Ticket #5034: improve-performance-of-digraph_preorderf-_ticket-_5034__.dpatch

File improve-performance-of-digraph_preorderf-_ticket-_5034__.dpatch, 99.5 KB (added by michalt, 3 years ago)

Changes the preorder and preorderF in Digraph module.

Line 
11 patch for repository http://darcs.haskell.org/ghc:
2
3Mon Mar 21 20:03:37 CET 2011  Michal Terepeta <michal.terepeta@gmail.com>
4  * Improve performance of Digraph.preorderF (ticket #5034).
5
6New patches:
7
8[Improve performance of Digraph.preorderF (ticket #5034).
9Michal Terepeta <michal.terepeta@gmail.com>**20110321190337
10 Ignore-this: 2d0cde2f1cdecd1e791a5916b76c4ba0
11] {
12hunk ./compiler/utils/Digraph.lhs 434
13 ------------------------------------------------------------
14 
15 \begin{code}
16-preorder            :: Tree a -> [a]
17-preorder (Node a ts) = a : preorderF ts
18+preorder' :: Tree a -> [a] -> [a]
19+preorder' (Node a ts) = (a :) . preorderF' ts
20 
21hunk ./compiler/utils/Digraph.lhs 437
22-preorderF           :: Forest a -> [a]
23-preorderF ts         = concat (map preorder ts)
24+preorderF' :: Forest a -> [a] -> [a]
25+preorderF' ts = foldr (.) id $ map preorder' ts
26+
27+preorderF :: Forest a -> [a]
28+preorderF ts = preorderF' ts []
29 
30 tabulate        :: Bounds -> [Vertex] -> Table Int
31 tabulate bnds vs = array bnds (zip vs [1..])
32}
33
34Context:
35
36[Add unboxedTupleTName to templateHaskellNames
37Ian Lynagh <igloo@earth.li>**20110317003406]
38[bindist comparison tool: Improve way-difference behaviour
39Ian Lynagh <igloo@earth.li>**20110316221338]
40[Bindist comparison tool: add --ignore-size-changes flag
41Ian Lynagh <igloo@earth.li>**20110316220748]
42[Bindist comparison tool: Handle differences in the library ways nicely
43Ian Lynagh <igloo@earth.li>**20110316214708
44 In particular, this makes it possible to compare release bindists (with
45 profiling files) and validate bindists (without them).
46]
47[Don't put includes/rts/Config.h in bindists twice
48Ian Lynagh <igloo@earth.li>**20110316190117
49 Ignore-this: b2d37c44557a84ae87675e795de209c
50]
51[Don't put 2 copies of the RTS libraries in the bindists
52Ian Lynagh <igloo@earth.li>**20110316172945]
53[Make include file paths a little prettier
54Ian Lynagh <igloo@earth.li>**20110316162045
55 This isn't important, but it stops us getting [...]/./[...] in the paths
56 in bindists.
57]
58[Initial implementation of bindist comparison tool
59Ian Lynagh <igloo@earth.li>**20110315162912
60 Ignore-this: c92609bde275ac3a61648aa26da97e12
61]
62[Stop explicitly asking for 10.5 support on OS X; fixes #5011
63Ian Lynagh <igloo@earth.li>**20110313140322
64 XCode 4 doesn't include the 10.5 SDK, so if we explicitly ask for it
65 then linking fails.
66]
67[Write the XCode version test differently
68Ian Lynagh <igloo@earth.li>**20110313134913
69 Using && causes problems on opensolaris, for an unknown reason.
70 http://www.haskell.org/pipermail/cvs-ghc/2011-March/060314.html
71]
72[Consistently use <sect1> etc rather than <section>; fixes #5009
73Ian Lynagh <igloo@earth.li>**20110311212958
74 It might be nicer to actually go the other way, and make everything
75 use <section> instead, but this fixes the incorrect numbering for
76 now, and we can look into whether <section> has any disadvantages
77 later.
78]
79[Drop dead core that was kept alive by RULES in CorePrep (#4962)
80Max Bolingbroke <batterseapower@hotmail.com>**20110219114726
81 Ignore-this: d1c8f1e4aa941911d21081932e14ec5c
82]
83[Optimise comparisons against min/maxBound (ticket #3744).
84Michal Terepeta <michal.terepeta@gmail.com>**20101027184054
85 Ignore-this: aacf39882c09784f04b7013de025f038
86 
87 This optimises away comparisons with minBound or maxBound
88 that are always false or always true.
89]
90[LLVM: Fix #4995, llvm mangler broken for large compiles
91David Terei <davidterei@gmail.com>**20110309215346
92 Ignore-this: 4fcef11ad0be0cff26c108bd05db0485
93]
94[Improve the XCode version detection
95Ian Lynagh <igloo@earth.li>**20110307225823
96 Amongst other improvements, we now handle 3-component versions
97 (like "3.1.4") correctly.
98]
99[Impredicative polymorphism no longer deprecated.
100Edward Z. Yang <ezyang@mit.edu>**20110306215916
101 Ignore-this: 49157eaae716879effb99c2638518265
102]
103[Fix doc formatting
104Ian Lynagh <igloo@earth.li>**20110306151532
105 Ignore-this: 4948eccebbf7637ff84db661a5138dd1
106]
107[configure: amend sanity check
108Sergei Trofimovich <slyfox@community.haskell.org>**20110301194018
109 Ignore-this: 4f40f8dd28bd436f02b487714b797ee7
110 
111 As we perform some mangling of original --build/--host/--target
112 params we should check the result of mangling against desired triplet,
113 not originally passed by user.
114 
115 Patch also adds mangled triplets to the output.
116]
117[configure: triplet: accept i486 CPU (and more), more vendors and OSes
118Sergei Trofimovich <slyfox@community.haskell.org>**20110301064108
119 Ignore-this: d508348bf04ad74afb069b8c743fd2f4
120 
121 Gentoo has interesting ports:
122 
123   - freebsd with following triplet:
124         ./configure --build=i686-gentoo-freebsd8 --host=i686-gentoo-freebsd8 --host=i686-gentoo-freebsd8
125     should be recognized as 'i386-unknown-freebsd'
126   - 'pc' vendor along with non-'i386' is very common:
127         ./configure --build=i486-pc-linux-gnu --i486-pc-linux-gnu --host=i486-pc-linux-gnu
128         ./configure --build=i686-pc-linux-gnu --i686-pc-linux-gnu --host=i686-pc-linux-gnu
129     should be recognized as 'i386-unknown-linux'
130 
131 Patch adds:
132 
133   - 'pc' vendor                 (maps to 'unknown')
134   - 'gentoo' vendor             (maps to 'unknown')
135   - 'i486', 'i586', 'i686' CPUs (maps to 'i386')
136   - 'freebsd8' OS               (maps to 'freebsd')
137]
138[Document -fwarn-missing-import-lists
139simonpj@microsoft.com**20110306000216
140 Ignore-this: 5a857935a775f608d8e00af9ad2b1633
141 
142 Also change the behaviour slightly, to warn only for
143 *unqualified* imports.  See Trac #4977.
144]
145[Completed the implementation of VECTORISE SCALAR
146Manuel M T Chakravarty <chak@cse.unsw.edu.au>**20110305123625
147 Ignore-this: 96714958893800c6ae750fb343b32e31
148 - The pragma {-# VECTORISE SCALAR foo #-} marks 'foo' as a
149   scalar function for for vectorisation and generates a
150   vectorised version by applying 'scalar_map' and friends.
151 - The set of scalar functions is not yet emitted into
152   interface files.  This will be added in a subsequent
153   patch via 'VectInfo'.
154]
155[Fix build, and improve ghc-cabal's errors
156Ian Lynagh <igloo@earth.li>**20110305153931
157 Ignore-this: eb0d1b7ee3dbc7be6a39c9ed3a787d
158]
159[Whitespace only
160Ian Lynagh <igloo@earth.li>**20110305150238]
161[Sanity check values that we wrap in single quotes
162Ian Lynagh <igloo@earth.li>**20110305150118
163 Make sure they contain no single quotes, leading spaces, or trailing spaces.
164]
165[Avoid some shell calls in the build system
166Ian Lynagh <igloo@earth.li>**20110305144825
167 The DEP_INCLUDE_DIRS and DEP_LIB_DIRS variables always contain
168 single-quote dirs, so we can use e.g.
169     $(subst $(space)',$(space)-L',$(space)$($1_$2_DEP_LIB_DIRS_SINGLE_QUOTED))
170 to add -L to the front of each dir. I've appended "_SINGLE_QUOTED" to
171 the variable names so we don't accidentally add bare directories to
172 them.
173]
174[Make -fno-enable-rewrite-rules work properly
175simonpj@microsoft.com**20110302114317
176 Ignore-this: cd133880e58d9af55cf4454e4c700d
177 
178 I'd failed to propagate the Opt_EnableRewriteRules flag properly,
179 which meant that -fno-enable-rewrite-rules didn't disable all
180 rewrites.  This patch fixes it.
181]
182[Comments only
183simonpj@microsoft.com**20110302113041
184 Ignore-this: e97a3ee359e0211b4fb5b5fbf3f5607d
185]
186[Restore SPECIALISE INSTANCE pragmas
187simonpj@microsoft.com**20110222173112
188 Ignore-this: 4c0a8fab4f9e619d988beb8b5269474f
189 
190 For some reason I'd commented out this code, which
191 meant that we weren't getting the the goodness of
192 the SPECIALISE INSTANCE pragamas in GHC.Real. And
193 that in turn killed performance in nofib 'power'
194]
195[GHC.Prim.threadStatus# now returns the cap number, and the value of TSO_LOCKED
196Simon Marlow <marlowsd@gmail.com>**20110301101025
197 Ignore-this: 49548cffd04bf6fdd3071158d9865034
198]
199[Improve GHCi line numbers in errors
200Ian Lynagh <igloo@earth.li>**20110227172012
201 Ignore-this: 52c2c2f7c83207e8c7b36fab6ebe52d8
202 When running commands from the user (as opposed to from a file), reset
203 the line number to 1 at the start of each command.
204]
205[:script file scripts in GHCi #1363
206Vivian McPhail <haskell.vivian.mcphail@gmail.com>**20110226073133
207 This patch adds the script command in GHCi
208 
209 A file is read and executed as a series of GHCi commands.
210 
211 Execution terminates on the first error.  The filename and
212 line number are included in the error.
213]
214[Turn off split objects on Darwin if XCode < 3.2 (#4013)
215Ian Lynagh <igloo@earth.li>**20110225184358
216 Ignore-this: 3e894673649fee957df2413ca82b1c99
217]
218[libffi: backport incorrect detection of selinux
219Sergei Trofimovich <slyfox@community.haskell.org>**20110208212140
220 Ignore-this: 31bdddb5683acae2bffa3e97effaefea
221 
222 This patch unbreaks ghci on GRSEC kernels hardened with
223 TPE (Trusted Path Execution) protection.
224 
225 TPE forbids mmap('rwx') files opened for writes:
226     fd = open (a_file_in_tmp, O_RDWR);
227     mmap (...,               PROT_READ | PROT_WRITE | PROT_EXEC, fd);
228 
229 while allows anonymous RWX mappings:
230     mmap (...MAP_ANONYMOUS , PROT_READ | PROT_WRITE | PROT_EXEC, -1);
231 
232 Thanks to klondike for finding it out.
233 
234 The result of a horrible typo.
235 
236 (unreleased yet) upstream also has the patch:
237 
238 http://github.com/atgreen/libffi/commit/eaf444eabc4c78703c0f98ac0197b1619c1b1bef
239]
240[Generate the OS X installer from a bindist, rather than from a source tree
241Ian Lynagh <igloo@earth.li>**20110223170016]
242[Use -h rather than -soname; fixes dynlibs on Solaris 10; trac #4973
243Ian Lynagh <igloo@earth.li>**20110222152656]
244[Fix another fundep error (fixes Trac #4969)
245simonpj@microsoft.com**20110221153239
246 Ignore-this: 9308a7c55346c9eaeaa0818903c8d67b
247 
248 If I had a pound for every hour Dimitrios and I have spent
249 making functional dependencies work right, we'd be rich!
250 
251 We had stupidly caused a 'wanted' to be rewritten by a 'derived', with
252 resulting abject failure.  As well as fixing the bug, this patch
253 refactors some more, adds useful assert and comments.
254]
255[New codegen: GC calling convention must use registers.
256Edward Z. Yang <ezyang@mit.edu>**20110218011745
257 Ignore-this: 13cdd38bb5cea0d244861f01f0004dee
258 
259 Previously, on register-deficient architectures like x86-32,
260 the new code generator would emit code for calls to stg_gc_l1,
261 stg_gc_d1 and stg_gc_f1 that pushed their single argument on
262 to the stack, while the functions themselves expected the
263 argument to live in L1, D1 and F1 (respectively).  This was
264 because cmmCall with the GC calling convention allocated real
265 registers, not virtual registers.
266 
267 This patch modifies the code for assigning registers/stack slots
268 to use the right calling convention for GC and adds an assertion
269 to ensure it did it properly.
270]
271[Added a VECTORISE pragma
272Manuel M T Chakravarty <chak@cse.unsw.edu.au>**20110220105032
273 Ignore-this: e3ed0e7caa45418452858a5a5937264
274 - Added a pragma {-# VECTORISE var = exp #-} that prevents
275   the vectoriser from vectorising the definition of 'var'.
276   Instead it uses the binding '$v_var = exp' to vectorise
277   'var'.  The vectoriser checks that the Core type of 'exp'
278   matches the vectorised Core type of 'var'.  (It would be
279   quite complicated to perform that check in the type checker
280   as the vectorisation of a type needs the state of the VM
281   monad.)
282 - Added parts of a related VECTORISE SCALAR pragma
283 - Documented -ddump-vect
284 - Added -ddump-vt-trace
285 - Some clean up
286]
287[Reenable object splitting on Darwin, now #4013 appears to be fixed
288Ian Lynagh <igloo@earth.li>**20110219191409
289 Ignore-this: e39e46beab8c8f2dc3369d5ec98610f2
290]
291[Clarify the "object splitting" variable names
292Ian Lynagh <igloo@earth.li>**20110219161508]
293[Fix #4867 (updated; corrects address calculation)
294gwright@antiope.com**20110214150924
295 Ignore-this: b68bcf1262f3507e1b9a7459139054bc
296 
297 This is a corrected fix for ticket #4867, "ghci displays negative floats
298 incorrectly".  The previous patch sometimes gave incorrect offset to values
299 in the __const section of the __TEXT segment.  The new patch arranges a zero
300 fixup for non-external, not-global offset table signed relocations.  This
301 is apparently what is required, though documentation on this point is scarce.
302 
303 With this change Doubles are negated properly, because the sign bit mask
304 is loaded from the correct offset.  This was tested both on HEAD and the 7.0
305 branch.
306 
307]
308[Scale columns in cost-centre-stack report to their contents
309Max Bolingbroke <batterseapower@hotmail.com>**20110216161340
310 Ignore-this: b7138527bb8c139ed554b9e6c9a9f37
311]
312[Change typechecker-trace output slightly
313simonpj@microsoft.com**20110217175904
314 Ignore-this: 4bf1c9dfa41c598765cd082a35bf193b
315]
316[Comments only
317simonpj@microsoft.com**20110217175840
318 Ignore-this: b9ea9687b41bfc062137a3ad6fafa3a6
319]
320[Fix Trac #4966
321simonpj@microsoft.com**20110217141000
322 Ignore-this: d4caaf910dfb86b1e469c5da3fe9cde2
323 
324 This is just a program that exploits overlapping
325 instances in a delicate way. The fix makes GHC
326 a bit more friendly towards such programs.
