Ticket #1845: avoid-mmap-on-powerpc-apple-darwin.patch

File avoid-mmap-on-powerpc-apple-darwin.patch, 162.5 KB (added by PHO, 4 years ago)
Line 
12 patches for repository http://darcs.haskell.org/ghc:
2
3Tue Sep 28 07:41:45 JST 2010  pho@cielonegro.org
4  * New member "archiveMemberName" for struct _ObjectCode
5 
6  struct _ObjectCode should be able to retain the name of archive members.
7  Though currently the only use of those names are for debugging outputs.
8
9Tue Sep 28 22:25:59 JST 2010  pho@cielonegro.org
10  * FIX #1845 (unconditional relative branch out of range)
11 
12  Don't use mmap on powerpc-apple-darwin as mmap doesn't support
13  reallocating but we need to allocate jump islands just after each
14  object images. Otherwise relative branches to jump islands can fail
15  due to 24-bits displacement overflow.
16-----BEGIN PGP SIGNED MESSAGE-----
17Hash: SHA1
18
19
20New patches:
21
22[New member "archiveMemberName" for struct _ObjectCode
23pho@cielonegro.org**20100927224145
24 Ignore-this: 628bc605e7dc4f0c4856c6f7ad23d9ee
25 
26 struct _ObjectCode should be able to retain the name of archive members.
27 Though currently the only use of those names are for debugging outputs.
28] {
29hunk ./rts/Linker.c 122
30 ObjectCode *objects = NULL;    /* initially empty */
31 
32 static HsInt loadOc( ObjectCode* oc );
33- -static ObjectCode* mkOc( char *path, char *image, int imageSize
34+static ObjectCode* mkOc( char *path, char *image, int imageSize,
35+                         char *archiveMemberName
36 #ifndef USE_MMAP
37 #ifdef darwin_HOST_OS
38                        , int misalignment
39hunk ./rts/Linker.c 1606
40 #endif // USE_MMAP
41 
42 static ObjectCode*
43- -mkOc( char *path, char *image, int imageSize
44+mkOc( char *path, char *image, int imageSize,
45+      char *archiveMemberName
46 #ifndef USE_MMAP
47 #ifdef darwin_HOST_OS
48     , int misalignment
49hunk ./rts/Linker.c 1631
50 
51    oc->image = image;
52    /* sigh, strdup() isn't a POSIX function, so do it the long way */
53- -   /* XXX What should this be for an archive? */
54    oc->fileName = stgMallocBytes( strlen(path)+1, "loadObj" );
55    strcpy(oc->fileName, path);
56 
57hunk ./rts/Linker.c 1634
58+   if (archiveMemberName) {
59+       oc->archiveMemberName = stgMallocBytes( strlen(archiveMemberName)+1, "loadObj" );
60+       strcpy(oc->archiveMemberName, archiveMemberName);
61+   }
62+   else {
63+       oc->archiveMemberName = NULL;
64+   }
65+
66    oc->fileSize          = imageSize;
67    oc->symbols           = NULL;
68    oc->sections          = NULL;
69hunk ./rts/Linker.c 1751
70        }
71 
72        if (isObject) {
73+           char *archiveMemberName;
74+
75            /* We can't mmap from the archive directly, as object
76               files need to be 8-byte aligned but files in .ar
77               archives are 2-byte aligned, and if we malloc the
78hunk ./rts/Linker.c 1763
79            n = fread ( image, 1, imageSize, f );
80            if (n != imageSize)
81                barf("loadObj: error whilst reading `%s'", path);
82- -           oc = mkOc(path, image, imageSize
83+
84+           archiveMemberName = stgMallocBytes(strlen(path) + fileNameSize + 3, "loadArchive(file)");
85+           sprintf(archiveMemberName, "%s(%.*s)", path, (int)fileNameSize, file);
86+
87+           oc = mkOc(path, image, imageSize, archiveMemberName
88 #ifndef USE_MMAP
89 #ifdef darwin_HOST_OS
90                     , 0
91hunk ./rts/Linker.c 1774
92 #endif
93 #endif
94                     );
95+
96+           stgFree(archiveMemberName);
97+
98            if (0 == loadOc(oc)) {
99                stgFree(file);
100                return 0;
101hunk ./rts/Linker.c 1920
102    fclose(f);
103 #endif /* USE_MMAP */
104 
105- -   oc = mkOc(path, image, fileSize
106+   oc = mkOc(path, image, fileSize, NULL
107 #ifndef USE_MMAP
108 #ifdef darwin_HOST_OS
109             , misalignment
110hunk ./rts/Linker.c 2025
111 unloadObj( char *path )
112 {
113     ObjectCode *oc, *prev;
114+    HsBool unloadedAnyObj = HS_BOOL_FALSE;
115 
116     ASSERT(symhash != NULL);
117     ASSERT(objects != NULL);
118hunk ./rts/Linker.c 2065
119            stgFree(oc->symbols);
120            stgFree(oc->sections);
121            stgFree(oc);
122- -         return 1;
123+
124+            /* This could be a member of an archive so continue
125+             * unloading other members. */
126+            unloadedAnyObj = HS_BOOL_TRUE;
127        }
128     }
129 
130hunk ./rts/Linker.c 2072
131- -    errorBelch("unloadObj: can't find `%s' to unload", path);
132- -    return 0;
133+    if (unloadedAnyObj) {
134+        return 1;
135+    }
136+    else {
137+        errorBelch("unloadObj: can't find `%s' to unload", path);
138+        return 0;
139+    }
140 }
141 
142 /* -----------------------------------------------------------------------------
143hunk ./rts/Linker.c 4661
144                              "scattered relocation entry: "
145                               "object file %s; entry type %ld; "
146                               "address %#lx\n",
147- -                              oc->fileName, scat->r_type, scat->r_address);
