported Ids, I noticed that the
556 Glorious Simplifier was repeatedly (and fruitlessly) simplifying the
557 same term.  It turned out to be easy to fix this, because I already
558 had a flag in the ApplyTo and Select constructors of SimplUtils.SimplCont.
559 
560 See Note [Avoid redundant simplification]
561]
562[Make the occurrence analyser deal correctly with RULES for imported Ids
563simonpj@microsoft.com**20101007094100
564 Ignore-this: 335b1cad013524e42b31e88c0a7a00f6
565 
566 This patch fixes a long-standing lurking bug, but it surfaced when I
567 was adding specialisation for imported Ids.
568 
569 See Note [ImpRuleUsage], which explains the issue.   The solution
570 seems more complicated than the problem really deserves, but I
571 could not think of a simpler way, so I just bit the bullet and
572 wrote the code.  Improvements welcome.
573]
574[Make warning-free
575simonpj@microsoft.com**20101007092007
576 Ignore-this: 4bae0c470a8a1f96d21990d1f3cf1f93
577]
578[This is just white-space and layout
579simonpj@microsoft.com**20101007091618
580 Ignore-this: 759c0335df70fce32558e967f140803a
581 
582 (At least, I don't think there is anything else.)
583]
584[Fix an ASSERT failure in FamInstEnv
585simonpj@microsoft.com**20101007091327
586 Ignore-this: a8c08ccb7ec2bc65864a674b5441539
587 
588 I added a lot of comments too, to explain the preconditions;
589 esp Note [FamInstEnv]
590]
591[Fix a looping bug in the new occur-check code
592simonpj@microsoft.com**20101007084104
593 Ignore-this: a02a2deafb9ec986ef1565f4596049ed
594]
595[Fix test T4235 with -O
596simonpj@microsoft.com**20101006155223
597 Ignore-this: f0fa0fe2f0c493e362d528d71f7a64e1
598 
599 The tag2Enum rule wasn't doing the right thing for
600 enumerations with a phantom type parameter, like
601    data T a = A | B
602]
603[Make warning-free
604simonpj@microsoft.com**20101006155033
605 Ignore-this: 221a3c95a6079c6ecc0468996a38b048
606]
607[Major bugfixing pass through the type checker
608dimitris@microsoft.com**20101006152854]
609[Typechecker performance fixes and flatten skolem bugfixing
610dimitris@microsoft.com**20101004130200
611 Ignore-this: 86721ba3f09479c146a0710796b43459
612]
613[Performance bug fixes
614dimitris@microsoft.com**20100923143918]
615[Fix Trac #4371: matching of view patterns
616simonpj@microsoft.com**20101006115316
617 Ignore-this: 494b28b91f1e6392b2f1521cda0e83b1
618]
619[Remove unused NoMatchContext construtor
620simonpj@microsoft.com**20101006115251
621 Ignore-this: 8985ff1dac51fb652bd65657a630a792
622]
623[Refactoring: mainly rename ic_env_tvs to ic_untch
624simonpj@microsoft.com**20101006102830
625 Ignore-this: 32999403a3f447e14b59cec7896027ff
626 
627 Plus remember to zonk the free_tvs in TcUnify.newImplication
628]
629[remove unnecessary/broken definition of mask_
630Simon Marlow <marlowsd@gmail.com>**20101002195118
631 Ignore-this: 4cdc9c95d40e01cdfe2d0ac411476603
632]
633[-fwarn-tabs: add "Warning" to the message
634Simon Marlow <marlowsd@gmail.com>**20101002195100
635 Ignore-this: 589a36daa3426ab51f2fb140e38df6c
636]
637[give a better error message in the non-threaded RTS for out-of-range FDs
638Simon Marlow <marlowsd@gmail.com>**20100929212916
639 Ignore-this: e94c9f390b8f79d24895a80f9d16c8d9
640 
641 # ./aw
642 aw: file descriptor 1027 out of range for select (0--1024).
643 Recompile with -threaded to work around this.
644]
645[Fix a very rare crash in GHCi
646Simon Marlow <marlowsd@gmail.com>**20101005144735
647 Ignore-this: dad1cd08934bae2ba47e72c0c000acfa
648 When a BCO with a zero-length bitmap was right at the edge of
649 allocated memory, we were reading a word of non-existent memory.
650 
651 This showed up as a segfault in T789(ghci) for me, but the crash was
652 extremely sensitive and went away with most changes.
653 
654 Also, optimised scavenge_large_bitmap a bit while I was in there.
655]
656[Using 'stdcall' when it is not supported is only a warning now (#3336)
657Simon Marlow <marlowsd@gmail.com>**20100924152445
658 Ignore-this: 66c5903a600a47485a7583535bb38455
659]
660[remove unnecessary stg_noForceIO (#3508)
661Simon Marlow <marlowsd@gmail.com>**20100924150202
662 Ignore-this: dec52de9cd9da7dcedae12b20691aba9
663]
664[Replace an outputStr with putStrLn calls; fixes #4332
665Ian Lynagh <igloo@earth.li>**20101003125707
666 Ignore-this: eb4b5f60d9d9d3f7dc203869927b28ba
667]
668[make test and fulltest targets in the main Makefile; fixes #4297
669Ian Lynagh <igloo@earth.li>**20100930224741
670 You can now run "make test" in the root, and the fast testsuite will be
671 run with cleaning enabled. It will also put the summary in
672 testsuite_summary.txt.
673]
674[Don't show the loaded packages in ":show packages"; fixes #4300
675Ian Lynagh <igloo@earth.li>**20100930210128
676 It's never worked properly, and the information is in ":show linker".
677]
678[Handle EXTRA_LIBRARIES when building programs
679Ian Lynagh <igloo@earth.li>**20100930192552
680 Ignore-this: 401a26e18d25dcaee010b13eaed8f011
681 And set hp2ps's EXTRA_LIBRARIES. Based on a patch from Sergei Trofimovich.
682]
683[Fix the doc directory on Windows
684Ian Lynagh <igloo@earth.li>**20100929133328]
685[Remove an unused import on Windows
686Ian Lynagh <igloo@earth.li>**20100929000024
687 Ignore-this: 2899e0e5a47122e637fb5c8aa0df52ab
688]
689[Use showCommandForUser when showing tracing commands
690Ian Lynagh <igloo@earth.li>**20100928235844
691 Ignore-this: 8a4a9c9f8a8029e708c4297b096b6ef1
692]
693[Fix hsc2hs docs: 'gcc' is now the default compiler, not 'ghc'; fixes #4341
694Ian Lynagh <igloo@earth.li>**20100928201938]
695[New member "archiveMemberName" for struct _ObjectCode
696pho@cielonegro.org**20100927224145
697 Ignore-this: 628bc605e7dc4f0c4856c6f7ad23d9ee
698 
699 struct _ObjectCode should be able to retain the name of archive members.
700 Though currently the only use of those names are for debugging outputs.
701]
702[Use an empty signal handler for SIGPIPE instead of SIG_IGN
703Simon Marlow <marlowsd@gmail.com>**20100925193548
704 Ignore-this: b4dc5de32740a7c5fd8fe4b3bfb1300f
705 
706 This is so that the SIGPIPE handler gets reset to the default
707 automatically on exec().
708]
709[Fix the TH deps
710Ian Lynagh <igloo@earth.li>**20100925210029
711 Ignore-this: 32b832301a3625d4ba70f84c5c4f94d2
712]
713[Check inplace doesn't exist before we try to create it
714Ian Lynagh <igloo@earth.li>**20100924191858
715 This fixes rerunning configure in a tree which already has an inplace
716 directory. Edward Z Yang ran into this; I guess whether it actually
717 fails depends on details of your installation, or we'd have run into
718 it sooner.
719]
720[Fix an egregious bug: INLINE pragmas on monomorphic Ids were being ignored
721simonpj@microsoft.com**20100924155815
722 Ignore-this: 38c6eec6710a92df7662a55fc5132c15
723 
724 I had do to some refactoring to make this work nicely
725 but now it does. I can't think how this escaped our
726 attention for so long!
727]
728[Eta expand only lambdas that bind a non-dictionary Id
729simonpj@microsoft.com**20100924155707
730 Ignore-this: 7cc265eaf6c0bb3fa12eb311d92594ac
731 
732 See Note [When to eta expand]. The idea is that dictionary
733 lambdas are invisible to the user, so we shouldn't eta
734 expand them.
735]
736[Add a comment
737simonpj@microsoft.com**20100924155620
738 Ignore-this: de210a1afdd40328824803e1d77b4d7f
739]
740[Add a debug print
741simonpj@microsoft.com**20100924155614
742 Ignore-this: 1a58b6d297fc77d6ded8eec7ea9f895d
743]
744[Just moving comments around
745simonpj@microsoft.com**20100924155600
746 Ignore-this: 96635b8e8c9d88b50d82938568152ef8
747]
748[use putStrLn instead of Haskeline's outputStrLn
749Simon Marlow <marlowsd@gmail.com>**20100924133154
750 Ignore-this: 7581ae11714a9a52e78ba098c3c216b3
751 use of the latter caused problems for Claus Reinke's macros that
752 redirect stdout.
753]
754[Change "OPTIONS" to "OPTIONS_GHC" in error messages; fixes #4327
755Ian Lynagh <igloo@earth.li>**20100924120423
756 Ignore-this: 1697c83a5c346db640c0a2e22c69ff55
757]
758[Add deps for TH uses in vector
759Ian Lynagh <igloo@earth.li>**20100923220244
760 Ignore-this: 54c3386b1c268821fcdd34b84bc8c6a4
761]
762[Bump Cabal dep
763Ian Lynagh <igloo@earth.li>**20100923143241]
764[Update Cabal's version number
765Ian Lynagh <igloo@earth.li>**20100923141719]
766[Build primitive with stage2
767Ian Lynagh <igloo@earth.li>**20100923140525
768 Ignore-this: 110a819b78a57629a7edf1d4facdc191
769]
770[Fix the Windows __chkstk build error (missing Linker symbol)
771Simon Marlow <marlowsd@gmail.com>**20100924113837
772 Ignore-this: 48f0907bb1bd5eaa0730b94a6bd94ea
773]
774[emit a helpful error message for missing DPH packages
775Simon Marlow <marlowsd@gmail.com>**20100923141957
776 Ignore-this: 55ff2ee90c94524e023e014243bfe5df
777]
778[Fix computation of installed packages
779simonpj@microsoft.com**20100924084737
780 Ignore-this: a597d2fa8be5135ba8ead6d2624b3d71
781 
782 This is a follow-on to Simon's patch yesterday, developed
783 with him.  It cleans up the computation of how packages
784 are installed, and installs the right ones.
785]
786[Fix braino in WwLib/Literal patch
787simonpj@microsoft.com**20100924070914
788 Ignore-this: f6eb3a42e10f8aa7920de541cdfe76d8
789]
790[For now, switch off incomplete-pattern warnings in containers
791simonpj@microsoft.com**20100923130117
792 Ignore-this: 7ffa58567f7a33aafe256492999da325
793 
794 Put it back on when my patch is applied to the containers repo.
795 (the one that removes two refuable lambdas)
796]
797[Make -funfolding-dict-threshold work properly
798simonpj@microsoft.com**20100923130032
799 Ignore-this: 417788f5b09d1d624f6b6371852c80c7
800 
801 and increase its default value. This makes overloaded functions
802 a bit keener to inline.  Which fixes Trac #4321
803]
804[Impredicative types is no longer deprecated
805simonpj@microsoft.com**20100923125910
806 Ignore-this: 2bbaeb38b5e8424551677c0add627683
807]
808[Do not make FunctionalDependencies force MonoLocalBinds
809simonpj@microsoft.com**20100923125900
810 Ignore-this: f4ae1fd07c87ec14f60bdfe3863ba7a9
811]
812[move CHECKED settings to the right place
813Simon Marlow <marlowsd@gmail.com>**20100923123558
814 Ignore-this: e00a0eb5855463cc9b953670b3bbf211
815]
816[turn off -Werror for primitive and vector
817Simon Marlow <marlowsd@gmail.com>**20100923122055
818 Ignore-this: 54d7b80f3f893385e1c3ef431e2a8a7b
819]
820[Add primitive and vector packages for DPH support
821Simon Marlow <marlowsd@gmail.com>**20100923104542
822 Ignore-this: c070d015385b0a0797394132dcbb7670
823 DPH is now using the public vector package instead of its internal
824 version.
825 
826 vector and primitive are not "boot" packages; they aren't required to
827 build GHC, but they are required to validate (because we include DPH
828 when validating).
829 
830 If you say './darcs-all get --no-dph' then you don't get DPH, vector,
831 or primitive.
832]
833[Refactoring and tidy up in the build system
834Simon Marlow <marlowsd@gmail.com>**20100923095642
835 Ignore-this: f7bf3a1fd160149d89b26f464b064fb1
836 
837 Instead of the ghc-stage and ghc-stage2-package files in a package, we
838 now have a list of these in ghc.mk.  There are other similar lists (of
839 boot-packages and non-installable packages), so this is not too bad,
840 and is simpler.
841 
842 While poking around in the top-level ghc.mk file I spotted various
843 opportunities to clean up and re-order some of the cruft that has
844 accumulated over time.
845]
846[Allow absent State# RealWorld arguments
847simonpj@microsoft.com**20100923111356
848 Ignore-this: c2d57633dec0293ebe6723ea3a4bb5df
849]
850[Add notSCCNote, and use it
851simonpj@microsoft.com**20100923105949
852 Ignore-this: c8cc758656558a7f366bf784d75f0304
853 
854 The point here is that SCCs get in the way of eta
855 expansion and we must treat them uniformly.
856]
857[Remove use of lambda with a refutable pattern
858simonpj@microsoft.com**20100923105901
859 Ignore-this: d7d48b94e5744717a838591a1cc79cf0
860]
861[Avoid ASSERT black hole
862simonpj@microsoft.com**20100923105820
863 Ignore-this: 5419d450871be22c8781ac3f0f40d76a
864 
865 When this ASSERT tripped in CoreToStg it tried to print out
866 too much, which tripped the asssertion again.  Result: an
867 infinite loop with no output at all.  Hard to debug!
868]
869[Rejig the absent-arg stuff for unlifted types
870simonpj@microsoft.com**20100923105732
871 Ignore-this: 69daa35816b948b0c4d259c73a5e928e
872 
873 This is what was giving the "absent entered" messages
874 See Note [Absent errors] in WwLib.  We now return a
875 suitable literal for absent values of unlifted type.
876]
877[Remove -fwarn-simple-patterns, and make -fwarn-incomplete-patterns include lambdas
878simonpj@microsoft.com**20100922133934
879 Ignore-this: e851a2fb0377e10c28c506f0bf14cc85
880 
881 This makes
882      \(x:xs) -> e
883 want when you have -fwarn-incomplete-patterns, which is consistent.
884]
885[Get rid of non-exhaustive lambda
886simonpj@microsoft.com**20100922133801
887 Ignore-this: 748b2d5b43b02b6591b81abe7c105cd6
888]
889[Fix an ASSERT failure with profiling
890simonpj@microsoft.com**20100922133741
891 Ignore-this: 170b2e94d6ee8cc7444cc4bb515328a0
892 
893 The problem arose with this kind of thing
894 
895    x = (,) (scc "blah" Nothing)
896 
897 Then 'x' is marked NoCafRefs by CoreTidy, becuase it has
898 arity 1, and doesn't mention any caffy things.
899 
900 That in turns means that CorePrep must not float out the
901 sat binding to give
902 
903   sat = scc "blah" Nothing
904   x = (,) sat
905 
906 Rather we must generate
907 
908   x = \eta. let sat = scc "blah" Nothing
909             in (,) sat eta
910 
911 URGH! This Caf stuff is such a mess.
912]
913[Remove an out of date paragraph from the user guide; fixes #4331
914Ian Lynagh <igloo@earth.li>**20100922225239]
915[Fix bindisttest when GhcProfiled = YES
916Ian Lynagh <igloo@earth.li>**20100921222634
917 Ignore-this: 47c620fd6bec745e3eb699d9f53441d8
918]
919[Fixes for when HADDOCK_DOCS=NO
920Ian Lynagh <igloo@earth.li>**20100921213916
921 Ignore-this: e0e069555c6db9d01a8ac70ba4dde591
922]
923[Bump version to 7.1
924Ian Lynagh <igloo@earth.li>**20100921195935
925 Ignore-this: 4563987e6885d5ef55995ec0fa0d5ae8
926]
927[Don't use -march=i686 on powerpc-apple-darwin
928Ian Lynagh <igloo@earth.li>**20100921193721
929 Thorikil ran into this when doing a PPC OS X build. We now also don't
930 use -m32 on PPC/OSX, but I don't think it should be necessary. We can
931 add it back if it does turn out to be.
932]
933[add a simple trace facility to the build system
934Simon Marlow <marlowsd@gmail.com>**20100921134729
935 Ignore-this: d23ea2d62a648d0711b4b07d98e1b79f
936 
937 saying
938 
939   make TRACE=1
940 
941 prints most of the macro calls and their arguments.  It's easy to
942 trace new macros; see rules/trace.mk.
943]
944[fix building with extra packages (packages were added to BUILD_DIRS twice)
945Simon Marlow <marlowsd@gmail.com>**20100921100153
946 Ignore-this: 4b425dff9777871ad5ba3e05e1d14483
947 Also add some comments about what extra-packages is doing
948]
949[add extra packages to $(EXTRA_PACKAGES), so we avoid installing them by default
950Simon Marlow <marlowsd@gmail.com>**20100920144307
951 Ignore-this: 3395825d911a8bf7ba8385518d8b517b
952]
953[Fix indexing error in archive loader
954Ian Lynagh <igloo@earth.li>**20100921121642]
955[Add some -Dl belches
956Ian Lynagh <igloo@earth.li>**20100921121624]
957[Add casts to fix warnings
958Ian Lynagh <igloo@earth.li>**20100921121714]
959[Add support for BSD-variant large filenames in .a archives
960Ian Lynagh <igloo@earth.li>**20100921000451]
961[Tell Cabal that we're not building GHCi libs if UseArchivesForGhci=YES
962Ian Lynagh <igloo@earth.li>**20100920230449]
963["UseArchivesForGhci = YES" on darwin
964Ian Lynagh <igloo@earth.li>**20100920211538]
965[Add a dependency that my OS X build has recently started tripping up over
966Ian Lynagh <igloo@earth.li>**20100920210239]
967[Add "Use archives for ghci" to --info output
968Ian Lynagh <igloo@earth.li>**20100920210523]
969[Implement archive loading for ghci
970Ian Lynagh <igloo@earth.li>**20100920201620]
971[Tweak gen_contents_index now dph may not be there
972Ian Lynagh <igloo@earth.li>**20100920201513]
973[Filter out the FFI library when loading package in ghci
974Ian Lynagh <igloo@earth.li>**20100920181032
975 The FFI GHCi import lib isn't needed as
976 compiler/ghci/Linker.lhs + rts/Linker.c link the
977 interpreted references to FFI to the compiled FFI.
978 We therefore filter it out so that we don't get
979 duplicate symbol errors.
980]
981[Loosen the conditions for -XUndecidableInstances; fixes Trac #4200
982simonpj@microsoft.com**20100919162623
983 Ignore-this: 2f4323e278b1ce9250549727ffd0aa1b
984]
985[Further improvements in error messages
986simonpj@microsoft.com**20100919153355
987 Ignore-this: b6fa0b11ae893df1a3ca68f78e427fa
988]
989[Add a flag -fwarn-missing-local-sigs, and improve -fwarn-mising-signatures
990simonpj@microsoft.com**20100919153327
991 Ignore-this: fda8dfca450054ea692be0ee30b01885
992 
993 The new flag prints out a warning if you have a local,
994 polymorphic binding that lacks a type signature. It's meant
995 to help with the transition to the new typechecker, which
996 discourages local let-generalisation.
997 
998 At the same time I moved the missing-signature code to TcHsSyn,
999 where it takes place as part of zonking.  That way the
1000 types are reported after all typechecking is complete,
1001 thereby fixing Trac #3696.  (It's even more important for
1002 local bindings, which is why I made the change.)
1003]
1004[Include the "stupid theta" in the type of $con2tag
1005simonpj@microsoft.com**20100919152201
1006 Ignore-this: d95fae78a0e66f48bbd5862573a11f4d
1007]
1008[Add a release note about the typechecker
1009Ian Lynagh <igloo@earth.li>**20100919132927]
1010[Enable shared libs on OpenBSD
1011Matthias Kilian <kili@outback.escape.de>**20100918205040
1012 Ignore-this: 729dd7ac0bba5d758f43bc31b541896
1013]
1014[Add separate functions for querying DynFlag and ExtensionFlag options
1015Ian Lynagh <igloo@earth.li>**20100918163815
1016 and remove the temporary DOpt class workaround.
1017]
1018[Fix mkUserGuidePart deps
1019Ian Lynagh <igloo@earth.li>**20100918145904
1020 We need to directly depend on the stage1 libs. The stage1 compiler lib
1021 doesn't depend on them.
1022]
1023[Fix build on cygwin: Normalise slashes in .depend files to be /
1024Ian Lynagh <igloo@earth.li>**20100918132328
1025 Ignore-this: 664f5ef4a41a4461eb34fe2ca7f2729a
1026]
1027[extra packages info is now read from packages file
1028Ian Lynagh <igloo@earth.li>**20100917224409
1029 rather than being repeated in the build system
1030]
1031[Tweak darcs-all
1032Ian Lynagh <igloo@earth.li>**20100917194435]
1033[Bump dependencies
1034Ian Lynagh <igloo@earth.li>**20100917183609]
1035[Library release notes for 7.0.1
1036Ian Lynagh <igloo@earth.li>**20100917174850]
1037[Fix overriding of implicit parameters in the solver
1038simonpj@microsoft.com**20100917140403
1039 Ignore-this: af76732309c7e2ca6b04f49327e9c14b
1040]
1041[Minor type printing amomaly
1042simonpj@microsoft.com**20100917140204
1043 Ignore-this: c90cb2e51421b4543a827e096051772e
1044]
1045[Spaces only
1046simonpj@microsoft.com**20100917140156
1047 Ignore-this: 7e34479502f7cb87d762355e40cbd012
1048]
1049[Minor refactoring
1050simonpj@microsoft.com**20100917140150
1051 Ignore-this: 6c0648b949b91b7e2f23c136b124faf2
1052]
1053[Add types of implicit parameters as untouchable
1054simonpj@microsoft.com**20100917140138
1055 Ignore-this: ba80740a557a9ba062dc7756e2320d17
1056 
1057 This is a tricky point:
1058    see Note [Implicit parameter untouchables]
1059]
1060[Better pretty printing of implicit parameters
1061simonpj@microsoft.com**20100917140054
1062 Ignore-this: 867dd67818a5bd687b2b6a1b59e15775
1063]
1064[Yet more error message improvement
1065simonpj@microsoft.com**20100917121206
1066 Ignore-this: 647fe8129d1d39d81e8249debd8df94e
1067]
1068[More error message wibbles
1069simonpj@microsoft.com**20100917094721
1070 Ignore-this: 8ec2f150b96b26af2e9ab7ac2b371fc7
1071]
1072[More error refactoring
1073simonpj@microsoft.com**20100917092834
1074 Ignore-this: 2d570ac0b9cc11305ddd33d093d11324
1075]
1076[Refactor type errors a bit
1077simonpj@microsoft.com**20100917080726
1078 Ignore-this: 33da4549373f585064e2ee22b50ad6ac
1079 
1080 Improves kind error messages in paticular
1081]
1082[Fix a very subtle shadowing bug in optCoercion
1083simonpj@microsoft.com**20100916170452
1084 Ignore-this: 9041cfb3c93e27a5e644e57815313aae
1085 
1086 See Note [Subtle shadowing in coercions]
1087 
1088 This is what was going wrong in Trac 4160.
1089]
1090[Fix bad error in tyVarsOfType
1091simonpj@microsoft.com**20100916170348
1092 Ignore-this: 67c8ce96a668cf6e3a38b82c893bcd81
1093 
1094 We weren't gathering the type variables free in the kind
1095 of a coercion binder!