327 
328 See Note [Overlap and deriving] in TcSimplify
329]
330[Use "on the spot" solving for fundeps
331simonpj@microsoft.com**20110217140921
332 Ignore-this: 99290ef5538819053af708f5f9464488
333 
334 When we spot an equality arising from a functional dependency,
335 we now use that equality (a "wanted") to rewrite the work-item
336 constraint right away.  This avoids two dangers
337 
338  Danger 1: If we send the original constraint on down the pipeline
339            it may react with an instance declaration, and in delicate
340           situations (when a Given overlaps with an instance) that
341           may produce new insoluble goals: see Trac #4952
342 
343  Danger 2: If we don't rewrite the constraint, it may re-react
344            with the same thing later, and produce the same equality
345            again --> termination worries.
346 
347 To achieve this required some refactoring of FunDeps.lhs (nicer
348 now!). 
349 
350 This patch also contains a couple of unrelated improvements
351 
352 * A bad bug in TcSMonad.nestImplicTcS whereby the Tcs tyvars
353   of an outer implication were not untouchable inside
354 
355 * Improved logging machinery for the type constraint solver;
356   use -ddump-cs-trace (probably with a wider default line width
357   -dppr-cols=200 or something)
358]
359[Increase exprIsDupable threshold a bit
360simonpj@microsoft.com**20110215143921
361 Ignore-this: cf5a73026aabaa9f2ca93d0d338a6f56
362 
363 Now that exprIsDupable is less aggressive, test MethSharing wasn't
364 doing enough inlining.  Increasing the threshold fixes the problem
365 but the real fix is in Trac #4960.
366]
367[Ensure exprIsCheap/exprIsExpandable deal with Cast properly
368simonpj@microsoft.com**20110215143655
369 Ignore-this: 13bd5540b048de02737070bfa66f1a71
370 
371 This bug was causing a Lint error on the stable branch. For some
372 reason it doesn't show up in HEAD, but it's still worth fixing.
373 
374 The point is that ((f `cast` co) a) is cheap if f has arity>1.
375 This was being gratuitously missed before.
376]
377[Do not treat absentError specially
378simonpj@microsoft.com**20110215143504
379 Ignore-this: 3a8f6d6eae17b7360f58de86ec934cde
380 
381 (This was part of an experiment I abandoned.)
382]
383[Fix exprIsDupable
384simonpj@microsoft.com**20110214111512
385 Ignore-this: 2b800410f8103590da167ae24e7b242d
386 
387 It turns out that exprIsDupable would return True for an expression of
388 *arbitrary* size, provide it was a nested bunch of applications in
389 which no function had more than three arguments.  That was never the
390 intention, and could give rise to massive code duplication. 
391 
392 This patch makes it much less aggressive.
393]
394[Better case-of-case transformation
395simonpj@microsoft.com**20110214111151
396 Ignore-this: 32360d2c7703d772c7a5ad84b019bf87
397 
398 The ThunkSplitting idea in WorkWrap wasn't working at all,
399 leading to Trac #4957.  The culprit is really the simplifier
400 which was combining the wrong case continuations. See
401 Note [Fusing case continuations] in Simplify.
402]
403[Cleaned up Expr and Vectorise
404keller@.cse.unsw.edu.au**20110215014434
405 Ignore-this: 11b78040a7d16dc6e5ce7950c73bf8a9
406]
407[Fixed two syntax errors
408keller@.cse.unsw.edu.au**20110214020531
409 Ignore-this: 1abeebe0171a4262ceacee89adc9246e
410]
411[Handling of recursive scalar functions in isScalarLam
412keller@cse.unsw.edu.au**20110214002945
413 Ignore-this: db522152971fdcc9c74124d06ef785bd
414]
415[Added handling of non-recursive module global functions to isScalar check
416keller@cse.unsw.edu.au**20110209042855
417 Ignore-this: feae370a9bcc30371ab62f3700471739
418]
419[
420keller@cse.unsw.edu.au**20110202051408
421 Ignore-this: 7e0e1fcd40315da9abe80e2c2eda877f
422]
423[Removed minor bug
424keller@cse.unsw.edu.au**20110202040826
425 Ignore-this: 1abdae6a0f39b2ebd002f2587b5d1f53
426]
427[added handling of data constructors to vectLam
428keller@cse.unsw.edu.au**20110201042807
429 Ignore-this: 511fba94f926bb3a1088b21e36cc1f8c
430]
431[pruneSparkQueue: check for tagged pointers
432Simon Marlow <marlowsd@gmail.com>**20110214123858
433 Ignore-this: 719f77302b0dff3ba97b66a0d75b2482
434 This was a bug in 6.12.3.  I think the problem no longer occurs due to
435 the way sparks are treated as weak pointers, but it doesn't hurt to
436 test for tagged pointers anyway: better to do the test than have a
437 subtle invariant.
438]
439[Fix Trac #4953: local let binders can have IdInfo with free names
440simonpj@microsoft.com**20110214140334
441 Ignore-this: c8e5dbc6f0270e45d9cf2edee34d0354
442 
443 Local let binders in IfaceExpr never used to have unfoldings,
444 but lately they can (becuase they can have an INLINE pragma).
445 We must take account of the variables mentioned in the pragma
446 when computing the fingerprint.
447]
448[Comments only
449simonpj@microsoft.com**20110214090703
450 Ignore-this: bf950e91e0c984186fea1100aa15d00c
451]
452[LLVM: Huge improvement to mangler speed.
453David Terei <davidterei@gmail.com>**20110213014406
454 Ignore-this: 4eeaa572dfe956c990895154bd942bb2
455 
456 The old llvm mangler was horrible! Very slow
457 due to bad design and code. New version is
458 linear complexity as it should be and far
459 lower coefficients. This fixes trac 4838.
460]
461[Fix platform detection in bindists
462Ian Lynagh <igloo@earth.li>**20110211184244
463 In a bindist, we generate files like the hsc2hs wrapper.
464 This means we need to have the right values for the variables like
465 CONF_GCC_LINKER_OPTS_STAGE1 which in turn means we need to know what
466 platform we're on.
467]
468[New plan: push unsolved wanteds inwards
469simonpj@microsoft.com**20110211174058
470 Ignore-this: ed40762e260dab75b5e51c696f9965fa
471 
472 This fixes Trac #4935.  See Note [Preparing inert set for implications].
473 Lots of comments, but not a lot of code is changed!
474]
475[Remove unnecessary import, plus white space
476simonpj@microsoft.com**20110211173925
477 Ignore-this: 963f455c3c8aee49009b5d5a02f835ac
478]
479[Fix small but egregious error: using un-zonked constraints in simplifyRule
480simonpj@microsoft.com**20110211173835
481 Ignore-this: 238586e420dbbb1be7f6368117cf6280
482 
483 This resulted in double unifications.  Fix is trivial.
484]
485[makeSolvedByInst is only called on wanteds
486simonpj@microsoft.com**20110211173415
487 Ignore-this: 36e6201ab59a082e6dc38e56bea99e29
488 
489 This patch just adds an assert error.
490]
491[Enable DTrace on Solaris; based on a patch from Karel Gardas
492Ian Lynagh <igloo@earth.li>**20110210155217
493 Ignore-this: 93eaf0e06c721c80c175aaee9a113e6
494]
495[Use DTrace whenever it's available
496Ian Lynagh <igloo@earth.li>**20110210153300
497 Ignore-this: 111c72bf20d6eaafd3e488196a89b2c
498 Now that we've stopped trying to support 64bit OS X 10.5, the DTrace
499 problems there don't matter.
500]
501[replace C++ comments with C comments (Solaris' DTrace fails on C++ comments)
502Karel Gardas <karel.gardas@centrum.cz>**20110112051829
503 Ignore-this: c229292227c7e2b512daf9129cb66aeb
504]
505[Fix #4867, ghci displays negative floats incorrectly
506gwright@antiope.com**20110209222423
507 Ignore-this: 1b3279fa5f6c4849ed6311275b1a466a
508 
509 This patch fixes the erroneous relocations that caused
510 the bug in ticket #4867.  External addresses and global
511 offset table entries were relocated correctly, but all other
512 relocations were incorrectly calculated.  This caused, for
513 example, bad references to constants stored in the __const
514 section of the __TEXT segment.
515 
516 This bug only affected OS X on 64-bit platforms.
517 
518]
519[Fix Array sizeof primops to use the correct offset (which happens to be 0, so it worked before anyway). Makes us more future-proof, at least
520Daniel Peebles <pumpkingod@gmail.com>**20110201063017
521 Ignore-this: 8e79c3f6f80c81b4160a31e80e4ed29d
522]
523[Add sizeof(Mutable)Array# primitives
524Daniel Peebles <pumpkingod@gmail.com>**20110126051554
525 Ignore-this: ae17d94dbb86d6e1ffa0a489da842f78
526]
527[fix TRY_ACQUIRE_LOCK on Windows.
528Simon Marlow <marlowsd@gmail.com>**20110210150035
529 Ignore-this: b48713585b6d65020bb42f952c2c54ac
530]
531[constant fold  (a + N) - M  and  (a - N) + M
532Simon Marlow <marlowsd@gmail.com>**20110210115608
533 Ignore-this: 312c33d5ff7d4288be8ab0a5c6939c0c
534]
535[Recursively call cmmMachOpFold on divides that we turned into shifts
536Simon Marlow <marlowsd@gmail.com>**20110208104345
537 Ignore-this: bca0db454127c5f2b64b81d6200fc000
538 There might be more simplification to do.
539]
540[Add unboxed tuple support to Template Haskell
541Ian Lynagh <igloo@earth.li>**20110210134528
542 Ignore-this: cf946570a9d16016debf8bccd21b2c79
543]
544[Allow TH brackets to contain things of any kind
545Ian Lynagh <igloo@earth.li>**20110209184459
546 Ignore-this: b23f53d201abc874cf1f15a4eddb3abb
547 You can now quasi-quote things with unboxed types, and unboxed tuples.
548]
549[Simpify constraints from a TH bracket eagerly
550simonpj@microsoft.com**20110209175003
551 Ignore-this: b341ea3d235af1b2e617107f238ae1d6
552 
553 See Trac #4949, where having a TH bracket implication
554 was messing things up.  Better to get rid of it right away.
555]
556[Typo in comment
557simonpj@microsoft.com**20110209174914
558 Ignore-this: 3590bf097a566c6f590a37cda2be9b4e
559]
560[Call the final build system phase "final" rather than ""
561Ian Lynagh <igloo@earth.li>**20110207142046
562 Ignore-this: cc87cf202cff1f1d8b105268dacdd63f
563]
564[Fix bug introduced in "Implement fuzzy matching for the Finder"
565Simon Marlow <marlowsd@gmail.com>**20110208090121
566 Ignore-this: 53e61080e7d7fb28f9187629fa20746a
567 The finder was reporting a hidden package when it meant a hidden
568 module, and vice versa (looks like a typo).
569]
570[Fix Trac #4945: another SpecConstr infelicity
571simonpj@microsoft.com**20110207102537
572 Ignore-this: c3ffbb640cdbbab32758e6130ae803bc
573 
574 Well, more a plain bug really, which led to SpecConstr
575 missing some obvious opportunities for specialisation.
576 
577 Thanks to Max Bolingbroke for spotting this.
578]
579[add missing initialisation of ws->todo_large_objects
580Simon Marlow <marlowsd@gmail.com>**20110204093148
581 Ignore-this: dc9a28f85aff97e0896d212d7b21ae30
582 Found-by: Valgrind.  Thanks Julian!
583]
584[Add -XNondecreasingIndentation to -XHaskell98 for backwards compatibility.
585Simon Marlow <marlowsd@gmail.com>**20110204084226
586 Ignore-this: 42f7ef8bfbfb8d58f61afa217af3ffea
587 The final straw was when I learned today that Happy broke.
588]
589[only the GHC repo is in git for now; add hoopl
590Simon Marlow <marlowsd@gmail.com>**20110203205705
591 Ignore-this: 1cf90e7e5b83aabef74ec5e24496464c
592]
593[Fix typo in SpecConstr that made it not work at all
594simonpj@microsoft.com**20110203172756
595 Ignore-this: b550d5c5b73ed13709ee2938c80a750f
596 
597 There was a terrible typo in this patch; I wrote "env"
598 instead of "env1".
599 
600    Mon Jan 31 11:35:29 GMT 2011  simonpj@microsoft.com
601      * Improve Simplifier and SpecConstr behaviour
602 
603 Anyway, this fix is essential to make it work properly.
604 Thanks to Max for spotting the problem (again).
605]
606[fix compacting GC
607Simon Marlow <marlowsd@gmail.com>**20110202170036
608 Ignore-this: e78c99b318586a7fccc2a8e36d9fbf88
609]
610[fix warning
611Simon Marlow <marlowsd@gmail.com>**20110202160415
612 Ignore-this: 99f65d20b38cce971b0eda6c53eab8d3
613]
614[GC refactoring and cleanup
615Simon Marlow <marlowsd@gmail.com>**20110202154955
616 Ignore-this: 96b5b5ec97d49e69617d0007ee7fe804
617 Now we keep any partially-full blocks in the gc_thread[] structs after
618 each GC, rather than moving them to the generation.  This should give
619 us slightly better locality (though I wasn't able to measure any
620 difference).
621 
622 Also in this patch: better sanity checking with THREADED.
623]
624[avoid adding HPC ticks to arrow constructs (fixes #1333)
625Ross Paterson <ross@soi.city.ac.uk>**20110202211425
626 Ignore-this: 2938850ebbb53d1bc6bf0399f68dd8e5
627]
628[Fix the profiling build
629Simon Marlow <marlowsd@gmail.com>**20110202132257
630 Ignore-this: cdf6de609c7eee47bd6c9e957276f12b
631]
632[A small GC optimisation
633Simon Marlow <marlowsd@gmail.com>**20110202123049
634 Ignore-this: 4119e33b0f40787fd9339ad1104b3b9e
635 Store the *number* of the destination generation in the Bdescr struct,
636 so that in evacuate() we don't have to deref gen to get it.
637 This is another improvement ported over from my GC branch.
638]
639[scheduleProcessInbox: use non-blocking acquire, and take the whole queue
640Simon Marlow <marlowsd@gmail.com>**20110202114907
641 Ignore-this: 12020d2751d355d1c006697351223d99
642 This is an improvement from my GC branch, that helps performance for
643 intensive message-passing communication between Capabilities.
644]
645[do a bit of by-hand CSE
646Simon Marlow <marlowsd@gmail.com>**20110202114417
647 Ignore-this: c0c90cd767c74f3eee8b7f8cbc08dfa0
648]
649[add a const
650Simon Marlow <marlowsd@gmail.com>**20110202114345
651 Ignore-this: d12300d69c91d7187aa1dd83a4a13ff9
652]
653[add TRY_ACQUIRE_LOCK()
654Simon Marlow <marlowsd@gmail.com>**20110202113242
655 Ignore-this: face9da80ce407b9013d5f73bb65c34d
656]
657[Remove the per-generation mutable lists
658Simon Marlow <marlowsd@gmail.com>**20110202112646
659 Ignore-this: 3e0cacbc8c8b6ddf7005d25b593d3357
660 Now that we use the per-capability mutable lists exclusively.
661]
662[+RTS -qw hasn't done anything since 7.0.1; remove the implementation & docs
663Simon Marlow <marlowsd@gmail.com>**20110201163727
664 Ignore-this: e6c6ba6b8a119d87efcd79310b4fb5d2
665 It is still (silently) accepted for backwards compatibility.
666]
667[comments
668Simon Marlow <marlowsd@gmail.com>**20110201085830
669 Ignore-this: 5a70e58a48aa60e5ed7afc6e908d150a
670]
671[Annotate thread stop events with the owner of the black hole
672Simon Marlow <marlowsd@gmail.com>**20110127164226
673 Ignore-this: a60cbe5cc50da911d58020775c513ed0
674 
675 So we can now get these in ThreadScope:
676 
677   19487000: cap 1: stopping thread 6 (blocked on black hole owned by thread 4)
678 
679 Note: needs an update to ghc-events.  Older ThreadScopes will just
680 ignore the new information.
