Ticket #1987: ghc1987

File ghc1987, 39.6 KB (added by felixmar, 6 years ago)
Line 
1DarcsURL: http://darcs.haskell.org/ghc
2MIME-Version: 1.0
3Content-Type: multipart/mixed; boundary="=_"
4
5--=_
6Content-Type: text/plain
7Content-Transfer-Encoding: quoted-printable
8
9Mon Dec 17 22:51:33 W. Europe Standard Time 2007  Felix Martini <fmartini@g=
10mail.com>
11  * (#1987) GHCi's config file in <AppData>\ghci folder on Windows
12
13--=_
14Content-Type: text/x-darcs-patch; name="__1987_-ghci_s-config-file-in-_appdata__ghci-folder-on-windows.dpatch"
15Content-Transfer-Encoding: quoted-printable
16Content-Description: A darcs patch for your repository!
17
18
19New patches:
20
21[(#1987) GHCi's config file in <AppData>\ghci folder on Windows
22Felix Martini <fmartini@gmail.com>**20071217215133] {
23hunk ./compiler/ghci/InteractiveUI.hs 333
24+#ifdef mingw32_HOST_OS
25+    let file =3D ".\\ghci.cfg"
26+#else
27hunk ./compiler/ghci/InteractiveUI.hs 337
28+#endif
29hunk ./compiler/ghci/InteractiveUI.hs 343
30-         either_hdl <- io (IO.try (openFile "./.ghci" ReadMode))
31+         either_hdl <- io (IO.try (openFile file ReadMode))
32hunk ./compiler/ghci/InteractiveUI.hs 350
33+#ifdef mingw32_HOST_OS
34+    either_dir <- io (IO.try (getAppUserDataDirectory "ghci"))
35+#else
36hunk ./compiler/ghci/InteractiveUI.hs 354
37+#endif
38hunk ./compiler/ghci/InteractiveUI.hs 360
39+#ifdef mingw32_HOST_OS
40+            let file =3D dir ++ "\\ghci.cfg"
41+#else
42hunk ./compiler/ghci/InteractiveUI.hs 364
43+#endif
44}
45
46Context:
47
48[Fix a bug in gen_contents_index
49Ian Lynagh <igloo@earth.li>**20071212121154
50 The library doc index thought that the docs were in $module.html, rather
51 than $package/$module.html.
52] =
53
54[Fix lifting of case expressions
55Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071215000837
56 =
57
58 We have to explicity check for empty arrays in each alternative as recursi=
59ve
60 algorithms wouldn't terminate otherwise.
61] =
62
63[Use (UArr Int) instead of PArray_Int# in vectorisation
64Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071215000739] =
65
66[Fix bug in VectInfo loading
67Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071214230914] =
68
69[Remove unused vectorisation built-in
70Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071214011015] =
71
72[Treat some standard data cons specially during vectorisation
73Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071213034855
74 =
75
76 This is a temporary hack which allows us to vectorise literals.
77] =
78
79[More vectorisation-related built ins
80Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071213034839] =
81
82[Track changes to package ndp
83Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071212062714] =
84
85[Add vectorisation built-ins
86Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071212040521] =
87
88[FIX #1963: catch Ctrl-C and clean up properly
89Simon Marlow <simonmar@microsoft.com>**20071213154056] =
90
91[Document the new threshold flags
92Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071214003003] =
93
94[Separate and optional size thresholds for SpecConstr and LiberateCase
95Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071214002719
96 =
97
98 This patch replaces -fspec-threshold by -fspec-constr-threshold and
99 -fliberate-case-threshold. The thresholds can be disabled by
100 -fno-spec-constr-threshold and -fno-liberate-case-threshold.
101] =
102
103[Make HscTypes.tyThingId respond not panic on ADataCon
104simonpj@microsoft.com**20071204152903] =
105
106[Use Unix format for RnPat (no other change)
107simonpj@microsoft.com**20071213140532] =
108
109[Improve free-variable handling for rnPat and friends (fixes Trac #1972)
110simonpj@microsoft.com**20071213140213
111 =
112
113 As well as fixing the immediate problem (Trac #1972) this patch does
114 a signficant simplification and refactoring of pattern renaming.
115 =
116
117 Fewer functions, fewer parameters passed....it's all good.  But it
118 took much longer than I expected to figure out.
119 =
120
121 The most significant change is that the NameMaker type does *binding*
122 as well as *making* and, in the matchNameMaker case, checks for unused
123 bindings as well.  This is much tider.
124 =
125
126 (No need to merge to the 6.8 branch, but no harm either.)
127 =
128
129 =
130
131] =
132
133[Allow more than 3 simplifier iterations to be run in phase 0
134Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071213040835
135 =
136
137 The number of iterations during the first run of phase 0 was erroneously
138 hardcoded to 3. It should be *at least* 3 (see comments in code) but can b=
139e
140 more.
141] =
142
143[Document -ddump-simpl-phases
144Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071213040822] =
145
146[New flag: -ddump-simpl-phases
147Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071213040644
148 =
149
150 This outputs the core after each simplifier phase (i.e., it produces less
151 information that -ddump-simpl-iterations).
152] =
153
154[Don't dump simplifier iterations with -dverbose-core2core
155Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071213034635
156 =
157
158 SimonPJ says this is the correct behaviour. We still have
159 -ddump-simpl-iterations.