148+                              OC_INFORMATIVE_FILENAME(oc),
149+                              scat->r_type,
150+                              scat->r_address);
151                         return 0;
152                      }
153 
154hunk ./rts/Linker.c 4700
155                         "with this r_length tag: "
156                          "object file %s; entry type %ld; "
157                          "r_length tag %ld; address %#lx\n",
158- -                         oc->fileName, scat->r_type, scat->r_length,
159+                         OC_INFORMATIVE_FILENAME(oc),
160+                         scat->r_type,
161+                         scat->r_length,
162                          scat->r_address);
163                     return 0;
164                }
165hunk ./rts/Linker.c 4712
166                barf("Don't know how to handle *PC-relative* Mach-O "
167                     "scattered relocation entry: "
168                      "object file %s; entry type %ld; address %#lx\n",
169- -                     oc->fileName, scat->r_type, scat->r_address);
170+                     OC_INFORMATIVE_FILENAME(oc),
171+                     scat->r_type,
172+                     scat->r_address);
173                return 0;
174            }
175 
176hunk ./rts/Linker.c 4768
177                 {
178                     barf("Can't handle this Mach-O relocation entry "
179                         "(not scattered): "
180- -                         "object file %s; entry type %ld; address %#lx\n",
181- -                         oc->fileName, reloc->r_type, reloc->r_address);
182+                         "object file %s; entry type %ld; address %#lx\n",
183+                         OC_INFORMATIVE_FILENAME(oc),
184+                         reloc->r_type,
185+                         reloc->r_address);
186                     return 0;
187                 }
188 
189hunk ./rts/Linker.c 4845
190                }
191                else if(reloc->r_type == PPC_RELOC_BR24)
192                {
193- -                    if((long)word > (long)0x01FFFFFF || (long)word < (long)0xFFE00000)
194+                    if((word & 0x03) != 0)
195+                        barf("%s: unconditional relative branch with a displacement "
196+                             "which isn't a multiple of 4 bytes: %#lx",
197+                             OC_INFORMATIVE_FILENAME(oc),
198+                             word);
199+
200+                    if((word & 0xFE000000) != 0xFE000000 &&
201+                       (word & 0xFE000000) != 0x00000000)
202                     {
203                         // The branch offset is too large.
204                         // Therefore, we try to use a jump island.
205hunk ./rts/Linker.c 4858
206                         if(jumpIsland == 0)
207                         {
208- -                            barf("unconditional relative branch out of range: "
209- -                                 "no jump island available");
210+                            barf("%s: unconditional relative branch out of range: "
211+                                 "no jump island available: %#lx",
212+                                 OC_INFORMATIVE_FILENAME(oc),
213+                                 word);
214                         }
215                         
216                         word = offsetToJumpIsland;
217hunk ./rts/Linker.c 4865
218- -                        if((long)word > (long)0x01FFFFFF || (long)word < (long)0xFFE00000)
219- -                            barf("unconditional relative branch out of range: "
220- -                                 "jump island out of range");
221+                        if((word & 0xFE000000) != 0xFE000000 &&
222+                           (word & 0xFE000000) != 0x00000000)
223+                            barf("%s: unconditional relative branch out of range: "
224+                                 "jump island out of range: %#lx",
225+                                 OC_INFORMATIVE_FILENAME(oc),
226+                                 word);
227                     }
228                    *wordPtr = (*wordPtr & 0xFC000003) | (word & 0x03FFFFFC);
229                    continue;
230hunk ./rts/Linker.c 4882
231                 barf("Can't handle Mach-O relocation entry (not scattered) "
232                       "with this r_length tag: "
233                       "object file %s; entry type %ld; "
234- -                      "r_length tag %ld; address %#lx\n",
235- -                      oc->fileName, reloc->r_type, reloc->r_length,
236+                      "r_length tag %ld; address %#lx\n",
237+                      OC_INFORMATIVE_FILENAME(oc),
238+                      reloc->r_type,
239+                      reloc->r_length,
240                       reloc->r_address);
241                 return 0;
242            }
243hunk ./rts/LinkerInternals.h 68
244     int        fileSize;
245     char*      formatName;            /* eg "ELF32", "DLL", "COFF", etc. */
246 
247+    /* If this object is a member of an archive, archiveMemberName is
248+     * like "libarchive.a(object.o)". Otherwise it's NULL.