681]
682[update debugging code for fragmentation
683Simon Marlow <marlowsd@gmail.com>**20110125111011
684 Ignore-this: a98cd31d2e48ae9bdc52f9d96424ce39
685]
686[Fix type checker error message
687simonpj@microsoft.com**20110201122920
688 Ignore-this: 7369cc5f8dae3d81621f580a8ddaf41e
689 
690 See Trac #4940. We had a message
691      The lambda expression `\ x -> x' has one argument one argument,
692 repeating the "one argument" part.  Easy fix.
693]
694[Some refactoring of SpecConstr
695simonpj@microsoft.com**20110201122841
696 Ignore-this: c2966091564a9ca4ceb27a9596d36b7d
697 
698 This was originally to improve the case when SpecConstr generated a
699 function with an unused argument (see Trac #4941), but I ended up
700 giving up on that.  But the refactoring is still an improvement.
701 
702 In particular I got rid of BothOcc, which was unused.
703]
704[Don't make join points when the case has only one non-bottom alternative
705simonpj@microsoft.com**20110201122637
706 Ignore-this: 333b9b62debbfc1338530ef4e710ccbb
707 
708 This fixes Trac #4930.  See Note [Bottom alternatives] in Simplify.lhs
709]
710[Improve Simplifier and SpecConstr behaviour
711simonpj@microsoft.com**20110131113529
712 Ignore-this: e5b96c97cee0950e558ddf15178bb6c9
713 
714 Trac #4908 identified a case where SpecConstr wasn't "seeing" a
715 specialisation it should easily get.  The solution was simple: see
716 Note [Add scrutinee to ValueEnv too] in SpecConstr.
717 
718 Then it turned out that there was an exactly analogous infelicity in
719 the mighty Simplifer too; see Note [Add unfolding for scrutinee] in
720 Simplify. This fix is good for Simplify even in the absence of the
721 SpecConstr change.  (It arose when I moved the binder- swap stuff to
722 OccAnall, not realising that it *remains* valuable to record info
723 about the scrutinee of a case expression.  The Note says why.
724 
725 Together these two changes are unconditionally good.  Better
726 simplification, better specialisation. Thank you Max.
727]
728[fix warning
729Simon Marlow <marlowsd@gmail.com>**20110131135951
730 Ignore-this: e893d9bfbabf1601133a1e09c50b908
731]
732[32-bit fix
733Simon Marlow <marlowsd@gmail.com>**20101013154200
734 Ignore-this: 7508977c263ed2cec321b40a8b5772a
735]
736[update to mingw gcc 4.5.2
737Simon Marlow <marlowsd@gmail.com>**20110119135053
738 Ignore-this: b9d5cb736a48a0adc5e35eb8a0c191cd
739]
740[count fizzled and GC'd sparks separately
741Simon Marlow <marlowsd@gmail.com>**20101111132727
742 Ignore-this: 4cb4b759aed06659b46cdf76e791e5c9
743]
744[count "dud" sparks (expressions that were already evaluated when sparked)
745Simon Marlow <marlowsd@gmail.com>**20101101124143
746 Ignore-this: ca94824c0e75da0b3688300e7285c7e6
747]
748[fix some shutdown memory leaks
749Simon Marlow <marlowsd@gmail.com>**20100820093133
750 Ignore-this: 3e7b80b5f4846d6c56319c150895953d
751]
752[fix DEBUG build
753Simon Marlow <marlowsd@gmail.com>**20110131123433
754 Ignore-this: f2e009eaa66a14a7c8ec6acc7a4bbdb1
755]
756[Fix formatting glitch in documentation
757simonpj@microsoft.com**20110128115400
758 Ignore-this: 6c410ed19956feac7e0cf68bb40b40b1
759]
760[Fix warnings
761Simon Marlow <marlowsd@gmail.com>**20110128103639
762 Ignore-this: aa7f2c9f9b91f9dabc7b5d5ea26121fd
763]
764[Merge in new code generator branch.
765Simon Marlow <marlowsd@gmail.com>**20110124121650
766 Ignore-this: 7762f21082cb84ec94daaeefd70f5ef2
767 This changes the new code generator to make use of the Hoopl package
768 for dataflow analysis.  Hoopl is a new boot package, and is maintained
769 in a separate upstream git repository (as usual, GHC has its own
770 lagging darcs mirror in http://darcs.haskell.org/packages/hoopl).
771 
772 During this merge I squashed recent history into one patch.  I tried
773 to rebase, but the history had some internal conflicts of its own
774 which made rebase extremely confusing, so I gave up. The history I
775 squashed was:
776 
777   - Update new codegen to work with latest Hoopl
778   - Add some notes on new code gen to cmm-notes
779   - Enable Hoopl lag package.
780   - Add SPJ note to cmm-notes
781   - Improve GC calls on new code generator.
782 
783 Work in this branch was done by:
784    - Milan Straka <fox@ucw.cz>
785    - John Dias <dias@cs.tufts.edu>
786    - David Terei <davidterei@gmail.com>
787 
788 Edward Z. Yang <ezyang@mit.edu> merged in further changes from GHC HEAD
789 and fixed a few bugs.
790]
791[Fix an egregious strictness analyser bug (Trac #4924)
792simonpj@microsoft.com**20110128080748
793 Ignore-this: 3bf533c3d30b45a8e78b1fec3d9634f
794 
795 The "virgin" flag was being threaded rather than treated
796 like an environment.  As a result, the second and subsequent
797 recursive definitions in a module were not getting a
798 correctly-initialised fixpoint loop, causing much worse
799 strictness analysis results.  Indeed the symptoms in
800 Trac #4924 were quite bizarre.
801 
802 Anyway, it's easily fixed.  Merge to stable branch.
803]
804[Refine incomplete-pattern checks (Trac #4905)
805simonpj@microsoft.com**20110127131304
806 Ignore-this: cf2e0852f20d1cadc6a2cba4272838f6
807 
808 The changes are:
809 
810 * New flag -fwarn-incomplete-uni-patterns, which checks for
811   incomplete patterns in (a) lambdas, (b) pattern bindings
812 
813 * New flag is not implied by -W or -Wall (too noisy; and many
814   libraries use incomplete pattern bindings)
815 
816 * Actually do the incomplete-pattern check for pattern bindings
817   (previously simply omitted)
818 
819 * Documentation for new flag
820]
821[Fix "make 1" etc following the build system changes
822Ian Lynagh <igloo@earth.li>**20110127001739
823 Ignore-this: 7ae0a41f2753d7740569f362a97ea5fb
824 The logic is now in mk/compiler-ghc.mk rather than being duplicated in
825 ghc/Makefile and compiler/Makefile.
826]
827[Fix vectorisation of recursive types
828Roman Leshchinskiy <rl@cse.unsw.edu.au>**20110126231843
829 Ignore-this: 983fc42a659be2e085da9b16f994aa2e
830]
831[Fix dependencies among specialisations for imported Ids
832simonpj@microsoft.com**20110126172112
833 Ignore-this: 364e09c11affe7bfe8f1b934ea28bbb6
834 
835 This was a subtle one (Trac #4903).  See
836   Note [Glom the bindings if imported functions are specialised]
837 in Speclialise.
838 
839 Fundamentally, a specialised binding for an imported Id was being
840 declared non-recursive, whereas in fact it can become recursive
841 via a RULE.  Once it's specified non-recurive the OccAnal pass
842 treats that as gospel -- and that in turn led to infinite inlining.
843 
844 Easily fixed by glomming all the specialised bindings in a Rec;
845 now the OccAnal will sort them out correctly.
846]
847[Fix bug in roughTopNames
848simonpj@microsoft.com**20110126171803
849 Ignore-this: eca8b144162f1bd94e2ccb433bca1e02
850 
851 roughTopNames was returning a name that in fact might be
852 "looked though" by the rule matcher. Result: a rule
853 that should match was being pre-emptively discarded.
854 
855 See Note [Care with roughTopName].
856 
857 Fixes a bug noticed by Pedro (Trac #4918).
858]
859[Comments only, plus a tiny bit of debug printing
860simonpj@microsoft.com**20110126171255
861 Ignore-this: f84364b2b90fc860e9289dd40d0395ac
862]
863[Comments only
864simonpj@microsoft.com**20110126171235
865 Ignore-this: 79059977f82aaac7f9714ad09e820ea9
866]
867[Look through type synonyms when computing orphans
868simonpj@microsoft.com**20110126171229
869 Ignore-this: 6dfc45dae3a94cdb0022b2d21d6e09f6
870 
871 I renamed functions tyClsNamesOfTypes to oprhNamesOfType,
872 because it's only used in that capacity, and we therefore
873 want to look through type synonyms.  Similarly exprOrphNames.
874 
875 This fixes Trac #4912.
876]
877[Bleat a bit more informatively in unionLists
878simonpj@microsoft.com**20110126171030
879 Ignore-this: 80b276aa3d7971c6d7802b5f6b522d2e
880]
881[Keep separate linker flags, for when we want to link with gcc or ld
882Ian Lynagh <igloo@earth.li>**20110124233121]
883[Fix validate on OS X 64
884Ian Lynagh <igloo@earth.li>**20110124183618]
885[Split main/GHC into GHC and GhcMake
886simonpj@microsoft.com**20110125161632
887 Ignore-this: 502ea034de77ecd81173161836d78287
888 
889 There are two things going on in main/GHC.hs.
890   * It's the root module of the GHC package
891   * It contains lots of stuff for --make
892 It is also gigantic (2.7k lines)
893 
894 This patch splits it into two
895   * GHC.hs is the root module for the GHC package
896       (1.3k lines)
897   * GhcMake.hs contains the stuff for --make
898       (1.4k lines)
899 
900 Happily the functional split divided it almost
901 exactly in half.
902 
903 This is a pure refactoring.  There should be no
904 behavioural change.
905]
906[Comments only
907simonpj@microsoft.com**20110125131115
908 Ignore-this: 7ec4e97a481d06894de940aba59c575d
909]
910[Fix Trac #3717 by making exprOkForSpeculation a bit cleverer
911simonpj@microsoft.com**20110125110525
912 Ignore-this: 13b606b05da69c29bf53aaf408fd602
913 
914 The main change here is to do with dropping redundant seqs.
915 See Note [exprOkForSpeculation: case expressions] in CoreUtils.
916]
917[Improve dataToTag# magic
918simonpj@microsoft.com**20110125110418
919 Ignore-this: 11fdb265e030dec4d5b13ed6b16c9761
920 
921 dataToTag# is a bit unsatisfactory because it requires
922 its argument to be evaluated, and we don't have a good
923 way to enforce that. This patch adds some comments, and
924 makes exprOkForSpeculation a bit less picky in the case
925 of dataToTag# (since the argument may, in fact, not be
926 eval'd).
927]
928[Fix Trac #4917: try a bit harder to unify on-the-fly
929simonpj@microsoft.com**20110125110112
930 Ignore-this: e96e0a19ab8517d4ba648efe91f6b379
931 
932 This is generally a modest improvement but, more important,
933 it fixes a "unify-under-forall" problem.  See Note [Avoid deferring].
934 
935 There's still a lurking unsatisfactory-ness in that we can't
936 defer arbitrary constraints that are trapped under a forall.
937]
938[DPH options updated
939Manuel M T Chakravarty <chak@cse.unsw.edu.au>**20110124043617
940 Ignore-this: 6b7d2949b75f9c923f279c1178d2d042
941 - '-Odph' is now '-O2 -fsimplifier-phases=3 -fsimplifier-iterations=20'
942 - The new option '-fdph-none' is the default; it indicates that no DPH
943   backend is selected and is the only valid option if the DPH libraries
944   are not installed.  If vectorisation is attempted with -fdph-none a
945   suitable error message is generated.
946 - Hence, '-fdph-par' (or '-fdph-seq') needs to be explicitly selected
947   when using vectorisation and when linking vectorised code.  (There
948   seems to be no elegant way to avoid that.)
949]
950[Add build system profiling to build system
951Ian Lynagh <igloo@earth.li>**20110123151408
952 Ignore-this: 75717810be32d60323980f9fd1baa853
953]
954[Fix ghci in stage3
955Ian Lynagh <igloo@earth.li>**20110123120232]
956[Remove use of non-existent $$(dir) variable in the rts ghc.mk
957Ian Lynagh <igloo@earth.li>**20110123021815]
958[Add some missing dependencies
959Ian Lynagh <igloo@earth.li>**20110123004208]
960[Tweak some deps to avoid multiple $(wildcard ...)s
961Ian Lynagh <igloo@earth.li>**20110123001045
962 Ignore-this: 38e53cb6f6b4f27c771ae0ed341f8958
963 Note that some things depending on the rts/includes header files now
964 depend on more files: They used to include depend on includes/*.h, but
965 now they also depend on header files in subdirectories. As far as I can
966 see this was a bug.
967]
968[Use := when assigning the result of $(wildcard ...)
969Ian Lynagh <igloo@earth.li>**20110122224532
970 Ignore-this: 67e2ca2ffbcffb5b7f55bd60c17fc6cf
971 Avoids repeated evaluations of things that need system calls etc
972]
973[Simplify the build system, and remove 2 phases
974Ian Lynagh <igloo@earth.li>**20110122190928
975 Ignore-this: 7b6184088befcbc44ea47b2f4abf85a9
976 From
977     http://hackage.haskell.org/trac/ghc/wiki/Building/Architecture/Idiom/PhaseOrdering
978 
979 Phase 0:
980     Includes: package-data.mk files for things built by the
981               bootstrapping compiler.
982     Builds:   the dependency files for hsc2hs and genprimopcode. We need
983               to do this now, as hsc2hs needs to be buildable in phase 1's
984               includes (so that we can make the hpc library's .hs source
985               files, which in turn is necessary for making its dependency
986               files), and genprimopcode needs to be buildable in phase 1's
987               includes (so that we can make the primop-*.hs-incl files,
988               which are sources for the stage1 compiler library, and thus
989               necessary for making its dependency files).
990 Phase 1:
991     Includes: dependency files for things built by the bootstrapping
992               compiler.
993     Builds:   package-data.mk files for everything else. Note that this
994               requires configuring the packages, which means telling cabal
995               which ghc to use, and thus the stage1 compiler gets built
996               during this phase.
997 Phase "":
998     Includes: dependency files for everything else.
999     Builds:   Everything else.
1000]
1001[Manually control more of the Cabal flags for the compiler and ghc packages
1002Ian Lynagh <igloo@earth.li>**20110121230552
1003 Ignore-this: 652b5f6327d246d7e2e47acbca614df2
1004 For some reason the Windows HEAD builder has started thinking the ghci
1005 flag should be on in stage 1. This should fix it, and generally make
1006 things a little more resilient.
1007]
1008[Remove some hardcoded makefile settings
1009Ian Lynagh <igloo@earth.li>**20110121230245
1010 Ignore-this: 6b1b68aebdfbe02c15518985d2ea7559
1011 Now that we used cabal to configure the ghc-bin package they are no
1012 longer needed.
1013]
1014[tweak newArray# documentation again
1015Simon Marlow <marlowsd@gmail.com>**20110119140633
1016 Ignore-this: ceee33428dbad7e0f5eabfa0a2590466
1017]
1018[Fix OSTYPE test
1019Ian Lynagh <igloo@earth.li>**20110120000308
1020 Ignore-this: 8fa5d5c03297cb507a166bd85675145c
1021]
1022[Comments only
1023simonpj@microsoft.com**20110119222247
1024 Ignore-this: ea531428e9093ecedb895735ed537791
1025]
1026[Add OSTYPE build-system variable, and use it
1027simonpj@microsoft.com**20110113155023
1028 Ignore-this: c4a75f0bb27a680924e57ca7075ec116
1029 
1030 The use is in install.mk.in, where we need to know when
1031 we're on Cygwin.