160] =
161
162["list --simple-output" should be quiet when there are no packages to list
163Simon Marlow <simonmar@microsoft.com>**20071212102230
164 =
165
166 Previously:
167 =
168
169 $ ghc-pkg list --user --simple-output
170 ghc-pkg: no matches
171 $
172 =
173
174 Now:
175 =
176
177 $ ghc-pkg list --user --simple-output
178 $
179] =
180
181[Fix vectorisation bug
182Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071206233015] =
183
184[Vectorisation-related built ins
185Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071206040829] =
186
187[Teach vectorisation about some temporary conversion functions
188Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071206032547] =
189
190[Vectorise case of unit correctly
191Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071205221305] =
192
193[Teach vectorisation about singletonP
194Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071205221240] =
195
196[Optimise desugaring of parallel array comprehensions
197Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071205221213] =
198
199[Teach vectorisation about tuple datacons
200Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071205050221] =
201
202[Track additions to package ndp
203Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071205042649] =
204
205[Track changes to package ndp
206Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071205033859] =
207
208[Improve pretty-printing of InstDecl
209simonpj@microsoft.com**20071210083053
210 =
211
212 Fixes Trac #1966. =
213
214 =
215
216] =
217
218[Comments only
219Pepe Iborra <mnislaih@gmail.com>**20071208204815] =
220
221[Refactoring only
222Pepe Iborra <mnislaih@gmail.com>**20071208195222
223 =
224
225 Suspensions in the Term datatype used for RTTI
226 always get assigned a Type, so there is no reason
227 to juggle around with a (Maybe Type) anymore. =
228
229 =
230
231] =
232
233[Change the format used by :print to show the content of references
234Pepe Iborra <mnislaih@gmail.com>**20071208193013
235     =
236
237     This comes as result of the short discussion linked below.
238     =
239
240     http://www.haskell.org/pipermail/cvs-ghc/2007-December/040049.html
241 =
242
243] =
244
245[Help the user when she tries to do :history without :trace
246Pepe Iborra <mnislaih@gmail.com>**20071208180918
247 =
248
249 Teach GHCi to show a "perhaps you forgot to use :trace?" when
250 it finds that the user is trying to retrieve an empty :history
251 =
252
253] =
254
255[Prevent the binding of unboxed things by :print
256Pepe Iborra <mnislaih@gmail.com>**20071208181830] =
257
258[Coercions from boxy splitters must be sym'ed in pattern matches
259Manuel M T Chakravarty <chak@cse.unsw.edu.au>**20071208105018] =
260
261[Properly keep track of whether normalising given or wanted dicts
262Manuel M T Chakravarty <chak@cse.unsw.edu.au>**20071207071302
263 - The information of whether given or wanted class dictionaries where
264   normalised by rewriting wasn't always correctly propagated in TcTyFuns,
265   which lead to malformed dictionary bindings.
266 - Also fixes a bug in TcPat.tcConPat where GADT equalities where emitted i=
267n
268   the wrong position in case bindings (which led to CoreLint failures).
269] =
270
271[TcPat.tcConPat uses equalities instead of GADT refinement
272Manuel M T Chakravarty <chak@cse.unsw.edu.au>**20071120071208
273 * This patch implements the use of equality constraints instead of GADT
274   refinements that we have been discussing for a while.
275 * It just changes TcPat.tcConPat.  It doesn't have any of the simplificati=
276on
277   and dead code removal that is possible due to this change.
278 * At the moment, this patch breaks a fair number of GADT regression tests.
279] =
280
281[Use installPackage for register --inplace as well as installing
282Ian Lynagh <igloo@earth.li>**20071207234652
283 We also need to do the GHC.Prim hack when registering inplace or the
284 tests that use it fail.
285] =
286
287[Fix the libraries Makefile
288Ian Lynagh <igloo@earth.li>**20071205125015
289     x && y
290 is not the same as
291     if x; then y; fi
292 as the latter doesn't fail when x fails
293] =
294
295[Copy hscolour.css into dist/... so it gets installed with the library docs
296Ian Lynagh <igloo@earth.li>**20071205013703] =
297
298[Add the hscolour.css from hscolour 1.8
299Ian Lynagh <igloo@earth.li>**20071205011733] =
300
301[BIN_DIST_INST_SUBDIR Needs to be defined in config.mk so ./Makefile can se=
302e it
303Ian Lynagh <igloo@earth.li>**20071207121317] =
304
305[#include ../includes/MachRegs.h rather than just MachRegs.h
306Ian Lynagh <igloo@earth.li>**20071205170335
307 This fixes building on NixOS. I'm not sure why it worked everywhere else,
308 but not on NixOS, before.
309] =
310
311[Fix bindist creation: readline/config.mk is gone
312Ian Lynagh <igloo@earth.li>**20071203123031] =
313
314[FIX #1843: Generate different instructions on PPC
315Ian Lynagh <igloo@earth.li>**20071203123237
316 The old ones caused lots of
317     unknown scattered relocation type 4
318 errors. Patch from Chris Kuklewicz.
319] =
320
321[Refactor gen_contents_index
322Ian Lynagh <igloo@earth.li>**20071207183538
323 Also fixes it with Solaris's sh, spotted by Christian Maeder
324] =
325
326[Use GHC.Exts rather than GHC.Prim
327Ian Lynagh <igloo@earth.li>**20071202234222] =
328
329[Alter the base:GHC.Prim hack in installPackage, following changes in base
330Ian Lynagh <igloo@earth.li>**20071202215719] =
331
332[Remove debug warning, and explain why
333simonpj@microsoft.com**20071207170507] =
334
335[comment only
336Simon Marlow <simonmar@microsoft.com>**20071206092422] =
337
338[comment typo
339Simon Marlow <simonmar@microsoft.com>**20071206092412] =
340
341[add Outputable instance for OccIfaceEq
342Simon Marlow <simonmar@microsoft.com>**20071206092403] =
343
344[Workaround for #1959: assume untracked names have changed
345Simon Marlow <simonmar@microsoft.com>**20071206092349
346 This fixes the 1959 test, but will do more recompilation than is
347 strictly necessary (but only when -O is on).  Still, more
348 recompilation is better than segfaults, link errors or other random
349 breakage.