249+     */
250+    char*      archiveMemberName;
251+
252     /* An array containing ptrs to all the symbol names copied from
253        this object into the global symbol hash table.  This is so that
254        we know which parts of the latter mapping to nuke when this
255hunk ./rts/LinkerInternals.h 115
256 
257 } ObjectCode;
258 
259+#define OC_INFORMATIVE_FILENAME(OC)             \
260+    ( (OC)->archiveMemberName ?                 \
261+      (OC)->archiveMemberName :                 \
262+      (OC)->fileName                            \
263+    )
264+
265 extern ObjectCode *objects;
266 
267 void exitLinker( void );
268}
269[FIX #1845 (unconditional relative branch out of range)
270pho@cielonegro.org**20100928132559
271 Ignore-this: a3d9726969704a52c4a4d61f43c37162
272 
273 Don't use mmap on powerpc-apple-darwin as mmap doesn't support
274 reallocating but we need to allocate jump islands just after each
275 object images. Otherwise relative branches to jump islands can fail
276 due to 24-bits displacement overflow.
277] {
278hunk ./rts/Linker.c 72
279 #include <sys/wait.h>
280 #endif
281 
282- -#if defined(linux_HOST_OS) || defined(freebsd_HOST_OS) || defined(dragonfly_HOST_OS) || defined(netbsd_HOST_OS) || defined(openbsd_HOST_OS) || defined(darwin_HOST_OS)
283+#if defined(linux_HOST_OS    ) || defined(freebsd_HOST_OS) || \
284+    defined(dragonfly_HOST_OS) || defined(netbsd_HOST_OS ) || \
285+    defined(openbsd_HOST_OS  ) || \
286+    ( defined(darwin_HOST_OS ) && !defined(powerpc_HOST_ARCH) )
287+/* Don't use mmap on powerpc-apple-darwin as mmap doesn't support
288+ * reallocating but we need to allocate jump islands just after each
289+ * object images. Otherwise relative branches to jump islands can fail
290+ * due to 24-bits displacement overflow.
291+ */
292 #define USE_MMAP
293 #include <fcntl.h>
294 #include <sys/mman.h>
295hunk ./rts/Linker.c 1682
296    size_t fileSize;
297    int isObject;
298    char tmp[12];
299+#if !defined(USE_MMAP) && defined(darwin_HOST_OS)
300+   int misalignment;
301+#endif
302 
303    IF_DEBUG(linker, debugBelch("loadArchive `%s'\n", path));
304 
305hunk ./rts/Linker.c 1764
306        if (isObject) {
307            char *archiveMemberName;
308 
309+#if defined(USE_MMAP)
310            /* We can't mmap from the archive directly, as object
311               files need to be 8-byte aligned but files in .ar
312               archives are 2-byte aligned, and if we malloc the
313hunk ./rts/Linker.c 1772
314               mmap some anonymous memory and use that. We could
315               do better here. */
316            image = mmapForLinker(imageSize, MAP_ANONYMOUS, -1);
317+#elif defined(darwin_HOST_OS)
318+           // See loadObj()
319+           misalignment = machoGetMisalignment(f);
320+           image = stgMallocBytes(imageSize + misalignment, "loadArchive(file)");
321+           image += misalignment;
322+#else
323+           image = stgMallocBytes(imageSize, "loadArchive(file)");
324+#endif
325            n = fread ( image, 1, imageSize, f );
326            if (n != imageSize)
327hunk ./rts/Linker.c 1782
328- -               barf("loadObj: error whilst reading `%s'", path);
329+               barf("loadArchive: error whilst reading `%s'", path);
330 
331            archiveMemberName = stgMallocBytes(strlen(path) + fileNameSize + 3, "loadArchive(file)");
332            sprintf(archiveMemberName, "%s(%.*s)", path, (int)fileNameSize, file);
333hunk ./rts/Linker.c 1790
334            oc = mkOc(path, image, imageSize, archiveMemberName
335 #ifndef USE_MMAP
336 #ifdef darwin_HOST_OS
337- -                    , 0
338+                    , misalignment
339 #endif
340 #endif
341                     );
342hunk ./rts/Linker.c 1851
343    int fd;
344 #else
345    FILE *f;
346+#  if defined(darwin_HOST_OS)
347+   int misalignment;
348+#  endif
349 #endif
350hunk ./rts/Linker.c 1855
351+
352    IF_DEBUG(linker, debugBelch("loadObj %s\n", path));
353 
354    initLinker();
355hunk ./rts/Linker.c 1928
356     // We calculate the correct alignment from the header before
357     // reading the file, and then we misalign image on purpose so
358     // that the actual sections end up aligned again.