1096]
1097[More assertions
1098simonpj@microsoft.com**20100916170310
1099 Ignore-this: 7fdcb53c99d791621a3d7e01ef454404
1100]
1101[Add more location info in CoreLint
1102simonpj@microsoft.com**20100916170229
1103 Ignore-this: 6558bab544b4f30189e0430668db87c3
1104]
1105[Print coercion variables as such (debugging change only)
1106simonpj@microsoft.com**20100916165944
1107 Ignore-this: c6d2001c1d8279a2288cb63bc339577d
1108]
1109[Remove pprTrace
1110simonpj@microsoft.com**20100915225935
1111 Ignore-this: 28185bbfa9732386f3c0f3eb4781a637
1112]
1113[Remove dead code dealing with type refinement
1114simonpj@microsoft.com**20100915223230
1115 Ignore-this: 62824b5c2ec1077c7642163352559621
1116]
1117[Use mkAppTy
1118simonpj@microsoft.com**20100915223205
1119 Ignore-this: e79e087b6a49219e9088846a1253a153
1120 
1121 Using AppTy in CoreLint was giving a bogus Lint failure
1122]
1123[Comments only
1124simonpj@microsoft.com**20100915221253
1125 Ignore-this: 3a45ea614188ccbb4a462de5cac96eda
1126]
1127[Extend eta reduction to work with casted arguments
1128simonpj@microsoft.com**20100915221229
1129 Ignore-this: 24b103dcdf70331211507af929789f86
1130 
1131 See Trac #4201, and
1132 Note [Eta reduction with casted arguments]
1133 
1134 Thanks to Louis Wasserman for suggesting this, and
1135 implementing an early version of the patch
1136]
1137[Allow "INLINEABLE" as a synonym
1138simonpj@microsoft.com**20100915154249
1139 Ignore-this: f41f80cb769e9acd5b463b170df698d0
1140]
1141[Documentation for INLINABLE
1142simonpj@microsoft.com**20100915154235
1143 Ignore-this: f942c02bcadc0d2d2f05b9369f93e280
1144]
1145[Implement TH reification of instances (Trac #1835)
1146simonpj@microsoft.com**20100915151242
1147 Ignore-this: 97dfa83db7da8f6cbd1b96801a57f8c5
1148 
1149 Accompanying patch for template-haskell package is reqd
1150]
1151[errno corresponding to ERROR_NO_DATA should be EPIPE (non-threaded RTS)
1152Simon Marlow <marlowsd@gmail.com>**20100915141809
1153 Ignore-this: 709c7280fbaa762e7071fb8796e8c01e
1154]
1155[Windows: use a thread-local variable for myTask()
1156Simon Marlow <marlowsd@gmail.com>**20100915120627
1157 Ignore-this: 13ffa4f19ebd319fe672af53af8d0b9a
1158 Which entailed fixing an incorrect #ifdef in Task.c
1159]
1160[Fix typo
1161Ian Lynagh <igloo@earth.li>**20100915140814]
1162[Add quotes in error message
1163simonpj@microsoft.com**20100915144724
1164 Ignore-this: c5158047c0aa41947a79e4c8edbe54f4
1165]
1166[Fix isDefaultInlinePragma
1167simonpj@microsoft.com**20100915144710
1168 Ignore-this: c9addf6bf811b23dc12603cf8521aa6c
1169]
1170[Implement INLINABLE pragma
1171simonpj@microsoft.com**20100915124442
1172 Ignore-this: 80a4ab2c2d65b27868dc9b2e954d6c6f
1173 
1174 Implements Trac #4299.  Documentation to come.
1175]
1176[Less voluminous error when derived code doesn't typecheck
1177simonpj@microsoft.com**20100915072301
1178 Ignore-this: eca7871dcc50c1070a0b530711adea27
1179]
1180[Improve pretty-printing of family instances
1181simonpj@microsoft.com**20100915123219
1182 Ignore-this: 25ec6bcc7e8a7f7c303b38ca201db90e
1183 
1184 Fixed Trac #4246
1185]
1186[Fix Trac #4240: -ddump-minimal-imports
1187simonpj@microsoft.com**20100915121937
1188 Ignore-this: ab85057cb829a42ea44a92f7b4af24a3
1189 
1190 See Note [Partial export] for the details.
1191 I also fixed one egregious bug that was just
1192 waiting to bite: we were using loadSysInterface
1193 instead of loadSrcInterface.
1194]
1195[Comments only
1196simonpj@microsoft.com**20100915105707
1197 Ignore-this: ab3a5f16f8260b7b8570e748bf97998a
1198]
1199[implement setThreadAffinity on Windows (#1741)
1200Simon Marlow <marlowsd@gmail.com>**20100914155844
1201 Ignore-this: a14c7b4ef812007042342d0a25478f0b
1202]
1203[COFF: cope with new debug sections in gcc 4.x (fixes ghciprog004)
1204Simon Marlow <marlowsd@gmail.com>**20100914153026
1205 Ignore-this: f340e40a2b0390836bc61bba144a04ed
1206 Also updated the object file parser to properly handle the overflow
1207 case for section names longer than 8 chars.
1208]
1209[eliminate clutter from make output
1210Simon Marlow <marlowsd@gmail.com>**20100915105712
1211 Ignore-this: bfa4480dd239dda2a02ac391b6a9219c
1212]
1213[rts_isProfiled should be a visible API (fixes T2615(dyn))
1214Simon Marlow <marlowsd@gmail.com>**20100915083941
1215 Ignore-this: b8ac09bb9d1a929bf45c6122f8485561
1216]
1217[Fix the "lost due to fragmentation" calculation
1218Simon Marlow <marlowsd@gmail.com>**20100914145945
1219 Ignore-this: cdffcc9f3061c3a33da5171be111fc43
1220 It was counting the space used by block descriptors as "lost"
1221]
1222[fix +RTS -S output: use peak_mblocks_allocated, now that mblocks can be freed
1223Simon Marlow <marlowsd@gmail.com>**20100914135030
1224 Ignore-this: 65d21e5f86d3ab6ab4d6c255180b6968
1225]
1226[Fix egregious bug in deeplyInstantiate
1227simonpj@microsoft.com**20100915070325
1228 Ignore-this: 22ede973038877af2673339aaf5de6cf
1229 
1230 This resulted in an infinite loop in applyTypeToArgs, in syb
1231]
1232[Improve HsSyn pretty printing
1233simonpj@microsoft.com**20100915070255
1234 Ignore-this: 7c8e2d86a482453c7e69e22bc31cb03f
1235]
1236[Remove (most of) the FiniteMap wrapper
1237Ian Lynagh <igloo@earth.li>**20100914201703
1238 We still have
1239     insertList, insertListWith, deleteList
1240 which aren't in Data.Map, and
1241     foldRightWithKey
1242 which works around the fold(r)WithKey addition and deprecation.
1243]
1244[Improve ASSERT
1245simonpj@microsoft.com**20100914113900
1246 Ignore-this: dbc0363be5924f543316e77f7d18dd77
1247]
1248[Comment on what an "enumeration" type is
1249simonpj@microsoft.com**20100914113850
1250 Ignore-this: c09c8591e3140f305d55fbf945adbf95
1251]
1252[Make absent-arg wrappers work for unlifted types (fix Trac #4306)
1253simonpj@microsoft.com**20100914113827
1254 Ignore-this: 1945e56779329e8b79780403710aba98
1255 
1256 Previously we were simply passing arguments of unlifted
1257 type to a wrapper, even if they were absent, which was
1258 stupid.
1259 
1260 See Note [Absent error Id] in WwLib.
1261]
1262[Comments only
1263simonpj@microsoft.com**20100914113641
1264 Ignore-this: 3191ce856c9b5d9700cedc9b149b8097
1265]
1266[Move error-ids to MkCore (from PrelRules)
1267simonpj@microsoft.com**20100914113635
1268 Ignore-this: c3d820db62ba6139dd7c96bf97e51bb5
1269 
1270 and adjust imports accordingly
1271]
1272[More wibbles to deriving error messages
1273simonpj@microsoft.com**20100914113523
1274 Ignore-this: bd2df662644961138fa209aec843a2aa
1275]
1276[Fix getThreadCPUTime()
1277Simon Marlow <marlowsd@gmail.com>**20100913153838
1278 Ignore-this: 950e048a5724086534b74c609c7d5ed
1279 ever since the patch "Check with sysconf _POSIX_THREAD_CPUTIME", it
1280 has been returning incorrect results, because the sysconf variable to
1281 check should have been _SC_THREAD_CPUTIME, not _POSIX_THREAD_CPUTIME.
1282]
1283[filter out the gcc-lib directory from the rts package's library-dirs
1284Simon Marlow <marlowsd@gmail.com>**20100913101259
1285 Ignore-this: 46dc1dccbfee8a65f9243e125eee117f
1286 fixes problems when building with GHC 6.10 on Windows
1287]
1288[Don't include GC time in heap profiles (#4225)
1289Simon Marlow <marlowsd@gmail.com>**20100913133852
1290 Ignore-this: 68ac48b004b311384b5996c6b33ba5cc
1291]
1292[Use clock_gettime (if available) to measure the process CPU time
1293Simon Marlow <marlowsd@gmail.com>**20100913133818
1294 Ignore-this: 8c9300df9b929bfc1db4713c9b6065b3
1295 This is much more accurate than getrusage, which was giving misleading
1296 results when trying to time very quick operations like a minor GC.
1297]
1298[make stg_arg_bitmaps public, and available via the GHCi linker (#3672)
1299Simon Marlow <marlowsd@gmail.com>**20100913105235
1300 Ignore-this: e18efd0bd77c521e5530fb59e93b5a42
1301]
1302[fix typo
1303Simon Marlow <marlowsd@gmail.com>**20100913105100
1304 Ignore-this: 6049eea21208864203b2d79db2edd143
1305]
1306[Update release notes and docs with LLVM info.
1307David Terei <davidterei@gmail.com>**20100914072135
1308 Ignore-this: 5b3d0e5c9d5da98ed6ae9c2e8e1f6f30
1309]
1310[Remove defaultExtensionFlags
1311Ian Lynagh <igloo@earth.li>**20100913165949
1312 The default should do into languageExtensions instead
1313]
1314[Improve crash message
1315simonpj@microsoft.com**20100913170407
1316 Ignore-this: 5c26a9979f18be8cd12cea823c9f4b5a
1317]
1318[Fix Trac #4302, plus a little refactoring
1319simonpj@microsoft.com**20100913170355
1320 Ignore-this: cf6886b587aa0e8d723362183625d946
1321]
1322[Fix build with 6.10
1323Ian Lynagh <igloo@earth.li>**20100913160048]
1324[Haddock fixes
1325simonpj@microsoft.com**20100913120510
1326 Ignore-this: f3157d6969f10d4cbd593000a477138b
1327]
1328[Remove two old junk files
1329simonpj@microsoft.com**20100913103426
1330 Ignore-this: ed7af5ef1b9592178909a8d876345302
1331]
1332[Super-monster patch implementing the new typechecker -- at last
1333simonpj@microsoft.com**20100913095048
1334 Ignore-this: 14d14a1e4d7a414f5ae8d9d89d1c6a4b
1335 
1336 This major patch implements the new OutsideIn constraint solving
1337 algorithm in the typecheker, following our JFP paper "Modular type
1338 inference with local assumptions". 
1339 
1340 Done with major help from Dimitrios Vytiniotis and Brent Yorgey.
1341 
1342]
1343[Fix simplifier statistics
1344simonpj@microsoft.com**20100909085441
1345 Ignore-this: 48e383655aafc912dea15c4d94382863
1346]
1347[Trace output
1348simonpj@microsoft.com**20100908170056
1349 Ignore-this: 4b67fa4b310fbf0a16b852686d2d3294
1350]
1351[Better debug output
1352simonpj@microsoft.com**20100908170047
1353 Ignore-this: 410cef00616dda7c0c162f65216e8ca3
1354]
1355[Add Outputable instance for OccEncl
1356simonpj@microsoft.com**20100908150510
1357 Ignore-this: 6362ef9028287d84f070eaf8963c1bfc
1358]
1359[Better simplifier counting
1360simonpj@microsoft.com**20100907214840
1361 Ignore-this: 9d4722703f8f47447e86a28c8c50e0ea
1362]
1363[Put liftStringName into the known-key names
1364simonpj@microsoft.com**20100906112415
1365 Ignore-this: 287064d14ff484da1a6dea6924bc9235
1366]
1367[Deprecate NoRelaxedPolyRec
1368simonpj@microsoft.com**20100903234519
1369 Ignore-this: 607217e77f6bc1b91bf57dfd8dd2b967
1370]
1371[Buglet in Core Lint
1372simonpj@microsoft.com**20100903234457
1373 Ignore-this: 277535d51b396d3b4b0265a0939c2d4
1374]
1375[Give seqId the right type
1376simonpj@microsoft.com**20100903093556
1377 Ignore-this: d1fc7a73dea160614a8d4ddc930f99cd
1378]
1379[Remove dead code
1380simonpj@microsoft.com**20100903093548
1381 Ignore-this: 92cc3f7651445aa349ee7f114d3ec758
1382]
1383[Comments and layout
1384simonpj@microsoft.com**20100903093502
1385 Ignore-this: 9987d1409e654992c1cb1be35cb87728
1386]
1387[Remove checkFreeness; no longer needed
1388simonpj@microsoft.com**20100902233227
1389 Ignore-this: c96a12ac9794290aa30402317d88c095
1390]
1391[Assert
1392simonpj@microsoft.com**20100902073642
1393 Ignore-this: 4be1ab2f6096665ae5ec7fdd1f025a67
1394]
1395[Add aserts
1396simonpj@microsoft.com**20100902073211
1397 Ignore-this: e1409441217fd070c5a7f9ee4cca99ab
1398]
1399[Wibbles
1400simonpj@microsoft.com**20100831113540
1401 Ignore-this: 903811ab493a7b560a62eb86fcf3ee25
1402]
1403[Wibble to allow phantom types in Enum
1404simonpj@microsoft.com**20100825112711
1405 Ignore-this: fdef1c50d92b4a3d46bbe4cbfd8a83ea
1406]
1407[Add HsCoreTy to HsType
1408simonpj@microsoft.com**20100824141845
1409 Ignore-this: 4ca742b099f9cc90af3167f1012dbba6
1410 
1411 The main thing here is to allow us to provide type
1412 signatures for 'deriving' bindings without pain.
1413]
1414[Comments
1415simonpj@microsoft.com**20100823223654
1416 Ignore-this: dd412a55839430c436902d8699d6900b
1417]
1418[Wibbles to error message
1419simonpj@microsoft.com**20100823163308
1420 Ignore-this: 4d6cd8e613762dca8135c2e3b09264ec
1421]
1422[Correct type signatures
1423simonpj@microsoft.com**20100823153045
1424 Ignore-this: 42942309221a443258246098f9c0a13b
1425]
1426[Add missing signatures
1427simonpj@microsoft.com**20100823112413
1428 Ignore-this: 8ee1ce40456306de469938c02df4fed5
1429]
1430[Add type signatures in "deriving" bindings
1431simonpj@microsoft.com**20100820234230
1432 Ignore-this: 4726b28968cf65ec16cb65b7e0e7303e
1433]
1434[Minor
1435dimitris@microsoft.com**20100820131021]
1436[Be a bit less aggressive in mark-many inside a cast
1437simonpj@microsoft.com**20100819104804
1438 Ignore-this: 3fd48fe7647ec7a58c2032cd86ca4d4f
1439]
1440[Wibble
1441simonpj@microsoft.com**20100818185738
1442 Ignore-this: d5c939311377c0d0c7244aa339193315
1443]
1444[Pretty printing change
1445simonpj@microsoft.com**20100818065436
1446 Ignore-this: 4f7e70976dbe52f95effb3e634dfef5d
1447]
1448[Remember to zonk FlatSkols!
1449simonpj@microsoft.com**20100811143555
1450 Ignore-this: 84f7f9dbda97f561a918c69308ddef9a
1451]
1452[De-polymorphise
1453simonpj@microsoft.com**20100730151217
1454 Ignore-this: a9304487b983e517a9083fd697f77576
1455]
1456[Work around missing type signature in Happy
1457simonpj@microsoft.com**20100730122405
1458 Ignore-this: 7f241a655d93c5ad7763a7ffe8db0c7a
1459 
1460 Happy generates
1461 
1462       notHappyAtAll = error "Blah"
1463 
1464 without a type signature, and currently the new
1465 typechecker doesn't generalise it.  This patch
1466 says "no monomorphism restriction" which makes it
1467 generalise again.
1468 
1469 Better would be to add a type sig to Happy's template
1470]
1471[Add two local type signatures
1472simonpj@microsoft.com**20100729152611
1473 Ignore-this: afa99bcc515469aa0990d44d8c18a9e6
1474]
1475[Second test from Simon's laptop
1476simonpj@microsoft.com**20100729091703
1477 Ignore-this: 4dc64cadae314a5a1b05cc5326918a83
1478]
1479[Test commit from Simon's laptop
1480simonpj@microsoft.com**20100729091344
1481 Ignore-this: 109eff835cc19e9f93799d12f09b0ba7
1482]
1483[Add OutsideIn flag
1484simonpj@microsoft.com**20100728075525
1485 Ignore-this: 69c2f5c3a15fa653f6da80598aa8d74d
1486]
1487[Layout only
1488simonpj@microsoft.com**20100727141539
1489 Ignore-this: 1a58a8fe80ba8bced18ae81a2efb9495
1490]
1491[Improvement to SimplUtils.mkLam
1492simonpj@microsoft.com**20100727131659
1493 Ignore-this: 739beaefa79baa7e0ebeb5b2b6d1ea91
1494]
1495[Give the correct kind to unsafeCoerce#
1496simonpj@microsoft.com**20100727131538
1497 Ignore-this: 6b787de3b398c6d7a61fa04fccd15fd6
1498]
1499[Suppress warnings about recursive INLINE in output of desugarer
1500simonpj@microsoft.com**20100727094549
1501 Ignore-this: a361f7238c0fcba526d46326722c42e
1502]
1503[Rename CorePrep.tryEtaReduce to tryEtaReducePrep
1504simonpj@microsoft.com**20100726231253
1505 Ignore-this: 4375ddace205745244ba224ae012252
1506 
1507 This avoids the name clash with the similar but
1508 not identical CoreUtils.tryEtaReduce
1509]
1510[Add a trace message
1511simonpj@microsoft.com**20100719211111
1512 Ignore-this: b5daebe46e50c8cf28cc693f84bbf099
1513]
1514[Don't use RelaxedPolyRec in the compiler; it's built in now
1515simonpj@microsoft.com**20100719170441
1516 Ignore-this: a2e4489cdf63478e46282a421ee7aec3
1517]
1518[Remove duplicated #defines for FreeBSD
1519Matthias Kilian <kili@outback.escape.de>**20100912181518
1520 Ignore-this: d16214fef8635c7c9ef4edec4e8e7896
1521]
1522[Don't fail with absolute silence
1523Matthias Kilian <kili@outback.escape.de>**20100912150506
1524 Ignore-this: 479e2321f39b263fa2d9f80491e5e9f7
1525]
1526[Add a release note: "-dynload wrapper" removed
1527Ian Lynagh <igloo@earth.li>**20100911195809]
1528[put back the conversion of warn-lazy-unlifted-bindings into an error until 7.2
1529Ian Lynagh <igloo@earth.li>**20100911193434
1530 I think we'll currently still have too many people with old versions of
1531 alex/happy to want to make this an error now.
1532]
1533[6.14 -> 7.0
1534Ian Lynagh <igloo@earth.li>**20100911192837]
1535[Add a couple more release notes
1536Ian Lynagh <igloo@earth.li>**20100911162059]
1537[Document -dsuppress-module-prefixes
1538Ian Lynagh <igloo@earth.li>**20100911162005]
1539[Enable -fregs-graph with -O2; fixes #2790
1540Ian Lynagh <igloo@earth.li>**20100910191301]
1541[Remove unused code
1542Ian Lynagh <igloo@earth.li>**20100909170207]
1543[Fix warnings
1544Ian Lynagh <igloo@earth.li>**20100909154348]
1545[Fix warnings
1546Ian Lynagh <igloo@earth.li>**20100909150957]
1547[Remove context completion
1548lykahb@gmail.com**20100901160153
1549 Ignore-this: dc61b259dcb7063f0c76f56788b5d2af
1550 Now completion suggests to remove only modules added to context before.
1551]
1552[avoid Foreign.unsafePerformIO
1553Ross Paterson <ross@soi.city.ac.uk>**20100909125531
1554 Ignore-this: 5cabeae4cffec8fc17ef7c0cabbea22a
1555]
1556[updates to the release notes
1557Simon Marlow <marlowsd@gmail.com>**20100909111450
1558 Ignore-this: a4d25ad8815c305b7e0f21fd4f6ee37b
1559]
1560[newAlignedPinnedByteArray#: avoid allocating an extra word sometimes
1561Simon Marlow <marlowsd@gmail.com>**20100909110805
1562 Ignore-this: 996a3c0460068ab2835b4920905b3e75
1563]
1564[Finish breaking up vectoriser utils
1565benl@ouroborus.net**20100909061311
1566 Ignore-this: 217fe1d58a3e8bb13200bcb81353a416
1567]
1568[Move VectType module to Vectorise tree
1569benl@ouroborus.net**20100909042451
1570 Ignore-this: 5af8cf394d4835911259ca3ffb6774c5
1571]
1572[Sort all the PADict/PData/PRDict/PRepr stuff into their own modules
1573benl@ouroborus.net**20100909035147
1574 Ignore-this: 53436329773347cad793adbd83e90a9e
1575]
1576[Break out Repr and PADict stuff for vectorisation of ADTs to their own modules
1577benl@ouroborus.net**20100909025759
1578 Ignore-this: d2b7d2f79332eda13416449742f7cf1c
1579]
1580[Break out conversion functions to own module
1581benl@ouroborus.net**20100909023332
1582 Ignore-this: 613f2666b6ca7f2f8876fcc1e4a59593
1583]
1584[Comments and formatting only
1585benl@ouroborus.net**20100909022117
1586 Ignore-this: c8e30139d730669e5db44f0ef491a588
1587]
1588[Remove "-dynload wrapper"; fixes trac #4275
1589Ian Lynagh <igloo@earth.li>**20100908213251]
1590[Don't set visibility on Windows
1591Ian Lynagh <igloo@earth.li>**20100905122442
1592 With gcc 4.5.0-1, using visibility hidden gives:
1593     error: visibility attribute not supported in this configuration; ignored
1594]
1595[Fix warnings on Windows
1596Ian Lynagh <igloo@earth.li>**20100905111201
1597 Ignore-this: c5cce63bb1e0c7a27271bed78d25fbc5
1598]
1599[Fix gcc wrapper for new mingw binaries
1600Ian Lynagh <igloo@earth.li>**20100905001807
1601 Ignore-this: f6acc8c911055ffce632bac138ccc939
1602]
1603[Don't pass our gcc options to stage0 ghc's gcc; they may not be suitable
1604Ian Lynagh <igloo@earth.li>**20100905103129]
1605[Update intree-mingw creation
1606Ian Lynagh <igloo@earth.li>**20100904225559]
1607[Update commands to build in-tree mingw
1608Ian Lynagh <igloo@earth.li>**20100904215112]
1609[Break out hoisting utils into their own module
1610benl@ouroborus.net**20100908074102
1611 Ignore-this: e3ba4ed0252a2def1ed88a9e14c58fea
1612]
1613[Break out closure utils into own module
1614benl@ouroborus.net**20100908072040
1615 Ignore-this: 216172b046ff101cf31a1753667a5383
1616]
1617[Move VectVar module to Vectorise tree
1618benl@ouroborus.net**20100908065904
1619 Ignore-this: 1fba5333d29927dba4275381e1a7f315
1620]
1621[Break out vectorisation of expressions into own module
1622benl@ouroborus.net**20100908065128
1623 Ignore-this: 6a952b80fb024b5291f166477eb1976
1624]
1625[Break out TyCon classifier into own module
1626benl@ouroborus.net**20100908063111
1627 Ignore-this: da754c4ef6960b4e152ea1bf8c04ab6f
1628]
1629[Break out vectorisation of TyConDecls into own module
1630benl@ouroborus.net**20100908052004
1631 Ignore-this: c0ab4fb2a05ca396efe348b384db1ebf
1632]
1633[Break out type vectorisation into own module
1634benl@ouroborus.net**20100907110311
1635 Ignore-this: 67bd70a21d16468daf68dd3ec1ff7d62
1636]
1637[Tidy up the ArchHasAdjustorSupport definition
1638Ian Lynagh <igloo@earth.li>**20100904144234]
1639[ppc: switch handling of 'foreign import wrapper' (FIW) to libffi
1640Sergei Trofimovich <slyfox@community.haskell.org>**20100829192859
1641 Ignore-this: 662ea926681ebea0759e2a04a38e82b7
1642 
1643 Joseph Jezak reported darcs-2.4.4 SIGSEGV in interactive mode in ghc-6.12.3.
1644 So I've concluded ppc also has rotten native adjustor. I don't have hardware
1645 to verify the patch (ticket #3516 should help to test it), but I think it will
1646 help (as similar patch helped for ia64 and ppc64).