350] =
351
352[FIX part of #1959: declaration versions were not being incremented correct=
353ly
354Simon Marlow <simonmar@microsoft.com>**20071206084556
355 We were building a mapping from ModuleName to [Occ] from the usage
356 list, using the usg_mod field as the key.  Unfortunately, due to a
357 very poor naming decision, usg_mod is actually the module version, not
358 the ModuleName.  usg_name is the ModuleName.  Since Version is also an
359 instance of Uniquable, there was no type error: all that happened was
360 lookups in the map never succeeded.  I shall rename the fields of
361 Usage in a separate patch.
362 =
363
364 This doesn't completely fix #1959, but it gets part of the way there.
365 =
366
367 I have to take partial blame as the person who wrote this fragment of
368 code in late 2006 (patch "Interface file optimisation and removal of
369 nameParent").
370] =
371
372[move FP_FIND_ROOT after the "GHC is required" check
373Simon Marlow <simonmar@microsoft.com>**20071205101814] =
374
375[FIX #1110: hackery also needed when running gcc for CPP
376Simon Marlow <simonmar@microsoft.com>**20071205150230] =
377
378[Teach :print to follow references (STRefs and IORefs)
379Pepe Iborra <mnislaih@gmail.com>**20071204105511
380 =
381
382 Prelude Data.IORef> :p l
383 l =3D (_t4::Maybe Integer) : (_t5::[Maybe Integer])
384 Prelude Data.IORef> p <- newIORef l
385 Prelude Data.IORef> :p p
386 p =3D GHC.IOBase.IORef (GHC.STRef.STRef {((_t6::Maybe Integer) :
387                                         (_t7::[Maybe Integer]))})
388 Prelude Data.IORef> :sp p
389 p =3D GHC.IOBase.IORef (GHC.STRef.STRef {(_ : _)})
390 =
391
392 =
393
394 I used braces to denote the contents of a reference.
395 Perhaps there is a more appropriate notation?
396] =
397
398[refactoring only
399Pepe Iborra <mnislaih@gmail.com>**20071202125400] =
400
401[Change --shared to -shared in Win32 DLL docs
402simonpj@microsoft.com**20071204154023] =
403
404[protect console handler against concurrent access (#1922)
405Simon Marlow <simonmar@microsoft.com>**20071204153918] =
406
407[Make eta reduction check more carefully for bottoms (fix Trac #1947)
408simonpj@microsoft.com**20071204145803
409 =
410
411 Eta reduction was wrongly transforming
412        f =3D \x. f x
413 to
414        f =3D f
415 =
416
417 Solution: don't trust f's arity information; instead look at its
418 unfolding.  See Note [Eta reduction conditions]
419 =
420
421 Almost all the new lines are comments!
422 =
423
424 =
425
426] =
427
428[Improve inlining for INLINE non-functions
429simonpj@microsoft.com**20071204114955
430        =
431
432 (No need to merge to 6.8, but no harm if a subsequent patch needs it.)
433 =
434
435 The proximate cause for this patch is to improve the inlining for INLINE
436 things that are not functions; this came up in the NDP project.  See
437 Note [Lone variables] in CoreUnfold.
438 =
439
440 This caused some refactoring that actually made things simpler.  In =
441
442 particular, more of the inlining logic has moved from SimplUtils to =
443
444 CoreUnfold, where it belongs.
445 =
446
447 =
448
449] =
450
451[fix race conditions in sandboxIO (#1583, #1922, #1946)
452Simon Marlow <simonmar@microsoft.com>**20071204114444
453 using the new block-inheriting forkIO (#1048)
454] =
455
456[:cd with no argument goes to the user's home directory
457Simon Marlow <simonmar@microsoft.com>**20071204113945
458 Seems better than getting a confusing 'cannot find directory' exception.
459] =
460
461[forkIO starts the new thread blocked if the parent is blocked (#1048)
462Simon Marlow <simonmar@microsoft.com>**20071204110947] =
463
464[Improve eta reduction, to reduce Simplifier iterations
465simonpj@microsoft.com**20071203150039
466 =
467
468 I finally got around to investigating why the Simplifier was sometimes
469 iterating so often.  There's a nice example in Text.ParserCombinators.Read=
470Prec,
471 which produced:
472 =
473
474 NOTE: Simplifier still going after 3 iterations; bailing out.  Size =3D 33=
4759
476 NOTE: Simplifier still going after 3 iterations; bailing out.  Size =3D 33=
4779
478 NOTE: Simplifier still going after 3 iterations; bailing out.  Size =3D 33=
4799
480 =
481
482 No progress is being made.  It turned out that an interaction between
483 eta-expansion, casts, and eta reduction was responsible. The change is
484 small and simple, in SimplUtils.mkLam: do not require the body to be
485 a Lam when floating the cast outwards.  =
486
487 =
488
489 I also discovered a missing side condition in the same equation, so fixing
490 that is good too.  Now there is no loop when compiling ReadPrec.
491 =
492
493 Should do a full nofib run though.
494 =
495
496] =
497
498[Don't default to stripping binaries when installing
499Ian Lynagh <igloo@earth.li>**20071202195817] =
500
501[Improve pretty-printing for Insts
502simonpj@microsoft.com**20071128173125] =
503
504[Reorganise TcSimplify (again); FIX Trac #1919
505simonpj@microsoft.com**20071128173146
506 =
507
508 This was a bit tricky.  We had a "given" dict like (d7:Eq a); then it got
509 supplied to reduceImplication, which did some zonking, and emerged with
510 a "needed given" (d7:Eq Int). That got everything confused.