359- -   misalignment = machoGetMisalignment(f);
360- -   image = stgMallocBytes(fileSize + misalignment, "loadObj(image)");
361- -   image += misalignment;
362- -#  else
363- -   image = stgMallocBytes(fileSize, "loadObj(image)");
364- -#  endif
365+    misalignment = machoGetMisalignment(f);
366+    image = stgMallocBytes(fileSize + misalignment, "loadObj(image)");
367+    image += misalignment;
368+#   else
369+    image = stgMallocBytes(fileSize, "loadObj(image)");
370+#   endif
371 
372    {
373        int n;
374hunk ./rts/Linker.c 2214
375      */
376     if( m > n ) // we need to allocate more pages
377     {
378+        errorBelch("%s: WARNING: Allocating jump islands separately from "
379+                   "the object image itself. This may interfere with "
380+                   "relative branches to them.",
381+                   OC_INFORMATIVE_FILENAME(oc));
382+
383         oc->symbol_extras = mmapForLinker(sizeof(SymbolExtra) * count,
384                                           MAP_ANONYMOUS, -1);
385     }
386hunk ./rts/Linker.c 5181
387     struct mach_header header;
388     int misalignment;
389     
390- -    fread(&header, sizeof(header), 1, f);
391- -    rewind(f);
392+    {
393+        int n = fread(&header, sizeof(header), 1, f);
394+        if (n != 1) {
395+            barf("machoGetMisalignment: can't read the Mach-O header");
396+        }
397+    }
398+    fseek(f, -sizeof(header), SEEK_CUR);
399 
400 #if x86_64_HOST_ARCH || powerpc64_HOST_ARCH
401     if(header.magic != MH_MAGIC_64) {
402hunk ./rts/Linker.c 5191
403- -        errorBelch("Bad magic. Expected: %08x, got: %08x.\n",
404- -                   MH_MAGIC_64, header->magic);
405- -        return 0;
406+        barf("Bad magic. Expected: %08x, got: %08x.",
407+             MH_MAGIC_64, header.magic);
408     }
409 #else
410     if(header.magic != MH_MAGIC) {
411hunk ./rts/Linker.c 5196
412- -        errorBelch("Bad magic. Expected: %08x, got: %08x.\n",
413- -                   MH_MAGIC, header->magic);
414- -        return 0;
415+        barf("Bad magic. Expected: %08x, got: %08x.",
416+             MH_MAGIC, header.magic);
417     }
418 #endif
419 
420}
421
422Context:
423
424[Use an empty signal handler for SIGPIPE instead of SIG_IGN
425Simon Marlow <marlowsd@gmail.com>**20100925193548
426 Ignore-this: b4dc5de32740a7c5fd8fe4b3bfb1300f
427 
428 This is so that the SIGPIPE handler gets reset to the default
429 automatically on exec().
430]
431[Fix the TH deps
432Ian Lynagh <igloo@earth.li>**20100925210029
433 Ignore-this: 32b832301a3625d4ba70f84c5c4f94d2
434]
435[Check inplace doesn't exist before we try to create it
436Ian Lynagh <igloo@earth.li>**20100924191858
437 This fixes rerunning configure in a tree which already has an inplace
438 directory. Edward Z Yang ran into this; I guess whether it actually
439 fails depends on details of your installation, or we'd have run into
440 it sooner.
441]
442[Fix an egregious bug: INLINE pragmas on monomorphic Ids were being ignored
443simonpj@microsoft.com**20100924155815
444 Ignore-this: 38c6eec6710a92df7662a55fc5132c15
445 
446 I had do to some refactoring to make this work nicely
447 but now it does. I can't think how this escaped our
448 attention for so long!
449]
450[Eta expand only lambdas that bind a non-dictionary Id
451simonpj@microsoft.com**20100924155707
452 Ignore-this: 7cc265eaf6c0bb3fa12eb311d92594ac
453 
454 See Note [When to eta expand]. The idea is that dictionary
455 lambdas are invisible to the user, so we shouldn't eta
456 expand them.
457]
458[Add a comment
459simonpj@microsoft.com**20100924155620
460 Ignore-this: de210a1afdd40328824803e1d77b4d7f
461]
462[Add a debug print
463simonpj@microsoft.com**20100924155614
464 Ignore-this: 1a58b6d297fc77d6ded8eec7ea9f895d
465]
466[Just moving comments around
467simonpj@microsoft.com**20100924155600
468 Ignore-this: 96635b8e8c9d88b50d82938568152ef8
469]
470[use putStrLn instead of Haskeline's outputStrLn
471Simon Marlow <marlowsd@gmail.com>**20100924133154
472 Ignore-this: 7581ae11714a9a52e78ba098c3c216b3
473 use of the latter caused problems for Claus Reinke's macros that
474 redirect stdout.