1032 
1033 This fixes the build on my Windows box, where I have
1034 both Msys and Cygwin.
1035]
1036[Remove an extraneous comma that stopped ghc-cabal from building
1037Ian Lynagh <igloo@earth.li>**20110119222359]
1038[Move some make variables around
1039Ian Lynagh <igloo@earth.li>**20110119221545
1040 Ignore-this: c57c93f39d72c3baef7c5f466861dd5b
1041]
1042[Remove a debugging 'info'
1043Ian Lynagh <igloo@earth.li>**20110119203305
1044 Ignore-this: ea912ba205eaae1d2bcf0cce7c13628d
1045]
1046[Move the PACKAGE_MAGIC evaluation inside package-data.mk
1047Ian Lynagh <igloo@earth.li>**20110119203229
1048 Ignore-this: 497c4e83ae75089c24d6c794c4e2891f
1049]
1050[Fix libraries/index.html's haddock dependency on Windows
1051Ian Lynagh <igloo@earth.li>**20110119172310]
1052[Add configure phases for the stage 3 compiler
1053Ian Lynagh <igloo@earth.li>**20110119130629]
1054[Include kfreebsdgnu in the list of Target Platforms.
1055Marco Silva <marcot@marcot.eti.br>**20110118222352
1056 Ignore-this: 759482baf33903b98cd837636a3f5328
1057]
1058[Fix documentation bug: newArray# accepts word count, not byte count.
1059Edward Z. Yang <ezyang@mit.edu>**20110118221834
1060 Ignore-this: 8daab134bf72a740b89d273fb4e983d5
1061]
1062[Update the location of libffi.dll.a
1063Ian Lynagh <igloo@earth.li>**20110118164225
1064 As far as I can see this has been wrong for some time, but only bit
1065 recently.
1066]
1067[Update the generics docs; pointed out by Christian Maeder
1068Ian Lynagh <igloo@earth.li>**20110117214632]
1069[ghc-cabal now adds the language flag being used
1070Ian Lynagh <igloo@earth.li>**20110117184833
1071 Ignore-this: 8198892ef7f8009561d3181425cde942
1072 This means we get -XHaskell98 added to the list of flags, just like we
1073 would if we were building with Cabal.
1074]
1075[Reinstate the OS X flags in the LDFLAGS etc variables
1076Ian Lynagh <igloo@earth.li>**20110117200540
1077 Ignore-this: 9261baa1843100f65b02fb91c1a0d225
1078 I expect this will fix:
1079 http://www.haskell.org/pipermail/cvs-ghc/2011-January/059098.html
1080]
1081[Add NondecreasingIndentation extension to ghc-bin
1082Ian Lynagh <igloo@earth.li>**20110117200427
1083 Ignore-this: b6b029ee6dfbda482c91d17e835f9000
1084]
1085[Change an "if ... else return ()" into a "when"
1086Ian Lynagh <igloo@earth.li>**20110117191714
1087 Ignore-this: 7de58b728e4fce7f86d7d24a3089e6c7
1088]
1089[Add NondecreasingIndentation to the list of extensions in ghc-pkg
1090Ian Lynagh <igloo@earth.li>**20110117190610
1091 Ignore-this: 20ce8144b7b64d1f67de2f6983717da3
1092]
1093[Add NondecreasingIndentation to the list of extensions in the ghc package
1094Ian Lynagh <igloo@earth.li>**20110117190404
1095 Ignore-this: 516b45e93c1b3bbb66da5414d9aabef1
1096]
1097[Fix deps on the ghc package
1098Ian Lynagh <igloo@earth.li>**20110117173010
1099 The standard libraries/$depname scheme doesn't apply, so we need to
1100 handle it specially.
1101]
1102[Tidy up gmp cleaning
1103Ian Lynagh <igloo@earth.li>**20110117121155
1104 Ignore-this: 61d9a57d14b70732f62d6b2c8d6d197a
1105]
1106[Remove redundant libraries/cabal-bin.hs
1107Ian Lynagh <igloo@earth.li>**20110116194919
1108 Ignore-this: 13b4a8d26fa06ec952351603c3bb40ee
1109]
1110[Turn off dtrace unless you override USE_DTRACE
1111Ian Lynagh <igloo@earth.li>**20110116180306
1112 Ignore-this: beafc2002091fa7f0e66666004c870a5
1113 There are problems with dtrace on 64bit 10.5. For now at least, we
1114 just turn dtrace off unless you override USE_DTRACE
1115]
1116[Simplify a bit of makefile
1117Ian Lynagh <igloo@earth.li>**20110116175218
1118 Ignore-this: 18f02e40e36eca2e2cab79c152c72541
1119]
1120[Tweak Windows phase ordering
1121Ian Lynagh <igloo@earth.li>**20110116173459
1122 Ignore-this: bb8a70741be4574edc149349acd0f4be
1123]
1124[Handle dependencies of programs on libraries correctly
1125Ian Lynagh <igloo@earth.li>**20110116155627]
1126[It's not clear if LDFLAGS flags will be given to gcc or ld,
1127Ian Lynagh <igloo@earth.li>**20110116151230
1128 Ignore-this: a6a2d0b1f550c922c32f6f252e4e3285
1129 and they accept different flags, so for now do nothing
1130]
1131[Fix cross-package dependency generation on Windows
1132Ian Lynagh <igloo@earth.li>**20110116150901
1133 Ignore-this: f78baaa7074ca36a6a4ff8a7e6f2e35
1134]
1135[Add some Windows-only CONFIGURE_PHASEs
1136Ian Lynagh <igloo@earth.li>**20110116150826
1137 Ignore-this: abf1bf498609107eb206b22d483613de
1138]
1139[Simplify, and future-proof, a dependency in the build system
1140Ian Lynagh <igloo@earth.li>**20110116020035
1141 Ignore-this: d089133430828d041b3601b1e9c8b22a
1142]
1143[Remove an unnecessary phase, and some unnecessary deps
1144Ian Lynagh <igloo@earth.li>**20110116015943
1145 Ignore-this: e649b072d006db5db97aee26d3753f65
1146 now that cross-package deps are tracked correctly.
1147]
1148[We can now pass -include-pkg-deps to the bootstrapping compiler
1149Ian Lynagh <igloo@earth.li>**20110116015714
1150 Ignore-this: bdfed941124bb93111f117800be5f2d8
1151]
1152[Remove some flags that are redundant now GHC gets configured by Cabal
1153Ian Lynagh <igloo@earth.li>**20110116003154
1154 Ignore-this: 43a023c5103b72c91d53cf3bed7a4c50
1155]
1156[Change some HC_OPTS var handling
1157Ian Lynagh <igloo@earth.li>**20110116003104
1158 Ignore-this: 629f4a3d37028f71a477c22ed4e8591e
1159 In particular, this means ghc gets built with -rtsopt, -threaded, etc again.
1160]
1161[Remove some unnecessary workarounds
1162Ian Lynagh <igloo@earth.li>**20110116002803
1163 Ignore-this: 5ecc62f765522c08c44aa0814c5b840e
1164 We can now rely on cross-package deps working properly, as we require
1165 GHC 6.12.
1166]
1167[Tidy up a bit
1168Ian Lynagh <igloo@earth.li>**20110116001121
1169 Ignore-this: a2baabc6da0cf2877507b7833d5b0fc7
1170]
1171[Build system improvements
1172Ian Lynagh <igloo@earth.li>**20110115231927
1173 Ignore-this: 92ea6514addc8aa8734d7e0eb61b50cb
1174 We no longer use dummy-ghc; instead we don't configure most packages
1175 until the stage1 compiler is available.
1176   
1177 We also now use Cabal for building the ghc-bin package.
1178 
1179 There are a couple more sanity checks too.
1180]
1181[Whitespace tweak
1182Ian Lynagh <igloo@earth.li>**20110115214149
1183 Ignore-this: 3e564566f311be473e94f6af609bdeaa
1184]
1185[Fix libffi build rules
1186Ian Lynagh <igloo@earth.li>**20110115202104
1187 Ignore-this: 57e1763d2079301b0165be7deba29c85
1188 Fixes a rare race when both libHSffi.a and libHSffi_p.a were being built
1189 at the same time:
1190 
1191 "cp" libffi/dist-install/build/libffi.a libffi/dist-install/build/libHSffi.a
1192 "cp" libffi/dist-install/build/libffi.a libffi/dist-install/build/libHSffi.a
1193 "cp" libffi/dist-install/build/libffi.so libffi/dist-install/build/libHSffi-ghc7.1.20110115.so
1194 cp: cannot create regular file `libffi/dist-install/build/libHSffi.a': File exists
1195]
1196[Fix Trac #4874: specialisation of INLINABLE things
1197simonpj@microsoft.com**20110114163227
1198 Ignore-this: b90543117ebddaf3bbeeaf0af0c18699
1199 
1200 Johan discovered that when INLINABLE things are specialised
1201 bad things can happen. This patch implements a hack -- but
1202 it's a simple hack and it solves the problem.
1203 
1204 See Note [Inline specialisations].
1205 
1206 The hack part is that really INLINABLE should not cause *any* loss
1207 optimisation, and it does; see Note [Don't w/w INLINABLE things] in
1208 WorkWrap.
1209]
1210[Comments only
1211simonpj@microsoft.com**20110114162959
1212 Ignore-this: f76d4d8f527c3fcd2598ec8cc5fd3049
1213]
1214[Fix a buglet in postInlineUnconditionally
1215simonpj@microsoft.com**20110114162927
1216 Ignore-this: 7a7b8610ef863907843d4ae36a8a1a3c
1217 
1218 Under obscure circumstances (actually only shown up when fixing something
1219 else) it was possible for a variable binding to be discarded although
1220 it was still used.  See Note [Top level and postInlineUnconditionally]
1221]
1222[cope with empty libraries/stamp directory (in git repo)
1223Simon Marlow <marlowsd@gmail.com>**20110114142406
1224 Ignore-this: 6e95c44368d784f86a0c1c1d1e24d810
1225]
1226[add .gitignore
1227Simon Marlow <marlowsd@gmail.com>**20110114142353
1228 Ignore-this: 23d7cabd2b04eedfe4c33ad94a120474
1229]
1230[Fix longstanding bug in C-- inlining for functions calls.
1231Edward Z. Yang <ezyang@mit.edu>**20110113130654
1232 Ignore-this: 79001003b1f3cc5005207ccfed980c21
1233]
1234[fix for remote repos without -r
1235Simon Marlow <marlowsd@gmail.com>**20110113131147
1236 Ignore-this: 3ddd8a4c616cad01a2dbdb500fb54279
1237]
1238[add a version of packages that stores all the repos in git
1239Simon Marlow <marlowsd@gmail.com>**20110113111733
1240 Ignore-this: fcca2eb2e753ee20bb5abce7f30f5205
1241]
1242[add the -r flag from darcs-all
1243Simon Marlow <marlowsd@gmail.com>**20110113111654
1244 Ignore-this: ada88377bd95ebb9c668dd48954f321e
1245]
1246[Make Template Haskell classInstances function return [ClassInstance]
1247simonpj@microsoft.com**20110113111421
1248 Ignore-this: d14381f0a94170965414dd8724188356
1249 
1250 This is a recently-introduce function, which was returning
1251 a [Name], being the names of the dfuns.  But what you really
1252 want (obviously!) is the ClassInstances, and we have a TH type
1253 for that.
1254 
1255 This is an API change, so don't merge into GHC 7.0.  But it's
1256 a new part of TH which is still settling down.
1257 
1258 Fixes Trac #4863.
1259]
1260[Improve the finder's error messages
1261simonpj@microsoft.com**20110113111233
1262 Ignore-this: ec4819b0a44af9fd03dc0a8b8e13699d
1263 
1264 I'd done all the work to add fuzzy-match suggestions, but they
1265 weren't really being used!  Here's what you get now
1266 
1267    module Foo where
1268     import Data.Lst
1269 
1270 Foo.hs:3:1:
1271     Failed to load interface for `Data.Lst'
1272     Perhaps you meant
1273       Data.List (from base)
1274       Data.List (needs flag -package haskell2010-1.0.0.0)
1275       Data.Int (needs flag -package haskell2010-1.0.0.0)
1276     Use -v to see a list of the files searched for.
1277]
1278[White space only
1279simonpj@microsoft.com**20110113093931
1280 Ignore-this: 4e46acca5241615a3283996052a634a
1281]
1282[Produce an error message, not a crash, for HsOpApp with non-var operator
1283simonpj@microsoft.com**20110112170719
1284 Ignore-this: df0f6f2e3318f9c33a714609019b0262
1285 
1286 Fixes Trac #4877.
1287]
1288[update to work with current packages file format
1289Simon Marlow <marlowsd@gmail.com>**20110112160224
1290 Ignore-this: da73498734aadbfbf0a31389a9dc44d
1291]
1292[In configure, test that GHC generates code for the correct platform (#4819)
1293Simon Marlow <marlowsd@gmail.com>**20110107163541
1294 Ignore-this: 29541d3896f9c9bcf791510edae70254
1295 Patch supplied by the bug reporter, tidied up by me.
1296 
1297 $ ./configure --with-ghc=$HOME/fp/bin/i386-unknown-linux/ghc --build=x86_64-unknown-linux
1298 checking for gfind... no
1299 checking for find... /usr/bin/find
1300 checking for sort... /usr/bin/sort
1301 checking for GHC version date... inferred 7.1.20110107
1302 checking version of ghc... 7.0.1
1303 checking build system type... x86_64-unknown-linux-gnu
1304 checking host system type... x86_64-unknown-linux-gnu
1305 checking target system type... x86_64-unknown-linux-gnu
1306 Host platform inferred as: i386-unknown-linux
1307 Target platform inferred as: i386-unknown-linux
1308 This GHC (/home/simonmar/fp/bin/i386-unknown-linux/ghc) does not generate code for the build platform
1309    GHC target platform    : i386-unknown-linux
1310    Desired build platform : x86_64-unknown-linux
1311]
1312[Major refactoring of the type inference engine
1313simonpj@microsoft.com**20110112145604
1314 Ignore-this: 6a7fc90c9b798e89505606726cc8090e
1315 
1316 This patch embodies many, many changes to the contraint solver, which
1317 make it simpler, more robust, and more beautiful.  But it has taken
1318 me ages to get right. The forcing issue was some obscure programs
1319 involving recursive dictionaries, but these eventually led to a
1320 massive refactoring sweep.
1321 
1322 Main changes are:
1323  * No more "frozen errors" in the monad.  Instead "insoluble
1324    constraints" are now part of the WantedConstraints type.
1325 
1326  * The WantedConstraint type is a product of bags, instead of (as
1327    before) a bag of sums.  This eliminates a good deal of tagging and
1328    untagging.
1329 
1330  * This same WantedConstraints data type is used
1331      - As the way that constraints are gathered
1332      - As a field of an implication constraint
1333      - As both argument and result of solveWanted
1334      - As the argument to reportUnsolved
1335 
1336  * We do not generate any evidence for Derived constraints. They are
1337    purely there to allow "impovement" by unifying unification
1338    variables.
1339 
1340  * In consequence, nothing is ever *rewritten* by a Derived
1341    constraint.  This removes, by construction, all the horrible
1342    potential recursive-dictionary loops that were making us tear our
1343    hair out.  No more isGoodRecEv search either. Hurrah!
1344 
1345  * We add the superclass Derived constraints during canonicalisation,
1346    after checking for duplicates.  So fewer superclass constraints
1347    are generated than before.
1348 
1349  * Skolem tc-tyvars no longer carry SkolemInfo.  Instead, the
1350    SkolemInfo lives in the GivenLoc of the Implication, where it
1351    can be tidied, zonked, and substituted nicely.  This alone is
1352    a major improvement.