511 =
512
513 I found a way to simplify matters significantly.  Now reduceContext
514        - first deals with methods/literals/dictionaries
515        - then deals with implications
516 Separating things in this way not only made the bug go away, but
517 eliminated the need for the recently-added "needed-givens" results returne=
518d
519 by checkLoop.  Hurrah.
520 =
521
522 It's still a swamp.  But it's a bit better.
523 =
524
525] =
526
527[FIX #1914: GHCi forgot all the modules that were loaded before an error
528Simon Marlow <simonmar@microsoft.com>**20071130130734] =
529
530[FIX #1744: ignore the byte-order mark at the beginning of a file
531Simon Marlow <simonmar@microsoft.com>**20071130101100] =
532
533[FIX Trac #1935: generate superclass constraints for derived classes
534simonpj@microsoft.com**20071128150541
535 =
536
537 This bug only reports a problem with phantom types, but actually
538 there was quite a long-standing and significant omission in the
539 constraint generation for derived classes.  See
540 Note [Superclasses of derived instance] in TcDeriv.
541 =
542
543 The test deriving-1935 tests both cases.
544 =
545
546 =
547
548] =
549
550[Print a bit more info in VarBinds (no need to merge)
551simonpj@microsoft.com**20071128150354] =
552
553[Check for duplicate bindings in CoreLint
554simonpj@microsoft.com**20071128150214] =
555
556[add comment
557Simon Marlow <simonmar@microsoft.com>**20071128111417] =
558
559[FIX #1916: don't try to convert float constants to int in CMM optimizer
560Bertram Felgenhauer <int-e@gmx.de>**20071122095513] =
561
562[give a more useful message when the static flags have not been initialised=
563 (#1938)
564Simon Marlow <simonmar@microsoft.com>**20071127135435] =
565
566[Rebuild utils with the stage1 compiler when making a bindist; fixes trac #=
5671860
568Ian Lynagh <igloo@earth.li>**20071127203959
569 This is a bit unpleasant, as "make binary-dist" really shouldn't actually
570 build anything, but it works.
571] =
572
573[Remove the --print-docdir flag
574Ian Lynagh <igloo@earth.li>**20071127195605
575 It wasn't doing the right thing for bindists. Let's rethink...
576] =
577
578[FIX #1925: the interpreter was not maintaining tag bits correctly
579Simon Marlow <simonmar@microsoft.com>**20071127122614
580 See comment for details
581] =
582
583[add missing instruction: ALLOC_AP_NOUPD
584Simon Marlow <simonmar@microsoft.com>**20071127122604] =
585
586[Check tag bits on the fun pointer of a PAP
587Simon Marlow <simonmar@microsoft.com>**20071126160420] =
588
589[canonicalise the path to HsColour
590Simon Marlow <simonmar@microsoft.com>**20071126141614] =
591
592[Consistently put www. on the front of haskell.org in URLs
593Ian Lynagh <igloo@earth.li>**20071126215256] =
594
595[Fix some more URLs
596Ian Lynagh <igloo@earth.li>**20071126214147] =
597
598[Tweak some URLs
599Ian Lynagh <igloo@earth.li>**20071126194148] =
600
601[Fix some links
602Ian Lynagh <igloo@earth.li>**20071126184406] =
603
604[Copy gmp stamps into bindists, so we don't try and rebuild gmp
605Ian Lynagh <igloo@earth.li>**20071125211919] =
606
607[On Windows, Delete the CriticalSection's we Initialize
608Ian Lynagh <igloo@earth.li>**20071125125845] =
609
610[On Windows, add a start menu link to the flag reference
611Ian Lynagh <igloo@earth.li>**20071125124429] =
612
613[Remove html/ from the paths we put in the start menu on Windows
614Ian Lynagh <igloo@earth.li>**20071125124150] =
615
616[MERGED: Make ":" in GHCi repeat the last command
617Ian Lynagh <igloo@earth.li>**20071125122020
618 Ian Lynagh <igloo@earth.li>**20071124231857
619  It used to be a synonym for ":r" in 6.6.1, but this wasn't documented or
620  known about by the developers. In 6.8.1 it was accidentally broken.
621  This patch brings it back, but as "repeat the last command", similar to
622  pressing enter in gdb. This is almost as good for people who want it to
623  reload, and means that it can also be used to repeat commands like :step.
624] =
625
626[MERGED: Put library docs in a $pkg, rather than $pkgid, directory; fixes t=
627rac #1864
628Ian Lynagh <igloo@earth.li>**20071124212305
629 Ian Lynagh <igloo@earth.li>**20071124171220
630] =
631
632[Don't make a library documentation prologue
633Ian Lynagh <igloo@earth.li>**20071124211943
634 It's far too large now, and no-one complained when 6.8.1 didn't have one.
635] =
636
637[Don't put package version numbers in links in index.html
638Ian Lynagh <igloo@earth.li>**20071124211629] =
639
640[Define install-strip in Makefile
641Ian Lynagh <igloo@earth.li>**20071124205037] =
642
643[Define install-strip in distrib/Makefile
644Ian Lynagh <igloo@earth.li>**20071124204803] =
645
646[Install gmp from bindists; fixes trac #1848
647Ian Lynagh <igloo@earth.li>**20071124185240] =
648
649[(native gen) fix code generated for GDTOI on x86_32
650Bertram Felgenhauer <int-e@gmx.de>**20071121063942
651 See trac #1910.
652] =
653
654[Copy the INSTALL hack from mk/config.mk.in into distrib/Makefile-bin-vars.=
655in
656Ian Lynagh <igloo@earth.li>**20071124163028
657 configure will set INSTALL to ./install-sh if it can't find it in the path=
658,
659 so we need to replace the . with the path to our root.