1647]
1648[Binary no longer has unusable UNPACK pragmas, so no need to turn of -Werror
1649Ian Lynagh <igloo@earth.li>**20100904133339]
1650[Don't haddock packages that we aren't going to install
1651Ian Lynagh <igloo@earth.li>**20100903231921]
1652[Give haddock per-package source entity paths; fixes #3810
1653Ian Lynagh <igloo@earth.li>**20100903221335]
1654[update for containers-0.4
1655Simon Marlow <marlowsd@gmail.com>**20100903105131
1656 Ignore-this: 556eac0e4926c9b8af6b66d7b069302c
1657]
1658[Fix for nursery resizing: the first block's back pointer should be NULL
1659Simon Marlow <marlowsd@gmail.com>**20100827102818
1660 Ignore-this: fb68938e3f1e291e3c9e5e8047f9dcd2
1661 I'm not sure if this could lead to a crash or not, but it upsets +RTS -DS
1662 Might be related to #4265
1663]
1664[Add some -no-user-package-conf flags
1665Ian Lynagh <igloo@earth.li>**20100902224726
1666 Stops user-installed packages breaking the build
1667]
1668[Fix warnings: Remove unused imports
1669Ian Lynagh <igloo@earth.li>**20100902204342]
1670[Finish breaking up VectBuiltIn and VectMonad, and add comments
1671benl@ouroborus.net**20100831100724
1672 Ignore-this: 65604f3d22d03433abc12f10be40050d
1673]
1674[Fix warnings
1675benl@ouroborus.net**20100830083746
1676 Ignore-this: 2a0e000985f694582a6f9a9261ff2739
1677]
1678[Break up vectoriser builtins module
1679benl@ouroborus.net**20100830070900
1680 Ignore-this: b86bd36a7875abdcf16763902ba2e637
1681]
1682[Move VectCore to Vectorise tree
1683benl@ouroborus.net**20100830053415
1684 Ignore-this: d5763ca6424285b39a58c7792f4a84a1
1685]
1686[Split out vectoriser environments into own module
1687benl@ouroborus.net**20100830050252
1688 Ignore-this: 5319111c74831394d2c29b9aedf5a766
1689]
1690[Comments and formatting to vectoriser, and split out varish stuff into own module
1691benl@ouroborus.net**20100830042722
1692 Ignore-this: d3f0c98ed8124dd0fca9a2ccea3e15fd
1693]
1694[Fix warnings
1695benl@ouroborus.net**20100830040340
1696 Ignore-this: d6cfad803ad4617e7fdaa62e4a895282
1697]
1698[Fix warning about multiply exported name
1699benl@ouroborus.net**20100830035243
1700 Ignore-this: 27ce2c1d22d9f99929d16a426343044e
1701]
1702[Vectorisation of method types
1703benl@ouroborus.net**20100830032941
1704 Ignore-this: 75614571d5c246a4906edb3b39ab1e0b
1705]
1706[Comments and formatting to vectoriser
1707benl@ouroborus.net**20100830032516
1708 Ignore-this: fe665b77108501c7960d858be3290761
1709]
1710[Implement -dsuppress-module-prefixes
1711benl@ouroborus.net**20100830032428
1712 Ignore-this: 2bb8bad9c60ef9044132bba118010687
1713]
1714[Whitespace only
1715benl@ouroborus.net**20100527045629
1716 Ignore-this: 4c160dfa77727e659817b6af9c84684a
1717]
1718[Disambiguate a function name
1719Ian Lynagh <igloo@earth.li>**20100828225827]
1720[users_guide.xml is now generated
1721Ian Lynagh <igloo@earth.li>**20100828225751]
1722[Pass more -pgm flags in the ghc wrapper; fixes #3863
1723Ian Lynagh <igloo@earth.li>**20100827204537]
1724[Add a new-IO manager release note
1725Ian Lynagh <igloo@earth.li>**20100827171616]
1726[Merge a duplicate release note
1727Ian Lynagh <igloo@earth.li>**20100827171511]
1728[Typo, spotted by Johan Tibell
1729Ian Lynagh <igloo@earth.li>**20100827153914]
1730[First pass at 6.14.1 release notes
1731Ian Lynagh <igloo@earth.li>**20100826220811]
1732[Fix typo
1733Ian Lynagh <igloo@earth.li>**20100824201330]
1734[FIX BUILD: add rts_isProfiled to the symbol table
1735Simon Marlow <marlowsd@gmail.com>**20100826094319
1736 Ignore-this: 9536ddb0a94721c8dec03a2a981cfa83
1737]
1738[Fix the DPH package cleaning/profiled mess even more (the build was broken)
1739Simon Marlow <marlowsd@gmail.com>**20100826084436
1740 Ignore-this: 49d7e4db2fb53b856c213c74c8969d82
1741]
1742[Remove the debugging memory allocator - valgrind does a better job
1743Simon Marlow <marlowsd@gmail.com>**20100824113537
1744 Ignore-this: a3731a83dc18b0fd0de49452e695a7ca
1745 
1746 I got fed up with the constant bogus output from the debugging memory
1747 allocator in RtsUtils.c.  One problem is that we allocate memory in
1748 constructors that then isn't tracked, because the debugging allocator
1749 hasn't been initialised yet.
1750 
1751 The bigger problem is that for a given piece of leaking memory it's
1752 impossible to find out where it was allocated; however Valgrind gives
1753 output like this:
1754 
1755 ==6967== 8 bytes in 1 blocks are still reachable in loss record 1 of 7
1756 ==6967==    at 0x4C284A8: malloc (vg_replace_malloc.c:236)
1757 ==6967==    by 0x4C28522: realloc (vg_replace_malloc.c:525)
1758 ==6967==    by 0x6745E9: stgReallocBytes (RtsUtils.c:213)
1759 ==6967==    by 0x68D812: setHeapAlloced (MBlock.c:91)
1760 ==6967==    by 0x68D8E2: markHeapAlloced (MBlock.c:116)
1761 ==6967==    by 0x68DB56: getMBlocks (MBlock.c:240)
1762 ==6967==    by 0x684F55: alloc_mega_group (BlockAlloc.c:305)
1763 ==6967==    by 0x6850C8: allocGroup (BlockAlloc.c:358)
1764 ==6967==    by 0x69484F: allocNursery (Storage.c:390)
1765 ==6967==    by 0x694ABD: allocNurseries (Storage.c:436)
1766 ==6967==    by 0x6944F2: initStorage (Storage.c:217)
1767 ==6967==    by 0x673E3C: hs_init (RtsStartup.c:160)
1768 
1769 which tells us exactly what the leaking bit of memory is.  So I don't
1770 think we need our own debugging allocator.
1771]
1772[free the entries in the thread label table on exit
1773Simon Marlow <marlowsd@gmail.com>**20100824112606
1774 Ignore-this: c9d577c06548cda80791e590e40d35b3
1775]
1776[Panic in the right way
1777simonpj@microsoft.com**20100825091614
1778 Ignore-this: e6ea4f6dfd2aea088828ea7a945ddd5f
1779]
1780[Fix the DPH/profiled make thing (again)
1781simonpj@microsoft.com**20100825091602
1782 Ignore-this: bc58fa48034ac40cf7be4170958ea29e
1783]
1784[Don't test for gcc flags before we've found gcc
1785Ian Lynagh <igloo@earth.li>**20100824131401]
1786[Change how the dblatex/lndir problem is worked around
1787Ian Lynagh <igloo@earth.li>**20100824130938
1788 Hack: dblatex normalises the name of the main input file using
1789 os.path.realpath, which means that if we're in a linked build tree,
1790 it find the real source files rather than the symlinks in our link
1791 tree. This is fine for the static sources, but it means it can't
1792 find the generated sources.
1793 
1794 We therefore also generate the main input file, so that it really
1795 is in the link tree, and thus dblatex can find everything.
1796]
1797[Clean the generated userguide sources
1798Ian Lynagh <igloo@earth.li>**20100824105827
1799 Ignore-this: 39b4f9702c688c053ed3273b20969597
1800]
1801[DPH should not even be built if GhcProfiled
1802simonpj@microsoft.com**20100823133439
1803 Ignore-this: 62acbf83de5b70ff6d27ab38ce9218ae
1804 
1805 It's not just when cleaning!
1806]
1807[The templateHaskellOk check should only run in stage2
1808simonpj@microsoft.com**20100823133353
1809 Ignore-this: f6dc9292923a1ca201953c5f58c0af3c
1810 
1811 Because rtsIsProfiled is only available in stage2
1812]
1813[Add a couple of missing tests for EAGER_BLACKHOLE
1814Simon Marlow <marlowsd@gmail.com>**20100823104654
1815 Ignore-this: 70c981b86370b0c7564b29b057650897
1816 This was leading to looping and excessive allocation, when the
1817 computation should have just blocked on the black hole. 
1818 
1819 Reported by Christian Höner zu Siederdissen <choener@tbi.univie.ac.at>
1820 on glasgow-haskell-users.
1821]
1822[Don't check for swept blocks in -DS.
1823Marco Túlio Gontijo e Silva <marcot@marcot.eti.br>**20100718225526
1824 Ignore-this: ad5dcf3c247bc19fbef5122c1142f3b2
1825 
1826 The checkHeap function assumed the allocated part of the block contained only
1827 alive objects and slops.  This was not true for blocks that are collected using
1828 mark sweep.  The code in this patch skip the test for this kind of blocks.
1829]
1830[Fix "darcs get"
1831Ian Lynagh <igloo@earth.li>**20100822183542]
1832[Add "darcs-all upstreampull"
1833Ian Lynagh <igloo@earth.li>**20100822163419
1834 This pulls from the upstream repos, for those packages which have
1835 upstreams
1836]
1837[Generate the bit in the user guide where we say what -fglasgow-exts does
1838Ian Lynagh <igloo@earth.li>**20100822155514
1839 Stops the docs going out of sync with the code.
1840]
1841[Factor out the packages file parsing in darcs-all
1842Ian Lynagh <igloo@earth.li>**20100822154813]
1843[Document --supported-extensions
1844Ian Lynagh <igloo@earth.li>**20100822134530]
1845[fix extraction of command stack of arguments of arrow "forms" (fixes #4236)
1846Ross Paterson <ross@soi.city.ac.uk>**20100822090022
1847 Ignore-this: a93db04ec4f20540642a19cdc67d1666
1848 
1849 The command stack was being extracted (by unscramble) with the outermost
1850 type first, contrary to the comment on the function.
1851]
1852[minor fix to comment
1853Ross Paterson <ross@soi.city.ac.uk>**20100822085838
1854 Ignore-this: 8d203ba2600eaf4cf21b043dcfa96cdc
1855]
1856[Add the RTS library path to the library search path
1857Ian Lynagh <igloo@earth.li>**20100820155523
1858 In case the RTS is being explicitly linked in. For #3807.
1859]
1860[Remove some duplication of C flags
1861Ian Lynagh <igloo@earth.li>**20100819233743
1862 We now use the CONF_CC_OPTS_STAGEn C flags in machdepCCOpts, rather than
1863 repeating them there.
1864]
1865[Set -fno-stack-protector in CONF_CC_OPTS_STAGE* rathre than extra-gcc-opts
1866Ian Lynagh <igloo@earth.li>**20100819233031
1867 The latter is only used when compiling .hc files, whereas we need it for
1868 .c files too.
1869]
1870[Give clearer errors for bad input in the packages file; suggested by pejo
1871Ian Lynagh <igloo@earth.li>**20100819232420]
1872[Set -march=i686 on OS X x86 in the configure variables
1873Ian Lynagh <igloo@earth.li>**20100819230939
1874 We used to set it only in machdepCCOpts, so this is more consistent
1875]
1876[Give each stage its own Config.hs
1877Ian Lynagh <igloo@earth.li>**20100819224709
1878 This also means the file is generated in a dist directory, not a
1879 source directory.
1880]
1881[Fix cleaning when GhcProfiled = YES
1882Ian Lynagh <igloo@earth.li>**20100819131346
1883 We need to include the DPH cleaning rules, even though we don't build DPH
1884 when GhcProfiled = YES.
1885]
1886[stgReallocBytes(DEBUG): don't fail when the ptr passed in is NULL
1887Simon Marlow <marlowsd@gmail.com>**20100817150836
1888 Ignore-this: 4b5063e65e01399f64a33f0d0555ff38
1889]
1890[Use make-command in rules/bindist.mk
1891Ian Lynagh <igloo@earth.li>**20100818191243
1892 Rather than it having its own specialised version
1893]
1894[Use make-command when installing packages
1895Ian Lynagh <igloo@earth.li>**20100818190600]
1896[Add _DATA_FILES to package-data.mk files
1897Ian Lynagh <igloo@earth.li>**20100818185801]
1898[Add a "make-command" utility Makefile function
1899Ian Lynagh <igloo@earth.li>**20100818183055]
1900[LLVM: Nicer format for lack of shared lib warning
1901David Terei <davidterei@gmail.com>**20100817145207
1902 Ignore-this: 753d45762601d87761614937a1bb6716
1903]
1904[fix FP_CHECK_ALIGNMENT for autoconf 2.66 (fixes #4252)
1905Ross Paterson <ross@soi.city.ac.uk>**20100816142442
1906 Ignore-this: cd784b8888d32b3b2cc2cc0969ec40f
1907 
1908 Recent versions of AS_LITERAL_IF don't like *'s.  Fix from
1909 
1910 http://blog.gmane.org/gmane.comp.sysutils.autoconf.general/month=20100701
1911]
1912[Refactor the command-line argument parsing (again)
1913simonpj@microsoft.com**20100816074453
1914 Ignore-this: 26dc9e37a88660a887a2e316ed7a9803
1915 
1916 This change allows the client of CmdLineParser a bit more flexibility,
1917 by giving him an arbitrary computation (not just a deprecation
1918 message) for each flag. 
1919 
1920 There are several clients, so there are lots of boilerplate changes.
1921 
1922 Immediate motivation: if RTS is not profiled, we want to make
1923 Template Haskell illegal.  That wasn't with the old setup.
1924]
1925[Add upstream repo to the packages file
1926Ian Lynagh <igloo@earth.li>**20100815154741]
1927[Make the "tag" column of the packages file always present
1928Ian Lynagh <igloo@earth.li>**20100815151657
1929 It makes the parsing simpler if we always have the same number of columns
1930]
1931[Disable object splitting on OSX; works around #4013
1932Ian Lynagh <igloo@earth.li>**20100815134759]
1933[Return memory to the OS; trac #698
1934Ian Lynagh <igloo@earth.li>**20100813170402]
1935[Reduce the xargs -s value we use on Windows
1936Ian Lynagh <igloo@earth.li>**20100812223721
1937 With 30000 I was getting:
1938     xargs: value for -s option should be < 28153
1939]
1940[LLVM: Enable shared lib support on Linux x64
1941David Terei <davidterei@gmail.com>**20100813191534
1942 Ignore-this: 642ed37af38e5f17d419bf4f09332671
1943]
1944[Re-do the arity calculation mechanism again (fix Trac #3959)
1945simonpj@microsoft.com**20100813161151
1946 Ignore-this: d4a2aa48150b503b20c25351a79decfb
1947 
1948 After rumination, yet again, on the subject of arity calculation,
1949 I have redone what an ArityType is (it's purely internal to the
1950 CoreArity module), and documented it better.  The result should
1951 fix #3959, and I hope the related #3961, #3983.
1952 
1953 There is lots of new documentation: in particular
1954  * Note [ArityType] 
1955    describes the new datatype for arity info
1956 
1957  * Note [State hack and bottoming functions]
1958    says how bottoming functions are dealt with, particularly
1959    covering catch# and Trac #3959
1960 
1961 I also found I had to be careful not to eta-expand single-method
1962 class constructors; see Note [Newtype classes and eta expansion].
1963 I think this part could be done better, but it works ok.
1964]
1965[Comments only
1966simonpj@microsoft.com**20100813161019
1967 Ignore-this: baf68300d8bc630bf0b7ab27647b33a0
1968]
1969[Modify FloatOut to fix Trac #4237
1970simonpj@microsoft.com**20100813163120
1971 Ignore-this: ffc8d00d4b7f0a8a785fcef312900413
1972 
1973 The problem was that a strict binding was getting floated
1974 out into a letrec. This only happened when profiling was
1975 on.  It exposed a fragility in the floating strategy.  This
1976 patch makes it more robust.  See
1977       Note [Avoiding unnecessary floating]
1978]
1979[Fix egregious bug in SetLevels.notWorthFloating
1980simonpj@microsoft.com**20100813161429
1981 Ignore-this: d22865f48d417e6a6b732de3dfba378f
1982 
1983 This bug just led to stupid code, which would
1984 later be optimised away, but better not to generate
1985 stupid code in the first place.
1986]
1987[Delete GhcLibProfiled
1988simonpj@microsoft.com**20100813140152
1989 Ignore-this: 2e1a3f677308be726bd022f45e2fd856
1990 
1991 Simon M and I looked at this, and we think GhcLibProfiled is
1992 (a) not needed (b) confusing.
1993 
1994 Ian should review.
1995 
1996 Really, if GhcProfiled is on we should also
1997 check that 'p' is in the GhcLibWays
1998]
1999[Do not build DPH when GhcProfiled (fixes #4172)
2000simonpj@microsoft.com**20100813140021
2001 Ignore-this: 9e20181643b456e841f845ae0cab0a9a
2002 
2003 Reason: DPH uses Template Haskell and TH doesn't work
2004 in a profiled compiler
2005]
2006[Fix Trac #4220
2007simonpj@microsoft.com**20100812131319
2008 Ignore-this: 33141cfd81627592150a9e5973411ff8
2009 
2010 For deriving Functor, Foldable, Traversable with empty
2011 data cons I just generate a null equation
2012    f _ = error "urk"
2013 
2014 There are probably more lurking (eg Enum) but this will do for now.
2015]
2016[Improve the Specialiser, fixing Trac #4203
2017simonpj@microsoft.com**20100812131133
2018 Ignore-this: 482afbf75165e24a80527a6e52080c07
2019 
2020 Simply fixing #4203 is a tiny fix: in case alterantives we should
2021 do dumpUDs *including* the case binder. 
2022 
2023 But I realised that we can do better and wasted far too much time
2024 implementing the idea.  It's described in
2025     Note [Floating dictionaries out of cases]
2026]
2027[Comments
2028simonpj@microsoft.com**20100812101456
2029 Ignore-this: 6362fe887d25688c12ef2c3cf5554ce4
2030]
2031[Comments only
2032simonpj@microsoft.com**20100812101439
2033 Ignore-this: 7ed2f5fc08811cbe9958c2309a9ed1fa
2034]
2035[Fix bug in linting of shadowed case-alternative binders
2036simonpj@microsoft.com**20100812101413
2037 Ignore-this: 9212a5e2c03421749f5935b3944ecf53
2038]
2039[Comments and spacing only
2040simonpj@microsoft.com**20100812101347
2041 Ignore-this: ed59a7dae7decb24470709dc1c118dbb
2042]
2043[Add more info to more parse error messages (#3811)
2044Ian Lynagh <igloo@earth.li>**20100809233108]
2045[Run finalizers *after* updating the stable pointer table (#4221)
2046Simon Marlow <marlowsd@gmail.com>**20100810133739
2047 Ignore-this: b0462f80dd64eac71e599d8a9f6dd665
2048 Silly bug really, we were running the C finalizers while the StablePtr
2049 table was still in a partially-updated state during GC, but finalizers
2050 are allowed to call freeStablePtr() (via hs_free_fun_ptr(), for
2051 example), and chaos ensues.
2052]
2053[Do the dependency-omitting for 'make 1' in a slightly different way
2054Simon Marlow <marlowsd@gmail.com>**20100810093446
2055 Ignore-this: af15edd3a1492cbd93111316b57e02e4
2056 
2057 I encountered a couple of things that broke after Ian's previous
2058 patch: one was my nightly build scripts that use 'make stage=2' at the
2059 top level, and the other is 'make fast' in libraries/base, which uses
2060 'stage=0' to avoid building any compilers.
2061 
2062 So my version of this patch is more direct: it just turns off the
2063 appropriate dependencies using a variable set by 'make 1', 'make 2',
2064 etc.
2065]
2066[Integrate new I/O manager, with signal support
2067Johan Tibell <johan.tibell@gmail.com>**20100724102355
2068 Ignore-this: eb092857a2a1b0ca966649caffe7ac2b
2069]
2070[Add DoAndIfThenElse support
2071Ian Lynagh <igloo@earth.li>**20100808194625]
2072[Make another parse error more informative
2073Ian Lynagh <igloo@earth.li>**20100808193340]
2074[Make a parse error say what it is failing to parse; part of #3811
2075Ian Lynagh <igloo@earth.li>**20100808155732]
2076[Send ghc progress output to stdout; fixes #3636
2077Ian Lynagh <igloo@earth.li>**20100808142542]
2078[Fix the HsColour test in the build system
2079Ian Lynagh <igloo@earth.li>**20100805155319
2080 Ignore-this: ba2752b04801a253e891b31e1914485d
2081]
2082[Fix the -lm configure test; fixes #4155
2083Ian Lynagh <igloo@earth.li>**20100805142508
2084 Ignore-this: 358b8b1074d2d22fb8d362ea6d8b80d6
2085]
2086[Don't restrict filenames in line pragmas to printable characters; fixes #4207
2087Ian Lynagh <igloo@earth.li>**20100805135011
2088 Ignore-this: e3d32312127165e40e6eaa919193d60b
2089 "printable" is ASCII-only, whereas in other locales we can get things like
2090 # 1 "<línea-de-orden>"
2091]
2092[Ensure extension flags are flattened in the Cmm phase
2093Ian Lynagh <igloo@earth.li>**20100805133614
2094 If we start with a .cmmcpp file then they don't get flattened in
2095 the CmmCpp phase, as we don't run that phase.
2096]
2097[Add "cmmcpp" as a Haskellish source suffix
2098Ian Lynagh <igloo@earth.li>**20100805132555]
2099[On amd64/OSX we don't need to be given memory in the first 31bits
2100Ian Lynagh <igloo@earth.li>**20100805120600
2101 Ignore-this: 42eb64e25ad4b66ae022884305e0297b
2102 as PIC is always on
2103]
2104[NCG: Don't worry about trying to re-freeze missing coalescences
2105benl@ouroborus.net**20100702053319
2106 Ignore-this: ea05cbee19b6c5c410db41292cbb64b0
2107]
2108[Make -rtsopts more flexible
2109Ian Lynagh <igloo@earth.li>**20100805011137
2110 The default is a new "some" state, which allows only known-safe flags
2111 that we want on by default. Currently this is only "--info".
2112]
2113[Test for (fd < 0) before trying to FD_SET it
2114Ian Lynagh <igloo@earth.li>**20100804173636]
2115[Remove "On by default" comments in DynFlags
2116Ian Lynagh <igloo@earth.li>**20100802110803
2117 Ignore-this: 2a51055277b5ce9f0e98e1438b212027
2118 These make less sense now we support multiple languges. The
2119 "languageExtensions" function gives the defaults.
2120]
2121[Fix build: Add newline to end of ghc-pkg/Main.hs
2122Ian Lynagh <igloo@earth.li>**20100801183206]
2123[Add a versions haddock binary for Windows
2124Ian Lynagh <igloo@earth.li>**20100801180917]
2125[ghc-pkg: don't fail, if a file is already removed
2126ich@christoph-bauer.net**20100725162606
2127 Ignore-this: 5501d6812c31f4da525c7fb24f6dcaed
2128]
2129[Remove push-all from file list in boot script (push-all no longer exists)
2130Ian Lynagh <igloo@earth.li>**20100801121841
2131 Ignore-this: eec130f06610d8728a57626682860a1a
2132]
2133[Add error checking to boot-pkgs script
2134Ian Lynagh <igloo@earth.li>**20100801121432
2135 Ignore-this: 8afd6663db443c774bad45d75bbfe950
2136]
2137[Add more error checking to the boot script
2138Ian Lynagh <igloo@earth.li>**20100801113628]
2139[Remove libHSrtsmain.a before creating it
2140Ian Lynagh <igloo@earth.li>**20100801005432
2141 Otherwise it isn't updated properly if rts/Main.c changes
2142]
2143[Expose the functions haddock needs even when haddock is disabled; #3558
2144Ian Lynagh <igloo@earth.li>**20100731115506]
2145[Always haddock by default
2146Ian Lynagh <igloo@earth.li>**20100730235001
2147 Revert this patch:
2148     Matthias Kilian <kili@outback.escape.de>**20090920181319
2149     Don't build haddock if HADDOC_DOCS = NO, and disable HADDOC_DOCS
2150         if GhcWithInterpreter = NO
2151     Haddock uses TcRnDriver.tcRnGetInfo, which is only available if
2152     GHCI is built. Set HADDOC_DOCS to NO if GhcWithInterpreter is NO,
2153     and disable the haddock build if HADDOC_DOCS = NO.
2154]
2155[Add a debugTrace for the phases that we run
2156Ian Lynagh <igloo@earth.li>**20100729201503]
2157[* Add StringPrimL as a constructor for Template Haskell (Trac #4168)
2158simonpj@microsoft.com**20100730131922
2159 Ignore-this: 520d0a0a14b499b299e8b2be8d148ff0
2160   
2161 There are already constructors for IntPrim, FloatPrim etc,
2162 so this makes it more uniform.
2163   
2164 There's a corresponding patch for the TH library
2165]
2166[Add thread affinity support for FreeBSD
2167Gabor Pali <pgj@FreeBSD.org>**20100720001409
2168 Ignore-this: 6c117b8219bfb45445089e82ee470410
2169 - Implement missing functions for setting thread affinity and getting real
2170   number of processors.
2171 - It is available starting from 7.1-RELEASE, which includes a native support
2172   for managing CPU sets.