1353 
1354  * Tidying is improved, so that we tend to get t1, t2, t3, rather
1355    than t1, t11, t111, etc
1356 
1357    Moreover, unification variables are always printed with a digit
1358    (thus a0, a1, etc), so that plain 'a' is available for a skolem
1359    arising from a type signature etc. In this way,
1360      (a) We quietly say which variables are unification variables,
1361          for those who know and care
1362      (b) Types tend to get printed as the user expects.  If he writes
1363              f :: a -> a
1364              f = ...blah...
1365          then types involving 'a' get printed with 'a', rather than
1366          some tidied variant.
1367 
1368  * There are significant improvements in error messages, notably
1369    in the "Cannot deduce X from Y" messages.
1370]
1371[Fix installation on cygwin
1372Ian Lynagh <igloo@earth.li>**20110111194838
1373 Ignore-this: fe923d0619da3bd3a34968106c92fdab
1374]
1375[Do dependency analysis when kind-checking type declarations
1376simonpj@microsoft.com**20110110110351
1377 Ignore-this: 17a8dee32694d3e1835cf7bb02d3abb5
1378 
1379 This patch fixes Trac #4875.  The main point is to do dependency
1380 analysis on type and class declarations, and kind-check them in
1381 dependency order, so as to improve error messages.
1382 
1383 This patch means that a few programs that would typecheck before won't
1384 typecheck any more; but before we were (naughtily) going beyond
1385 Haskell 98 without any language-extension flags, and Trac #4875
1386 convinces me that doing so is a Bad Idea.
1387 
1388 Here's an example that won't typecheck any more
1389        data T a b = MkT (a b)
1390        type F k = T k Maybe
1391 
1392 If you look at T on its own you'd default 'a' to kind *->*;
1393 and then kind-checking would fail on F.
1394 
1395 But GHC currently accepts this program beause it looks at
1396 the *occurrences* of T.
1397]
1398[Move imports around (no change in behaviour)
1399simonpj@microsoft.com**20110110105647
1400 Ignore-this: d618cabbc52be7d7968de1e0bdd44082
1401]
1402[Make fuzzy matching a little less eager for short identifiers
1403simonpj@microsoft.com**20110107102855
1404 Ignore-this: a753643e88433d74b44a480cc0f4170c
1405 
1406 For single-character identifiers we now don't make any suggestions
1407 See comments in Util.fuzzyLookup
1408]
1409[Fix Trac #4870: get the inlining for an imported INLINABLE Id
1410simonpj@microsoft.com**20110105002712
1411 Ignore-this: 60c0192eb48590c2e6868d15ba8f84ce
1412 
1413 We need the unfolding even for a *recursive* function (indeed
1414 that's the point) and I was using the wrong function to get it
1415 (idUnfolding rather than realIdUnfolding).
1416]
1417[Rejig the includes/ installation rules
1418Ian Lynagh <igloo@earth.li>**20110109181158
1419 They're a little nicer now, and a regression in the cygwin build is
1420 fixed (the $i in the destination wasn't surviving being passed through
1421 cygpath).
1422]
1423[Make DESTDIR an absolute path when installing; fixes #4883
1424Ian Lynagh <igloo@earth.li>**20110108171635]
1425[Add utils/ghc-cabal/Makefile
1426Ian Lynagh <igloo@earth.li>**20110108144049]
1427[Remove redundant import
1428Ian Lynagh <igloo@earth.li>**20110108130047
1429 Ignore-this: 1c7fdec77b48319c845c9593b5fb94af
1430]
1431[Improve error message of :set in ghci (ticket #4190).
1432Michal Terepeta <michal.terepeta@gmail.com>**20101130211505
1433 Ignore-this: ccc8a0816a900ba8c4a966285a465b23
1434]
1435[Improve error message when importing data constructors (ticket #4058).
1436Michal Terepeta <michal.terepeta@gmail.com>**20101127211338
1437 Ignore-this: 3289a08f0391dd90dfef2e0403a04ccd
1438]
1439[catch SIGTSTP and save/restore terminal settings (#4460)
1440Simon Marlow <marlowsd@gmail.com>**20110107124042
1441 Ignore-this: 38f7f27bf75178899f466404c048241d
1442 As far as I can tell, it is the responsibility of the program to save
1443 and restore its own terminal settings across a suspend/foreground, the
1444 shell doesn't do it (which seems odd).  So I've added a signal handler
1445 for SIGTSTP to the RTS which will save and restore the terminal
1446 settings iff we modified them with hSetBuffering or hSetEcho (we
1447 already restore them at exit time in these cases).
1448]
1449[comment updates
1450Simon Marlow <marlowsd@gmail.com>**20110107094236
1451 Ignore-this: c2b30b0c98645e2847a2749c7fdc167f
1452]
1453[On Cygwin, use a Cygwin-style path for /bin/install's destination
1454Ian Lynagh <igloo@earth.li>**20110106223030
1455 
1456 cygwin's /bin/install doesn't set file modes correctly if the
1457 destination path is a C: style path:
1458 
1459 $ /bin/install -c -m 644 foo /cygdrive/c/cygwin/home/ian/foo2
1460 $ /bin/install -c -m 644 foo c:/cygwin/home/ian/foo3
1461 $ ls -l foo*
1462 -rw-r--r-- 1 ian None 0 2011-01-06 18:28 foo
1463 -rw-r--r-- 1 ian None 0 2011-01-06 18:29 foo2
1464 -rwxrwxrwx 1 ian None 0 2011-01-06 18:29 foo3
1465 
1466 This causes problems for bindisttest/checkBinaries.sh which then
1467 thinks that e.g. the userguide HTML files are binaries.
1468 
1469 We therefore use a /cygdrive path if we are on cygwin
1470]
1471[Fix mkUserGuidePart program name on Windows
1472Ian Lynagh <igloo@earth.li>**20110106143707]
1473[add comment to remind people to update driver/gcc/gcc.c
1474Simon Marlow <marlowsd@gmail.com>**20110106152402
1475 Ignore-this: c07d7ac11eb9221ef821f78aab1807cb
1476]
1477[use Win32 CreateProcess() rather than mingw spawnv() (#4531)
1478Simon Marlow <marlowsd@gmail.com>**20110106133834
1479 Ignore-this: 4c0947853549dad034622c044391af6c
1480]
1481[update paths now that we upgraded gcc to 4.5.0
1482Simon Marlow <marlowsd@gmail.com>**20110106133729
1483 Ignore-this: f8f9bcad984fdd472e0ae958b66bea9d
1484]
1485[fix markup
1486Simon Marlow <marlowsd@gmail.com>**20110106093152
1487 Ignore-this: 555b6e39ae6b5a177b03c5edffc169ab
1488]
1489[fix up multi-line GHCi patch (#4316)
1490Simon Marlow <marlowsd@gmail.com>**20110105154548
1491 Ignore-this: 53d5d489bd2a792c01f2cc56a11f3ce6
1492]
1493[multiline commands in GHCi #4316
1494Vivian McPhail <haskell.vivian.mcphail@gmail.com>**20101105051308
1495 This patch adds support for multiline commands in GHCi.
1496 
1497 The first line of input is lexed.  If there is an active
1498 layout context once the lexer reaches the end of file, the
1499 user is prompted for more input.
1500 
1501 Multiline input is exited by an empty line and can be escaped
1502 with a user interrupt.
1503 
1504 Multiline mode is toggled with `:set +m`
1505]
1506[Replace a #if with a Haskell conditional
1507Ian Lynagh <igloo@earth.li>**20110105183011
1508 Ignore-this: f08f3a4356586efab2725ad8704b2eba
1509]
1510[Whitespace only in X86.Ppr
1511Ian Lynagh <igloo@earth.li>**20110105171124]
1512[Fix error compiling AsmCodeGen.lhs for PPC Mac (unused makeFar addr)
1513naur@post11.tele.dk**20101219213555
1514 Ignore-this: ab25d5f2e2ebe163547d5babaf4b1dbf
1515]
1516[Define cTargetArch and start to use it rather than ifdefs
1517Ian Lynagh <igloo@earth.li>**20110104220013
1518 Using Haskell conditionals means the compiler sees all the code, so
1519 there should be less rot of code specific to uncommon arches. Code
1520 for other platforms should still be optimised away, although if we want
1521 to support targetting other arches then we'll need to compile it
1522 for-real anyway.
1523]
1524[Fix error compiling AsmCodeGen.lhs for PPC Mac (rtsPackageId)
1525naur@post11.tele.dk**20101219212530
1526 Ignore-this: 946f6d3e0d3c3ddf2dc07b85e1f82d85
1527]
1528[Rename the c*Platform variables to c*PlatformString
1529Ian Lynagh <igloo@earth.li>**20110104210250]
1530[Fix #4829 (build does not respect --with-gcc option)
1531gwright@antiope.com**20101221133233
1532 Ignore-this: 37918feb82f911c2beb75915b6e8b97b
1533 
1534 This patch fixes what seems to be the last problem with the --with-gcc
1535 option.  On OS X, we need to pass the path to gcc to dtrace as the
1536 preprocessor.  (Internally, dtrace on OS X sets the default preprocessor
1537 to /usr/bin/gcc.)  ATM, dtrace is only supported on OS X, so we don't
1538 need any conditionalization.  If dtrace is ported to other platforms,
1539 we might need to change this. However, usage on other platforms will
1540 probably be similar to OS X, since many of Apple's changes are to
1541 use the gnu toolchain instead of the Sun toolchain.
1542   
1543]
1544[Drop a seven years old workaround for happy
1545Matthias Kilian <kili@outback.escape.de>**20101231192343
1546 Ignore-this: a9348c91292c113bd967464fbe859f1f
1547]
1548[Add gcc and ld flags to --info output
1549Ian Lynagh <igloo@earth.li>**20101220173520]
1550[Fix Trac #4525: report type errors in terms of the immediate type synonym
1551simonpj@microsoft.com**20101224082520
1552 Ignore-this: a3bd076bfe0e1c6f575b106f77f326c6
1553 
1554 This small change means that if you have
1555      type Age = Int
1556 and you try to unify Age and Bool, you'll get a complaint about
1557 not matching Age and Bool, rather than Int and Bool.  See the notes
1558 with Trac #4525
1559]
1560[Comments only
1561simonpj@microsoft.com**20101224082310
1562 Ignore-this: 1f69fa3244663b653607093efcdf7b0
1563]
1564[Implement fuzzy matching for the Finder
1565simonpj@microsoft.com**20101222175400
1566 Ignore-this: 4dfbbc07bcb59c5f4cee9a902c89d63e
1567 
1568 ..so that you get a more helpful message when
1569 you mis-spell a module name in an 'import'.
1570 
1571 Validates, but not fully tested.
1572 
1573 Based on Max's patch in Trac #2442, but heavily refactored.
1574]
1575[Implement fuzzy matching for the renamer
1576simonpj@microsoft.com**20101222175306
1577 Ignore-this: 66478736249de793a61612f184d484b0
1578 
1579 ...so that you get helpful suggestions when you mis-spell a name
1580 Based on Max's patch in Trac #2442, but heavily refactored.
1581]
1582[Add fuzzyLookup, a variant of fuzzyMatch
1583simonpj@microsoft.com**20101222175124
1584 Ignore-this: f0eafaf275b9edffee176f2fb4effe2f
1585 
1586 Plus, I changed quite a bit of layout to make the lines shorter.
1587]
1588[White space only
1589simonpj@microsoft.com**20101222175001
1590 Ignore-this: ddabada2042f4529e83d1c1ecb052306
1591]
1592[Layout and white space only
1593simonpj@microsoft.com**20101222174950
1594 Ignore-this: bf4e4fd9d39714d0461ab799d6b8ed91
1595]
1596[Tidy up rebindable syntax for MDo
1597simonpj@microsoft.com**20101222132210
1598 Ignore-this: b40ae8709e5a39d75f2b2813169af215
1599 
1600 For a long time an 'mdo' expression has had a SyntaxTable
1601 attached to it.  However, we're busy deprecating SyntaxTables
1602 in favour of rebindable syntax attached to individual Stmts,
1603 and MDoExpr was totally inconsistent with DoExpr in this
1604 regard.
1605 
1606 This patch tidies it all up.  Now there's no SyntaxTable on
1607 MDoExpr, and 'modo' is generally handled much more like 'do'.
1608 
1609 There is resulting small change in behaviour: now MonadFix is
1610 required only if you actually *use* recursion in mdo. This
1611 seems consistent with the implicit dependency analysis that
1612 is done for mdo.
1613 
1614 Still to do:
1615   * Deal with #4148 (this patch is on the way)
1616   * Get rid of the last remaining SyntaxTable on HsCmdTop
1617]
1618[Make the occurrence analyser track preInlineUnconditionally
1619simonpj@microsoft.com**20101222131156
1620 Ignore-this: 82edb06bcca6106327c2cce9d78c4e61
1621 
1622 This fixes a somewhat obscure situation in which an
1623 over-optimistic use of "occurs once" led to an infinite
1624 sequence of simplifier iterations.  Se Note [Cascading inlines]
1625 for the details.
1626 
1627 This showed up when compiling rather large DPH programs, which
1628 run lots of iterations of the simplifier, which in turn made
1629 compilation take much longer than necessary.
1630]
1631[Make mkDFunUnfolding more robust
1632simonpj@microsoft.com**20101222130854
1633 Ignore-this: 10bb4168a7080c843f6613043354151b
1634 
1635 It now uses tcSplitDFunTy, which is designed for the purpose and
1636 allows arbitrary argument types to the dfun, rather than
1637 tcSplitSigmaTy.  This generality is used in DPH, which has
1638 internally-generated dfuns with impliciation-typed arguments.
1639 
1640 To do this I had to make tcSplitDFunTy return the number of
1641 arguments, so there are some minor knock-on effects in other
1642 modules.
1643]
1644[Count allocations more accurately
1645Simon Marlow <marlowsd@gmail.com>**20101221152956
1646 Ignore-this: 33a4ed3a77bf35f232aa5c9078e8e380
1647 The allocation stats (+RTS -s etc.) used to count the slop at the end
1648 of each nursery block (except the last) as allocated space, now we
1649 count the allocated words accurately.  This should make allocation
1650 figures more predictable, too.
1651 
1652 This has the side effect of reducing the apparent allocations by a
1653 small amount (~1%), so remember to take this into account when looking
1654 at nofib results.
1655]
1656[Add a simple arity analyser
1657simonpj@microsoft.com**20101221165800
1658 Ignore-this: d5f3a9f56404d61bb7f374c875b42c49
1659 
1660 I've wanted to do this for ages, but never gotten around to
1661 it.  The main notes are in Note [Arity analysis] in SimplUtils.
1662 
1663 The motivating example for arity analysis is this:
1664 
1665   f = \x. let g = f (x+1)
1666           in \y. ...g...
1667 
1668 What arity does f have?  Really it should have arity 2, but a naive
1669 look at the RHS won't see that.  You need a fixpoint analysis which
1670 says it has arity "infinity" the first time round.
1671 
1672 This makes things more robust to the way in which you write code.  For
1673 example, see Trac #4474 which is fixed by this change.
1674 
1675 Not a huge difference, but worth while:
1676 
1677         Program           Size    Allocs   Runtime   Elapsed
1678 --------------------------------------------------------------------------------
1679             Min          -0.4%     -2.2%    -10.0%    -10.0%
1680             Max          +2.7%     +0.3%     +7.1%     +6.9%
1681  Geometric Mean          -0.3%     -0.2%     -2.1%     -2.2%
1682 
1683 I don't really believe the runtime numbers, because the machine was
1684 busy, but the bottom line is that not much changes, and what does
1685 change reliably (allocation and size) is in the right direction.