660] =
661
662[Make install-sh executable /before/ we try to find it
663Ian Lynagh <igloo@earth.li>**20071124162450] =
664
665[Document --info in the +RTS -? help
666Ian Lynagh <igloo@earth.li>**20071123204352] =
667
668[MERGED: If we have hscolour then make source code links in teh haddock doc=
669s
670Ian Lynagh <igloo@earth.li>**20071123233113
671 Fri Nov 23 13:15:59 PST 2007  Ian Lynagh <igloo@earth.li>
672] =
673
674[Tidy and trim the type environment in mkBootModDetails
675simonpj@microsoft.com**20071123153519
676 =
677
678 Should fix Trac #1833
679 =
680
681 We were failing to trim the type envt in mkBootModDetails, so several
682 functions all called (*), for example, were getting into the interface.
683 Result chaos.  It only actually bites when we do the retyping-loop thing,
684 which is why it's gone so long without a fix.
685 =
686
687 =
688
689] =
690
691[refactor: HscNothing and boot modules do not need desugaring
692Simon Marlow <simonmar@microsoft.com>**20071123135237] =
693
694[FIX #1910: fix code generated for GDTOI on x86_32
695Bertram Felgenhauer <int-e@gmx.de>*-20071121102627] =
696
697[Properly ppr InstEqs in wanteds of implication constraints
698Manuel M T Chakravarty <chak@cse.unsw.edu.au>**20071122093002] =
699
700[FIX #1910: fix code generated for GDTOI on x86_32
701Bertram Felgenhauer <int-e@gmx.de>**20071121102627] =
702
703[Add built-in Double operations to vectorisation
704Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071122002517] =
705
706[Teach vectorisation about Double
707Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071121054932] =
708
709[Vectorise polyexprs with notes
710Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071121053102] =
711
712[Make rebindable do-notation behave as advertised
713simonpj@microsoft.com**20071121174914
714 =
715
716 Adopt Trac #1537.  The patch ended up a bit bigger than I expected,
717 so I suggest we do not merge this into the 6.8 branch.  But there
718 is no funadamental reason why not.
719 =
720
721 With this patch, rebindable do-notation really does type as if you
722 had written the original (>>) and (>>=3D) operations in desguared form.
723 =
724
725 I ended up refactoring some of the (rather complicated) error-context
726 stuff in TcUnify, by pushing an InstOrigin into tcSubExp and its
727 various calls. That means we could get rid of tcFunResTy, and the
728 SubCtxt type.  This should improve error messages slightly
729 in complicated situations, because we have an Origin to hand
730 to instCall (in the (isSigmaTy actual_ty) case of tc_sub1).
731 =
732
733 Thanks to Pepe for the first draft of the patch.
734 =
735
736] =
737
738[Add DEBUG-only flag -dsuppress-uniques to suppress printing of uniques
739simonpj@microsoft.com**20071116152446
740 =
741
742 This is intended only for debugging use: it makes it easier to
743 compare two variants without the variations between uniques mattering.
744 =
745
746 (Of course, you can't actually feed the output to the C compiler
747 or assembler and expect anything sensible to happen!)
748 =
749
750] =
751
752[Add -dcore-lint when validating libraries
753simonpj@microsoft.com**20071105164733] =
754
755[Fix Trac #1913: check data const for derived types are in scope
756simonpj@microsoft.com**20071121151428
757 =
758
759 When deriving an instance, the data constructors should all be in scope.
760 This patch checks the condition.
761 =
762
763 =
764
765] =
766
767[Fix Trac #1909: type of map in docs
768simonpj@microsoft.com**20071120160152] =
769
770[Move file locking into the RTS, fixing #629, #1109
771Simon Marlow <simonmar@microsoft.com>**20071120140859
772 File locking (of the Haskell 98 variety) was previously done using a
773 static table with linear search, which had two problems: the array had
774 a fixed size and was sometimes too small (#1109), and performance of
775 lockFile/unlockFile was suboptimal due to the linear search.
776 Also the algorithm failed to count readers as required by Haskell 98
777 (#629).
778 =
779
780 Now it's done using a hash table (provided by the RTS).  Furthermore I
781 avoided the extra fstat() for every open file by passing the dev_t and
782 ino_t into lockFile.  This and the improvements to the locking
783 algorithm result in a healthy 20% or so performance increase for
784 opening/closing files (see openFile008 test).
785] =
786
787[FIX Trac #1825: standalone deriving Typeable
788simonpj@microsoft.com**20071120125732
789 =
790
791 Standalone deriving of typeable now requires you to say
792        instance Typeable1 Maybe
793 which is exactly the shape of instance decl that is generated
794 by a 'deriving( Typeable )' clause on the data type decl.
795 =
796
797 This is a bit horrid, but it's the only consistent way, at least
798 for now.  If you say something else, the error messages are helpful.
799 =
800
801 MERGE to 6.8 branch
802 =
803
804] =
805
806[FIX #1715: egregious bug in ifaceDeclSubBndrs
807simonpj@microsoft.com**20071120111723
808 =
809
810 ifaceDeclSubBndrs didn't have an IfaceSyn case; but with type
811 families an IfaceSyn can introduce subordinate binders.  Result:
812 chaos.
813 =
814
815 The fix is easy though.  Merge to 6.8 branch.
816 =
817
818 =
819
820] =
821
822[Always do 'setup makefile' before building each library
823Simon Marlow <simonmar@microsoft.com>**20071120103329
824 This forces preprocessing to happen, which is necessary if any of the
825 .hsc files have been modified.  Without this change, a 'setup
826 makefile' would be required by hand after a .hsc file changed.