2173 - Add __BSD_VISIBLE, since it is required for certain types to be visible in
2174   addition to POSIX & C99.
2175]
2176[Disable symbol visibility pragmas for FreeBSD
2177Ian Lynagh <igloo@earth.li>**20100729012507
2178 Do not use GCC pragmas for controlling visibility, because it causes
2179 "undefined reference" errors at link-time.  The true reasons are
2180 unknown, however FreeBSD 8.x includes GCC 4.2.1 in the base system,
2181 which might be buggy.
2182]
2183[Fix numeric escape sequences parsing
2184Anton Nikishaev <anton.nik@gmail.com>**20100721194208
2185 Ignore-this: dd71935b1866b5624f7975c45ad519a1
2186 This fixes trac bug #1344
2187]
2188[Explicitly give the right path to perl when making the OS X installer; #4183
2189Ian Lynagh <igloo@earth.li>**20100728163030]
2190[Set -fno-stack-protector in extra-gcc-opts; fixes #4206
2191Ian Lynagh <igloo@earth.li>**20100728161957
2192 We were using it only when building the RTS, and only on certain
2193 platforms. However, some versions of OS X need the flag, while others
2194 don't support it, so we now test for it properly.
2195]
2196[Make PersistentLinkerState fields strict; fixes #4208
2197Ian Lynagh <igloo@earth.li>**20100727201911
2198 Ignore-this: fc5cfba48cd16624f6bb15a7a03a3b4
2199 We modify fields a lot, so we retain the old value if they aren't forced.
2200]
2201[Don't rebuild dependency files unnecessarily when doing "make 1" etc
2202Ian Lynagh <igloo@earth.li>**20100726211512
2203 Ignore-this: d91a729e5113aa964cc67768e92e57ef
2204]
2205[LLVM: If user specifies optlo, don't use '-O' levels
2206David Terei <davidterei@gmail.com>**20100726105650
2207 Ignore-this: e05e103b09d1de937540ffad7983f88e
2208]
2209[Flatten flags for ghci's :show
2210Ian Lynagh <igloo@earth.li>**20100725135320]
2211[Add support for Haskell98 and Haskell2010 "languages"
2212Ian Lynagh <igloo@earth.li>**20100724230121]
2213[Rename "language" varibles etc to "extension", and add --supported-extensions
2214Ian Lynagh <igloo@earth.li>**20100724223624]
2215[Separate language option handling into 2 phases
2216Ian Lynagh <igloo@earth.li>**20100724212013
2217 We now first collect the option instructions (from the commandline,
2218 from pragmas in source files, etc), and then later flatten them into
2219 the list of enabled options. This will enable us to use different
2220 standards (H98, H2010, etc) as a base upon which to apply the
2221 instructions, when we don't know what the base will be when we start
2222 collecting instructions.
2223]
2224[Separate the language flags from the other DynFlag's
2225Ian Lynagh <igloo@earth.li>**20100724133103
2226 Ignore-this: 47bb8d42e621e47016b66c7472bd6cb5
2227]
2228[Set stage-specific CC/LD opts in the bindist configure.ac
2229Ian Lynagh <igloo@earth.li>**20100724113748
2230 Ignore-this: f06926d185a35ddd05490ca4a257e992
2231]
2232[Use different CC/LD options for different stages
2233Ian Lynagh <igloo@earth.li>**20100723223059]
2234[Add some error belchs to the linker, when we find bad magic numbers
2235Ian Lynagh <igloo@earth.li>**20100723200822]
2236[Add some more linker debugging prints
2237Ian Lynagh <igloo@earth.li>**20100723180237]
2238[When (un)loading an object fails, say which object in teh panic
2239Ian Lynagh <igloo@earth.li>**20100723162649]
2240[Add a release note: GHCi import syntax
2241Ian Lynagh <igloo@earth.li>**20100721193647]
2242[Deprecate NewQualifiedOperators extension (rejected by H')
2243Ian Lynagh <igloo@earth.li>**20100719150909
2244 Ignore-this: 6e7e3bedc5360c5975f73497b3e6cba5
2245]
2246[LLVM: Allow optlc and optlo to override default params for these systools
2247David Terei <davidterei@gmail.com>**20100722181631
2248 Ignore-this: e60af7941996f7170fb3bfb02a002082
2249]
2250[LLVM: Code and speed improvement to dominateAllocs pass.
2251David Terei <davidterei@gmail.com>**20100721143654
2252 Ignore-this: 9fb7058c8a2afc005521298c7b8d0036
2253]
2254[Comments only
2255simonpj@microsoft.com**20100721144257
2256 Ignore-this: b3091ddcd1df271eb85fe90978ab7adc
2257]
2258[Fix inlining for default methods
2259simonpj@microsoft.com**20100721144248
2260 Ignore-this: 61a11a8f741f775000c6318aae4b3191
2261 
2262 This was discombobulated by a patch a week ago;
2263 now fixed, quite straightforwardly.  See
2264 Note [Default methods and instances]
2265]
2266[Allow reification of existentials and GADTs
2267simonpj@microsoft.com**20100721090437
2268 Ignore-this: 20f1ccd336cc25aff4d4d67a9ac2211a
2269 
2270 It turns out that TH.Syntax is rich enough to express even GADTs,
2271 provided we express them in equality-predicate form.  So for
2272 example
2273 
2274   data T a where
2275      MkT1 :: a -> T [a]
2276      MkT2 :: T Int
2277 
2278 will appear in TH syntax like this
2279 
2280   data T a = forall b. (a ~ [b]) => MkT1 b
2281            | (a ~ Int) => MkT2
2282 
2283 While I was at it I also improved the reification of types,
2284 so that we use TH.TupleT and TH.ListT when we can.
2285]
2286[add numSparks# primop (#4167)
2287Simon Marlow <marlowsd@gmail.com>**20100720153746
2288 Ignore-this: f3f925e7de28f3f895213aefbdbe0b0f
2289]
2290[LLVM: Decrease max opt level used under OSX to avoid bug
2291David Terei <davidterei@gmail.com>**20100720160938
2292 Ignore-this: 34b0b3550f00b27b00ad92f8232745e5
2293 
2294 Currently, many programs compiled with GHC at -O2 and LLVM
2295 set to -O3 will segfault (only under OSX). Until this issue
2296 is fixed I have simply 'solved' the segfault by lowering
2297 the max opt level for LLVM used to -O2 under OSX.
2298 
2299 All these recent changes to OSX should mean its finally as
2300 stable as Linux and Windows.
2301]
2302[LLVM: Fix OSX to work again with TNTC disabled.
2303David Terei <davidterei@gmail.com>**20100720160845
2304 Ignore-this: 8dc98139cfa536b2a64aa364d040b581
2305]
2306[LLVM: Fix printing of local vars so LLVM works with -fnew-codegen
2307David Terei <davidterei@gmail.com>**20100720160302
2308 Ignore-this: d883c433dfaed67921a8c5360e1f9f6a
2309]
2310[Use a separate mutex to protect all_tasks, avoiding a lock-order-reversal
2311Simon Marlow <marlowsd@gmail.com>**20100716150832
2312 Ignore-this: ffbdb4ee502e0f724d57acb9bfbe9d92
2313 In GHC 6.12.x I found a rare deadlock caused by this
2314 lock-order-reversal:
2315 
2316 AQ cap->lock
2317   startWorkerTask
2318     newTask
2319       AQ sched_mutex
2320 
2321 scheduleCheckBlackHoles
2322   AQ sched_mutex
2323    unblockOne_
2324     wakeupThreadOnCapabilty
2325       AQ cap->lock
2326 
2327 so sched_mutex and cap->lock are taken in a different order in two
2328 places.
2329 
2330 This doesn't happen in the HEAD because we don't have
2331 scheduleCheckBlackHoles, but I thought it would be prudent to make
2332 this less likely to happen in the future by using a different mutex in
2333 newTask.  We can clearly see that the all_tasks mutex cannot be
2334 involved in a deadlock, becasue we never call anything else while
2335 holding it.
2336]
2337['make fast' in a package does not build any compilers
2338Simon Marlow <marlowsd@gmail.com>**20100715125904
2339 Ignore-this: f27e70faf3944831dad16e89a4e273da
2340]
2341[LLVM: Fix up botched last commit
2342David Terei <davidterei@gmail.com>**20100719104823
2343 Ignore-this: a32e0f6a38cb9e02527eb8ca69b3eb59
2344]
2345[LLVM: Fix warning introduce in last commit.
2346David Terei <davidterei@gmail.com>**20100719103411
2347 Ignore-this: e9c92a9402aff50d60ab26e6ad441bfc
2348]
2349[LLVM: Use mangler to fix up stack alignment issues on OSX
2350David Terei <davidterei@gmail.com>**20100718231000
2351 Ignore-this: 9f6e8cb855269cb3a5ac1a23480d0e71
2352]
2353[Fix #4195 (isGadtSyntaxTyCon returns opposite result)
2354illissius@gmail.com**20100715134134
2355 Ignore-this: a90403f893030432b5c15d743647f350
2356]
2357[Update to time 1.2.0.3
2358Ian Lynagh <igloo@earth.li>**20100717181810
2359 Ignore-this: 1ccb4801a73f399e6718ce556543ede1
2360]
2361[Reorder RTS --info output
2362Ian Lynagh <igloo@earth.li>**20100717162356]
2363[Fix unreg prof build: Define CCS_SYSTEM in stg/MiscClosures.h
2364Ian Lynagh <igloo@earth.li>**20100717142832
2365 Ignore-this: 9675f3f51b6dac40483155344e7f45b6
2366]
2367[Make mkDerivedConstants as a stage 1 program
2368Ian Lynagh <igloo@earth.li>**20100717000827
2369 Ignore-this: 5357403461b209b8606f1d33defb51cf
2370 This way it gets the defines for the right platform when cross-compiling
2371]
2372[Don't generate Haskell dependencies if we don't have any Haskell sources
2373Ian Lynagh <igloo@earth.li>**20100717000800
2374 Ignore-this: 454abd0358f535b7e789327125c9206c
2375]
2376[Link programs that have no Haskell objects with gcc rather than ghc
2377Ian Lynagh <igloo@earth.li>**20100716235303
2378 Ignore-this: f65588b69675edea616cc434e769b0a4
2379]
2380[Use gcc to build C programs for stages >= 1
2381Ian Lynagh <igloo@earth.li>**20100716223703
2382 Ignore-this: 9f843a4e17285cda582117504707f9e7
2383]
2384[Add platform info to "ghc --info" output
2385Ian Lynagh <igloo@earth.li>**20100716141953]
2386[Tidy up Config.hs generation
2387Ian Lynagh <igloo@earth.li>**20100716140630]
2388[Fix HC porting test in makefiles
2389Ian Lynagh <igloo@earth.li>**20100716010808
2390 Ignore-this: 6052c1dd022a6108ab2236a299ee1d84
2391 Now that we are trying to support cross compilation, we can't use
2392     "$(TARGETPLATFORM)" != "$(HOSTPLATFORM)"
2393 as a test for HC-porting.
2394]
2395[Change a BUILD var to a HOST var
2396Ian Lynagh <igloo@earth.li>**20100716002558]
2397[Remove an unnecessary #include
2398Ian Lynagh <igloo@earth.li>**20100715233930
2399 Ignore-this: dcede249de6be7e3c9305c9279c2ca07
2400]
2401[Split up some make commands, so that errors aren't overlooked
2402Ian Lynagh <igloo@earth.li>**20100715152237
2403 Ignore-this: fb69b0a25d9ca71dae5e75d38db675cd
2404 When we ask make to run "a | b", if a fails then the pipeline might
2405 still exit successfuly.
2406]
2407[Remove an unnecessary #include
2408Ian Lynagh <igloo@earth.li>**20100715143000
2409 Ignore-this: 4e098cac5dda2dd595ca0a0f5121853c
2410]
2411[Simplify some more CPP __GLASGOW_HASKELL__ tests
2412Ian Lynagh <igloo@earth.li>**20100715142500]
2413[Remove some code only used with GHC 6.11.*
2414Ian Lynagh <igloo@earth.li>**20100715141720]
2415[__GLASGOW_HASKELL__ >= 609 is now always true
2416Ian Lynagh <igloo@earth.li>**20100715141544]
2417[Correct the values in ghc_boot_platform.h
2418Ian Lynagh <igloo@earth.li>**20100714223717
2419 Ignore-this: 4c99116f7ac73fadbd6d16807f57a693
2420]
2421[Change some TARGET checks to HOST checks
2422Ian Lynagh <igloo@earth.li>**20100714184715]
2423[LLVM: Add inline assembly to binding.
2424David Terei <davidterei@gmail.com>**20100714152530
2425 Ignore-this: 72a7b5460c128ed511e8901e5889fe2b
2426]
2427[LLVM: Fix mistype in last commit which broke TNTC under win/linux.
2428David Terei <davidterei@gmail.com>**20100714153339
2429 Ignore-this: 302d7957e3dded80368ebade5312ab35
2430]
2431[Remove unnecessary #include
2432Ian Lynagh <igloo@earth.li>**20100713153704
2433 Ignore-this: c37d3127b1dc68f59270c07173994c28
2434]
2435[Change some TARGET tests to HOST tests in the RTS
2436Ian Lynagh <igloo@earth.li>**20100713141034
2437 Which was being used seemed to be random
2438]
2439[LLVM: Add in new LLVM mangler for implementing TNTC on OSX
2440David Terei <davidterei@gmail.com>**20100713183243
2441 Ignore-this: 394fb74d7f9657d8b454bd0148d24bf7
2442]
2443[Refactor where an error message is generated
2444simonpj@microsoft.com**20100713115733
2445 Ignore-this: f94467856238586fcbbe48537141cf78
2446]
2447[Comments only
2448simonpj@microsoft.com**20100713115703
2449 Ignore-this: 5815442c4e69b9ec331b34242a596253
2450]
2451[Comments on data type families
2452simonpj@microsoft.com**20100713115640
2453 Ignore-this: 90a333bb7f7d64a49fb7dd180d893f6b
2454]
2455[Fix Trac #T4136: take care with nullary symbol constructors
2456simonpj@microsoft.com**20100707135945
2457 Ignore-this: 2a717a24fefcd593ea41c23dad351db0
2458 
2459 When a nullary constructor is a symbol eg (:=:) we need
2460 to take care.  Annoying.
2461]
2462[Fix Trac #4127 (and hence #4173)
2463simonpj@microsoft.com**20100707123125
2464 Ignore-this: 98bb6d0f7182b59f8c93596c61f9785d
2465 
2466 The change involves a little refactoring, so that the default
2467 method Ids are brought into scope earlier, before the value
2468 declarations are compiled.  (Since a value decl may contain
2469 an instance decl in a quote.)
2470 
2471 See Note [Default method Ids and Template Haskell] in
2472 TcTyClsDcls.
2473]
2474[Fix second bug in Trac #4127
2475simonpj@microsoft.com**20100701140124
2476 Ignore-this: c8d1cc27364fe9ee5a52acb1ecb5cdd9
2477 
2478 This bug concerned the awkward shadowing we do for
2479 Template Haskell declaration brackets.  Lots of
2480 comments in
2481 
2482   Note [Top-level Names in Template Haskell decl quotes]
2483]
2484[ia64: switch handling of 'foreign import wrapper' (FIW) to libffi
2485Sergei Trofimovich <slyfox@community.haskell.org>**20100709213922
2486 Ignore-this: fd07687e0089aebabf62de85d2be693
2487 
2488 I tried to build darcs-2.4.4 with ghc-6.12.3 and got coredumps when darcs is used
2489 in interactive mode. I tried test from ticket #3516 and found out FIW code is broken.
2490 Instead of fixing it I just switched to libffi. Result built successfully, passed
2491 'foreign import wrapper' test from ticket #3516 and builds working darcs.
2492]
2493[* storage manager: preserve upper address bits on 64bit machines (thanks to zygoloid)
2494Sergei Trofimovich <slyfox@community.haskell.org>**20100709115917
2495 Ignore-this: 9f1958a19992091ddc2761c389ade940
2496 
2497 Patch does not touch amd64 as it's address lengts is 48 bits at most, so amd64 is unaffected.
2498 
2499 the issue: during ia64 ghc bootstrap (both 6.10.4 and 6.12.3) I
2500 got the failure on stage2 phase:
2501     "inplace/bin/ghc-stage2"   -H32m -O -H64m -O0 -w ...
2502     ghc-stage2: internal error: evacuate: strange closure type 15
2503         (GHC version 6.12.3 for ia64_unknown_linux)
2504         Please report this as a GHC bug:  http://www.haskell.org/ghc/reportabug
2505     make[1]: *** [libraries/dph/dph-base/dist-install/build/Data/Array/Parallel/Base/Hyperstrict.o] Aborted
2506 
2507 gdb backtrace (break on 'barf'):
2508 Breakpoint 1 at 0x400000000469ec31: file rts/RtsMessages.c, line 39.
2509 (gdb) run -B/var/tmp/portage/dev-lang/ghc-6.12.3/work/ghc-6.12.3/inplace/bin --info
2510 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
2511 [Thread debugging using libthread_db enabled]
2512 
2513 Breakpoint 1, barf (s=0x40000000047915b0 "evacuate: strange closure type %d") at rts/RtsMessages.c:39
2514 39        va_start(ap,s);
2515 (gdb) bt
2516 #0  barf (s=0x40000000047915b0 "evacuate: strange closure type %d") at rts/RtsMessages.c:39
2517 #1  0x400000000474a1e0 in evacuate (p=0x6000000000147958) at rts/sm/Evac.c:756
2518 #2  0x40000000046d68c0 in scavenge_srt (srt=0x6000000000147958, srt_bitmap=7) at rts/sm/Scav.c:348
2519 ...
2520 
2521 > 16:52:53 < zygoloid> slyfox: i'm no ghc expert but it looks like HEAP_ALLOCED_GC(q)
2522 >                      is returning true for a FUN_STATIC closure
2523 > 17:18:43 < zygoloid> try: p HEAP_ALLOCED_miss((unsigned long)(*p) >> 20, *p)
2524 > 17:19:12 < slyfox> (gdb) p HEAP_ALLOCED_miss((unsigned long)(*p) >> 20, *p)
2525 > 17:19:12 < slyfox> $1 = 0
2526 > 17:19:40 < zygoloid> i /think/ that means the mblock_cache is broken
2527 > 17:22:45 < zygoloid> i can't help further. however i am suspicious that you seem to have pointers with similar-looking low 33
2528 >                      bits and different high 4 bits, and it looks like such pointers get put into the same bucket in
2529 >                      mblock_cache.
2530 ...
2531 > 17:36:16 < zygoloid> slyfox: try changing the definition of MbcCacheLine to StgWord64, see if that helps
2532 > 17:36:31 < zygoloid> that's in includes/rts/storage/MBlock.h
2533 And it helped!
2534]
2535[Fixing link failure of compiler on ia64 ('-Wl,' prefixed value passed directly to ld)
2536Sergei Trofimovich <slyfox@community.haskell.org>**20100708180943
2537 Ignore-this: ced99785e1f870ee97e5bec658e2504f
2538 
2539     /usr/bin/ld -Wl,--relax -r -o dist-stage1/build/HSghc-6.10.4.o \
2540                                   dist-stage1/build/BasicTypes.o dist-stage1/build/DataCon.o ...
2541     /usr/bin/ld: unrecognized option '-Wl,--relax'
2542 
2543 If we just drop '-Wl,' part it will not help as '-r' and '--relax' are incompatible.
2544 
2545 Looks like '-Wl,--relax' was skipped by earlier binutils' ld as unknown option.
2546 Removing ia64 specific path.
2547]
2548[LLVM: Allow getelementptr to use LlvmVar for indexes.
2549David Terei <davidterei@gmail.com>**20100712152529
2550 Ignore-this: 9e158d9b89a86bca8abf11d082328278
2551]
2552[Move all the warning workarounds to one place
2553Ian Lynagh <igloo@earth.li>**20100710161723]
2554[xhtml is now warning-free
2555Ian Lynagh <igloo@earth.li>**20100710144635]
2556[Move a bit of build system code
2557Ian Lynagh <igloo@earth.li>**20100709224534]
2558[adapt to the new async exceptions API
2559Simon Marlow <marlowsd@gmail.com>**20100709125238
2560 Ignore-this: 55d845e40b9daed3575c1479d8dda1d5
2561]
2562[quiet some new spewage
2563Simon Marlow <marlowsd@gmail.com>**20100709091521
2564 Ignore-this: de7f91976bbc9789e6fd7091f05c25c0
2565]
2566[New asynchronous exception control API (ghc parts)
2567Simon Marlow <marlowsd@gmail.com>**20100708144851
2568 Ignore-this: 56320c5fc61ae3602d586609387aae22
2569 
2570 As discussed on the libraries/haskell-cafe mailing lists
2571   http://www.haskell.org/pipermail/libraries/2010-April/013420.html
2572 
2573 This is a replacement for block/unblock in the asychronous exceptions
2574 API to fix a problem whereby a function could unblock asynchronous
2575 exceptions even if called within a blocked context.
2576 
2577 The new terminology is "mask" rather than "block" (to avoid confusion
2578 due to overloaded meanings of the latter).
2579 
2580 In GHC, we changed the names of some primops:
2581 
2582   blockAsyncExceptions#   -> maskAsyncExceptions#
2583   unblockAsyncExceptions# -> unmaskAsyncExceptions#
2584   asyncExceptionsBlocked# -> getMaskingState#
2585 
2586 and added one new primop:
2587 
2588   maskUninterruptible#
2589 
2590 See the accompanying patch to libraries/base for the API changes.
2591]
2592[remove outdated comment
2593Simon Marlow <marlowsd@gmail.com>**20100708100840
2594 Ignore-this: afb2e9f6fe1f1acda51b0cbdf2637176
2595]
2596[remove 'mode: xml' emacs settings (#2208)
2597Simon Marlow <marlowsd@gmail.com>**20100708100817
2598 Ignore-this: 3a8d997fb90e01ca88dc47fb95feeba0
2599]
2600[typo in comment
2601Simon Marlow <marlowsd@gmail.com>**20100616111359
2602 Ignore-this: d3ef9288d6d6b9ab3bacbe09e0d9801c
2603]
2604[Win32 getProcessElapsedTime: use a higher-resolution time source
2605Simon Marlow <marlowsd@gmail.com>**20100708093223
2606 Ignore-this: 821989d4ff7ff2bff40cee71a881521c
2607 QueryPerformanceCounter() on Windows gives much better resolution than
2608 GetSystemTimeAsFileTime().
2609]
2610[alpha: switch handling of 'foreign import wrapper' (FIW) to libffi
2611Sergei Trofimovich <slyfox@community.haskell.org>**20100708065318
2612 Ignore-this: ddee15876737a6aa7f6dabc8ff79ce0d
2613 
2614 I tried to build ghc-6.12.3 and found out FIW part of code
2615 does not compile anymore. It uses absent functions under #ifdef.
2616 Instead of fixing it I just switched to libffi. Result built successfully
2617 and passed 'foreign import wrapper' test I wrote for trac ticket #3516.
2618 
2619 I didn't try to build -HEAD yet, but this patch only removes code, so
2620 it should not make -HEAD worse.
2621]
2622[Reorder the CPP flags so -optP can override the platform defines
2623Ian Lynagh <igloo@earth.li>**20100708203523]
2624[Add docs for DatatypeContexts extension
2625Ian Lynagh <igloo@earth.li>**20100707230907
2626 Ignore-this: 8158f03b35a2d7442a75fe85d6f1b1c7
2627]
2628[Make datatype contexts an extension (on by default) (DatatypeContexts)
2629Ian Lynagh <igloo@earth.li>**20100707212529
2630 Ignore-this: 6885ff510a0060610eeeba65122caef5
2631]
2632[LLVM: Fix various typos in comments
2633David Terei <davidterei@gmail.com>**20100707220448
2634 Ignore-this: 1ba3e722f150492da2f9d485c5795e80
2635]
2636[Handle haddock headers when looking for LANGUAGE/OPTIONS_GHC pragmas
2637Ian Lynagh <igloo@earth.li>**20100707120423
2638 Ignore-this: a75aa67690284a6cee3e62c943d4fd01
2639]
2640[Make pragState call mkPState, rather than duplicating everything
2641Ian Lynagh <igloo@earth.li>**20100706173007
2642 Ignore-this: 61fe24b99dbe7a42efff1a9dd703a75c
2643 This also means that extsBitmap gets set, whereas is was just being set
2644 to 0 before.
2645]
2646[LLVM: Add alias type defenitions to LlvmModule.
2647David Terei <davidterei@gmail.com>**20100707142053
2648 Ignore-this: eee6ad5385563ccf08e664d2634a03f2
2649]
2650[LLVM: Use packed structure type instead of structure type
2651David Terei <davidterei@gmail.com>**20100707120320
2652 Ignore-this: a06e0359d182291b81cae56993ca385e
2653 
2654 The regular structure type adds padding to conform to the platform ABI,
2655 which causes problems with structures storing doubles under windows since
2656 we don't conform to the platform ABI there. So we use packed structures
2657 instead now that don't do any padding.