475]
476[Change "OPTIONS" to "OPTIONS_GHC" in error messages; fixes #4327
477Ian Lynagh <igloo@earth.li>**20100924120423
478 Ignore-this: 1697c83a5c346db640c0a2e22c69ff55
479]
480[Add deps for TH uses in vector
481Ian Lynagh <igloo@earth.li>**20100923220244
482 Ignore-this: 54c3386b1c268821fcdd34b84bc8c6a4
483]
484[Bump Cabal dep
485Ian Lynagh <igloo@earth.li>**20100923143241]
486[Update Cabal's version number
487Ian Lynagh <igloo@earth.li>**20100923141719]
488[Build primitive with stage2
489Ian Lynagh <igloo@earth.li>**20100923140525
490 Ignore-this: 110a819b78a57629a7edf1d4facdc191
491]
492[Fix the Windows __chkstk build error (missing Linker symbol)
493Simon Marlow <marlowsd@gmail.com>**20100924113837
494 Ignore-this: 48f0907bb1bd5eaa0730b94a6bd94ea
495]
496[emit a helpful error message for missing DPH packages
497Simon Marlow <marlowsd@gmail.com>**20100923141957
498 Ignore-this: 55ff2ee90c94524e023e014243bfe5df
499]
500[Fix computation of installed packages
501simonpj@microsoft.com**20100924084737
502 Ignore-this: a597d2fa8be5135ba8ead6d2624b3d71
503 
504 This is a follow-on to Simon's patch yesterday, developed
505 with him.  It cleans up the computation of how packages
506 are installed, and installs the right ones.
507]
508[Fix braino in WwLib/Literal patch
509simonpj@microsoft.com**20100924070914
510 Ignore-this: f6eb3a42e10f8aa7920de541cdfe76d8
511]
512[For now, switch off incomplete-pattern warnings in containers
513simonpj@microsoft.com**20100923130117
514 Ignore-this: 7ffa58567f7a33aafe256492999da325
515 
516 Put it back on when my patch is applied to the containers repo.
517 (the one that removes two refuable lambdas)
518]
519[Make -funfolding-dict-threshold work properly
520simonpj@microsoft.com**20100923130032
521 Ignore-this: 417788f5b09d1d624f6b6371852c80c7
522 
523 and increase its default value. This makes overloaded functions
524 a bit keener to inline.  Which fixes Trac #4321
525]
526[Impredicative types is no longer deprecated
527simonpj@microsoft.com**20100923125910
528 Ignore-this: 2bbaeb38b5e8424551677c0add627683
529]
530[Do not make FunctionalDependencies force MonoLocalBinds
531simonpj@microsoft.com**20100923125900
532 Ignore-this: f4ae1fd07c87ec14f60bdfe3863ba7a9
533]
534[move CHECKED settings to the right place
535Simon Marlow <marlowsd@gmail.com>**20100923123558
536 Ignore-this: e00a0eb5855463cc9b953670b3bbf211
537]
538[turn off -Werror for primitive and vector
539Simon Marlow <marlowsd@gmail.com>**20100923122055
540 Ignore-this: 54d7b80f3f893385e1c3ef431e2a8a7b
541]
542[Add primitive and vector packages for DPH support
543Simon Marlow <marlowsd@gmail.com>**20100923104542
544 Ignore-this: c070d015385b0a0797394132dcbb7670
545 DPH is now using the public vector package instead of its internal
546 version.
547 
548 vector and primitive are not "boot" packages; they aren't required to
549 build GHC, but they are required to validate (because we include DPH
550 when validating).
551 
552 If you say './darcs-all get --no-dph' then you don't get DPH, vector,
553 or primitive.
554]
555[Refactoring and tidy up in the build system
556Simon Marlow <marlowsd@gmail.com>**20100923095642
557 Ignore-this: f7bf3a1fd160149d89b26f464b064fb1
558 
559 Instead of the ghc-stage and ghc-stage2-package files in a package, we
560 now have a list of these in ghc.mk.  There are other similar lists (of
561 boot-packages and non-installable packages), so this is not too bad,
562 and is simpler.
563 
564 While poking around in the top-level ghc.mk file I spotted various
565 opportunities to clean up and re-order some of the cruft that has
566 accumulated over time.
567]
568[Allow absent State# RealWorld arguments
569simonpj@microsoft.com**20100923111356
570 Ignore-this: c2d57633dec0293ebe6723ea3a4bb5df
571]
572[Add notSCCNote, and use it
573simonpj@microsoft.com**20100923105949
574 Ignore-this: c8cc758656558a7f366bf784d75f0304
575 
576 The point here is that SCCs get in the way of eta
577 expansion and we must treat them uniformly.
578]
579[Remove use of lambda with a refutable pattern
580simonpj@microsoft.com**20100923105901
581 Ignore-this: d7d48b94e5744717a838591a1cc79cf0
582]
583[Avoid ASSERT black hole
584simonpj@microsoft.com**20100923105820
585 Ignore-this: 5419d450871be22c8781ac3f0f40d76a
586 
587 When this ASSERT tripped in CoreToStg it tried to print out
588 too much, which tripped the asssertion again.  Result: an
589 infinite loop with no output at all.  Hard to debug!