827 Fortunately 'setup makefile' isn't much extra work, and I've made it
828 not overwrite GNUmakefile if it hasn't changed, which avoids
829 recalculating the dependencies each time.
830] =
831
832[FIX #1847 (improve :browse! docs, fix unqual)
833claus.reinke@talk21.com**20071108013147
834 =
835
836 - add example to docs, explain how to interpret =
837
838   output of `:browse! Data.Maybe`
839 - print unqualified names according to current =
840
841   context, not the context of the target module
842 =
843
844] =
845
846[Track changes to package ndp
847Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071120033716] =
848
849[Temporary hack for passing PArrays from unvectorised to vectorised code
850Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071120024545] =
851
852[Bind NDP stuff to [:.:] arrays
853Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071119020302] =
854
855[Don't treat enumerations specially during vectorisation for the moment
856Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071119013729] =
857
858[Fix bugs in vectorisation of case expressions
859Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071119013714] =
860
861[More built-in NDP combinators
862Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071119012205] =
863
864[New vectorisation built-ins
865Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071118051940] =
866
867[Fix bug in conversion unvect/vect
868Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071118051926] =
869
870[Extend built-in vectorisation environments
871Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071118045219] =
872
873[Fix bug in generation of environments for vectorisation
874Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071118045203] =
875
876[Add builtin var->var mapping to vectorisation
877Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071118042605] =
878
879[Extend vectorisation built-in mappings with datacons
880Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071118034351] =
881
882[Change representation of parallel arrays of enumerations
883Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071118033355] =
884
885[Add vectorisation-related builtin
886Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071118031513] =
887
888[Teach vectorisation about Bool
889Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071117042714] =
890
891[Incomplete support for boxing during vectorisation
892Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071117040739] =
893
894[Make sure some TyCons always vectorise to themselves
895Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071117040537] =
896
897[Simple conversion vectorised -> unvectorised
898Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071117023029] =
899
900[Fix bug in case vectorisation
901Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071117015014] =
902
903[Vectorisation of algebraic case expressions
904Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071116074814] =
905
906[More vectorisation-related built-ins
907Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071116061831] =
908
909[Vectorisation utilities
910Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071116051037] =
911
912[Add vectorisation built-ins
913Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071116050959] =
914
915[Fix vectorisation of binders in case expressions
916Roman Leshchinskiy <rl@cse.unsw.edu.au>**20071116021833] =
917
918[Two small typos in the flags summary (merge to 6.8 branch)
919simonpj@microsoft.com**20071119134639] =
920
921[Improve the situation for Trac #959: civilised warning instead of a trace =
922msg
923simonpj@microsoft.com**20071119122938
924 =
925
926 This doesn't fix the root cause of the bug, but it makes the report
927 more civilised, and points to further info.
928 =
929
930] =
931
932[FIX Trac #1806: test for correct arity for datacon in infix pattern patch
933simonpj@microsoft.com**20071119114301
934 =
935
936 Happily the fix is easy; pls merge
937 =
938
939] =
940
941[Accept x86_64-*-freebsd* as well as amd64-*-freebsd* in configure.ac
942Ian Lynagh <igloo@earth.li>**20071117154502
943 Patch from Brian P. O'Hanlon
944] =
945
946[Attempt at fixing #1873, #1360
947Simon Marlow <simonmar@microsoft.com>**20071116152148
948 =
949
950 I think I figured out a reasonable way to manage the GHCi context,
951 comments welcome.
952 =
953
954 Rule 1: external package modules in the context are persistent.  That
955 is, when you say 'import Data.Maybe' it survives over :load, :add,
956 :reload and :cd.
957 =
958
959 Rule 2: :load and :add remove all home-package modules from the
960 context and add the rightmost target, as a *-module if possible.  This
961 is as before, and makes sense for :load because we're starting a new
962 program; the old home-package modules don't make sense any more.  For
963 :add, it usually does what you want, because the new target will
964 become the context.
965 =
966
967 Rule 3: any modules from the context that fail to load during a
968 :reload are remembered, and re-added to the context at the next
969 successful :reload.
970 =
971
972 Claus' suggestion about adding the "remembered" modules to the prompt
973 prefixed with a ! is implemented but commented out.  I couldn't
974 decide whether it was useful or confusing.
975 =
976
977 One difference that people might notice is that after a :reload where
978 there were errors, GHCi would previously dump you in the most recent
979 module that it loaded.  Now it dumps you in whatever subset of the
980 current context still makes sense, and in the common case that will
981 probably be {Prelude}.
982] =
983
984[Wibble to fix Trac #1901 (shorten messsage slightly)
985simonpj@microsoft.com**20071116150341] =
986
987[Improve links from flag reference to the relevant section; and improve doc=
988 of RankN flags
989simonpj@microsoft.com**20071116145816] =
990
991[FIX Trac #1901: check no existential context in H98 mode
992simonpj@microsoft.com**20071116145609] =
993
994[Improve documentation of data type declarations (Trac #1901)
995simonpj@microsoft.com**20071116081841] =
996
997[Change the command-line semantics for query commands
998Simon Marlow <simonmar@microsoft.com>**20071116132046
999 =
1000
1001 From the help text:
1002 =
1003
1004   Commands that query the package database (list, latest, describe,
1005   field) operate on the list of databases specified by the flags
1006   --user, --global, and --package-conf.  If none of these flags are
1007   given, the default is --global --user.
1008 =
1009
1010 This makes it possible to query just a single database (e.g. the
1011 global one without the user one), which needed tricks to accomplish
1012 before.