1686]
1687[Miscellaneous tidying up and refactoring
1688simonpj@microsoft.com**20101221161931
1689 Ignore-this: 7706d3065e6fc1defafe1cb8975b9969
1690]
1691[Comments only
1692simonpj@microsoft.com**20101221161918
1693 Ignore-this: 3e269a62da5cbec72d3e4b8328689628
1694]
1695[Single-method classes are implemented with a newtype
1696simonpj@microsoft.com**20101221161911
1697 Ignore-this: 4ca00f0b367fbeb8146146bc53116eb7
1698 
1699 This patch changes things so that such classes rely on the coercion
1700 mechanism for inlining (since the constructor is really just a cast)
1701 rather than on the dfun mechanism, therby removing some needless
1702 runtime indirections.
1703]
1704[For single-method classes use newtypes
1705simonpj@microsoft.com**20101101080736
1706 Ignore-this: d3851f92eb2385501411da57066b775e
1707 
1708 This clears up an awkward hack for exprIsConApp_maybe, and
1709 works better too.  See Note [Single-method classes] in
1710 TcInstDcls.
1711]
1712[boundTaskExiting: don't set task->stopped unless this is the last call (#4850)
1713Simon Marlow <marlowsd@gmail.com>**20101221115807
1714 Ignore-this: 7e1b990aa08b3ea9cdaa9385d8e41e48
1715 The bug in this case was that we had a worker thread making a foreign
1716 call which invoked a callback (in this case it was performGC, I
1717 think).  When the callback ended, boundTaskExiting() was setting
1718 task->stopped, but the Task is now per-OS-thread, so it is shared by
1719 the worker that made the original foreign call.  When the foreign call
1720 returned, because task->stopped was set, the worker was not placed on
1721 the queue of spare workers.  Somehow the worker woke up again, and
1722 found the spare_workers queue empty, which lead to a crash.
1723 
1724 Two bugs here: task->stopped should not have been set by
1725 boundTaskExiting (this broke when I split the Task and InCall structs,
1726 in 6.12.2), and releaseCapabilityAndQueueWorker() should not be
1727 testing task->stopped anyway, because it should only ever be called
1728 when task->stopped is false (this is now an assertion).
1729]
1730[releaseCapabilityAndQueueWorker: task->stopped should be false (#4850)
1731Simon Marlow <marlowsd@gmail.com>**20101221114911
1732 Ignore-this: b9c430a4bc9d2e0c7f4140d6d6971eae
1733]
1734[Fix Windows build
1735Simon Marlow <marlowsd@gmail.com>**20101221102101
1736 Ignore-this: f4773e06d030a335c9ac721af193b8d2
1737]
1738[raiseExceptionHelper: update tso->stackobj->sp before calling threadStackOverflow (#4845)
1739Simon Marlow <marlowsd@gmail.com>**20101221101411
1740 Ignore-this: 48495131fcc8c548882a470c2509f9f5
1741]
1742[add 'make re2' for rebuilding stage2 (similarly re1 and re3)
1743Simon Marlow <marlowsd@gmail.com>**20101221100254
1744 Ignore-this: 5c0afe3810b66a5b6e53a3a0fe933945
1745]
1746[fix warning
1747Simon Marlow <marlowsd@gmail.com>**20101216160415
1748 Ignore-this: 54a0eedfa5b7fc15c31dffffb1b10aad
1749]
1750[Small improvement to CorePrep
1751simonpj@microsoft.com**20101220123715
1752 Ignore-this: d0490225ed1895a1a5b97d786ed44260
1753 
1754 This change avoids unnecessary bindings. Example
1755 
1756      foo (let fn = \x.blah in
1757           in fn)
1758 
1759 We were generating something stupid like
1760 
1761     let fn = \x.blah in
1762     let fn' = \eta. fn eta
1763     in foo fn
1764 
1765 Now we don't.  The change is quite small.
1766 
1767 Thanks to Ben for showing me an example of this happening.
1768]
1769[Fix warnings
1770Ian Lynagh <igloo@earth.li>**20101219202711
1771 Ignore-this: 898015b086f684de5371bf97a23b9e2e
1772]
1773[Small refactoring
1774Ian Lynagh <igloo@earth.li>**20101219194032]
1775[Drop GhcWithLlvmCodeGen configuration bits
1776Matthias Kilian <kili@outback.escape.de>**20101219180239
1777 Ignore-this: 815ed46be7650792f85807c232edfcc
1778 The LLVM code generator is always built unconditionally, so both the
1779 configuration variable in mk/config.mk.in as well as the string in
1780 compilerInfo can be removed.
1781]
1782[Pass --hoogle to haddock; fixes trac #4521
1783Ian Lynagh <igloo@earth.li>**20101219125243]
1784[vectoriser: don't always pass superclass dictionaries to PA dfuns
1785Roman Leshchinskiy <rl@cse.unsw.edu.au>**20101218234838
1786 Ignore-this: 77c71976db8fc63aeb83f4abdba994d8
1787 
1788 This is just a guess at how this should work.
1789]
1790[vectoriser: delete dead code
1791Roman Leshchinskiy <rl@cse.unsw.edu.au>**20101218125350
1792 Ignore-this: 437eea71ad15ad5dc7902e596597c577
1793]
1794[vectoriser: adapt to new superclass story part I (dictionary construction)
1795Roman Leshchinskiy <rl@cse.unsw.edu.au>**20101218114953
1796 Ignore-this: 29c9aa46a1622beaae1dcefc4c482a30
1797]
1798[Replace uses of the old try function with the new one
1799Ian Lynagh <igloo@earth.li>**20101218230827
1800 Ignore-this: 5dd6c1a4142405aa1aab3fc4ec07eea6
1801]
1802[Replace uses of the old catch function with the new one
1803Ian Lynagh <igloo@earth.li>**20101218213350]
1804[Create ~/.ghc/ if it doesn't already exist; fixes trac #4522
1805Ian Lynagh <igloo@earth.li>**20101218184925]
1806[Document GADTSyntax extension
1807Ian Lynagh <igloo@earth.li>**20101218150121]
1808[Implement GADTSyntax extension
1809Ian Lynagh <igloo@earth.li>**20101218144550]
1810[Whitespace-only in rts/Linker.c
1811Ian Lynagh <igloo@earth.li>**20101217234124]
1812[Add some casts to fix warnings; patch from Greg Wright
1813Ian Lynagh <igloo@earth.li>**20101217223811]
1814[Put an up-to-date Makefile in docs/Makefile
1815Ian Lynagh <igloo@earth.li>**20101217223707
1816 It doesn't do anything useful yet, but it works with the new build system
1817]
1818[do not compile part of shared lib RTS with -fno-PIC on Solaris
1819Karel Gardas <karel.gardas@centrum.cz>**20101217085133
1820 Ignore-this: 8c8dbb45cac0578a58a3557f1e03c66
1821]
1822[provide shared libraries support on i386-unknown-solaris2 platform
1823Karel Gardas <karel.gardas@centrum.cz>**20101217084617
1824 Ignore-this: b6079c6a39a71200a1ee863573e40828
1825]
1826[fix CPP detection of Solaris in NCG
1827Karel Gardas <karel.gardas@centrum.cz>**20101217084510
1828 Ignore-this: 9d1ce59d469294eab1f0cbc697e48d69
1829]
1830[Fix checkBinaries on OS X
1831Ian Lynagh <igloo@earth.li>**20101216201121]
1832[Remove redundant HpcMap and HpcSet wrappers around Data.{Map,Set}
1833Ian Lynagh <igloo@earth.li>**20101216190605]
1834[Use "-perm -u+x" rather than "-executable" to find executables
1835Ian Lynagh <igloo@earth.li>**20101216145235
1836 On Windows, -executable is matching the html docs.
1837]
1838[Remove a debugging print
1839Ian Lynagh <igloo@earth.li>**20101216011459]
1840[__GLASGOW_HASKELL__ >= 604 is now always true
1841Ian Lynagh <igloo@earth.li>**20101215214656]
1842[Remove more dead code now we require GHC >= 6.12
1843Ian Lynagh <igloo@earth.li>**20101215213715]
1844[refactor and tidy up the section on RTS options
1845Simon Marlow <marlowsd@gmail.com>**20101216123151
1846 Ignore-this: 9cdafd687351d8a3ff879b64347f85d3
1847]
1848[Related to #4826: Some minor tweaks to the wording of the User Guide, section 4.16
1849Orphi <MathematicalOrchid@hotmail.com>**20101209170440
1850 Ignore-this: c3d942d58594be7d4c2eb4dc3a22f19
1851]
1852[FIX #4826 partial: Add -rtsopts and -with-rtsopts to User Guide section 4.11.6
1853Orphi <MathematicalOrchid@hotmail.com>**20101209165152
1854 Ignore-this: 2fc1c0abbb783695773ab0f9c013bbaa
1855]
1856[FIX #4826 partially: Change -f to -? in User Guide section F4.16
1857Orphi <MathematicalOrchid@hotmail.com>**20101209144148
1858 Ignore-this: 73410b350e80c8943ae722dec8dea44b
1859]
1860[fix #3910
1861Simon Marlow <marlowsd@gmail.com>**20101216114452
1862 Ignore-this: 410e95e188344a523520e192a3fb58ea
1863]
1864[remove an optimisation that wasn't
1865Simon Marlow <marlowsd@gmail.com>**20101215152656
1866 Ignore-this: e8413f58e8292c6e7463087d885b3a7d
1867]
1868[fix a warning
1869Simon Marlow <marlowsd@gmail.com>**20101216105723
1870 Ignore-this: ed6024378021a698ce638267ed3e21ab
1871]
1872[use EXTERN_INLINE instead of STATIC_INLINE to avoid some gcc warnings
1873Simon Marlow <marlowsd@gmail.com>**20101216105709
1874 Ignore-this: d4e1586cf318883a8e611b55df7fbf10
1875]
1876[remove dead code
1877Simon Marlow <marlowsd@gmail.com>**20101216104944
1878 Ignore-this: 97a04a3e37c1b28abc222a28bab3d17d
1879]
1880[fix retainer profiling: add missing case for TSO
1881Simon Marlow <marlowsd@gmail.com>**20101216103900
1882 Ignore-this: 11bda81ac159f638d719c1f6177702fb
1883]
1884[add a missing STACK case
1885Simon Marlow <marlowsd@gmail.com>**20101216102100
1886 Ignore-this: ac1c036b5cbf4209b1d10b6ab1c83f27
1887]
1888[Remove code that is dead now that we need >= 6.12 to build
1889Ian Lynagh <igloo@earth.li>**20101215201006]
1890[fix for large stack allocations
1891Simon Marlow <marlowsd@gmail.com>**20101215152419
1892 Ignore-this: d9aca17d68bd99214c126989a2318e79
1893]
1894[Implement stack chunks and separate TSO/STACK objects
1895Simon Marlow <marlowsd@gmail.com>**20101215120843
1896 Ignore-this: 73fa9460314d4a4e54456af12bef7960
1897 
1898 This patch makes two changes to the way stacks are managed:
1899 
1900 1. The stack is now stored in a separate object from the TSO.
1901 
1902 This means that it is easier to replace the stack object for a thread
1903 when the stack overflows or underflows; we don't have to leave behind
1904 the old TSO as an indirection any more.  Consequently, we can remove
1905 ThreadRelocated and deRefTSO(), which were a pain.
1906 
1907 This is obviously the right thing, but the last time I tried to do it
1908 it made performance worse.  This time I seem to have cracked it.
1909 
1910 2. Stacks are now represented as a chain of chunks, rather than
1911    a single monolithic object.
1912 
1913 The big advantage here is that individual chunks are marked clean or
1914 dirty according to whether they contain pointers to the young
1915 generation, and the GC can avoid traversing clean stack chunks during
1916 a young-generation collection.  This means that programs with deep
1917 stacks will see a big saving in GC overhead when using the default GC
1918 settings.
1919 
1920 A secondary advantage is that there is much less copying involved as
1921 the stack grows.  Programs that quickly grow a deep stack will see big
1922 improvements.
1923 
1924 In some ways the implementation is simpler, as nothing special needs
1925 to be done to reclaim stack as the stack shrinks (the GC just recovers
1926 the dead stack chunks).  On the other hand, we have to manage stack
1927 underflow between chunks, so there's a new stack frame
1928 (UNDERFLOW_FRAME), and we now have separate TSO and STACK objects.
1929 The total amount of code is probably about the same as before.
1930 
1931 There are new RTS flags:
1932 
1933    -ki<size> Sets the initial thread stack size (default 1k)  Egs: -ki4k -ki2m
1934    -kc<size> Sets the stack chunk size (default 32k)
1935    -kb<size> Sets the stack chunk buffer size (default 1k)
1936 
1937 -ki was previously called just -k, and the old name is still accepted
1938 for backwards compatibility.  These new options are documented.
1939]
1940[comments on SRC_HC_OPTS (#4829)
1941Simon Marlow <marlowsd@gmail.com>**20101214101340
1942 Ignore-this: e2bdec00f07b68e82837e77a4faf6514
1943]
1944[fix another sanity error, and refactor/tidy up
1945Simon Marlow <marlowsd@gmail.com>**20101209163919
1946 Ignore-this: d5ce953ac78e90fc0e22cd9848d26e2e
1947]
1948[Fix a bug in functorLikeTraverse, which was giving wrong answer for tuples
1949simonpj@microsoft.com**20101215123725
1950 Ignore-this: 560220e92429b5b1a6197a62f94a4ff2
1951 
1952 This bug led to Trac #4816, which is hereby fixed
1953]
1954[Improve printing for -ddump-deriv
1955simonpj@microsoft.com**20101215121955
1956 Ignore-this: 3181c948c4c2471bd99b32c5ee487a1e
1957]
1958[Tighten up what it means to be an "enumeration data constructor"
1959simonpj@microsoft.com**20101215121927
1960 Ignore-this: 459b3f9f7994a13094ed87b0768b33a8
1961 
1962 See Note [Enumeration types] in TyCon, and comments in Trac #4528
1963]
1964[Allow enumerations to have phantom arguments.
1965simonpj@microsoft.com**20101215121817
1966 Ignore-this: 32ef8cb869e6e38c2e43b3ae87b1b9a8
1967 
1968 The bytecode generator was being too eager.
1969 Fixes Trac #4528, or rather, a near variant.
1970]
1971[Instance declaration overlap allowed if *either* has -XOverlappingInstances
1972simonpj@microsoft.com**20101214180500
1973 Ignore-this: f1b1492541a7e0464ebc6adb45510a2e
1974 
1975 This satisfies Trac #3877.  Documentation is changed too.
1976 I'm not sure if this should go in 7.0.2.
1977]
1978[Fix Trac #4841: behave right with TypeSynonymInstances and NoFlexibleInstances
1979simonpj@microsoft.com**20101214174755
1980 Ignore-this: dccd707fdca84904b7885170a296ecb6
1981 
1982 When we have TypeSynonymInstances without FlexibleInstances we should still
1983 insist on a H98-style instance head, after looking through the synonym.
1984 
1985 This patch also make FlexibleInstances imply TypeSynonymInstances.  Anything
1986 else is a bit awkward, and not very useful.
1987 
1988]
1989[Fix Trac #3731: more superclass subtlety (sigh)
1990simonpj@microsoft.com**20101214180344
1991 Ignore-this: f4168e59f3164303ba7be022ba19c37b
1992 
1993 I will add more comments, but I want to commit this tonight,
1994 so the overnight builds get it.
1995]
1996[Less verbose debug print
1997simonpj@microsoft.com**20101214180248
1998 Ignore-this: e405e8545763e913155abe43daf7e36c
1999]
2000[Wibble to InstEnv.instanceHead
2001simonpj@microsoft.com**20101214082939
2002 Ignore-this: 851db517f8638a0aeb7ad461298f7e9f
2003 
2004 Fixes an accidental glitch in T1835
2005]
2006[Remove dead code now that we require the bootstrapping compiler be >= 6.12
2007Ian Lynagh <igloo@earth.li>**20101214011011]
2008[GHC 6.12 is now needed to build the HEAD
2009Ian Lynagh <igloo@earth.li>**20101214010923]
2010[Add libstdc++-4.5.0-1-mingw32-dll-6.tar.lzma to mingw tarballs
2011Ian Lynagh <igloo@earth.li>**20101213223153]
2012[Fix recursive superclasses (again).  Fixes Trac #4809.