590]
591[Rejig the absent-arg stuff for unlifted types
592simonpj@microsoft.com**20100923105732
593 Ignore-this: 69daa35816b948b0c4d259c73a5e928e
594 
595 This is what was giving the "absent entered" messages
596 See Note [Absent errors] in WwLib.  We now return a
597 suitable literal for absent values of unlifted type.
598]
599[Remove -fwarn-simple-patterns, and make -fwarn-incomplete-patterns include lambdas
600simonpj@microsoft.com**20100922133934
601 Ignore-this: e851a2fb0377e10c28c506f0bf14cc85
602 
603 This makes
604      \(x:xs) -> e
605 want when you have -fwarn-incomplete-patterns, which is consistent.
606]
607[Get rid of non-exhaustive lambda
608simonpj@microsoft.com**20100922133801
609 Ignore-this: 748b2d5b43b02b6591b81abe7c105cd6
610]
611[Fix an ASSERT failure with profiling
612simonpj@microsoft.com**20100922133741
613 Ignore-this: 170b2e94d6ee8cc7444cc4bb515328a0
614 
615 The problem arose with this kind of thing
616 
617    x = (,) (scc "blah" Nothing)
618 
619 Then 'x' is marked NoCafRefs by CoreTidy, becuase it has
620 arity 1, and doesn't mention any caffy things.
621 
622 That in turns means that CorePrep must not float out the
623 sat binding to give
624 
625   sat = scc "blah" Nothing
626   x = (,) sat
627 
628 Rather we must generate
629 
630   x = \eta. let sat = scc "blah" Nothing
631             in (,) sat eta
632 
633 URGH! This Caf stuff is such a mess.
634]
635[Remove an out of date paragraph from the user guide; fixes #4331
636Ian Lynagh <igloo@earth.li>**20100922225239]
637[Fix bindisttest when GhcProfiled = YES
638Ian Lynagh <igloo@earth.li>**20100921222634
639 Ignore-this: 47c620fd6bec745e3eb699d9f53441d8
640]
641[Fixes for when HADDOCK_DOCS=NO
642Ian Lynagh <igloo@earth.li>**20100921213916
643 Ignore-this: e0e069555c6db9d01a8ac70ba4dde591
644]
645[Bump version to 7.1
646Ian Lynagh <igloo@earth.li>**20100921195935
647 Ignore-this: 4563987e6885d5ef55995ec0fa0d5ae8
648]
649[Don't use -march=i686 on powerpc-apple-darwin
650Ian Lynagh <igloo@earth.li>**20100921193721
651 Thorikil ran into this when doing a PPC OS X build. We now also don't
652 use -m32 on PPC/OSX, but I don't think it should be necessary. We can
653 add it back if it does turn out to be.
654]
655[add a simple trace facility to the build system
656Simon Marlow <marlowsd@gmail.com>**20100921134729
657 Ignore-this: d23ea2d62a648d0711b4b07d98e1b79f
658 
659 saying
660 
661   make TRACE=1
662 
663 prints most of the macro calls and their arguments.  It's easy to
664 trace new macros; see rules/trace.mk.
665]
666[fix building with extra packages (packages were added to BUILD_DIRS twice)
667Simon Marlow <marlowsd@gmail.com>**20100921100153
668 Ignore-this: 4b425dff9777871ad5ba3e05e1d14483
669 Also add some comments about what extra-packages is doing
670]
671[add extra packages to $(EXTRA_PACKAGES), so we avoid installing them by default
672Simon Marlow <marlowsd@gmail.com>**20100920144307
673 Ignore-this: 3395825d911a8bf7ba8385518d8b517b
674]
675[Fix indexing error in archive loader
676Ian Lynagh <igloo@earth.li>**20100921121642]
677[Add some -Dl belches
678Ian Lynagh <igloo@earth.li>**20100921121624]
679[Add casts to fix warnings
680Ian Lynagh <igloo@earth.li>**20100921121714]
681[Add support for BSD-variant large filenames in .a archives
682Ian Lynagh <igloo@earth.li>**20100921000451]
683[Tell Cabal that we're not building GHCi libs if UseArchivesForGhci=YES
684Ian Lynagh <igloo@earth.li>**20100920230449]
685["UseArchivesForGhci = YES" on darwin
686Ian Lynagh <igloo@earth.li>**20100920211538]
687[Add a dependency that my OS X build has recently started tripping up over
688Ian Lynagh <igloo@earth.li>**20100920210239]
689[Add "Use archives for ghci" to --info output
690Ian Lynagh <igloo@earth.li>**20100920210523]
691[Implement archive loading for ghci
692Ian Lynagh <igloo@earth.li>**20100920201620]
693[Tweak gen_contents_index now dph may not be there
694Ian Lynagh <igloo@earth.li>**20100920201513]
695[Filter out the FFI library when loading package in ghci
696Ian Lynagh <igloo@earth.li>**20100920181032
697 The FFI GHCi import lib isn't needed as
698 compiler/ghci/Linker.lhs + rts/Linker.c link the
699 interpreted references to FFI to the compiled FFI.
700 We therefore filter it out so that we don't get
701 duplicate symbol errors.