1013] =
1014
1015[use "ghc-pkg latest --global" instead of "ghc-pkg list --simple-output"
1016Simon Marlow <simonmar@microsoft.com>**20071116122018
1017 The former now does the right thing: it uses the global database only,
1018 and picks the most recent package with the given name.
1019] =
1020
1021[Disallow installing packages whose names  differ in case only.
1022Simon Marlow <simonmar@microsoft.com>**20071116121153
1023 --force overrides.  Requested by Duncan Coutts, with a view to
1024 treating package names as case-insensitive in the future.
1025] =
1026
1027[FIX BUILD (with GHC 6.2.x): update .hi-boot file
1028Simon Marlow <simonmar@microsoft.com>**20071116101227] =
1029
1030[FIX #1828: installing to a patch with spaces in =
1031
1032Simon Marlow <simonmar@microsoft.com>**20071115155747
1033 We have to pass the path to gcc when calling windres, which itself
1034 might have spaces in.  Furthermore, we have to pass the path to gcc's
1035 tools to gcc.  This means getting the quoting right, and after much
1036 experimentation and reading of the windres sources I found something
1037 that works: passing --use-temp-files to windres makes it use its own
1038 implementation of quoting instead of popen(), and this does what we
1039 want.  Sigh.
1040] =
1041
1042[on Windows, install to a directory with spaces (test for #1828)
1043Simon Marlow <simonmar@microsoft.com>**20071115155327] =
1044
1045[FIX #1679: crash on returning from a foreign call
1046Simon Marlow <simonmar@microsoft.com>**20071115131635
1047 We forgot to save a pointer to the BCO over the foreign call.  Doing
1048 enough allocation and GC during the call could provoke a crash.
1049] =
1050
1051[Avoid the use of unversioned package dependencies
1052Simon Marlow <simonmar@microsoft.com>**20071115103249
1053 Fortunately "ghc-pkg list $pkg --simple-output" is a good way to add
1054 the version number.
1055] =
1056
1057[FIX #1596 (remove deprecated --define-name)
1058Simon Marlow <simonmar@microsoft.com>**20071114165323
1059 Also remove the old command-line syntax for ghc-pkg, which was not
1060 documented.  Do not merge.
1061] =
1062
1063[FIX #1837: remove deprecated support for unversioned dependencies (do not =
1064merge)
1065Simon Marlow <simonmar@microsoft.com>**20071114161044
1066 =
1067
1068] =
1069
1070[wibble
1071Pepe Iborra <mnislaih@gmail.com>**20071114233356] =
1072
1073[Make pprNameLoc more robust in absence of loc information
1074Pepe Iborra <mnislaih@gmail.com>**20071114233343] =
1075
1076[Try to manage the size of the text rendered for ':show bindings'
1077Pepe Iborra <mnislaih@gmail.com>**20071114231601] =
1078
1079[Make the Term ppr depth aware
1080Pepe Iborra <mnislaih@gmail.com>**20071114183417] =
1081
1082[Use paragraph fill sep where possible
1083Pepe Iborra <mnislaih@gmail.com>**20071114181233] =
1084
1085[Make SpecConstr work again
1086simonpj@microsoft.com**20071115084242
1087 =
1088
1089 In a typo I'd written env instead of env', and as a result RULES are
1090 practically guaranteed not to work in a recursive group.  This pretty
1091 much kills SpecConstr in its tracks!
1092 =
1093
1094 Well done Kenny Lu for spotting this.  The fix is easy.
1095 =
1096
1097 Merge into 6.8 please.
1098 =
1099
1100 =
1101
1102 =
1103
1104] =
1105
1106[Documentation only - fix typo in flags reference
1107Tim Chevalier <chevalier@alum.wellesley.edu>**20071115055748] =
1108
1109[Avoid making Either String an instance of Monad in the Haddock parser
1110David Waern <david.waern@gmail.com>**20071114204050] =
1111
1112[FIX 1463 (implement 'ghc-pkg find-module')
1113claus.reinke@talk21.com**20071109162652
1114 =
1115
1116 - the ticket asks for a module2package lookup in ghc-pkg
1117   (this would be useful to have in cabal, as well)
1118 =
1119
1120 - we can now ask which packages expose a module we need,
1121   eg, when preparing a cabal file or when getting errors
1122   after package reorganisations:
1123 =
1124
1125   $ ./ghc-pkg-inplace find-module Var
1126   c:/fptools/ghc/driver/package.conf.inplace:
1127       (ghc-6.9.20071106)
1128   =
1129
1130   $ ./ghc-pkg-inplace find-module Data.Sequence
1131   c:/fptools/ghc/driver/package.conf.inplace:
1132       containers-0.1
1133 =
1134
1135 - implemented as a minor variation on listPackages
1136 =
1137
1138 (as usual, it would be useful if one could combine =
1139
1140 multiple queries into one)
1141 =
1142
1143] =
1144
1145[remove --define-name from the --help usage message (#1596)
1146Simon Marlow <simonmar@microsoft.com>**20071114153417
1147 =
1148
1149] =
1150
1151[FIX #1837: emit deprecated message for unversioned dependencies
1152Simon Marlow <simonmar@microsoft.com>**20071114153010] =
1153
1154[Fix #782, #1483, #1649: Unicode GHCi input
1155Simon Marlow <simonmar@microsoft.com>**20071114151411
1156 GHCi input is now treated universally as UTF-8, except for the Windows
1157 console where we do the correct conversion from the current code
1158 page (see System.Win32.stringToUnicode).