2658]
2659[Make mkPState and pragState take their arguments in the same order
2660Ian Lynagh <igloo@earth.li>**20100706172611]
2661[Remove an out-of-date comment
2662Ian Lynagh <igloo@earth.li>**20100706172217
2663 Ignore-this: 710ebd7d2dc01c1b0f1e58a5b6f85701
2664]
2665[LLVM: Stop llvm saving stg caller-save regs across C calls
2666David Terei <davidterei@gmail.com>**20100705162629
2667 Ignore-this: 28b4877b31b9358e682e38fc54b88658
2668 
2669 This is already handled by the Cmm code generator so LLVM is simply
2670 duplicating work. LLVM also doesn't know which ones are actually live
2671 so saves them all which causes a fair performance overhead for C calls
2672 on x64. We stop llvm saving them across the call by storing undef to
2673 them just before the call.
2674]
2675[LLVM: Add in literal undefined value to binding
2676David Terei <davidterei@gmail.com>**20100705161544
2677 Ignore-this: 95d8361b11584aaeec44c30e76916470
2678]
2679[LLVM: Add a literal NULL value to binding
2680David Terei <davidterei@gmail.com>**20100705161308
2681 Ignore-this: 9507b4b12c1157498704a9d1e5860f12
2682 
2683 Patch from Erik de Castro Lopo <erikd@mega-nerd.com>.
2684]
2685[refactor import declaration support (#2362)
2686Simon Marlow <marlowsd@gmail.com>**20100705104557
2687 Ignore-this: ee034ac377078a7e92bfada1907c86a0
2688]
2689[Disable dynamic linking optimisations on OS X
2690Simon Marlow <marlowsd@gmail.com>**20100705103014
2691 Ignore-this: b04420d3705c51112797758d17b2e40c
2692 To improve performance of the RTS when dynamically linked on x86, I
2693 previously disabled -fPIC for certain critical modules (the GC, and a
2694 few others).  However, build reports suggest that the dynamic linker
2695 on OS X doesn't like this, so I'm disabling this optimsation on that
2696 platform.
2697]
2698[trac #2362 (full import syntax in ghci)
2699amsay@amsay.net**20100625032632
2700 Ignore-this: a9d0859d84956beb74e27b797431bf9c
2701 'import' syntax is seperate from ':module' syntax
2702]
2703[Simplify ghc-pkg's Cabal dependencies
2704Ian Lynagh <igloo@earth.li>**20100704184155
2705 We no longer support building with a compiler that doesn't come with
2706 base 4.
2707]
2708[Use Cabal to configure the dist-install ghc-pkg; fixes trac #4156
2709Ian Lynagh <igloo@earth.li>**20100704132612]
2710[Remove dead code (standalone deriving flag no longer used in parser)
2711Ian Lynagh <igloo@earth.li>**20100701162058]
2712[LLVM: Use the inbounds keyword for getelementptr instructions.
2713David Terei <davidterei@gmail.com>**20100702160511
2714 Ignore-this: 3708e658a4c82b78b1402393f4405541
2715]
2716[threadPaused: fix pointer arithmetic
2717Simon Marlow <marlowsd@gmail.com>**20100701085046
2718 Ignore-this: b78210e5d978f18ffd235f1c78a55a23
2719 Noticed by Henrique Ferreiro <hferreiro@udc.es>, thanks!
2720]
2721[LLVM: Change more operations to use getelementptr
2722David Terei <davidterei@gmail.com>**20100701161856
2723 Ignore-this: fb24eb124e203f50680c6fec3ff9fe7d
2724]
2725[Add the haskell2010 package
2726Simon Marlow <marlowsd@gmail.com>**20100630125532
2727 Ignore-this: e9b011313f283a8ff2fcda7d029a01f
2728]
2729[LLVM: Use getelementptr instruction for a lot of situations
2730David Terei <davidterei@gmail.com>**20100630181157
2731 Ignore-this: 34d314dd8dffad9bdcffdc525261a49d
2732 
2733 LLVM supports creating pointers in two ways, firstly through
2734 pointer arithmetic (by casting between pointers and ints)
2735 and secondly using the getelementptr instruction. The second way
2736 is preferable as it gives LLVM more information to work with.
2737 
2738 This patch changes a lot of pointer related code from the first
2739 method to the getelementptr method.
2740]
2741[remove out of date comments; point to the wiki
2742Simon Marlow <marlowsd@gmail.com>**20100625100313
2743 Ignore-this: 95f363a373534b9471b1818102ec592d
2744]
2745[NCG: allocatableRegs is only giving us 8 SSE regs to allocate to
2746benl@ouroborus.net**20100629054321
2747 Ignore-this: b3e0fa0b4ce988a0258dc12261989ee0
2748]
2749[LLVM: Use intrinsic functions for pow, sqrt, sin, cos
2750David Terei <davidterei@gmail.com>**20100628182949
2751 Ignore-this: 98a0befaca3fe2b36d710d8ff9f062c4
2752 
2753 Instead of calling the C library for these Cmm functions
2754 we use intrinsic functions provided by llvm. LLVM will
2755 then either create a compile time constant if possible, or
2756 use a cpu instruction or as a last resort call the C
2757 library.
2758]
2759[LLVM: Fix test '2047' under linux-x64
2760David Terei <davidterei@gmail.com>**20100628165256
2761 Ignore-this: 41735d4f431a430db636621650ccd71e
2762]
2763[LLVM: Fix test 'ffi005' under linux-x64
2764David Terei <davidterei@gmail.com>**20100628155355
2765 Ignore-this: 841f3142c63cc898ac4c3f89698a837e
2766]
2767[LLVM: Update to use new fp ops introduced in 2.7
2768David Terei <davidterei@gmail.com>**20100628144037
2769 Ignore-this: 5dd2e5964e3c039d297ed586841e706b
2770]
2771[Add noalias and nocapture attributes to pointer stg registers
2772David Terei <davidterei@gmail.com>**20100628115120
2773 Ignore-this: 492a1e723cb3a62498d240d7de92dd7
2774 
2775 At the moment this gives a very slight performance boost of around 1 - 2%.
2776 Future changes to the generated code though so that pointers are kept as
2777 pointers more often instead of being cast to integer types straight away
2778 should hopefully improve the benefit this brings.
2779 
2780]
2781[during shutdown, only free the heap if we waited for foreign calls to exit
2782Simon Marlow <marlowsd@gmail.com>**20100628090536
2783 Ignore-this: d545384a4f641d701455d08ef1217479
2784]
2785[Fix typo in -ddump-pass's document.
2786shelarcy <shelarcy@gmail.com>**20100620070759
2787 Ignore-this: f4f1ddb53f147949e948147d89190c37
2788]
2789[Add #undefs for posix source symbols when including papi.h
2790dmp@rice.edu**20100624163514
2791 Ignore-this: 8a1cba21b880d12a75a75f7e96882053
2792 
2793 Validation fails when validating with PAPI support (i.e. GhcRtsWithPapi  = YES
2794 in validate.mk).  The problem is that the posix symbols are defined by a header
2795 included from papi.h. Compilation then fails because these symbols are
2796 redefined in PosixSource.h.
2797 
2798 This patch adds an undefine for the posix symbols after including papi.h and
2799 before including PosixSource.h. The #undefines are localized to Papi.c since
2800 that is the only case where they are getting defined twice.
2801]
2802[Use machdepCCOpts in runPhase_MoveBinary; fixes trac #3952
2803Ian Lynagh <igloo@earth.li>**20100625220953]
2804[LLVM: Fix bug with calling tail with empty list
2805David Terei <davidterei@gmail.com>**20100625115729
2806 Ignore-this: 46b4b32c8d92372a2d49794a96fe1613
2807]
2808[Fix warnings
2809benl@ouroborus.net**20100624091339
2810 Ignore-this: 5ba4bbd6abb9c9d1fb8c5d21ab73f218
2811]
2812[NCG: Comments and formatting only
2813benl@ouroborus.net**20100624083121
2814 Ignore-this: 86002e72c30d06bcc876d8c49f4caa5a
2815]
2816[NCG: Do the actual reversing of SCCs
2817benl@ouroborus.net**20100624082717
2818 Ignore-this: 12d2027ea118e751fbb48b27126150ef
2819]
2820[NCG: Fix dumping of graphs in regalloc stats for graph allocator
2821benl@ouroborus.net**20100624082625
2822 Ignore-this: 2b971bc9e0318099a9afb0e0db135730
2823]
2824[NCG: Reverse SCCs after each round in the graph allocator
2825benl@ouroborus.net**20100624082437
2826 Ignore-this: f0152e4039d6f16f7b5a99b286538116
2827]
2828[NCG: Don't actually complain on unreachable code blocks
2829benl@ouroborus.net**20100624081445
2830 Ignore-this: e7335ae6120917cb858c38c7c6da8e24
2831]
2832[NCG: Do explicit check for precondition of computeLiveness
2833benl@ouroborus.net**20100624080747
2834 Ignore-this: e7053c4e5e4c3c746b5ebf016913424a
2835 
2836  computeLiveness requires the SCCs of blocks to be in reverse dependent
2837  order, and if they're not it was silently giving bad liveness info,
2838  yielding a bad allocation.
2839 
2840  Now it complains, loudly.
2841]
2842[NCG: Fix off-by-one error in realRegSqueeze
2843benl@ouroborus.net**20100623095813
2844 Ignore-this: ab0698686d4c250da8e207f734f8252d
2845]
2846[NCG: Handle stripping of liveness info from procs with no blocks (like stg_split_marker)
2847benl@ouroborus.net**20100623091209
2848 Ignore-this: c0319b6cc62ec713afe4eb03790406e3
2849]
2850[NCG: Emit a warning on unreachable code block instead of panicing
2851benl@ouroborus.net**20100623085002
2852 Ignore-this: d20314b79e3c31e764ed4cd97290c696
2853]
2854[NCG: Remember to keep the entry block first when erasing liveness info
2855Ben.Lippmeier@anu.edu.au**20090917104429
2856 Ignore-this: 1b0c1df19d622858d50ffb6a01f2cef0
2857]
2858[NCG: Refactor representation of code with liveness info
2859Ben.Lippmeier@anu.edu.au**20090917090730
2860 Ignore-this: 2aebb3b02ebd92e547c5abad9feb0f0d
2861 
2862  * I've pushed the SPILL and RELOAD instrs down into the
2863    LiveInstr type to make them easier to work with.
2864 
2865  * When the graph allocator does a spill cycle it now just
2866    re-annotates the LiveCmmTops instead of converting them
2867    to NatCmmTops and back.
2868 
2869  * This saves working out the SCCS again, and avoids rewriting
2870    the SPILL and RELOAD meta instructions into real machine
2871    instructions.
2872]
2873[NCG: Add sanity checking to linear allocator
2874Ben.Lippmeier@anu.edu.au**20090917090335
2875 Ignore-this: 5a442be8b5087d04bc8b58dffa9ea080
2876 If there are are unreachable basic blocks in the native code then the
2877 linear allocator might loop. Detect this case and panic instead.
2878]
2879[NCG: Refactor LiveCmmTop to hold a list of SCCs instead of abusing ListGraph
2880Ben.Lippmeier@anu.edu.au**20090917060332
2881 Ignore-this: 3fec8d69ed0f760e53a202f873d5d9cb
2882]
2883[NCG: Allow the liveness map in a LiveInfo to be Nothing
2884Ben.Lippmeier@anu.edu.au**20090917043937
2885 Ignore-this: 5f82422d54d1b0ffc0589eb7e82fb7a4
2886]
2887[NCG: Also show the result of applying coalesings with -ddump-asm-regalloc-stages
2888Ben.Lippmeier.anu.edu.au**20090917034427
2889 Ignore-this: 76bd6d5ca43adb2167cb25832cbaa80b
2890]
2891[Fix panic when running "ghc -H"; trac #3364
2892Ian Lynagh <igloo@earth.li>**20100624234011
2893 The problem is that showing SDoc's looks at the static flags global
2894 variables, but those are panics while we are parsing the static flags.
2895 We work around this by explicitly using a fixed prettyprinter style.
2896]
2897[Allow for stg registers to have pointer type in llvm BE.
2898David Terei <davidterei@gmail.com>**20100621175839
2899 Ignore-this: fc09b1a8314aef0bde945c77af1124fb
2900 
2901 Before all the stg registers were simply a bit type or
2902 floating point type but now they can be declared to have
2903 a pointer type to one of these. This will allow various
2904 optimisations in the future in llvm since the type is
2905 more accurate.
2906]
2907[Add support for parameter attributes to the llvm BE binding
2908David Terei <davidterei@gmail.com>**20100624111744
2909 Ignore-this: 77f3c0c7bf8f81c4a154dc835ae7bcba
2910 
2911 These allow annotations of the code produced by the backend
2912 which should bring some perforamnce gains. At the moment
2913 the attributes aren't being used though.
2914]
2915[Cast some more nats to StgWord to be on the safe side
2916Simon Marlow <marlowsd@gmail.com>**20100624105700
2917 Ignore-this: e6176683856f9872fdeb2358bb065bb8
2918 And add a comment about the dangers of int overflow
2919]
2920[comments only
2921Simon Marlow <marlowsd@gmail.com>**20100624105105
2922 Ignore-this: fc8f762f4c3a5ffca2f8da2bc63ac2a4
2923]
2924[Fix an arithmetic overflow bug causing crashes with multi-GB heaps
2925Simon Marlow <marlowsd@gmail.com>**20100624104654
2926 Ignore-this: 67210755aa098740ff5230347be0fd5d
2927]
2928[Add support for collecting PAPI native events
2929dmp@rice.edu**20100622195953
2930 Ignore-this: 7269f9c4dfb2912a024eb632200fcd1
2931 
2932 This patch extends the PAPI support in the RTS to allow collection of native
2933 events. PAPI can collect data for native events that are exposed by the
2934 hardware beyond the PAPI present events. The native events supported on your
2935 hardware can found by using the papi_native_avail tool.
2936 
2937 The RTS already allows users to specify PAPI preset events from the command
2938 line. This patch extends that support to allow users to specify native events.
2939 The changes needed are:
2940 
2941 1) New option (#) for the RTS PAPI flag for native events. For example, to
2942    collect the native event 0x40000000, use ./a.out +RTS -a#0x40000000 -sstderr
2943 
2944 2) Update the PAPI_FLAGS struct to store whether the user specified event is a
2945    papi preset or a native event
2946 
2947 3) Update init_countable_events function to add the native events after parsing
2948    the event code and decoding the name using PAPI_event_code_to_name
2949 
2950]
2951[Don't warn about unused bindings with parents in .hs-boot files; trac #3449
2952Ian Lynagh <igloo@earth.li>**20100624110351]
2953[fix the home_imps filter to allow for 'import "this" <module>'
2954Simon Marlow <marlowsd@gmail.com>**20100621125535
2955 Ignore-this: da4e605b0513afc32a4e7caa921a2c76
2956 In the PackageImports extension, import "this" means "import from the
2957 current package".
2958]
2959[Use the standard C wrapper code for the ghc-$version.exe wrapper
2960Ian Lynagh <igloo@earth.li>**20100622202859
2961 Ignore-this: 60cd3e6db3afb63e6ba9e2db3b033580
2962]
2963[Don't rely on "-packagefoo" working; use "-package foo" instead
2964Ian Lynagh <igloo@earth.li>**20100622202547]
2965[Remove unnecessary C #includes
2966Ian Lynagh <igloo@earth.li>**20100622172919]
2967[Make the ghci.exe wrapper call the right ghc.exe
2968Ian Lynagh <igloo@earth.li>**20100622172247]
2969[More updates to datalayout description in llvm BE
2970David Terei <davidterei@gmail.com>**20100622165339
2971 Ignore-this: b0c604fe7673b0aa7c7064694d574437
2972]
2973[Remove LlvmAs phase as the llvm opt tool now handles this phase
2974David Terei <davidterei@gmail.com>**20100622144044
2975 Ignore-this: b9fd8f959702b6af014e2fa654bede3
2976 
2977 This phase originally invoked the llvm-as tool that turns a textual
2978 llvm assembly file into a bit code file for the rest of llvm to deal
2979 with. Now the llvm opt tool can do this itself, so we don't need to
2980 use llvm-as anymore.
2981]
2982[Update datalayout info in llvm BE
2983David Terei <davidterei@gmail.com>**20100622123457
2984 Ignore-this: 89b043d211225dcd819f30549afe1840
2985]
2986[Fix handling of float literals in llvm BE
2987David Terei <davidterei@gmail.com>**20100622121642
2988 Ignore-this: a3b5f382ad4b5a426ad4b581664506fa
2989]
2990[Declare some top level globals to be constant when appropriate
2991David Terei <davidterei@gmail.com>**20100621174954
2992 Ignore-this: 44832f65550d4f995d11c01cc1affef5
2993 
2994 This involved removing the old constant handling mechanism
2995 which was fairly hard to use. Now being constant or not is
2996 simply a property of a global variable instead of a separate
2997 type.
2998]
2999[Reduce the number of passes over the cmm in llvm BE
3000David Terei <davidterei@gmail.com>**20100621125220
3001 Ignore-this: cb2f4e54e8d0f982d5087fbeee35c73c
3002]
3003[Fix negate op not working for -0 in llvm backend
3004David Terei <davidterei@gmail.com>**20100621123606
3005 Ignore-this: c5d76e5cffa781fed074137851b1347f
3006]
3007[ROLLBACK: picCCOpts: -dynamic should not entail -optc-fPIC
3008Simon Marlow <marlowsd@gmail.com>**20100621100409
3009 Ignore-this: f2fac7df33d3919199befc59bd455414
3010 and add a comment to explain why it was wrong.  This fixes the dyn
3011 test failures that sprang up recently.
3012]
3013[Check files are really created in libffi
3014Ian Lynagh <igloo@earth.li>**20100620163724
3015 when we think that the libffi build creates them, so they just depend
3016 on the libffi build stamp.
3017]
3018[Improve the missing-import-list warning
3019Ian Lynagh <igloo@earth.li>**20100620124320
3020 Ignore-this: 551e5fdf2dfb56b49d249e0cebaa6115
3021]
3022[Tweak missing-import-list warning
3023Ian Lynagh <igloo@earth.li>**20100620122622
3024 Ignore-this: 360cdf59ae13d66ded181129325506c4
3025]
3026[trac #1789 (warnings for missing import lists)
3027amsay@amsay.net**20100618150649
3028 Ignore-this: b0b0b1e048fbca0817c1e6fade1153fa
3029]
3030[Refix docs for sizeofByteArray#/sizeofMutableByteArray# (#3800)
3031Ian Lynagh <igloo@earth.li>**20100620103749]
3032[Remove some old commented out code
3033Ian Lynagh <igloo@earth.li>**20100620000459]
3034[SET_ARR_HDR's last argument is now a number of bytes, rather than words
3035Ian Lynagh <igloo@earth.li>**20100619235214
3036 This avoids unnecessary work and potential loss of information
3037]
3038[Replace an (incorrect) bytes-to-words calculation with ROUNDUP_BYTES_TO_WDS
3039Ian Lynagh <igloo@earth.li>**20100619234310]
3040[FIX #38000 Store StgArrWords payload size in bytes
3041Antoine Latter <aslatter@gmail.com>**20100101183346
3042 Ignore-this: 7bf3ab4fc080c46311fc10b179361bb6
3043]
3044[Add win32 datalayout support to llvm backend
3045David Terei <davidterei@gmail.com>**20100618131733
3046 Ignore-this: 4b7bffaa8ef38c628ab852c1a6c1c009
3047]
3048[Remove unused 'ddump-opt-llvm' flag
3049David Terei <davidterei@gmail.com>**20100618101237
3050 Ignore-this: f78467496d986897e49d82646ee2907e
3051]
3052[generate "movl lbl(%reg1), %reg2" instructions, better codegen for -fPIC
3053Simon Marlow <marlowsd@gmail.com>**20100618082258
3054 Ignore-this: a25567ebff9f575303ddc8f2deafebbf
3055]
3056[joinToTargets: fix a case of panic "handleComponent cyclic"
3057Simon Marlow <marlowsd@gmail.com>**20100618082147
3058 Ignore-this: 765baeefbb5a41724004acd92405cecc
3059]
3060[comment typo
3061Simon Marlow <marlowsd@gmail.com>**20100618082102
3062 Ignore-this: e495610b7dd5ec30b02938638b56cb7
3063]
3064[Add support of TNTC to llvm backend
3065David Terei <davidterei@gmail.com>**20100618093205
3066 Ignore-this: 2c27d21668374a5b0d5e844882c69439
3067 
3068 We do this through a gnu as feature called subsections,
3069 where you can put data/code into a numbered subsection
3070 and those subsections will be joined together in descending
3071 order by gas at compile time.
3072]
3073[Don't automatically insert a -fvia-C flag in an unregisterised compiler
3074Ian Lynagh <igloo@earth.li>**20100617190901
3075 Ignore-this: eb25a9a338fade9e17c153da7c5f27e9
3076 The default object mode is already HscC, so it's unnecessary, and
3077 -fvia-C generates a deprecated flag warning now.
3078]
3079[In PosixSource.h, conditionally define things based on platform
3080Ian Lynagh <igloo@earth.li>**20100617174912
3081 This may not be ideal, but it should get GHC building on all platforms
3082 again.
3083]
3084[disable -fPIC for the GC for performance reasons
3085Simon Marlow <marlowsd@gmail.com>**20100617140025
3086 Ignore-this: c7c152bbff71ef7891eaee8ff39fc281
3087 see comment for details
3088]
3089[picCCOpts: -dynamic should not entail -optc-fPIC
3090Simon Marlow <marlowsd@gmail.com>**20100617115259
3091 Ignore-this: d71e42bd56e4bd107d2c431b801855e5
3092]
3093[Make getAllocations() visible
3094Simon Marlow <marlowsd@gmail.com>**20100617113259
3095 Ignore-this: 1b7fb38a01358c0acbe8987df07d23f2
3096]
3097[Fix the symbol visibility pragmas
3098Simon Marlow <marlowsd@gmail.com>**20100617105758
3099 Ignore-this: 76552500865473a1dbebbc1cc2def9f0
3100]
3101[pick up changes to $(GhcStage1HcOpts) without re-configuring the ghc package
3102Simon Marlow <marlowsd@gmail.com>**20100616124718
3103 Ignore-this: afb56d5560c813051285607fefb15493
3104]
3105[Fix bindisttest Makefile
3106Ian Lynagh <igloo@earth.li>**20100616205611
3107 Ignore-this: 39cd352152422f378572fc3859c5a377
3108]
3109[Remove some more unused make variables
3110Ian Lynagh <igloo@earth.li>**20100616180519]
3111[Convert some more variable names to FOO_CMD, for consistency
3112Ian Lynagh <igloo@earth.li>**20100616175916]
3113[Rename some variables from FOO to FOO_CMD
3114Ian Lynagh <igloo@earth.li>**20100616161108
3115 This fixes a problem with commands like gzip, where if $GZIP is exported
3116 in the environment, then when make runs a command it'll put the Makefile
3117 variable's value in the environment. But gzip treats $GZIP as arguments
3118 for itself, so when we run gzip it thinks we're giving it "gzip" as an
3119 argument.
3120]
3121[Make the "show" target work anywhere in the build tree
3122Ian Lynagh <igloo@earth.li>**20100616122910
3123 Ignore-this: 299d40cbe16112accd9f14e56fa12158
3124]
3125[Change ghc-pwd's license to a string Cabal recognises
3126Ian Lynagh <igloo@earth.li>**20100615204015
3127 Ignore-this: c935b6ad7f605aab0168997a90b40fc6
3128]
3129[fix warning
3130Simon Marlow <marlowsd@gmail.com>**20100604205933
3131 Ignore-this: 2aaa4ed6a8b9ae1e39adc4696aaf14a3
3132]
3133[--install-signal-handles=no does not affect the timer signal (#1908)
3134Simon Marlow <marlowsd@gmail.com>**20100527214627
3135 Ignore-this: b0c51f1abdb159dc360662485095a11a
3136]
3137[Small optimisation: allocate nursery blocks contiguously
3138Simon Marlow <marlowsd@gmail.com>**20100509194928
3139 Ignore-this: e650e99e9ea9493d2efb245d565beef4
3140 This lets automatic prefetching work better, for a tiny performance boost
3141]
3142[fix -fforce-recomp setting: module is PrimOp, not PrimOps
3143Simon Marlow <marlowsd@gmail.com>**20100507084507
3144 Ignore-this: f76e0d9b643682ec0e8fb7d91afdea68
3145]
3146[it should be an error to use relative directories (#4134)
3147Simon Marlow <marlowsd@gmail.com>**20100615151740
3148 Ignore-this: 2068021701832e018ca41b22877921d5
3149]
3150[missing include-dirs or library-dirs is only a warning now (#4104)
3151Simon Marlow <marlowsd@gmail.com>**20100615151702
3152 Ignore-this: e3114123cef147bbd28ccb64581a1afb
3153]
3154[fix #3822: desugaring case command in arrow notation
3155Ross Paterson <ross@soi.city.ac.uk>**20100615225110
3156 Ignore-this: 477d6c460b4174b94b4cd113fa5b9d19
3157 
3158 Get the set of free variables from the generated case expression:
3159 includes variables in the guards and decls that were missed before,
3160 and is also a bit simpler.