702]
703[Loosen the conditions for -XUndecidableInstances; fixes Trac #4200
704simonpj@microsoft.com**20100919162623
705 Ignore-this: 2f4323e278b1ce9250549727ffd0aa1b
706]
707[Further improvements in error messages
708simonpj@microsoft.com**20100919153355
709 Ignore-this: b6fa0b11ae893df1a3ca68f78e427fa
710]
711[Add a flag -fwarn-missing-local-sigs, and improve -fwarn-mising-signatures
712simonpj@microsoft.com**20100919153327
713 Ignore-this: fda8dfca450054ea692be0ee30b01885
714 
715 The new flag prints out a warning if you have a local,
716 polymorphic binding that lacks a type signature. It's meant
717 to help with the transition to the new typechecker, which
718 discourages local let-generalisation.
719 
720 At the same time I moved the missing-signature code to TcHsSyn,
721 where it takes place as part of zonking.  That way the
722 types are reported after all typechecking is complete,
723 thereby fixing Trac #3696.  (It's even more important for
724 local bindings, which is why I made the change.)
725]
726[Include the "stupid theta" in the type of $con2tag
727simonpj@microsoft.com**20100919152201
728 Ignore-this: d95fae78a0e66f48bbd5862573a11f4d
729]
730[Add a release note about the typechecker
731Ian Lynagh <igloo@earth.li>**20100919132927]
732[Enable shared libs on OpenBSD
733Matthias Kilian <kili@outback.escape.de>**20100918205040
734 Ignore-this: 729dd7ac0bba5d758f43bc31b541896
735]
736[Add separate functions for querying DynFlag and ExtensionFlag options
737Ian Lynagh <igloo@earth.li>**20100918163815
738 and remove the temporary DOpt class workaround.
739]
740[Fix mkUserGuidePart deps
741Ian Lynagh <igloo@earth.li>**20100918145904
742 We need to directly depend on the stage1 libs. The stage1 compiler lib
743 doesn't depend on them.
744]
745[Fix build on cygwin: Normalise slashes in .depend files to be /
746Ian Lynagh <igloo@earth.li>**20100918132328
747 Ignore-this: 664f5ef4a41a4461eb34fe2ca7f2729a
748]
749[extra packages info is now read from packages file
750Ian Lynagh <igloo@earth.li>**20100917224409
751 rather than being repeated in the build system
752]
753[Tweak darcs-all
754Ian Lynagh <igloo@earth.li>**20100917194435]
755[Bump dependencies
756Ian Lynagh <igloo@earth.li>**20100917183609]
757[Library release notes for 7.0.1
758Ian Lynagh <igloo@earth.li>**20100917174850]
759[Fix overriding of implicit parameters in the solver
760simonpj@microsoft.com**20100917140403
761 Ignore-this: af76732309c7e2ca6b04f49327e9c14b
762]
763[Minor type printing amomaly
764simonpj@microsoft.com**20100917140204
765 Ignore-this: c90cb2e51421b4543a827e096051772e
766]
767[Spaces only
768simonpj@microsoft.com**20100917140156
769 Ignore-this: 7e34479502f7cb87d762355e40cbd012
770]
771[Minor refactoring
772simonpj@microsoft.com**20100917140150
773 Ignore-this: 6c0648b949b91b7e2f23c136b124faf2
774]
775[Add types of implicit parameters as untouchable
776simonpj@microsoft.com**20100917140138
777 Ignore-this: ba80740a557a9ba062dc7756e2320d17
778 
779 This is a tricky point:
780    see Note [Implicit parameter untouchables]
781]
782[Better pretty printing of implicit parameters
783simonpj@microsoft.com**20100917140054
784 Ignore-this: 867dd67818a5bd687b2b6a1b59e15775
785]
786[Yet more error message improvement
787simonpj@microsoft.com**20100917121206
788 Ignore-this: 647fe8129d1d39d81e8249debd8df94e
789]
790[More error message wibbles
791simonpj@microsoft.com**20100917094721
792 Ignore-this: 8ec2f150b96b26af2e9ab7ac2b371fc7
793]
794[More error refactoring
795simonpj@microsoft.com**20100917092834
796 Ignore-this: 2d570ac0b9cc11305ddd33d093d11324
797]
798[Refactor type errors a bit
799simonpj@microsoft.com**20100917080726
800 Ignore-this: 33da4549373f585064e2ee22b50ad6ac
801 
802 Improves kind error messages in paticular
803]
804[Fix a very subtle shadowing bug in optCoercion
805simonpj@microsoft.com**20100916170452
806 Ignore-this: 9041cfb3c93e27a5e644e57815313aae
807 
808 See Note [Subtle shadowing in coercions]
809 
810 This is what was going wrong in Trac 4160.
811]
812[Fix bad error in tyVarsOfType
813simonpj@microsoft.com**20100916170348
814 Ignore-this: 67c8ce96a668cf6e3a38b82c893bcd81
815 
816 We weren't gathering the type variables free in the kind
817 of a coercion binder!