1159 =
1160
1161 That leaves non-UTF-8 locales on Unix as unsupported, but (a) we only
1162 accept source files in UTF-8 anyway, and (b) UTF-8 is quite ubiquitous
1163 as the default locale.
1164 =
1165
1166] =
1167
1168[Fix build
1169David Waern <david.waern@gmail.com>**20071114125842
1170 I had forgot to update HaddockLex.hi-boot-6, so the build with 6.2.2 =
1171
1172 failed. This fixes that.
1173] =
1174
1175[FIX Trac 1662: actually check for existentials in proc patterns
1176simonpj@microsoft.com**20071114112930
1177 =
1178
1179 I'd fixed the bug for code that should be OK, but had forgotten to =
1180
1181 make the test for code that should be rejected! =
1182
1183 =
1184
1185 Test is arrowfail004
1186 =
1187
1188] =
1189
1190[FIX Trac 1888; duplicate INLINE pragmas
1191simonpj@microsoft.com**20071114104701
1192 =
1193
1194 There are actually three things here
1195 - INLINE pragmas weren't being pretty-printed properly
1196 - They were being classified into too-narrow boxes by eqHsSig
1197 - They were being printed in to much detail by hsSigDoc
1198 =
1199
1200 All easy.  Test is rnfail048.
1201 =
1202
1203] =
1204
1205[Run the -frule-check pass more often (when asked)
1206simonpj@microsoft.com**20071114104632] =
1207
1208[GHCi debugger: added a new flag, -fno-print-binding-contents
1209Pepe Iborra <mnislaih@gmail.com>**20071113174539
1210 =
1211
1212 The contents of bindings show at breakpoints and by :show bindings
1213 is rendered using the same printer that :print uses.
1214 But sometimes the output it gives spans over too many lines and the
1215 user may want to be able to disable it.
1216] =
1217
1218[Fix Trac 1865: GHCi debugger crashes with :print
1219Pepe Iborra <mnislaih@gmail.com>**20071113170113] =
1220
1221[Replaced two uses of head b explicit pattern matching
1222Pepe Iborra <mnislaih@gmail.com>**20071013113136] =
1223
1224[Print binding contents in :show bindings
1225Pepe Iborra <mnislaih@gmail.com>**20071006123952] =
1226
1227[ Leftovers from the 1st GHCi debugger prototype
1228Pepe Iborra <mnislaih@gmail.com>**20071004204718] =
1229
1230[Following an indirection doesn't count as a RTTI step
1231Pepe Iborra <mnislaih@gmail.com>**20070928091941] =
1232
1233[FIX #1653 (partially): add -X flags to completion for :set
1234Simon Marlow <simonmar@microsoft.com>**20071113153257] =
1235
1236[Merge from Haddock: Add <<url>> for images
1237David Waern <david.waern@gmail.com>**20071112220537
1238 A merge of this patch:
1239 =
1240
1241   Mon Aug  7 16:22:14 CEST 2006  Simon Marlow <simonmar@microsoft.com>
1242     * Add <<url>> for images
1243     Submitted by: Lennart Augustsson
1244 =
1245
1246 Please merge to the 6.8.2 branch.
1247] =
1248
1249[Improve documentation of INLINE, esp its interactions with other transform=
1250ations
1251simonpj@microsoft.com**20071112160240] =
1252
1253[Comment re Trac #1220
1254simonpj@microsoft.com**20071112154109] =
1255
1256[Merge from Haddock: Modify lexing of /../
1257David Waern <david.waern@gmail.com>**20071112023856
1258 =
1259
1260   Tue Aug 28 11:19:54 CEST 2007  Simon Marlow <simonmar@microsoft.com>
1261     * Modify lexing of /../ =
1262
1263     This makes /../ more like '..', so that a single / on a line doesn't
1264     trigger a parse error.  This should reduce the causes of accidental
1265     parse errors in Haddock comments; apparently stray / characters are
1266     a common source of failures.
1267 =
1268
1269 Please merge this to the 6.8.2 branch.
1270] =
1271
1272[Merge from Haddock: allow blank lines inside code blocks
1273David Waern <david.waern@gmail.com>**20071112013439
1274 =
1275
1276   Tue Jan  9 14:14:34 CET 2007  Simon Marlow <simonmar@microsoft.com>
1277     * allow blank lines inside a @...@ code block
1278 =
1279
1280 Please merge this to the 6.8.2 branch
1281] =
1282
1283[Merge of a patch from the old Haddock branch:
1284David Waern <david.waern@gmail.com>**20071112013143
1285 =
1286
1287   Fri Jan  5 12:13:41 CET 2007  Simon Marlow <simonmar@microsoft.com>
1288     * Fix up a case of extra vertical space after a code block
1289 =
1290
1291 Please merge this to the 6.8.2 branch
1292] =
1293
1294[Remove ex-extralibs from libraries/Makefile
1295Ian Lynagh <igloo@earth.li>**20071111213618] =
1296
1297[Remove the X11 and HGL libraries from extralibs
1298Ian Lynagh <igloo@earth.li>**20071111213447
1299 Don Stewart, X11 maintainer, requested we remove X11, and HGL depends on i=
1300t
1301 on Linux (and we don't try to build HGL on Windows).
1302] =
1303
1304[arrows is no longer an extralib
1305Ian Lynagh <igloo@earth.li>**20071027123656] =
1306
1307[Turn -fprint-bind-result off by default
1308Ian Lynagh <igloo@earth.li>**20071111001126] =
1309
1310[TAG 2007-11-11
1311Ian Lynagh <igloo@earth.li>**20071111161540] =
1312
1313Patch bundle hash:
1314cc5521704708d436ff2882ba67eede0977f29fc3
1315
1316--=_--
1317
1318.
1319