2013simonpj@microsoft.com**20101213171511
2014 Ignore-this: b91651397918fd8f0183812f9a070073
2015 
2016 This patch finally deals with the super-delicate question of
2017 superclases in possibly-recursive dictionaries.  The key idea
2018 is the DFun Superclass Invariant (see TcInstDcls):
2019 
2020      In the body of a DFun, every superclass argument to the
2021      returned dictionary is
2022        either   * one of the arguments of the DFun,
2023        or       * constant, bound at top level
2024 
2025 To establish the invariant, we add new "silent" superclass
2026 argument(s) to each dfun, so that the dfun does not do superclass
2027 selection internally.  There's a bit of hoo-ha to make sure that
2028 we don't print those silent arguments in error messages; a knock
2029 on effect was a change in interface-file format.
2030 
2031 A second change is that instead of the complex and fragile
2032 "self dictionary binding" in TcInstDcls and TcClassDcl,
2033 using the same mechanism for existential pattern bindings.
2034 See Note [Subtle interaction of recursion and overlap] in TcInstDcls
2035 and Note [Binding when looking up instances] in InstEnv.
2036 
2037 Main notes are here:
2038 
2039   * Note [Silent Superclass Arguments] in TcInstDcls,
2040     including the DFun Superclass Invariant
2041 
2042 Main code changes are:
2043 
2044   * The code for MkId.mkDictFunId and mkDictFunTy
2045 
2046   * DFunUnfoldings get a little more complicated;
2047     their arguments are a new type DFunArg (in CoreSyn)
2048 
2049   * No "self" argument in tcInstanceMethod
2050   * No special tcSimplifySuperClasss
2051   * No "dependents" argument to EvDFunApp
2052 
2053 IMPORTANT
2054    It turns out that it's quite tricky to generate the right
2055    DFunUnfolding for a specialised dfun, when you use SPECIALISE
2056    INSTANCE.  For now I've just commented it out (in DsBinds) but
2057    that'll lose some optimisation, and I need to get back to
2058    this.
2059]
2060[Doing the smart canonicalization only if we are not simplifying a Rule LHS.
2061dimitris@microsoft.com**20101210132221
2062 Also, same thing now applies for adding superclasses.
2063 
2064]
2065[Moved canonicalisation inside solveInteract
2066dimitris@microsoft.com**20101209141215
2067 
2068 Moreover canonicalisation now is "clever", i.e. it never canonicalizes a class
2069 constraint if it can already discharge it from some other inert or previously
2070 encountered constraints. See Note [Avoiding the superclass explosion]
2071 
2072]
2073[GHCi linker: Assume non-Haskell libraries are dynamic libs
2074Ian Lynagh <igloo@earth.li>**20101213124930
2075 Ignore-this: aa153a8f6e309c7b3dae7e46bb7a9583
2076 This works around a segfault we get when trying to load libiconv.a on
2077 some platforms.
2078]
2079[Add --version support to ghc-cabal
2080Ian Lynagh <igloo@earth.li>**20101212213600
2081 Ignore-this: ef696dcb1b96a23765f9f18e75a56f5
2082]
2083[Don't link the GHC RTS into our C-only programs
2084Ian Lynagh <igloo@earth.li>**20101210185402
2085 Ignore-this: 56f620f7eb16a03e7497a161bc48458e
2086]
2087[Build a copy of ghc-cabal with the in-tree compiler, for the bindist
2088Ian Lynagh <igloo@earth.li>**20101210181123]
2089[Add a test that all programs in the bindist were built with the right GHC
2090Ian Lynagh <igloo@earth.li>**20101210161218
2091 They should use the GHC from the build tree, not the bootstrapping compiler.
2092]
2093[Fix Trac #4534: renamer bug
2094simonpj@microsoft.com**20101210084530
2095 Ignore-this: 8163bfa3a56344cfe89ad17c62e9655d
2096   
2097 The renamer wasn't attaching the right used-variables to a
2098 TransformStmt constructor.
2099 
2100 The real modification is in RnExpr; the rest is just
2101 pretty-printing and white space.
2102]
2103[White space only
2104simonpj@microsoft.com**20101210084255
2105 Ignore-this: 3fcf8a4fc8c15052c379a135951d53ea
2106]
2107[Comments only
2108simonpj@microsoft.com**20101210084116
2109 Ignore-this: 55bb1de129b1c9513751885eaa84b884
2110]
2111[Make the case-to-let transformation a little less eager
2112simonpj@microsoft.com**20101208172251
2113 Ignore-this: 55eaa1b5753af31aeb32ec792cb6b662
2114 
2115 See Note [Case elimination: lifted case].
2116 Thanks to Roman for identifying this case.
2117]
2118[warning fix: don't redefine BLOCKS_PER_MBLOCK
2119Simon Marlow <marlowsd@gmail.com>**20101210094002
2120 Ignore-this: cadba57f1c38f5e2af1de37d0a79c7ee
2121]
2122[Only reset the event log if logging is turned on (addendum to #4512)
2123Simon Marlow <marlowsd@gmail.com>**20101210093951
2124 Ignore-this: c9f85f0de2b11a37337672fba59aecc6
2125]
2126[allocate enough room for the longer filename (addendum to #4512)
2127Simon Marlow <marlowsd@gmail.com>**20101210093906
2128 Ignore-this: 270dc0219d98f1e0f9e006102ade7087
2129]
2130[Fix Windows build: move rtsTimerSignal to the POSIX-only section
2131Simon Marlow <marlowsd@gmail.com>**20101210090045
2132 Ignore-this: aa1844b70b9f1a44447787c4bbe10d44
2133]
2134[Default the value of -dppr-cols when the static flags aren't initialised yet
2135Ben Lippmeier <benl@ouroborus.net>**20101210060154
2136 Ignore-this: 4cea29085ef904f379a8829714c9e180
2137 If GHC's command line options are bad then the options parser uses the
2138 pretty printer before the -dppr-cols flag has been read.
2139]
2140[Defensify naked read in LLVM mangler
2141Ben Lippmeier <benl@ouroborus.net>**20101210045922
2142 Ignore-this: 1373f597863851bd03e7a7254558ed04
2143]
2144[Formatting only
2145Ben Lippmeier <benl@ouroborus.net>**20101210042600
2146 Ignore-this: 20bbcd95c70b59094d0bb8a63e459103
2147]
2148[Always ppr case alts on separate lines
2149Ben Lippmeier <benl@ouroborus.net>**20101208070508
2150 Ignore-this: 7e2edd57a61427621aeb254aef84f0f7
2151]
2152[Add -dppr-colsN to set width of dumps
2153Ben Lippmeier <benl@ouroborus.net>**20101208070245
2154 Ignore-this: edc64fee6c373b895bb80b83b549ce1a
2155]
2156[Add -dppr-case-as-let to print "strict lets" as actual lets
2157Ben Lippmeier <benl@ouroborus.net>**20101208065548
2158 Ignore-this: eb1d122dbd73b5263cae3a9f8259a838
2159]
2160[Suppress more info with -dsuppress-idinfo
2161Ben Lippmeier <benl@ouroborus.net>**20101208063037
2162 Ignore-this: 5e8213d7b8d2905e245917aa3e83efc5
2163]
2164[Implement -dsuppress-type-signatures
2165Ben Lippmeier <benl@ouroborus.net>**20101208062814
2166 Ignore-this: 34dbefe5f8d7fe58ecb26d6a748d1c71
2167]
2168[Add more suppression flags
2169Ben Lippmeier <benl@ouroborus.net>**20101208020723
2170 Ignore-this: b010ba9789a2fde6b815f33494fcc23c
2171  -dsuppress-all
2172  -dsuppress-type-applications
2173  -dsuppress-idinfo
2174]
2175[fix ticket number (#4505)
2176Simon Marlow <marlowsd@gmail.com>**20101209120404
2177 Ignore-this: 5769c5ce2a8d69d62d977a9ae138ec23
2178]
2179[fix warnings
2180Simon Marlow <marlowsd@gmail.com>**20101209115844
2181 Ignore-this: ffff37feb2abbfc5bd12940c7007c208
2182]
2183[Catch too-large allocations and emit an error message (#4505)
2184Simon Marlow <marlowsd@gmail.com>**20101209114005
2185 Ignore-this: c9013ab63dd0bd62ea045060528550c6
2186 
2187 This is a temporary measure until we fix the bug properly (which is
2188 somewhat tricky, and we think might be easier in the new code
2189 generator).
2190 
2191 For now we get:
2192 
2193 ghc-stage2: sorry! (unimplemented feature or known bug)
2194   (GHC version 7.1 for i386-unknown-linux):
2195         Trying to allocate more than 1040384 bytes.
2196 
2197 See: http://hackage.haskell.org/trac/ghc/ticket/4550
2198 Suggestion: read data from a file instead of having large static data
2199 structures in the code.
2200]
2201[Export the value of the signal used by scheduler (#4504)
2202Dmitry Astapov <dastapov@gmail.com>**20101208183755
2203 Ignore-this: 427bf8c2469283fc7a6f759440d07d87
2204]
2205[Tweak the "sorry" message a bit
2206Simon Marlow <marlowsd@gmail.com>**20101208163212
2207 Ignore-this: aa1ce5bc3c27111548204b740572efbe
2208 
2209 -              "sorry! (this is work in progress)\n"
2210 +              "sorry! (unimplemented feature or known bug)\n"
2211]
2212[:unset settings support
2213Boris Lykah <lykahb@gmail.com>**20101123190132
2214 Ignore-this: 5e97c99238f5d2394592858c34c004d
2215 Added support for settings [args, prog, prompt, editor and stop].
2216 Now :unset supports the same set of options as :set.
2217]
2218[Fix Windows memory freeing: add a check for fb == NULL; fixes trac #4506
2219Ian Lynagh <igloo@earth.li>**20101208152349
2220 Also added a few comments, and a load of code got indented 1 level deeper.
2221]
2222[Fixes for #4512: EventLog.c - provides ability to terminate event logging, Schedule.c - uses them in forkProcess.
2223Dmitry Astapov <dastapov@gmail.com>**20101203133950
2224 Ignore-this: 2da7f215d6c22708a18291a416ba8881
2225]
2226[Make CPPFLAGS variables, as well as CFLAGS and LDFLAGS
2227Ian Lynagh <igloo@earth.li>**20101207010033
2228 Ignore-this: 2fc1ca1422aae1988d0fe1d29a8485d9
2229 This fixes the "does unsetenv return void" test in the unix package on
2230 OS X, if I tell it to make 10.4-compatible binaries. The test uses
2231 CPPFLAGS but not CFLAGS, so it thought it returned int (as it was
2232 in 10.5-mode), but the C compiler (using CFLAGS, so in 10.4 mode)
2233 thought it returned void.
2234 
2235 I also added CONF_LD_OPTS_STAGE$3 to the list of things in LDFLAGS,
2236 which looks like an accidental ommission.
2237]
2238[Add a configure message
2239Ian Lynagh <igloo@earth.li>**20101206215201]
2240[Link even programs containing no Haskell modules with GHC
2241Ian Lynagh <igloo@earth.li>**20101206203329
2242 I don't remember why we made it use gcc instead, but going back to
2243 using ghc doesn't seem to break anything, and should fix the build
2244 on OS X 10.6.
2245]
2246[Correct the stage that the includes/ tools are built in
2247Ian Lynagh <igloo@earth.li>**20101206203125]
2248[Tweak the cleaning of inplace/; fixes trac #4320
2249Ian Lynagh <igloo@earth.li>**20101205212048]
2250[Close .ghci files after reading them; fixes trac #4487
2251Ian Lynagh <igloo@earth.li>**20101205205301]
2252[Fix the behaviour of :history for ticks surrounding top level functions
2253pepeiborra@gmail.com**20101203202346
2254 Ignore-this: 8059d4859c52c0c9a235b937cb8cde1d
2255]
2256[Don't warn of duplicate exports in case of module exports.
2257Michal Terepeta <michal.terepeta@gmail.com>**20101127212116
2258 Ignore-this: ea225d517826f971c400bbb68d1405b8
2259 
2260 But only when the module exports refer to different modules.
2261 See ticket #4478.
2262]
2263[Fix whitespace/layout in RnNames.
2264Michal Terepeta <michal.terepeta@gmail.com>**20101030171303
2265 Ignore-this: 707a7955fc4fc51683cc5a1dfe57f93
2266]
2267[Tell gcc to support back to OS X 10.5
2268Ian Lynagh <igloo@earth.li>**20101203201558
2269 Ignore-this: f02d70e5b9cce50137981c6cb2b62a18
2270]
2271[Make RelaxedLayout off by default
2272Ian Lynagh <igloo@earth.li>**20101202140808
2273 I suspect this is a vary rarely used extension to the official layout
2274 rule.
2275]
2276[Fix up TcInstDcls
2277simonpj@microsoft.com**20101203180758
2278 Ignore-this: 9311aeb4ee67c799704afec90b5982d0
2279 
2280 I really don't know how this module got left out of my last
2281 patch, namely
2282   Thu Dec  2 12:35:47 GMT 2010  simonpj@microsoft.com
2283   * Re-jig simplifySuperClass (again)
2284 
2285 I suggest you don't pull either the patch above, or this
2286 one, unless you really have to.  I'm not fully confident
2287 that it works properly yet.  Ran out of time. Sigh.
2288]
2289[throwTo: report the why_blocked value in the barf()
2290Simon Marlow <marlowsd@gmail.com>**20101203094840
2291 Ignore-this: 3b167c581be1c51dfe3586cc6359e1d0
2292]
2293[handle ThreadMigrating in throwTo() (#4811)
2294Simon Marlow <marlowsd@gmail.com>**20101203094818
2295 Ignore-this: 8ef8cb7fd3b50a27f83c29968131d461
2296 If a throwTo targets a thread that has just been created with
2297 forkOnIO, then it is possible the exception strikes while the thread
2298 is still in the process of migrating.  throwTo() didn't handle this
2299 case, but it's fairly straightforward.
2300]
2301[removeThreadFromQueue: stub out the link field before returning (#4813)
2302Simon Marlow <marlowsd@gmail.com>**20101202160838
2303 Ignore-this: 653ae17bc1120d7f4130da94665002a1
2304]
2305[small tidyup
2306Simon Marlow <marlowsd@gmail.com>**20101126140620
2307 Ignore-this: 70b1d5ed4c81a7b29dd5980a2d84aae1
2308]
2309[Fix a recomp bug: make classes/datatypes depend directly on DFuns (#4469)
2310Simon Marlow <marlowsd@gmail.com>**20101202122349
2311 Ignore-this: 61c765583bb1d97caa88cf9b4f45b87c
2312 And remove the old mechanism of recording dfun uses separately,
2313 because it didn't work.
2314 
2315 This wiki page describes recompilation avoidance and fingerprinting.
2316 I'll update it to describe the new method and what went wrong with the
2317 old method:
2318 
2319 http://hackage.haskell.org/trac/ghc/wiki/Commentary/Compiler/RecompilationAvoidance
2320]
2321[make a panic message more informative and suggest -dcore-lint (see #4534)
2322Simon Marlow <marlowsd@gmail.com>**20101201151706
2323 Ignore-this: 2a10761925d6f9f52675948baa30f7a
2324]
2325[Re-jig simplifySuperClass (again)
2326simonpj@microsoft.com**20101202123547
2327 Ignore-this: fe4062b8988258f6748ebd8fbd6515b5
2328 
2329 This fixes the current loop in T3731, and will fix other
2330 reported loops.  The loops show up when we are generating
2331 evidence for superclasses in an instance declaration.
2332 
2333 The trick is to make the "self" dictionary simplifySuperClass
2334 depend *explicitly* on the superclass we are currently trying
2335 to build.  See Note [Dependencies in self dictionaries] in TcSimplify.