3161]
3162[Deprecate the -fvia-C flag; trac #3232
3163Ian Lynagh <igloo@earth.li>**20100615151836
3164 Ignore-this: c2452b2648bf7e44546465c1b964fce
3165]
3166[Avoid using the new ~~ perl operator in the mangler
3167Ian Lynagh <igloo@earth.li>**20100615151236
3168 Ignore-this: 709a7ba4e514b1596841b3ba7e5c6cc
3169]
3170[stmAddInvariantToCheck: add missing init of invariant->lock (#4057)
3171Simon Marlow <marlowsd@gmail.com>**20100615123643
3172 Ignore-this: 3b132547fa934cecf71a846db2a5f70e
3173]
3174[Add new LLVM code generator to GHC. (Version 2)
3175David Terei <davidterei@gmail.com>**20100615094714
3176 Ignore-this: 4dd2fe5854b64a3f0339d484fd5c238
3177 
3178 This was done as part of an honours thesis at UNSW, the paper describing the
3179 work and results can be found at:
3180 
3181 http://www.cse.unsw.edu.au/~pls/thesis/davidt-thesis.pdf
3182 
3183 A Homepage for the backend can be found at:
3184 
3185 http://hackage.haskell.org/trac/ghc/wiki/Commentary/Compiler/Backends/LLVM
3186 
3187 Quick summary of performance is that for the 'nofib' benchmark suite, runtimes
3188 are within 5% slower than the NCG and generally better than the C code
3189 generator.  For some code though, such as the DPH projects benchmark, the LLVM
3190 code generator outperforms the NCG and C code generator by about a 25%
3191 reduction in run times.
3192 
3193]
3194[Fix Trac #4127: build GlobalRdrEnv in GHCi correctly
3195simonpj@microsoft.com**20100615070626
3196 Ignore-this: d907e3bfa7882878cea0af172aaf6e84
3197 
3198 GHCi was building its GlobalRdrEnv wrongly, so that the
3199 gre_par field was bogus.  That in turn fooled the renamer.
3200 The fix is easy: use the right function!  Namely, call
3201 RnNames.gresFromAvail rather than availsToNameSet.
3202]
3203[Comments, and improvement to pretty-printing of HsGroup
3204simonpj@microsoft.com**20100615070409
3205 Ignore-this: ec8358f2485370b20226a97ec84e9024
3206]
3207[Don't reverse bindings in rnMethodBinds (fix Trac #4126)
3208simonpj@microsoft.com**20100614163935
3209 Ignore-this: a6ffbb5af6f51b142ed0aeae8ee5e3a9
3210]
3211[Fix Trac #4120: generate a proper coercion when unifying forall types
3212simonpj@microsoft.com**20100614134311
3213 Ignore-this: 601592bb505305f1954cbe730f168da4
3214 
3215 This was just a blatant omission, which hasn't come up before.
3216 Easily fixed, happily.
3217]
3218[Use mkFunTy to ensure that invariants are respected
3219simonpj@microsoft.com**20100614134159
3220 Ignore-this: 67dcada7a4e8d9927581cd77af71b6f
3221]
3222[Remove redundant debug code
3223simonpj@microsoft.com**20100601154151
3224 Ignore-this: e6ff11c04c631cf6aac73788cbcf02b5
3225]
3226[Fix Trac #4099: better error message for type functions
3227simonpj@microsoft.com**20100531140413
3228 Ignore-this: 3f53ca98cf770577818b9c0937482577
3229 
3230 Now we only want about "T is a type function and might not be
3231 injective" when matchin (T x) against (T y), which is the case
3232 that is really confusing.
3233]
3234[Gruesome fix in CorePrep to fix embarassing Trac #4121
3235simonpj@microsoft.com**20100614132726
3236 Ignore-this: fe82d15474afaac3e6133adfd7a7e055
3237 
3238 This is a long-lurking bug that has been flushed into
3239 the open by other arity-related changes.  There's a
3240 long comment
3241 
3242      Note [CafInfo and floating]
3243 
3244 to explain. 
3245 
3246 I really hate the contortions we have to do through to keep correct
3247 CafRef information on top-level binders.  The Right Thing, I believe,
3248 is to compute CAF and arity information later, and merge it into the
3249 interface-file information when the latter is generated.
3250 
3251 But for now, this hackily fixes the problem.
3252]
3253[Fix a bug in CorePrep that meant output invariants not satisfied
3254simonpj@microsoft.com**20100531150013
3255 Ignore-this: d34eb36d8877d3caf1cf2b20de426abd
3256 
3257 In cpePair I did things in the wrong order so that something that
3258 should have been a CprRhs wasn't.  Result: a crash in CoreToStg.
3259 Fix is easy, and I added more informative type signatures too.
3260]
3261[Robustify the treatement of DFunUnfolding
3262simonpj@microsoft.com**20100531145332
3263 Ignore-this: 8f5506ada4d89f6ab8ad1e8c3ffb09ba
3264 
3265 See Note [DFun unfoldings] in CoreSyn.  The issue here is that
3266 you can't tell how many dictionary arguments a DFun needs just
3267 from looking at the Arity of the DFun Id: if the dictionary is
3268 represented by a newtype the arity might include the dictionary
3269 and value arguments of the (single) method.
3270 
3271 So we need to record the number of arguments need by the DFun
3272 in the DFunUnfolding itself.  Details in
3273    Note [DFun unfoldings] in CoreSyn
3274]
3275[Fix spelling in comment
3276simonpj@microsoft.com**20100614132259
3277 Ignore-this: bbf0d55f2e5f10ef9c74592c12f9201c
3278]
3279[Update docs on view patterns
3280simonpj@microsoft.com**20100614074801
3281 Ignore-this: 8617b9078800d4942d71f142a5b6c831
3282]
3283[Fix printing of splices; part of #4124
3284Ian Lynagh <igloo@earth.li>**20100613154838
3285 Just putting parens around non-atomic expressions isn't sufficient
3286 for splices, as only the $x and $(e) forms are valid input.
3287]
3288[In ghci, catch IO exceptions when calling canonicalizePath
3289Ian Lynagh <igloo@earth.li>**20100613134627
3290 We now get an exception if the path doesn't exist
3291]
3292[Whitespace only
3293Ian Lynagh <igloo@earth.li>**20100612213119]
3294[Whitespace only
3295Ian Lynagh <igloo@earth.li>**20100612165450]
3296[Update ghci example output in user guide; patch from YitzGale in #4111
3297Ian Lynagh <igloo@earth.li>**20100612162250]
3298[Fix #4131 missing UNTAG_CLOSURE in messageBlackHole()
3299benl@ouroborus.net**20100611044614]
3300[messageBlackHole: fix deadlock bug caused by a missing 'volatile'
3301Simon Marlow <marlowsd@gmail.com>**20100610080636
3302 Ignore-this: 3cda3054bb45408aa9bd2d794b69c938
3303]
3304[Pass --no-tmp-comp-dir to Haddock (see comment)
3305Simon Marlow <marlowsd@gmail.com>**20100604083214
3306 Ignore-this: bfa4d74038637bd149f4d878b4eb8a87
3307]
3308[Track changes to DPH libs
3309Roman Leshchinskiy <rl@cse.unsw.edu.au>**20100607052903
3310 Ignore-this: 4dbc3f8418af3e74b3fc4f9a9dfe7764
3311]
3312[Track changes to DPH libs
3313Roman Leshchinskiy <rl@cse.unsw.edu.au>**20100607012642
3314 Ignore-this: 5d4e498171a3c57ab02621bfaea82cff
3315]
3316[In ghc-pkg, send warnings to stderr
3317Ian Lynagh <igloo@earth.li>**20100606161726
3318 Ignore-this: 56927d13b5e1c1ce2752734f0f9b665b
3319]
3320[Re-add newlines to enable layout for multi-line input.
3321Ian Lynagh <igloo@earth.li>**20100602180737
3322 Patch from Adam Vogt <vogt.adam@gmail.com>
3323 Partial fix for #3984
3324]
3325[Don't use unnecessary parens when printing types (Fix Trac 4107)
3326simonpj@microsoft.com**20100604110143
3327 Ignore-this: a833714ab13013c4345b222f4e87db1d
3328 
3329    f :: Eq a => a -> a
3330 rather than
3331    f :: (Eq a) => a -> a
3332]
3333[Track DPH library changes
3334Roman Leshchinskiy <rl@cse.unsw.edu.au>**20100604005728
3335 Ignore-this: 32bc2fbea6ad975e89545d4c42fd7c30
3336]
3337[fix --source-entity option passed to Haddock: we needed to escape a #
3338Simon Marlow <marlowsd@gmail.com>**20100603125459
3339 Ignore-this: d52ae6188b510c482bcebb23f0e553ae
3340]
3341[__stg_EAGER_BLACKHOLE_INFO -> __stg_EAGER_BLACKHOLE_info (#4106)
3342Simon Marlow <marlowsd@gmail.com>**20100602091419
3343 Ignore-this: 293315ac8f86fd366b8d61992ecc7961
3344]
3345[Add xhtml package (a new dependency of Haddock; not installed/shipped)
3346Simon Marlow <marlowsd@gmail.com>**20100602090101
3347 Ignore-this: af0ac8b91abe98f7fdb624ea0a4dee20
3348]
3349[Use UserInterrupt rather than our own Interrupted exception (#4100)
3350Simon Marlow <marlowsd@gmail.com>**20100602082345
3351 Ignore-this: 1909acf2f452593138b9f85024711714
3352]
3353[Add the global package DB to ghc --info (#4103)
3354Simon Marlow <marlowsd@gmail.com>**20100602082233
3355 Ignore-this: fd5c0e207e70eb0f62606c45dc5b8124
3356]
3357[rts/sm/GC.c: resize_generations(): Remove unneeded check of number of generations.
3358Marco Túlio Gontijo e Silva <marcot@debian.org>**20100528115612
3359 Ignore-this: 6f1bea62917c01c7adac636146132c97
3360 
3361 This "if" is inside another "if" which checks for RtsFlags.GcFlags.generations
3362 > 1, so testing this again is redundant, assuming the number of generations
3363 won't change during program execution.
3364]
3365[rts/sm/BlockAlloc.c: Small comment correction.
3366Marco Túlio Gontijo e Silva <marcot@debian.org>**20100526205839
3367 Ignore-this: bd2fcd4597cc872d80b0e2eeb1c3998a
3368]
3369[rts/sm/GC.c: Annotate constants.
3370Marco Túlio Gontijo e Silva <marcot@debian.org>**20100526205707
3371 Ignore-this: f232edb89383564d759ed890a18f602f
3372]
3373[includes/rts/storage/GC.h: generation_: n_words: Improve comment.
3374Marco Túlio Gontijo e Silva <marcot@debian.org>**20100526204615
3375 Ignore-this: f5d5feefa8f7b552303978f1804fea23
3376]
3377[Add PPC_RELOC_LOCAL_SECTDIFF support; patch from PHO in #3654
3378Ian Lynagh <igloo@earth.li>**20100601204211
3379 Ignore-this: 51293b7041cdce3ce7619ef11cf7ceb
3380]
3381[powerpc-apple-darwin now supports shared libs
3382Ian Lynagh <igloo@earth.li>**20100601173325]
3383[Vectoriser: only treat a function as scalar if it actually computes something
3384Roman Leshchinskiy <rl@cse.unsw.edu.au>**20100601045630
3385 Ignore-this: e5d99a6ddb62052e3520094a5af47552
3386]
3387[Add a release notes file for 6.14.1
3388Ian Lynagh <igloo@earth.li>**20100530171117
3389 Ignore-this: 1941e6d3d1f4051b69ca2f17a1cf84d6
3390]
3391[Check dblatex actually creates the files we tell it to
3392Ian Lynagh <igloo@earth.li>**20100530171043
3393 Ignore-this: ccc72caea2313be05cbac59bb54c0603
3394 If it fails, it still exits successfully.
3395]
3396[Add darwin to the list of OSes for which we use mmap
3397Ian Lynagh <igloo@earth.li>**20100529145016
3398 Ignore-this: a86d12a3334aaaafc86f7af9dbb0a7ae
3399 Patch from Barney Stratford
3400]
3401[Simplify the CPP logic in rts/Linker.c
3402Ian Lynagh <igloo@earth.li>**20100529144929
3403 Ignore-this: 1288f5b752cc1ab8b1c90cfd0ecfdf68
3404]
3405[Fix validate on OS X
3406Ian Lynagh <igloo@earth.li>**20100529154726]
3407[OS X x86_64 fix from Barney Stratford
3408Ian Lynagh <igloo@earth.li>**20100529122440]
3409[OS X 64 installer fixes from Barney Stratford
3410Ian Lynagh <igloo@earth.li>**20100528234935]
3411[fix warning
3412Simon Marlow <marlowsd@gmail.com>**20100525155812
3413 Ignore-this: f34eee3fe3d89579fd8d381c91ced750
3414]
3415[Fix doc bugs (#4071)
3416Simon Marlow <marlowsd@gmail.com>**20100525155728
3417 Ignore-this: aa25be196de567de360075022a1942f7
3418]
3419[Make sparks into weak pointers (#2185)
3420Simon Marlow <marlowsd@gmail.com>**20100525150435
3421 Ignore-this: feea0bb5006007b82c932bc3006124d7
3422 The new strategies library (parallel-2.0+, preferably 2.2+) is now
3423 required for parallel programming, otherwise parallelism will be lost.
3424]
3425[If you say 'make' or 'make stage=2' here, pretend we're in the ghc dir
3426Simon Marlow <marlowsd@gmail.com>**20100525085301
3427 Ignore-this: 78b740337aa460915c812cbbcdae5321
3428]
3429[Another attempt to get these #defines right
3430Simon Marlow <marlowsd@gmail.com>**20100525154313
3431 Ignore-this: 460ca0c47d81cd25eae6542114f67899
3432 Apparently on Solaris it is an error to omit _ISOC99_SOURCE when using
3433 _POSIX_C_SOURCE==200112L.
3434]
3435[Add configure flags for the location of GMP includes/library; fixes #4022
3436Ian Lynagh <igloo@earth.li>**20100525221616
3437 Ignore-this: fc3060caf995d07274ec975eeefbdf3e
3438]
3439[Refactor pretty printing of TyThings to fix Trac #4015
3440simonpj@microsoft.com**20100525153126
3441 Ignore-this: 8f15053b7554f62caa84201d2e4976d2
3442]
3443[When haddocking, we need the dependencies to have been built
3444Ian Lynagh <igloo@earth.li>**20100525145830
3445 as haddock loads the .hi files with the GHC API.
3446]
3447[Fix profiling output; spotted by jlouis
3448Ian Lynagh <igloo@earth.li>**20100525111217
3449 We were outputing the number of words allocated in a column titled "bytes".
3450]
3451[Improve printing of TyThings; fixes Trac #4087
3452simonpj@microsoft.com**20100525114045
3453 Ignore-this: da2a757a533454bba80b9b77cc5a771
3454]
3455[Spelling in comments
3456simonpj@microsoft.com**20100525114001
3457 Ignore-this: 270f3da655e526cf04e27db7a01e29c0
3458]
3459[Refactor (again) the handling of default methods
3460simonpj@microsoft.com**20100525113910
3461 Ignore-this: 6686f6cdb878d57abf6b49fec64fcbb1
3462 
3463 This patch fixes Trac #4056, by
3464 
3465  a) tidying up the treatment of default method names
3466  b) removing the 'module' argument to newTopSrcBinder
3467 
3468 The details aren't that interesting, but the result
3469 is much tidier. The original bug was a 'nameModule' panic,
3470 caused by trying to find the module of a top-level name.
3471 But TH quotes generate Internal top-level names that don't
3472 have a module, and that is generally a good thing. 
3473 
3474 Fixing that in turn led to the default-method refactoring,
3475 which also makes the Name for a default method be handled
3476 in the same way as other derived names, generated in BuildTyCl
3477 via a call newImplicitBinder.  Hurrah.
3478]
3479[Don't do SpecConstr on NOINLINE things (Trac #4064)
3480simonpj@microsoft.com**20100525112807
3481 Ignore-this: 452be0a2cef0042fb67275c2827b5f72
3482 
3483 Since the RULE from specialising gets the same Activation as
3484 the inlining for the Id itself there's no point in specialising
3485 a NOINLINE thing, because the rule will be permanently switched
3486 off.
3487 
3488 See Note [Transfer activation] in SpecConstr
3489 and Note [Auto-specialisation and RULES] in Specialise.
3490]
3491[Change our #defines to work on FreeBSD too
3492Simon Marlow <marlowsd@gmail.com>**20100524105828
3493 Ignore-this: b23ede46211e67859206c0ec57d6a86f
3494 With glibc, things like _POSIX_C_SOURCE and _ISOC99_SOURCE are
3495 additive, but on FreeBSD they are mutually exclusive.  However, it
3496 turns out we only need to define _POSIX_C_SOURCE and _XOPEN_SOURCE to
3497 get all the C99 stuff we need too, so there's no need for any #ifdefs.
3498 
3499 Submitted by: Gabor PALI <pgj@FreeBSD.org>
3500]
3501[Add a missing UNTAG_CLOSURE, causing bus errors on Sparc
3502Simon Marlow <marlowsd@gmail.com>**20100524105547
3503 Ignore-this: a590b5391d6f05d50c8c088456c3c166
3504 We just about got away with this on x86 which isn't
3505 alignment-sensitive.  The result of the memory load is compared
3506 against a few different values, but there is a fallback case that
3507 happened to be the right thing when the pointer was tagged.  A good
3508 bug to find, nonetheless.
3509]
3510[Add wiki links
3511Simon Marlow <marlowsd@gmail.com>**20100520095953
3512 Ignore-this: c22f126cde166e6207922b2eb51d29e3
3513]
3514[the 'stage=0' trick to disable all compiler builds stopped working; fix it
3515Simon Marlow <marlowsd@gmail.com>**20100520104455
3516 Ignore-this: bb6fae9056471612c8dbf06916188c33
3517]
3518[Comments and formatting only
3519benl@ouroborus.net**20100524014021
3520 Ignore-this: 64579c38154728b632e358bec751cc0b
3521]
3522[Core prettyprinter fixes. Patch from Tim Chevalier. Fixes #4085
3523Ian Lynagh <igloo@earth.li>**20100522225048]
3524[Fix the RTS debug_p build
3525Ian Lynagh <igloo@earth.li>**20100522163127]
3526[Unset $CFLAGS for "GNU non-executable stack" configure test; fixes #3889
3527Ian Lynagh <igloo@earth.li>**20100521165005
3528 With gcc 4.4 we get
3529     Error: can't resolve `.note.GNU-stack' {.note.GNU-stack section} - `.Ltext0' {.text section}
3530 when running gcc with the -g flag. To work around this we unset
3531 CFLAGS when running the test.
3532]
3533[Don't run "set -o igncr" before configuring libffi
3534Ian Lynagh <igloo@earth.li>**20100520162918
3535 Ignore-this: 489fa94df23f2adf4ff63c8ede2c0794
3536 It used to make the build work on cygwin, but now it breaks it instead:
3537     config.status: creating include/Makefile
3538     gawk: ./confLqjohp/subs.awk:1: BEGIN {\r
3539     gawk: ./confLqjohp/subs.awk:1: ^ backslash not last character on line
3540     config.status: error: could not create include/Makefile
3541     make[2]: *** [libffi/stamp.ffi.configure-shared] Error 1
3542     make[1]: *** [all] Error 2
3543]
3544[Stop passing -Wl,-macosx_version_min to gcc
3545Ian Lynagh <igloo@earth.li>**20100520154003
3546 Fixes a build failure on OS X 10.6. When linking
3547     rts/dist/build/libHSrts-ghc6.13.20100519.dylib
3548 we got
3549     ld: symbol dyld_stub_binding_helper not defined (usually in crt1.o/dylib1.o/bundle1.o)
3550     collect2: ld returned 1 exit status
3551]
3552[Fix build on FreeBSD; patch from Gabor PALI
3553Ian Lynagh <igloo@earth.li>**20100519140552]
3554[Fix package shadowing order (#4072)
3555Simon Marlow <marlowsd@gmail.com>**20100519104617
3556 Ignore-this: 26ea5e4bb5dff18618b807a54c7d6ebb
3557 
3558 Later packages are supposed to shadow earlier ones in the stack,
3559 unless the ordering is overriden with -package-id flags.
3560 Unfortunately an earlier fix for something else had sorted the list of
3561 packages so that it was in lexicographic order by installedPackageId,
3562 and sadly our test (cabal/shadow) didn't pick this up because the
3563 lexicographic ordering happened to work for the test.  I've now fixed
3564 the test so it tries both orderings.
3565]
3566[Set more env variables when configuring libffi
3567Ian Lynagh <igloo@earth.li>**20100518185014
3568 We now tell it where to find ld, nm and ar
3569]
3570[Set the location of ar to be the in-tree ar on Windows
3571Ian Lynagh <igloo@earth.li>**20100518181556]
3572[Change another / to </> to avoid building paths containing \/
3573Ian Lynagh <igloo@earth.li>**20100518172015
3574 This will hopefully fix #2889.
3575]
3576[Fix #4074 (I hope).
3577Simon Marlow <marlowsd@gmail.com>**20100518113214
3578 Ignore-this: 73cd70f5bc6f5add5247b61985c03fc1
3579 
3580 1. allow multiple threads to call startTimer()/stopTimer() pairs
3581 2. disable the timer around fork() in forkProcess()
3582 
3583 A corresponding change to the process package is required.
3584]
3585[we don't have a gcc-lib in LIB_DIR any more
3586Simon Marlow <marlowsd@gmail.com>**20100401102351
3587 Ignore-this: f41acd2d8f8e6763aa8bd57a0b44a7e4
3588]
3589[In validate, use gmake if available; based on a patch from Gabor PALI
3590Ian Lynagh <igloo@earth.li>**20100517200654]
3591[Remove duplicate "./configure --help" output; fixes #4075
3592Ian Lynagh <igloo@earth.li>**20100516141206]
3593[Update various 'sh boot's to 'perl boot'
3594Ian Lynagh <igloo@earth.li>**20100516122609
3595 Spotted by Marco Túlio Gontijo e Silva
3596]
3597[add missing initialisation for eventBufMutex
3598Simon Marlow <marlowsd@gmail.com>**20100514094943
3599 Ignore-this: 7f75594a8cb54fbec5aebd46bb959f45
3600]
3601[Undo part of #4003 patch
3602Simon Marlow <marlowsd@gmail.com>**20100513142017
3603 Ignore-this: cb65db86a38a7e5ccee9f779e489d104
3604 We still need the workaround for when compiling HEAD with 6.12.2
3605 
3606]
3607[fix !TABLES_NEXT_TO_CODE
3608Simon Marlow <marlowsd@gmail.com>**20100510151934
3609 Ignore-this: fccb859b114bef1c3122c98e60af51
3610]
3611[looksLikeModuleName: allow apostrophe in module names (#4051)
3612Simon Marlow <marlowsd@gmail.com>**20100510094741
3613 Ignore-this: df9348f3ba90608bec57257b47672985
3614]
3615[add the proper library dependencies for GhcProfiled=YES
3616Simon Marlow <marlowsd@gmail.com>**20100506122118
3617 Ignore-this: 6236993aa308ab5b5e1e5ea5f65982a
3618]
3619[Fix Trac #4003: fix the knot-tying in checkHiBootIface
3620simonpj@microsoft.com**20100511075026
3621 Ignore-this: a9ce2a318386fdc8782848df84592002
3622 
3623 I had incorrectly "optimised" checkHiBootIface so that it forgot
3624 to update the "knot-tied" type environment.
3625 
3626 This patch fixes the HEAD
3627]
3628[Re-engineer the derived Ord instance generation code (fix Trac #4019)
3629simonpj@microsoft.com**20100510133333
3630 Ignore-this: 8fe46e4dad27fbee211a7928acf372c2
3631   
3632 As well as fixing #4019, I rejigged the way that Ord instances are
3633 generated, which should make them faster in general.  See the
3634 Note [Generating Ord instances].
3635 
3636 I tried to measure the performance difference from this change, but
3637 the #4019 fix only removes one conditional branch per iteration, and
3638 I couldn't measure a consistent improvement.  But still, tihs is
3639 better than before.
3640]
3641[Make arity of INLINE things consistent
3642simonpj@microsoft.com**20100510133005
3643 Ignore-this: 15e7abf803d1dcb3f4ca760d2d939d0d
3644 
3645 We eta-expand things with INLINE pragmas;
3646 see Note [Eta-expanding INLINE things].
3647 
3648 But I eta-expanded it the wrong amount when the function
3649 was overloaded.  Ooops.
3650]
3651[Compacting GC fix, we forgot to thread the new bq field of StgTSO.