2336 
2337 That in turn means that EvDFunApp needs a dependency-list, used
2338 when chasing dependencies in isGoodRecEv.
2339]
2340[A little refactoring (remove redundant argument passed to isGoodRecEv)
2341simonpj@microsoft.com**20101202123110
2342 Ignore-this: e517c5c12109a230f08dafb4d1e386df
2343]
2344[Make rebindable if-then-else a little more permissive
2345simonpj@microsoft.com**20101202122540
2346 Ignore-this: ddb552cfe307607b42d1e4baf4e3bf21
2347 
2348 See Note [Rebindable syntax for if].  Fixes Trac #4798.
2349 Thanks to Nils Schweinsberg <mail@n-sch.de>
2350]
2351[Improve error message (Trac #4799)
2352simonpj@microsoft.com**20101202102706
2353 Ignore-this: d9896e4d182936de1f256c820b96a8cf
2354]
2355[Fix a nasty bug in RULE matching: Trac #4814
2356simonpj@microsoft.com**20101202102618
2357 Ignore-this: ba058ad46a02bd2faf3a14de93fd19c6
2358 
2359 See Note [Matching lets], which explains it all in detail.
2360 It took me a day to think of a nice way to fix the bug,
2361 but I think the result is quite respectable. Subtle, though.
2362]
2363[Rename -XPArr to -XParallelArrays
2364Ben Lippmeier <benl@ouroborus.net>**20101130075415
2365 Ignore-this: 21b37680a7f25800d1200b59ad0b6b39
2366]
2367[FIX #1845 (unconditional relative branch out of range)
2368pho@cielonegro.org**20101130143014
2369 Ignore-this: df234bd8ad937104c455656fe3c33732
2370 
2371 Don't use mmap on powerpc-apple-darwin as mmap doesn't support
2372 reallocating but we need to allocate jump islands just after each
2373 object images. Otherwise relative branches to jump islands can fail
2374 due to 24-bits displacement overflow.
2375]
2376[rts/Linker.c (loadArchive):
2377pho@cielonegro.org**20101130142700
2378 Ignore-this: bc84f9369ce5c2d289440701b7a3a2ab
2379 
2380 This routine should be aware of Mach-O misalignment of malloc'ed memory regions.
2381]
2382[rts/Linker.c (machoGetMisalignment):
2383pho@cielonegro.org**20101130123355
2384 Ignore-this: 75425600049efd587e9873578e26392f
2385 
2386 Use fseek(3) instead of rewind(3) to move the file position indicator back to the initial position. Otherwise we can't use this function in loadArchive().
2387]
2388[rts/Linker.c (ocFlushInstructionCache):
2389pho@cielonegro.org**20101130121425
2390 Ignore-this: 1e2c207e4b1d17387617ec5d645204b7
2391 
2392 I found this function causes a segfault when ocAllocateSymbolExtras() has allocated a separate memory region for jump islands.
2393]
2394[Remove NewQualifiedOperators
2395Ian Lynagh <igloo@earth.li>**20101201181117
2396 The extension was rejected by Haskell', and deprecated in 7.0.
2397]
2398[fix ref to utils/ext-core, which moved to Hackage (extcore package)
2399Simon Marlow <marlowsd@gmail.com>**20101201092147
2400 Ignore-this: 272a7daaa335ef60bcc645db70b4d68b
2401]
2402[fix floating-point/FFI section: fenv is C99, not POSIX
2403Simon Marlow <marlowsd@gmail.com>**20101201092119
2404 Ignore-this: ce8b3edd428e4f77691dd739b5b4ae73
2405]
2406[Fixed some 'unused vars' warnings
2407keller@cse.unsw.edu.au**20101130013425
2408 Ignore-this: 35790d443faa23b87e4ba442e62376a3
2409]
2410[vectScalarLam handles int, float, and double now
2411keller@cse.unsw.edu.au**20101129231043
2412 Ignore-this: 6d67bdc8dd8577184040e791e6f3d0
2413]
2414[Handling of lets, letrec and case when checking if a lambda expr needs to be vectorised
2415keller@cse.unsw.edu.au**20101115051225
2416 Ignore-this: 1db6ed63d7b3f6d093e019322b407ff7
2417]
2418[Document the behaviour of fenv.h functions with GHC (#4391)
2419Simon Marlow <marlowsd@gmail.com>**20101126125336
2420 Ignore-this: bc4eab49428d567505a28add6fed90f1
2421]
2422[Remove the no-ghci-lib warning in ghc-pkg
2423Ian Lynagh <igloo@earth.li>**20101127235805
2424 GHCi libs are no longer necessary, as we can use the .a or .so versions
2425 instead.
2426]
2427[Add GNU-variant support to the .a parser, and other improvements/tidyups
2428Ian Lynagh <igloo@earth.li>**20101127223945]
2429[Re-indent only
2430Ian Lynagh <igloo@earth.li>**20101127191646]
2431[Improve linker debugging for archive files
2432Ian Lynagh <igloo@earth.li>**20101127190907]
2433[Always enable the archive-loading code
2434Ian Lynagh <igloo@earth.li>**20101127173000
2435 If the GHCi .o lib doesn't exist, load the .a instead
2436]
2437[Inherit the ForceSpecConstr flag in non-recursive nested bindings
2438Roman Leshchinskiy <rl@cse.unsw.edu.au>**20101127125025
2439 Ignore-this: 401391eae25cefcb4afaba2e357decc1
2440 
2441 This makes sure that join points are fully specialised in loops which are
2442 marked as ForceSpecConstr.
2443]
2444[Document -ddump-rule-firings and -ddump-rule-rewrites
2445Roman Leshchinskiy <rl@cse.unsw.edu.au>**20101127123528
2446 Ignore-this: beade2efe0cd767c0ce9d4f45a3380ba
2447]
2448[New flag -dddump-rule-rewrites
2449Roman Leshchinskiy <rl@cse.unsw.edu.au>**20101127122022
2450 Ignore-this: c0ef5b8a199fbd1ef020258d2cde85a3
2451 
2452 Now, -ddump-rule-firings only shows the names of the rules that fired (it would
2453 show "before" and "after" with -dverbose-core2core previously) and
2454 -ddump-rule-rewrites always shows the "before" and "after" bits, even without
2455 -dverbose-core2core.
2456]
2457[Acutally, wild-card variables *can* have occurrences
2458simonpj@microsoft.com**20101126162409
2459 Ignore-this: 544bffed75eeccef03a1097f98524eea
2460 
2461 This patch removes the Lint test, and comments why
2462]
2463[Tidy up the handling of wild-card binders, and make Lint check it
2464simonpj@microsoft.com**20101126133210
2465 Ignore-this: 9e0be9f7867d53046ee5b0e478a0f433
2466 
2467 See Note [WildCard binders] in SimplEnv.  Spotted by Roman.
2468]
2469[Substitution should just substitute, not optimise
2470simonpj@microsoft.com**20101125172356
2471 Ignore-this: 657628d9b6796ceb5f915c43d56e4a06
2472 
2473 This was causing Trac #4524, by optimising
2474      (e |> co)  to   e
2475 on the LHS of a rule.  Result, the template variable
2476 'co' wasn't bound any more.
2477 
2478 Now that substition doesn't optimise, it seems sensible to call
2479 simpleOptExpr rather than substExpr when substituting in the
2480 RHS of rules.  Not a big deal either way.
2481]
2482[Make SpecConstr "look through" identity coercions
2483simonpj@microsoft.com**20101125172138
2484 Ignore-this: c1cc585ed890a7702c33987e971e0af6
2485]
2486[Comment only
2487simonpj@microsoft.com**20101125172011
2488 Ignore-this: 3c7be8791badd00dcca9610ebb8981d1
2489]
2490[White space only
2491simonpj@microsoft.com**20101101080748
2492 Ignore-this: f7133fc6b22ae263c6672543a8534a6f
2493]
2494[Keep a maximum of 6 spare worker threads per Capability (#4262)
2495Simon Marlow <marlowsd@gmail.com>**20101125135729
2496 Ignore-this: a020786569656bf2f3a1717b65d463bd
2497]
2498[Unicide OtherNumber category should be allowed in identifiers (#4373)
2499Simon Marlow <marlowsd@gmail.com>**20101115095444
2500 Ignore-this: e331b6ddb17550163ee91bd283348800
2501]
2502[vectoriser: fix warning
2503Ben Lippmeier <benl@ouroborus.net>**20101126044036
2504 Ignore-this: e1a66bb405bf2f3f56b42c3b13fd4bf3
2505]
2506[vectoriser: fix warning
2507Ben Lippmeier <benl@ouroborus.net>**20101126042950
2508 Ignore-this: df8dd25bcfb3946c2974b13953a2f2c7
2509]
2510[vectoriser: take class directly from the instance tycon
2511Ben Lippmeier <benl@ouroborus.net>**20101126042900
2512 Ignore-this: 626a416717a5a059f39e53f4ec95fc66
2513]
2514[vectoriser: comments only
2515Ben Lippmeier <benl@ouroborus.net>**20101125073201
2516 Ignore-this: 8846ea8895307083bd1ebbc5d7fb1c5
2517]
2518[vectoriser: follow changes in mkClass
2519Ben Lippmeier <benl@ouroborus.net>**20101125062349
2520 Ignore-this: d5018cc022686d4272e126ca9a12283a
2521]
2522[vectoriser: tracing wibbles
2523Ben Lippmeier <benl@ouroborus.net>**20101125062332
2524 Ignore-this: c2024d8f03bc03bee2851f4f1c139fd5
2525]
2526[mkDFunUnfolding wants the type of the dfun to be a PredTy
2527benl@ouroborus.net**20100914062939
2528 Ignore-this: 7aa6e6b140746184cf00355b50c83b66
2529]
2530[vectoriser: fix conflicts
2531Ben Lippmeier <benl@ouroborus.net>**20101125060904
2532 Ignore-this: cc3decab1affada8629ca3818b76b3bf
2533]
2534[Comments and formatting only
2535benl@ouroborus.net**20100914062903
2536 Ignore-this: b0fc25f0952cafd56cc25353936327d4
2537]
2538[Comments and formatting to type environment vectoriser
2539benl@ouroborus.net**20100909080405
2540 Ignore-this: ab8549d53f845c9d82ed9a525fda3906
2541]
2542[Don't mix implicit and explicit layout
2543Ian Lynagh <igloo@earth.li>**20101124231514]
2544[Whitespace only
2545Ian Lynagh <igloo@earth.li>**20101124230655]
2546[Separate NondecreasingIndentation out into its own extension
2547Ian Lynagh <igloo@earth.li>**20101124220507]
2548[Add another GHC layout rule relaxation to RelaxedLayout
2549Ian Lynagh <igloo@earth.li>**20101124205957]
2550[Remove an unused build system variable: GhcDir
2551Ian Lynagh <igloo@earth.li>**20101124140455]
2552[Remove unused build system variable: GhcHasEditline
2553Ian Lynagh <igloo@earth.li>**20101124140415]
2554[Remove unused variables from the build system: HBC, NHC, MKDEPENDHS
2555Ian Lynagh <igloo@earth.li>**20101124140052]
2556[Remove references to Haskell 98
2557Ian Lynagh <igloo@earth.li>**20101123233536
2558 They are no longer right, as we have Haskell' generating new Haskell
2559 standards.
2560]
2561[Tweak a configure test
2562Ian Lynagh <igloo@earth.li>**20101123170621]
2563[Add a configure test for the visibility hidden attribute
2564Ian Lynagh <igloo@earth.li>**20101123170541]
2565[sanity: fix places where we weren't filling fresh memory with 0xaa
2566Simon Marlow <marlowsd@gmail.com>**20101029092843
2567 Ignore-this: 2cb18f7f5afcaf33371aeffce67e218f
2568]
2569[Just some alpha renaming
2570Ian Lynagh <igloo@earth.li>**20101121144455
2571 Ignore-this: d5e807c5470840efc199e29f7d50804c
2572]
2573[Fix bug #3165 (:history throws irrefutable pattern failed)
2574pepeiborra@gmail.com**20101115223623
2575 Ignore-this: 73edf56e502b4d0385bc044133b27946
2576 
2577 I ran across this bug and took the time to fix it, closing
2578 a long time due TODO in InteractiveEval.hs
2579 
2580 Instead of looking around to find the enclosing declaration
2581 of a tick, this patch makes use of the information already collected during the
2582 coverage desugaring phase
2583]
2584[For bindists, build ghc-pwd with stage 1
2585Ian Lynagh <igloo@earth.li>**20101121183520
2586 Ignore-this: a3b5c8b78c81ec1b6d5fbf23da346ff5
2587 rather then the bootstrapping compiler. This fixes problems where the
2588 bootstrapping compiler dynamically links against libraries not on the
2589 target machine.
2590]
2591[Makefile tweak
2592Ian Lynagh <igloo@earth.li>**20101121183342
2593 Ignore-this: cd55a2819c1a5fd36da1bc7a75d2ded1
2594]
2595[Fix a makefile include ordering sanity check
2596Ian Lynagh <igloo@earth.li>**20101121174916
2597 Ignore-this: d0bdd41c4b618944d04ecb4f54fdd0f1
2598]
2599[Add an extension for GHC's layout-rule relaxations
2600Ian Lynagh <igloo@earth.li>**20101120215340
2601 Still TODO: Add the other relaxation (#1060) and update the alternative
2602 layout rule to use the extension.
2603]
2604[Tweak the bindist configure.ac.in
2605Ian Lynagh <igloo@earth.li>**20101120173735]
2606[configure.ac tweaks
2607Ian Lynagh <igloo@earth.li>**20101120170245]
2608[When testing the bindist, tell it where gcc is
2609Ian Lynagh <igloo@earth.li>**20101120155920
2610 The location isn't baked into the bindist, as it may differ from
2611 machine to machine.
2612]
2613[Comments only
2614simonpj@microsoft.com**20101119100153
2615 Ignore-this: 7abd5d965ea805770449d6f8dadbb921
2616]
2617[ForceSpecConstr now forces specialisation even for arguments which aren't scrutinised
2618Roman Leshchinskiy <rl@cse.unsw.edu.au>**20101118212839
2619 Ignore-this: db45721d29a694e53746f8b76513efa4
2620]
2621[Move the superclass generation to the canonicaliser
2622simonpj@microsoft.com**20101118120533
2623 Ignore-this: 5e0e525402a240b709f2b8104c1682b2
2624 
2625 Doing superclass generation in the canonicaliser (rather than
2626 TcInteract) uses less code, and is generally more efficient.
2627 
2628 See Note [Adding superclasses] in TcCanonical.
2629 
2630 Fixes Trac #4497.
2631]
2632[Fix the generation of in-scope variables for IfaceLint check
2633simonpj@microsoft.com**20101118090057
2634 Ignore-this: bbcdba61ddf89d07fe69ca99c2017e3f
2635]
2636[Comments only
2637simonpj@microsoft.com**20101118090034
2638 Ignore-this: fa2936d35a0f7be4e4535ea9e2b7bf7b
2639]
2640[Omit bogus test for -XDeriveFunctor
2641simonpj@microsoft.com**20101118090028
2642 Ignore-this: a534243011809ebbb788b910961601c5
2643 
2644 It was duplicated in the case of 'deriving( Functor )'
2645 and wrong for 'deriving( Foldable )'
2646]
2647[Improve error message on advice from a user
2648simonpj@microsoft.com**20101118085306
2649 Ignore-this: bd4f3858ff24e602e985288f27d536f3
2650 
2651 See Trac #4499
2652]
2653[TAG 2010-11-18
2654Ian Lynagh <igloo@earth.li>**20101118011554
2655 Ignore-this: ccadbe7fadd1148d2ee3caa8c8821ec5
2656]
2657Patch bundle hash:
2658087471f2e34533a476f408812b8c550e70ffc0b2