3652Simon Marlow <marlowsd@gmail.com>**20100510082325
3653 Ignore-this: a079c8446e2ad53efff6fd95d0f3ac80
3654]
3655[Add version constraints for the boot packages; fixes trac #3852
3656Ian Lynagh <igloo@earth.li>**20100509175051
3657 When using the bootstrapping compiler, we now explicitly constrain
3658 the version of boot packages (Cabal, extensible-exceptions, etc) to the
3659 in-tree version, so that the build system is less fragile should the
3660 user have a newer version installed for the bootstrapping compiler.
3661]
3662[Don't include inter-package dependencies when compiling with stage 0; #4031
3663Ian Lynagh <igloo@earth.li>**20100509130511
3664 This fixes a problem when building with GHC 6.12 on Windows, where
3665 dependencies on stage 0 (bootstrapping compiler) packages have absolute
3666 paths c:/ghc/..., and make gets confused by the colon.
3667]
3668[Add a ghc.mk for bindisttest/
3669Ian Lynagh <igloo@earth.li>**20100508223911]
3670[Move some make variables around so they are available when cleaning
3671Ian Lynagh <igloo@earth.li>**20100508212405]
3672[Optimise checkremove a bit
3673Ian Lynagh <igloo@earth.li>**20100508202006]
3674[Improve the bindisttest Makefile
3675Ian Lynagh <igloo@earth.li>**20100508195450]
3676[Add tools to test that cleaning works properly
3677Ian Lynagh <igloo@earth.li>**20100508194105]
3678[Tweak the ghc-pkg finding code
3679Ian Lynagh <igloo@earth.li>**20100508125815
3680 It now understand the ghc-stage[123] names we use in-tree, and it won't
3681 go looking for any old ghc-pkg if it can't find the one that matches
3682 ghc.
3683]
3684[Add a way to show what cleaning would be done, without actually doing it
3685Ian Lynagh <igloo@earth.li>**20100508122438]
3686[Tidy up the "rm" flags in the build system
3687Ian Lynagh <igloo@earth.li>**20100508115745]
3688[Correct install-name for dynamic Darwin rts
3689pho@cielonegro.org**20100508151155
3690 Ignore-this: 6d31716c8c113dcb46e9cb925c4201df
3691]
3692[PIC support for PowerPC
3693pho@cielonegro.org**20100508143900
3694 Ignore-this: 3673859a305398c4acae3f4d7c997615
3695 
3696 PPC.CodeGen.getRegister was not properly handling PicBaseReg.
3697 It seems working with this patch, but I'm not sure this change is correct.
3698]
3699[Fix makefile loop (#4050)
3700pho@cielonegro.org**20100507140707
3701 Ignore-this: 3a1cb13d0600977e74d17ac26cbef83d
3702 
3703 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.
3704]
3705[Fix crash in nested callbacks (#4038)
3706Simon Marlow <marlowsd@gmail.com>**20100507093222
3707 Ignore-this: cade85e361534ce711865a4820276388
3708 Broken by "Split part of the Task struct into a separate struct
3709 InCall".
3710]
3711[Add $(GhcDynamic) knob, set to YES to get stage2 linked with -dynamic
3712Simon Marlow <marlowsd@gmail.com>**20100428205241
3713 Ignore-this: 1db8bccf92099785ecac39aebd27c92d
3714 Default currently NO.
3715 
3716 Validate passed with GhcDynamic=YES on x86/Linux here.
3717 
3718 The compiler is currently slower on x86 when linked -dynamic,
3719 because the GC inner loop has been adversely affected by -fPIC, I'm
3720 looking into how to fix it.
3721]
3722[omit "dyn" from the way appended to the __stginit label
3723Simon Marlow <marlowsd@gmail.com>**20100428204914
3724 Ignore-this: 14183f3defa9f2bde68fda6729b740bc
3725 When GHCi is linked dynamically, we still want to be able to load
3726 non-dynamic object files.
3727]
3728[improvements to findPtr(), a neat hack for browsing the heap in gdb
3729Simon Marlow <marlowsd@gmail.com>**20100506115427
3730 Ignore-this: ac57785bb3e13b97a5945f753f068738
3731]
3732[Fix +RTS -G1
3733Simon Marlow <marlowsd@gmail.com>**20100506110739
3734 Ignore-this: 86a5de39a94d3331a4ee1213f82be497
3735]
3736[Enable the "redundant specialise pragmas" warning; fixes trac #3855
3737Ian Lynagh <igloo@earth.li>**20100506175351]
3738[Find the correct external ids when there's a wrapper
3739simonpj@microsoft.com**20100506164135
3740 Ignore-this: 636266407b174b05b2b8646cc73062c0
3741 
3742 We were failing to externalise the wrapper id for a function
3743 that had one.
3744]
3745[Add a comment about pattern coercions
3746simonpj@microsoft.com**20100506164027
3747 Ignore-this: 17428089f3df439f65d892e23e8ed61a
3748]
3749[Comments only
3750simonpj@microsoft.com**20100506163829
3751 Ignore-this: 169167b6463873ab173cc5750c5be469
3752]
3753[Make a missing name in mkUsageInfo into a panic
3754simonpj@microsoft.com**20100506163813
3755 Ignore-this: b82ff1b8bf89f74f146db7cb5cc4c4d7
3756 
3757 We really want to know about this!
3758]
3759[Refactoring of hsXxxBinders
3760simonpj@microsoft.com**20100506163737
3761 Ignore-this: 97c6667625262b160f9746f7bea1c980
3762 
3763 This patch moves various functions that extract the binders
3764 from a HsTyClDecl, HsForeignDecl etc into HsUtils, and gives
3765 them consistent names.
3766]
3767[Fix Trac #3966: warn about useless UNPACK pragmas
3768simonpj@microsoft.com**20100506163337
3769 Ignore-this: 5beb24b686eda6113b614dfac8490df1
3770 
3771 Warning about useless UNPACK pragmas wasn't as easy as I thought.
3772 I did quite a bit of refactoring, which improved the code by refining
3773 the types somewhat.  In particular notice that in DataCon, we have
3774 
3775     dcStrictMarks   :: [HsBang]
3776     dcRepStrictness :: [StrictnessMarks]
3777 
3778 The former relates to the *source-code* annotation, the latter to
3779 GHC's representation choice.
3780]
3781[Make tcg_dus behave more sanely; fixes a mkUsageInfo panic
3782simonpj@microsoft.com**20100506162719
3783 Ignore-this: d000bca15b0e127e297378ded1bfb81b
3784 
3785 The tcg_dus field used to contain *uses* of type and class decls,
3786 but not *defs*.  That was inconsistent, and it really went wrong
3787 for Template Haskell bracket.  What happened was that
3788  foo = [d| data A = A
3789                   f :: A -> A
3790                   f x = x |]
3791 would find a "use" of A when processing the top level of the module,
3792 which in turn led to a mkUsageInfo panic in MkIface.  The cause was
3793 the fact that the tcg_dus for the nested quote didn't have defs for
3794 A.
3795]
3796[Add a HsExplicitFlag to SpliceDecl, to improve Trac #4042
3797simonpj@microsoft.com**20100506161523
3798 Ignore-this: e4e563bac2fd831cc9e94612f5b4fa9d
3799 
3800 The issue here is that
3801 
3802     g :: A -> A
3803     f
3804     data A = A
3805 
3806 is treated as if you'd written $(f); that is the call of
3807 f is a top-level Template Haskell splice.  This patch
3808 makes sure that we *first* check the -XTemplateHaskellFlag
3809 and bleat about a parse error if it's off.  Othewise we
3810 get strange seeing "A is out of scope" errors.
3811]
3812[Change an assert to a warn
3813simonpj@microsoft.com**20100506161111
3814 Ignore-this: 739a4fb4c7940376b0f2c8ad52a1966c
3815 
3816 This is in the constraint simplifier which I'm about
3817 to rewrite, so I'm hoping the assert isn't fatal!
3818]
3819[Tidy up debug print a little
3820simonpj@microsoft.com**20100506161027
3821 Ignore-this: bd5492878e06bee1cddcbb3fc4df66d8
3822]
3823[Remove useless UNPACK pragmas
3824simonpj@microsoft.com**20100506161012
3825 Ignore-this: 3e5ab1a7cf58107034412a798bc214e5
3826]
3827[Add WARNM2 macro, plus some refactoring
3828simonpj@microsoft.com**20100506160808
3829 Ignore-this: 2ab4f1f0b5d94be683036e77aec09255
3830]
3831[Use -Wwarn for the binary package, becuase it has redundant UNPACK pragmas
3832simonpj@microsoft.com**20100506160750
3833 Ignore-this: cf0d3a11473e28bfce9602e716e69a5f
3834]
3835[Fix Trac #3966: warn about unused UNPACK pragmas
3836simonpj@microsoft.com**20100409201812
3837 Ignore-this: c96412596b39c918b5fb9b3c39ce2119
3838]
3839[Fix Trac #3953: fail earlier when using a bogus quasiquoter
3840simonpj@microsoft.com**20100409201748
3841 Ignore-this: ef48e39aa932caed538643985234f043
3842]
3843[Fix Trac #3965: tighten conditions when deriving Data
3844simonpj@microsoft.com**20100409184420
3845 Ignore-this: 96f7d7d2da11565d26b465d7d0497ac9
3846 
3847 It's tricky to set up the context for a Data instance.  I got it wrong
3848 once, and fixed it -- hence the "extra_constraints" in
3849 TcDeriv.inferConstraints. 
3850 
3851 But it still wasn't right!  The tricky bit is that dataCast1 is only
3852 generated when T :: *->*, and dataCast2 when T :: *->*->*. (See
3853 the code in TcGenDeriv for dataCastX.
3854]
3855[Fix Trac #3964: view patterns in DsArrows
3856simonpj@microsoft.com**20100409165557
3857 Ignore-this: d823c182831d5e2e592e995b16180e2f
3858 
3859 Just a missing case; I've eliminated the catch-all so
3860 that we get a warning next time we extend HsPat
3861]
3862[Fix Trac #3955: renamer and type variables
3863simonpj@microsoft.com**20100409163710
3864 Ignore-this: bd5ec64d76c0f583bf5f224792bf294c
3865 
3866 The renamer wasn't computing the free variables of a type declaration
3867 properly.  This patch refactors a bit, and makes it more robust,
3868 fixing #3955 and several other closely-related bugs.  (We were
3869 omitting some free variables and that could just possibly lead to a
3870 usage-version tracking error.
3871]
3872[Layout only
3873simonpj@microsoft.com**20100409163506
3874 Ignore-this: 1f14990b5aa0b9821b84452fb34e9f41
3875]
3876[Give a better deprecated message for INCLUDE pragmas; fixes #3933
3877Ian Lynagh <igloo@earth.li>**20100506130910
3878 We now have a DeprecatedFullText constructor, so we can override the
3879 "-#include is deprecated: " part of the warning.
3880]
3881[De-haddock a comment that confuses haddock
3882Ian Lynagh <igloo@earth.li>**20100506123607]
3883[Fix comment to not confuse haddock
3884Ian Lynagh <igloo@earth.li>**20100506113642]
3885[Detect EOF when trying to parse a string in hp2ps
3886Ian Lynagh <igloo@earth.li>**20100506000830]
3887[Make the demand analyser sdd demands for strict constructors
3888simonpj@microsoft.com**20100505200936
3889 Ignore-this: eb32632adbc354eb7a5cf884c263e0d3
3890 
3891 This opportunity was spotted by Roman, and is documented in
3892 Note [Add demands for strict constructors] in DmdAnal.
3893]
3894[Fix interaction of exprIsCheap and the lone-variable inlining check
3895simonpj@microsoft.com**20100505200723
3896 Ignore-this: f3cb65085c5673a99153d5d7b6559ab1
3897 
3898 See Note [Interaction of exprIsCheap and lone variables] in CoreUnfold
3899 
3900 This buglet meant that a nullary definition with an INLINE pragma
3901 counter-intuitively didn't get inlined at all.  Roman identified
3902 the bug.
3903]
3904[Matching cases in SpecConstr and Rules
3905simonpj@microsoft.com**20100505200543
3906 Ignore-this: f5c28c780fbf8badce84c6fdc9aa1779
3907 
3908 This patch has zero effect.  It includes comments,
3909 a bit of refactoring, and a tiny bit of commment-out
3910 code go implement the "matching cases" idea below.
3911 
3912 In the end I've left it disabled because while I think
3913 it does no harm I don't think it'll do any good either.
3914 But I didn't want to lose the idea totally. There's
3915 a thread called "Storable and constant memory" on
3916 the libraries@haskell.org list (Apr 2010) about it.
3917 
3918 Note [Matching cases]
3919 ~~~~~~~~~~~~~~~~~~~~~
3920 {- NOTE: This idea is currently disabled.  It really only works if
3921          the primops involved are OkForSpeculation, and, since
3922         they have side effects readIntOfAddr and touch are not.
3923         Maybe we'll get back to this later .  -}
3924   
3925 Consider
3926    f (case readIntOffAddr# p# i# realWorld# of { (# s#, n# #) ->
3927       case touch# fp s# of { _ ->
3928       I# n# } } )
3929 This happened in a tight loop generated by stream fusion that
3930 Roman encountered.  We'd like to treat this just like the let
3931 case, because the primops concerned are ok-for-speculation.
3932 That is, we'd like to behave as if it had been
3933    case readIntOffAddr# p# i# realWorld# of { (# s#, n# #) ->
3934    case touch# fp s# of { _ ->
3935    f (I# n# } } )
3936]
3937[Comments only
3938simonpj@microsoft.com**20100504163629
3939 Ignore-this: 3be12df04714aa820bce706b5dc8a9cb
3940]
3941[Comments only
3942simonpj@microsoft.com**20100504163529
3943 Ignore-this: 791e2fd39c7d880ce1dc80ebdf3a5398
3944]
3945[Comments only
3946simonpj@microsoft.com**20100504163457
3947 Ignore-this: f19e9ffeb3d65770b1595bca5f97a59d
3948]
3949[Comments only (about type families)
3950simonpj@microsoft.com**20100417145032
3951 Ignore-this: dd39425ef2155d52dbf55a4d5fd97cb8
3952]
3953[Fix hp2ps when the .hp file has large string literals
3954Ian Lynagh <igloo@earth.li>**20100505191921]
3955[In build system, call package-config after including package data
3956Ian Lynagh <igloo@earth.li>**20100504225035
3957 Otherwise the $1_$2_HC_OPTS variable gets clobbered.
3958]
3959[runghc: flush stdout/stderr on an exception (#3890)
3960Simon Marlow <marlowsd@gmail.com>**20100505133848
3961 Ignore-this: 224c1898cec64cb1c94e0d7033e7590e
3962]
3963[Remove the Unicode alternative for ".." (#3894)
3964Simon Marlow <marlowsd@gmail.com>**20100505121202
3965 Ignore-this: 2452cd67281667106f9169747b6d784f
3966]
3967[tidyup; no functional changes
3968Simon Marlow <marlowsd@gmail.com>**20100505115015
3969 Ignore-this: d0787e5cdeef1dee628682fa0a46019
3970]
3971[Make the running_finalizers flag task-local
3972Simon Marlow <marlowsd@gmail.com>**20100505114947
3973 Ignore-this: 345925d00f1dca203941b3c5d84c90e1
3974 Fixes a bug reported by Lennart Augustsson, whereby we could get an
3975 incorrect error from the RTS about re-entry from a finalizer,
3976]
3977[add a MAYBE_GC() in killThread#, fixes throwto003(threaded2) looping
3978Simon Marlow <marlowsd@gmail.com>**20100505114746
3979 Ignore-this: efea04991d6feed04683a42232fc85da
3980]
3981[Allow filepath-1.2.*
3982Simon Marlow <marlowsd@gmail.com>**20100505101139
3983 Ignore-this: 1b5580cd9cd041ec48f40cd37603326a
3984]
3985[BlockedOnMsgThrowTo is possible in resurrectThreads (#4030)
3986Simon Marlow <marlowsd@gmail.com>**20100505094534
3987 Ignore-this: ac24a22f95ffeaf480187a1620fdddb2
3988]
3989[Don't raise a throwTo when the target is masking and BlockedOnBlackHole
3990Simon Marlow <marlowsd@gmail.com>**20100505094506
3991 Ignore-this: 302616931f61667030d77ddfbb02374e
3992]
3993[Fix build with GHC 6.10
3994Ian Lynagh <igloo@earth.li>**20100504180302
3995 In GHC 6.10, intersectionWith is (a -> b -> a) instead of (a -> b -> c),
3996 so we need to jump through some hoops to get the more general type.
3997]
3998[The libffi patches are no longer needed
3999Ian Lynagh <igloo@earth.li>**20100504171603]
4000[Use the in-tree windres; fixes trac #4032
4001Ian Lynagh <igloo@earth.li>**20100504170941]
4002[Print unfoldings on lambda-bound variables
4003Simon PJ <simonpj@microsoft.com>**20100503181822
4004 Ignore-this: 2fd5a7502cc6273d96258e0914f0f8cd
4005 
4006 ...in the unusual case where they have one;
4007 see Note [Case binders and join points] in Simplify.lhs
4008]
4009[Replace FiniteMap and UniqFM with counterparts from containers.
4010Milan Straka <fox@ucw.cz>**20100503171315
4011 Ignore-this: a021972239163dbf728284b19928cebb
4012 
4013 The original interfaces are kept. There is small performance improvement:
4014 - when compiling for five nofib, we get following speedups:
4015     Average                -----           -2.5%
4016     Average                -----           -0.6%
4017     Average                -----           -0.5%
4018     Average                -----           -5.5%
4019     Average                -----          -10.3%
4020 - when compiling HPC ten times, we get:
4021     switches                          oldmaps   newmaps
4022     -O -fasm                          117.402s  116.081s (98.87%)
4023     -O -fasm -fregs-graph             119.993s  118.735s (98.95%)
4024     -O -fasm -fregs-iterative         120.191s  118.607s (98.68%)
4025]
4026[Make the demand analyser take account of lambda-bound unfoldings
4027Simon PJ <simonpj@microsoft.com>**20100503151630
4028 Ignore-this: 2ee8e27d4df2debfc79e6b8a17c32bc1
4029 
4030 This is a long-standing lurking bug. See Note [Lamba-bound unfoldings]
4031 in DmdAnal.
4032 
4033 I'm still not really happy with this lambda-bound-unfolding stuff.
4034]
4035[Fix dynamic libs on OS X, and enable them by default
4036Ian Lynagh <igloo@earth.li>**20100503150302]
4037[Switch back to using bytestring from the darcs repo; partially fixes #3855
4038Ian Lynagh <igloo@earth.li>**20100502113458]
4039[Fix some cpp warnings when building on FreeBSD; patch from Gabor PALI
4040Ian Lynagh <igloo@earth.li>**20100428150700]
4041[Fix "make 2"
4042Ian Lynagh <igloo@earth.li>**20100427162212
4043 The new Makefile logic was enabling the stage 1 rules when stage=2,
4044 so "make 2" was rebuilding stage 1.
4045]
4046[Inplace programs depend on their shell wrappers
4047Ian Lynagh <igloo@earth.li>**20100427160038]
4048[--make is now the default (#3515), and -fno-code works with --make (#3783)
4049Simon Marlow <marlowsd@gmail.com>**20100427122851
4050 Ignore-this: 33330474fa4703f32bf9997462b4bf3c
4051 If the command line contains any Haskell source files, then we behave
4052 as if --make had been given.
4053 
4054 The meaning of the -c flag has changed (back): -c now selects one-shot
4055 compilation, but stops before linking.  However, to retain backwards
4056 compatibility, -c is still allowed with --make, and means the same as
4057 --make -no-link.  The -no-link flag has been un-deprecated.
4058 
4059 -fno-code is now allowed with --make (#3783); the fact that it was
4060 disabled before was largely accidental, it seems.  We also had some
4061 regressions in this area: it seems that -fno-code was causing a .hc
4062 file to be emitted in certain cases.  I've tidied up the code, there
4063 was no need for -fno-code to be a "mode" flag, as far as I can tell.
4064 
4065 -fno-code does not emit interface files, nor does it do recompilation
4066 checking, as suggested in #3783.  This would make Haddock emit
4067 interface files, for example, and I'm fairly sure we don't want to do
4068 that.  Compiling with -fno-code is pretty quick anyway, perhaps we can
4069 get away without recompilation checking.
4070]
4071[remove duplicate docs for -e in --help output (#4010)
4072Simon Marlow <marlowsd@gmail.com>**20100426140642
4073 Ignore-this: 187ff893ba8ffa0ec127867a7590e38d
4074]
4075[workaround for #4003, fixes HEAD build with 6.12.2
4076Simon Marlow <marlowsd@gmail.com>**20100426103428
4077 Ignore-this: c4bc445dc8052d4e6efef3f1daf63562
4078]
4079[Make sure all the clean rules are always included
4080Ian Lynagh <igloo@earth.li>**20100424181823
4081 In particular, this fixes a problem where stage3 bits weren't being cleaned
4082]
4083[Correct the name of the amd64/FreeBSD platform in PlatformSupportsSharedLibs
4084Ian Lynagh <igloo@earth.li>**20100424132830
4085 We weren't getting sharedlibs on amd64/FreeBSD because of this
4086]
4087[Include DPH docs in bindists
4088Ian Lynagh <igloo@earth.li>**20100424123101]
4089[reinstate eta-expansion during SimplGently, to fix inlining of sequence_
4090Simon Marlow <marlowsd@gmail.com>**20100423124853
4091 Ignore-this: 4fa0fd5bafe0d6b58fc81076f50d5f8d
4092]
4093[fix 64-bit value for W_SHIFT, which thankfully appears to be not used
4094Simon Marlow <marlowsd@gmail.com>**20100422213605
4095 Ignore-this: 525c062d2456c224ec8d0e083edd3b55
4096]
4097[Add missing constant folding and optimisation for unsigned division
4098Simon Marlow <marlowsd@gmail.com>**20100422213443
4099 Ignore-this: fb10d1cda0852fab0cbcb47247498fb3
4100 Noticed by Denys Rtveliashvili <rtvd@mac.com>, see #4004
4101]
4102[Fix the GHC API link in the main doc index.html
4103Ian Lynagh <igloo@earth.li>**20100422213226]
4104[Give the right exit code in darcs-all
4105Ian Lynagh <igloo@earth.li>**20100421171339
4106 Our END block was calling system, which alters $?. So now we save and
4107 restore it.
4108]
4109[Use StgWord64 instead of ullong
4110Ian Lynagh <igloo@earth.li>**20100421162336
4111 This patch also fixes ullong_format_string (renamed to showStgWord64)
4112 so that it works with values outside the 32bit range (trac #3979), and
4113 simplifies the without-commas case.
4114]
4115[Implement try10Times in Makefile
4116Ian Lynagh <igloo@earth.li>**20100420165909
4117 Avoid using seq, as FreeBSD has jot instead.
4118]
4119[Fix crash in non-threaded RTS on Windows
4120Simon Marlow <marlowsd@gmail.com>**20100420122125
4121 Ignore-this: 28b0255a914a8955dce02d89a7dfaca
4122 The tso->block_info field is now overwritten by pushOnRunQueue(), but
4123 stg_block_async_info was assuming that it still held a pointer to the
4124 StgAsyncIOResult.  We must therefore save this value somewhere safe
4125 before putting the TSO on the run queue.
4126]
4127[Expand the scope of the event_buf_mutex to cover io_manager_event
4128Simon Marlow <marlowsd@gmail.com>**20100420122026
4129 Ignore-this: 185a6d84f7d4a35997f10803f6dacef1
4130 I once saw a failure that I think was due to a race on
4131 io_manager_event, this should fix it.
4132]
4133[Flags -auto and -auto-all operate only on functions not marked INLINE.
4134Milan Straka <fox@ucw.cz>**20100331191050
4135 Ignore-this: 3b63580cfcb3c33d62ad697c36d94d05
4136]
4137[Spelling correction for LANGUAGE pragmas
4138Max Bolingbroke <batterseapower@hotmail.com>**20100413192825
4139 Ignore-this: 311b51ba8d43f6c7fd32f48db9a88dee
4140]
4141[Update the user guide so it talks about the newer "do rec" notation everywhere
4142Ian Lynagh <igloo@earth.li>**20100416205416
4143 Some of the problems highlighted in trac #3968.
4144]
4145[Fix typo
4146Ian Lynagh <igloo@earth.li>**20100416205412]
4147[Fix Trac #3950: unifying types of different kinds
4148simonpj@microsoft.com**20100412151845
4149 Ignore-this: d145b9de5ced136ef2c39f3ea4a04f4a
4150 
4151 I was assuming that the unifer only unified types of the
4152 same kind, but now we can "defer" unsolved constraints that
4153 invariant no longer holds.  Or at least is's more complicated
4154 to ensure. 
4155 
4156 This patch takes the path of not assuming the invariant, which
4157 is simpler and more robust.  See
4158 Note [Mismatched type lists and application decomposition]
4159]
4160[Fix Trac #3943: incorrect unused-variable warning
4161simonpj@microsoft.com**20100412151630
4162 Ignore-this: 52459f2b8b02c3cb120abe674dc9a060
4163 
4164 In fixing this I did the usual little bit of refactoring
4165]
4166[Convert boot and boot-pkgs to perl
4167Ian Lynagh <igloo@earth.li>**20100415143919
4168 This stops us having to worry about sh/sed/... portability.
4169]
4170[Use $(MAKE), not make, when recursively calling make
4171Ian Lynagh <igloo@earth.li>**20100415121453]
4172[Remove the ghc_ge_609 makefile variables
4173Ian Lynagh <igloo@earth.li>**20100412235658
4174 They are now guaranteed to be YES
4175]
4176[Increase the minimum version number required to 6.10 in configure.ac
4177Ian Lynagh <igloo@earth.li>**20100412235313]
4178[The bootstrapping compiler is now required to be > 609
4179Ian Lynagh <igloo@earth.li>**20100409161046]
4180[Handle IND_STATIC in isRetainer
4181Ian Lynagh <igloo@earth.li>**20100409104207
4182 IND_STATIC used to be an error, but at the moment it can happen
4183 as isAlive doesn't look through IND_STATIC as it ignores static
4184 closures. See trac #3956 for a program that hit this error.
4185]
4186[Add Data and Typeable instances to HsSyn
4187David Waern <david.waern@gmail.com>**20100330011020
4188 Ignore-this: c3f2717207b15539fea267c36b686e6a
4189 
4190 The instances (and deriving declarations) have been taken from the ghc-syb
4191 package.
4192]
4193[Fix for derefing ThreadRelocated TSOs in MVar operations
4194Simon Marlow <marlowsd@gmail.com>**20100407092824
4195 Ignore-this: 94dd7c68a6094eda667e2375921a8b78
4196]
4197[sanity check fix
4198Simon Marlow <marlowsd@gmail.com>**20100407092746
4199 Ignore-this: 9c18cd5f5393e5049015ca52e62a1269
4200]
4201[get the reg liveness right in the putMVar# heap check
4202Simon Marlow <marlowsd@gmail.com>**20100407092724
4203 Ignore-this: b1ba07a59ecfae00e9a1f8391741abc
4204]
4205[initialise the headers of MSG_BLACKHOLE objects properly
4206Simon Marlow <marlowsd@gmail.com>**20100407081712
4207 Ignore-this: 183dcd0ca6a395d08db2be12b02bdd79
4208]
4209[initialise the headers of MVAR_TSO_QUEUE objects properly
4210Simon Marlow <marlowsd@gmail.com>**20100407081514
4211 Ignore-this: 4b4a2f30cf2fb69ca4128c41744687bb
4212]
4213[undo debugging code
4214Simon Marlow <marlowsd@gmail.com>**20100406142740
4215 Ignore-this: 323c2248f817b6717c19180482fc4b00
4216]
4217[putMVar#: fix reg liveness in the heap check
4218Simon Marlow <marlowsd@gmail.com>**20100406135832
4219 Ignore-this: cddd2c7807ac7612c9b2c4c0d384d284
4220]
4221[account for the new BLACKHOLEs in the GHCi debugger
4222Simon Marlow <marlowsd@gmail.com>**20100406133406
4223 Ignore-this: 4d4aeb4bbada3f50dc1fb0123f565e8f
4224]
4225[don't forget to deRefTSO() in tryWakeupThread()
4226Simon Marlow <marlowsd@gmail.com>**20100406130411
4227 Ignore-this: 171d57c4f8653835dec0b69f9be9881c
4228]
4229[Fix bug in popRunQueue
4230Simon Marlow <marlowsd@gmail.com>**20100406091453
4231 Ignore-this: 9d3cec8f18f5c5cbd51751797386eb6f
4232]
4233[fix bug in migrateThread()
4234Simon Marlow <marlowsd@gmail.com>**20100401105840
4235 Ignore-this: 299bcf0d1ea0f8865f3e845eb93d2ad3
4236]
4237[Remove the IND_OLDGEN and IND_OLDGEN_PERM closure types
4238Simon Marlow <marlowsd@gmail.com>**20100401093519
4239 Ignore-this: 95f2480c8a45139835eaf5610217780b
4240 These are no longer used: once upon a time they used to have different
4241 layout from IND and IND_PERM respectively, but that is no longer the
4242 case since we changed the remembered set to be an array of addresses
4243 instead of a linked list of closures.
4244]
4245[Change the representation of the MVar blocked queue
4246Simon Marlow <marlowsd@gmail.com>**20100401091605
4247 Ignore-this: 20a35bfabacef2674df362905d7834fa
4248 
4249 The list of threads blocked on an MVar is now represented as a list of
4250 separately allocated objects rather than being linked through the TSOs
4251 themselves.  This lets us remove a TSO from the list in O(1) time
4252 rather than O(n) time, by marking the list object.  Removing this
4253 linear component fixes some pathalogical performance cases where many
4254 threads were blocked on an MVar and became unreachable simultaneously
4255 (nofib/smp/threads007), or when sending an asynchronous exception to a
4256 TSO in a long list of thread blocked on an MVar.
4257 
4258 MVar performance has actually improved by a few percent as a result of
4259 this change, slightly to my surprise.
4260 
4261 This is the final cleanup in the sequence, which let me remove the old
4262 way of waking up threads (unblockOne(), MSG_WAKEUP) in favour of the
4263 new way (tryWakeupThread and MSG_TRY_WAKEUP, which is idempotent).  It
4264 is now the case that only the Capability that owns a TSO may modify
4265 its state (well, almost), and this simplifies various things.  More of
4266 the RTS is based on message-passing between Capabilities now.
4267]
4268[eliminate some duplication with a bit of CPP
4269Simon Marlow <marlowsd@gmail.com>**20100330154355
4270 Ignore-this: 838f7d341f096ca14c86ab9c81193e36
4271]
4272[Make ioManagerDie() idempotent
4273Simon Marlow <marlowsd@gmail.com>**20100401100705
4274 Ignore-this: a5996b43cdb2e2d72e6e971d7ea925fb
4275 Avoids screeds of "event buffer overflowed; event dropped" in
4276 conc059(threaded1).
4277]
4278[Move a thread to the front of the run queue when another thread blocks on it
4279Simon Marlow <marlowsd@gmail.com>**20100329144521
4280 Ignore-this: c518ff0d41154680edc811d891826a29
4281 This fixes #3838, and was made possible by the new BLACKHOLE
4282 infrastructure.  To allow reording of the run queue I had to make it
4283 doubly-linked, which entails some extra trickiness with regard to
4284 GC write barriers and suchlike.
4285]
4286[remove non-existent MUT_CONS symbols
4287Simon Marlow <marlowsd@gmail.com>**20100330152600
4288 Ignore-this: 885628257a9d03f2ece2a754d993014a
4289]
4290[change throwTo to use tryWakeupThread rather than unblockOne
4291Simon Marlow <marlowsd@gmail.com>**20100329144613
4292 Ignore-this: 10ad4965e6c940db71253f1c72218bbb
4293]
4294[tiny GC optimisation
4295Simon Marlow <marlowsd@gmail.com>**20100329144551
4296 Ignore-this: 9e095b9b73fff0aae726f9937846ba92
4297]
4298[New implementation of BLACKHOLEs
4299Simon Marlow <marlowsd@gmail.com>**20100329144456
4300 Ignore-this: 96cd26793b4e6ab9ddd0d59aae5c2f1d
4301 
4302 This replaces the global blackhole_queue with a clever scheme that
4303 enables us to queue up blocked threads on the closure that they are
4304 blocked on, while still avoiding atomic instructions in the common
4305 case.
4306 
4307 Advantages:
4308 
4309  - gets rid of a locked global data structure and some tricky GC code
4310    (replacing it with some per-thread data structures and different
4311    tricky GC code :)
4312 
4313  - wakeups are more prompt: parallel/concurrent performance should
4314    benefit.  I haven't seen anything dramatic in the parallel
4315    benchmarks so far, but a couple of threading benchmarks do improve
4316    a bit.
4317 
4318  - waking up a thread blocked on a blackhole is now O(1) (e.g. if
4319    it is the target of throwTo).
4320 
4321  - less sharing and better separation of Capabilities: communication
4322    is done with messages, the data structures are strictly owned by a
4323    Capability and cannot be modified except by sending messages.
4324 
4325  - this change will utlimately enable us to do more intelligent
4326    scheduling when threads block on each other.  This is what started
4327    off the whole thing, but it isn't done yet (#3838).
4328 
4329 I'll be documenting all this on the wiki in due course.
4330 
4331]
4332[Fix warnings (allow pushOnRunQueue() to not be inlined)
4333Simon Marlow <marlowsd@gmail.com>**20100401114559
4334 Ignore-this: f40bfbfad70a5165a946d11371605b7d
4335]
4336[remove out of date comment
4337Simon Marlow <marlowsd@gmail.com>**20100401105853
4338 Ignore-this: 26af88dd418ee0bcda7223b3b7e4e8d2
4339]
4340[tidy up spacing in stderr traces
4341Simon Marlow <marlowsd@gmail.com>**20100326163122
4342 Ignore-this: 16558b0433a274be217d4bf39aa4946
4343]
4344[Fix an assertion that was not safe when running in parallel
4345Simon Marlow <marlowsd@gmail.com>**20100325143656
4346 Ignore-this: cad08fb8900eb3a475547af0189fcc47
4347]
4348[Never jump directly to a thunk's entry code, even if it is single-entry
4349Simon Marlow <marlowsd@gmail.com>**20100325114847
4350 Ignore-this: 938da172c06a97762ef605c8fccfedf1
4351 I don't think this fixes any bugs as we don't have single-entry thunks
4352 at the moment, but it could cause problems for parallel execution if
4353 we ever did re-introduce update avoidance.
4354]
4355[Rename forgotten -dverbose-simpl to -dverbose-core2core in the docs.
4356Milan Straka <fox@ucw.cz>**20100331153626
4357 Ignore-this: 2da58477fb96e1cfb80f37dddd7c422c
4358]
4359[Add -pa and -V to the documentation of time profiling options.
4360Milan Straka <fox@ucw.cz>**20100329191121
4361 Ignore-this: be74d216481ec5a19e5f40f85e6e3d65
4362]
4363[Keep gcc 4.5 happy
4364Simon Marlow <marlowsd@gmail.com>**20100330120425
4365 Ignore-this: 7811878cc2bd1ce9cfbb5bf102fe3454
4366]
4367[Fix warning compiling Linker.c for PPC Mac
4368naur@post11.tele.dk**20100403182355
4369 Ignore-this: e2d2448770c9714ce17dd6cf3e297063
4370 The warning message eliminated is:
4371 > rts/Linker.c:4756:0:
4372 >      warning: nested extern declaration of 'symbolsWithoutUnderscore'
4373]
4374[Fix error compiling AsmCodeGen.lhs for PPC Mac (mkRtsCodeLabel)
4375naur@post11.tele.dk**20100403181656
4376 Ignore-this: deb7524ea7852a15a2ac0849c8c82f74
4377 The error messages eliminated are:
4378 > compiler/nativeGen/AsmCodeGen.lhs:875:31:
4379 >     Not in scope: `mkRtsCodeLabel'
4380 > compiler/nativeGen/AsmCodeGen.lhs:879:31:
4381 >     Not in scope: `mkRtsCodeLabel'
4382 > compiler/nativeGen/AsmCodeGen.lhs:883:31:
4383 >     Not in scope: `mkRtsCodeLabel'
4384]
4385[Fix error compiling AsmCodeGen.lhs for PPC Mac (DestBlockId)
4386naur@post11.tele.dk**20100403180643
4387 Ignore-this: 71e833e94ed8371b2ffabc2cf80bf585
4388 The error message eliminated is:
4389 > compiler/nativeGen/AsmCodeGen.lhs:637:16:
4390 >     Not in scope: data constructor `DestBlockId'
4391]
4392[Fix boot-pkgs's sed usage to work with Solaris's sed
4393Ian Lynagh <igloo@earth.li>**20100401153441]
4394[Pass "-i org.haskell.GHC" to packagemaker when building the OS X installer
4395Ian Lynagh <igloo@earth.li>**20100331144707
4396 This seems to fix this failure:
4397 [...]
4398 ** BUILD SUCCEEDED **
4399 rm -f -f GHC-system.pmdoc/*-contents.xml
4400 /Developer/usr/bin/packagemaker -v --doc GHC-system.pmdoc\
4401              -o /Users/ian/to_release/ghc-6.12.1.20100330/GHC-6.12.1.20100330-i386.pkg
4402 2010-03-31 15:08:15.695 packagemaker[13909:807] Setting to : 0 (null)
4403 2010-03-31 15:08:15.709 packagemaker[13909:807] Setting to : 0 org.haskell.glasgowHaskellCompiler.ghc.pkg
4404 2010-03-31 15:08:15.739 packagemaker[13909:807] relocate: (null) 0
4405 2010-03-31 15:08:15.740 packagemaker[13909:807] *** Terminating app due to uncaught exception 'NSInvalidArgumentException', reason: '*** -[NSXMLDocument initWithXMLString:options:error:]: nil argument'
4406 2010-03-31 15:08:15.741 packagemaker[13909:807] Stack: (
4407     2511962091,
4408     2447007291,
4409     2511961547,
4410     2511961610,
4411     2432803204,
4412     453371,
4413     447720,
4414     436209,
4415     435510,
4416     9986,
4417     9918
4418 )
4419 make[1]: *** [framework-pkg] Trace/BPT trap
4420 make: *** [framework-pkg] Error 2
4421]
4422[Use machdepCCOpts when compiling the file to toggle -(no-)rtsopts
4423Ian Lynagh <igloo@earth.li>**20100331161302
4424 Should fix toggling on OS X "Snow Leopard". Diagnosed by Roman Leshchinskiy.
4425]
4426[Avoid a non-portable use of tar reported by Roman Leshchinskiy
4427Ian Lynagh <igloo@earth.li>**20100330145802]
4428[Don't install EXTRA_PACKAGES by default
4429Simon Marlow <marlowsd@gmail.com>**20100330142714
4430 Ignore-this: d4cc8f87a6de8d9d1d6dc9b77130b3
4431]
4432[fix a non-portable printf format
4433Simon Marlow <marlowsd@gmail.com>**20100330134437
4434 Ignore-this: d41c23c54ec29654cb2049de1e588570
4435]
4436[avoid single quote in #error
4437Simon Marlow <marlowsd@gmail.com>**20100330120346
4438 Ignore-this: 663f39e7a27fead2f648fbf22d345bb4
4439]
4440[use FMT_Word64 instead of locally-defined version
4441Simon Marlow <marlowsd@gmail.com>**20100330114650
4442 Ignore-this: 82697b8095dffb3a8e196c687006ece0
4443]
4444[remove old/unused DotnetSupport and GhcLibsWithUnix
4445Simon Marlow <marlowsd@gmail.com>**20100330123732
4446 Ignore-this: c68814868b3671abdc369105bbeafe6c
4447]
4448[fix return type cast in f.i.wrapper when using libffi (#3516)
4449Simon Marlow <marlowsd@gmail.com>**20100329154220
4450 Ignore-this: f898eb8c9ae2ca2009e539735b92c438
4451 
4452 Original fix submitted by
4453   Sergei Trofimovich <slyfox@community.haskell.org>
4454 modified by me:
4455  - exclude 64-bit types
4456  - compare uniques, not strings
4457  - #include "ffi.h" is conditional
4458]
4459[libffi: install 'ffitarget.h' header as sole 'ffi.h' is unusable
4460Simon Marlow <marlowsd@gmail.com>**20100329135734
4461 Ignore-this: f9b555ea289d8df1aa22cb6faa219a39
4462 Submitted by: Sergei Trofimovich <slyfox@community.haskell.org>
4463 Re-recorded against HEAD.
4464]
4465[avoid a fork deadlock (see comments)
4466Simon Marlow <marlowsd@gmail.com>**20100329132329
4467 Ignore-this: 3377f88b83bb3b21e42d7fc5f0d866f
4468]
4469[tidy up the end of the all_tasks list after forking
4470Simon Marlow <marlowsd@gmail.com>**20100329132253
4471 Ignore-this: 819d679875be5f344e816210274d1c29
4472]
4473[Add a 'setKeepCAFs' external function (#3900)
4474Simon Marlow <marlowsd@gmail.com>**20100329110036
4475 Ignore-this: ec532a18cad4259a09847b0b9ae2e1d2
4476]
4477[Explicitly check whether ar supports the @file syntax
4478Ian Lynagh <igloo@earth.li>**20100329123325
4479 rather than assuming that all GNU ar's do.
4480 Apparently OpenBSD's older version doesn't.
4481]
4482[Fix the format specifier for Int64/Word64 on Windows
4483Ian Lynagh <igloo@earth.li>**20100327182126
4484 mingw doesn't understand %llu/%lld - it treats them as 32-bit rather
4485 than 64-bit. We use %I64u/%I64d instead.
4486]
4487[Fix the ghci startmenu item
4488Ian Lynagh <igloo@earth.li>**20100326235934
4489 I'm not sure what changed, but it now doesn't work for me without
4490 the "Start in" field being set.
4491]
4492[Fix paths to docs in "Start Menu" entries in Windows installer; fixes #3847
4493Ian Lynagh <igloo@earth.li>**20100326155917]
4494[Add a licence file for the Windows installer to use
4495Ian Lynagh <igloo@earth.li>**20100326155130]
4496[Add gcc-g++ to the inplace mingw installation; fixes #3893
4497Ian Lynagh <igloo@earth.li>**20100326154714]
4498[Add the licence file to the Windows installer. Fixes #3934
4499Ian Lynagh <igloo@earth.li>**20100326152449]
4500[Quote the paths to alex and happy in configure
4501Ian Lynagh <igloo@earth.li>**20100325143449
4502 Ignore-this: d6d6e1a250f88985bbeea760e63a79db
4503]
4504[Use </> rather than ++ "/"
4505Ian Lynagh <igloo@earth.li>**20100325133237
4506 This stops us generating paths like
4507     c:\foo\/ghc460_0/ghc460_0.o
4508 which windres doesn't understand.
4509]
4510[Append $(exeext) to utils/ghc-pkg_dist_PROG
4511Ian Lynagh <igloo@earth.li>**20100324233447
4512 Fixes bindist creation
4513]
4514[A sanity check
4515Simon Marlow <marlowsd@gmail.com>**20100325110500
4516 Ignore-this: 3b3b76d898c822456857e506b7531e65
4517]
4518[do_checks: do not set HpAlloc if the stack check fails
4519Simon Marlow <marlowsd@gmail.com>**20100325110328
4520 Ignore-this: 899ac8c29ca975d03952dbf4608d758
4521 
4522 This fixes a very rare heap corruption bug, whereby
4523 
4524  - a context switch is requested, which sets HpLim to zero
4525    (contextSwitchCapability(), called by the timer signal or
4526    another Capability).
4527 
4528  - simultaneously a stack check fails, in a code fragment that has
4529    both a stack and a heap check.
4530 
4531 The RTS then assumes that a heap-check failure has occurred and
4532 subtracts HpAlloc from Hp, although in fact it was a stack-check
4533 failure and retreating Hp will overwrite valid heap objects.  The bug
4534 is that HpAlloc should only be set when Hp has been incremented by the
4535 heap check.  See comments in rts/HeapStackCheck.cmm for more details.
4536 
4537 This bug is probably incredibly rare in practice, but I happened to be
4538 working on a test that triggers it reliably:
4539 concurrent/should_run/throwto001, compiled with -O -threaded, args 30
4540 300 +RTS -N2, run repeatedly in a loop.
4541]
4542[comments and formatting only
4543Simon Marlow <marlowsd@gmail.com>**20100325104617
4544 Ignore-this: c0a211e15b5953bb4a84771bcddd1d06
4545]
4546[Change how perl scripts get installed; partially fixes #3863
4547Ian Lynagh <igloo@earth.li>**20100324171422
4548 We now regenerate them when installing, which means the path for perl
4549 doesn't get baked in
4550]
4551[Pass the location of gcc in the ghc wrapper script; partially fixes #3863
4552Ian Lynagh <igloo@earth.li>**20100324171408
4553 This means we don't rely on baking a path to gcc into the executable
4554]
4555[Quote the ar path in configure
4556Ian Lynagh <igloo@earth.li>**20100324162043]
4557[Remove unused cUSER_WAY_NAMES cUSER_WAY_OPTS
4558Ian Lynagh <igloo@earth.li>**20100324145048]
4559[Remove unused cCONTEXT_DIFF
4560Ian Lynagh <igloo@earth.li>**20100324145013]
4561[Remove unused cEnableWin32DLLs
4562Ian Lynagh <igloo@earth.li>**20100324144841]
4563[Remove unused cGHC_CP
4564Ian Lynagh <igloo@earth.li>**20100324144656]
4565[Fix the build for non-GNU-ar
4566Ian Lynagh <igloo@earth.li>**20100324132907]
4567[Tweak the Makefile code for making .a libs; fixes trac #3642
4568Ian Lynagh <igloo@earth.li>**20100323221325
4569 The main change is that, rather than using "xargs ar" we now put
4570 all the filenames into a file, and do "ar @file". This means that
4571 ar adds all the files at once, which works around a problem where
4572 files with the same basename in a later invocation were overwriting
4573 the existing file in the .a archive.
4574]
4575[Enable shared libraries on Windows; fixes trac #3879
4576Ian Lynagh <igloo@earth.li>**20100320231414
4577 Ignore-this: c93b35ec5b7a7fa6ddb286d17a616216
4578]
4579[Add the external core PDF to the new build system
4580Ian Lynagh <igloo@earth.li>**20100321161909]
4581[Allow specifying $threads directly when validating
4582Ian Lynagh <igloo@earth.li>**20100321112835]
4583[Remove LazyUniqFM; fixes trac #3880
4584Ian Lynagh <igloo@earth.li>**20100320213837]
4585[UNDO: slight improvement to scavenging ...
4586Simon Marlow <marlowsd@gmail.com>**20100319153413
4587 Ignore-this: f0ab581c07361f7b57eae02dd6ec893c
4588 
4589 Accidnetally pushed this patch which, while it validates, isn't
4590 correct.
4591 
4592 rolling back:
4593 
4594 Fri Mar 19 11:21:27 GMT 2010  Simon Marlow <marlowsd@gmail.com>
4595   * slight improvement to scavenging of update frames when a collision has occurred
4596 
4597     M ./rts/sm/Scav.c -19 +15
4598]
4599[slight improvement to scavenging of update frames when a collision has occurred
4600Simon Marlow <marlowsd@gmail.com>**20100319112127
4601 Ignore-this: 6de2bb9614978975f17764a0f259d9bf
4602]
4603[Don't install the utf8-string package
4604Ian Lynagh <igloo@earth.li>**20100317212709]
4605[Don't use -Bsymbolic when linking the RTS
4606Ian Lynagh <igloo@earth.li>**20100316233357
4607 This makes the RTS hooks work when doing dynamic linking
4608]
4609[Fix Trac #3920: Template Haskell kinds
4610simonpj@microsoft.com**20100317123519
4611 Ignore-this: 426cac7920446e04f3cc30bd1d9f76e2
4612 
4613 Fix two places where we were doing foldl instead of foldr
4614 after decomposing a Kind.  Strange that the same bug appears
4615 in two quite different places!
4616]
4617[copy_tag_nolock(): fix write ordering and add a write_barrier()
4618Simon Marlow <marlowsd@gmail.com>**20100316143103
4619 Ignore-this: ab7ca42904f59a0381ca24f3eb38d314
4620 
4621 Fixes a rare crash in the parallel GC.
4622 
4623 If we copy a closure non-atomically during GC, as we do for all
4624 immutable values, then before writing the forwarding pointer we better
4625 make sure that the closure itself is visible to other threads that
4626 might follow the forwarding pointer.  I imagine this doesn't happen
4627 very often, but I just found one case of it: in scavenge_stack, the
4628 RET_FUN case, after evacuating ret_fun->fun we then follow it and look
4629 up the info pointer.
4630]
4631[Add sliceP mapping to vectoriser builtins
4632benl@ouroborus.net**20100316060517
4633 Ignore-this: 54c3cafff584006b6fbfd98124330aa3
4634]
4635[Comments only
4636benl@ouroborus.net**20100311064518
4637 Ignore-this: d7dc718cc437d62aa5b1b673059a9b22
4638]
4639[TAG 2010-03-16
4640Ian Lynagh <igloo@earth.li>**20100316005137
4641 Ignore-this: 234e3bc29e2f26cc59d7b03d780cc352
4642]
4643Patch bundle hash:
464440b4eb7782c9771b705e2738abb70ecf26d31e1f
4645-----BEGIN PGP SIGNATURE-----
4646Version: GnuPG v1.4.10 (Darwin)
4647
4648iD8DBQFMuwIJFOecpxqG73IRArxKAJ4nBxAXYBoF1LqasHbsbJqvJTxYrACgj6rx
4649nfHy4rBz36eIQ5uVRzvnmvw=
4650=T9KU
4651-----END PGP SIGNATURE-----