818]
819[More assertions
820simonpj@microsoft.com**20100916170310
821 Ignore-this: 7fdcb53c99d791621a3d7e01ef454404
822]
823[Add more location info in CoreLint
824simonpj@microsoft.com**20100916170229
825 Ignore-this: 6558bab544b4f30189e0430668db87c3
826]
827[Print coercion variables as such (debugging change only)
828simonpj@microsoft.com**20100916165944
829 Ignore-this: c6d2001c1d8279a2288cb63bc339577d
830]
831[Remove pprTrace
832simonpj@microsoft.com**20100915225935
833 Ignore-this: 28185bbfa9732386f3c0f3eb4781a637
834]
835[Remove dead code dealing with type refinement
836simonpj@microsoft.com**20100915223230
837 Ignore-this: 62824b5c2ec1077c7642163352559621
838]
839[Use mkAppTy
840simonpj@microsoft.com**20100915223205
841 Ignore-this: e79e087b6a49219e9088846a1253a153
842 
843 Using AppTy in CoreLint was giving a bogus Lint failure
844]
845[Comments only
846simonpj@microsoft.com**20100915221253
847 Ignore-this: 3a45ea614188ccbb4a462de5cac96eda
848]
849[Extend eta reduction to work with casted arguments
850simonpj@microsoft.com**20100915221229
851 Ignore-this: 24b103dcdf70331211507af929789f86
852 
853 See Trac #4201, and
854 Note [Eta reduction with casted arguments]
855 
856 Thanks to Louis Wasserman for suggesting this, and
857 implementing an early version of the patch
858]
859[Allow "INLINEABLE" as a synonym
860simonpj@microsoft.com**20100915154249
861 Ignore-this: f41f80cb769e9acd5b463b170df698d0
862]
863[Documentation for INLINABLE
864simonpj@microsoft.com**20100915154235
865 Ignore-this: f942c02bcadc0d2d2f05b9369f93e280
866]
867[Implement TH reification of instances (Trac #1835)
868simonpj@microsoft.com**20100915151242
869 Ignore-this: 97dfa83db7da8f6cbd1b96801a57f8c5
870 
871 Accompanying patch for template-haskell package is reqd
872]
873[errno corresponding to ERROR_NO_DATA should be EPIPE (non-threaded RTS)
874Simon Marlow <marlowsd@gmail.com>**20100915141809
875 Ignore-this: 709c7280fbaa762e7071fb8796e8c01e
876]
877[Windows: use a thread-local variable for myTask()
878Simon Marlow <marlowsd@gmail.com>**20100915120627
879 Ignore-this: 13ffa4f19ebd319fe672af53af8d0b9a
880 Which entailed fixing an incorrect #ifdef in Task.c
881]
882[Fix typo
883Ian Lynagh <igloo@earth.li>**20100915140814]
884[Add quotes in error message
885simonpj@microsoft.com**20100915144724
886 Ignore-this: c5158047c0aa41947a79e4c8edbe54f4
887]
888[Fix isDefaultInlinePragma
889simonpj@microsoft.com**20100915144710
890 Ignore-this: c9addf6bf811b23dc12603cf8521aa6c
891]
892[Implement INLINABLE pragma
893simonpj@microsoft.com**20100915124442
894 Ignore-this: 80a4ab2c2d65b27868dc9b2e954d6c6f
895 
896 Implements Trac #4299.  Documentation to come.
897]
898[Less voluminous error when derived code doesn't typecheck
899simonpj@microsoft.com**20100915072301
900 Ignore-this: eca7871dcc50c1070a0b530711adea27
901]
902[Improve pretty-printing of family instances
903simonpj@microsoft.com**20100915123219
904 Ignore-this: 25ec6bcc7e8a7f7c303b38ca201db90e
905 
906 Fixed Trac #4246
907]
908[Fix Trac #4240: -ddump-minimal-imports
909simonpj@microsoft.com**20100915121937
910 Ignore-this: ab85057cb829a42ea44a92f7b4af24a3
911 
912 See Note [Partial export] for the details.
913 I also fixed one egregious bug that was just
914 waiting to bite: we were using loadSysInterface
915 instead of loadSrcInterface.
916]
917[Comments only
918simonpj@microsoft.com**20100915105707
919 Ignore-this: ab3a5f16f8260b7b8570e748bf97998a
920]
921[implement setThreadAffinity on Windows (#1741)
922Simon Marlow <marlowsd@gmail.com>**20100914155844
923 Ignore-this: a14c7b4ef812007042342d0a25478f0b
924]
925[COFF: cope with new debug sections in gcc 4.x (fixes ghciprog004)
926Simon Marlow <marlowsd@gmail.com>**20100914153026
927 Ignore-this: f340e40a2b0390836bc61bba144a04ed
928 Also updated the object file parser to properly handle the overflow
929 case for section names longer than 8 chars.
930]
931[eliminate clutter from make output
932Simon Marlow <marlowsd@gmail.com>**20100915105712
933 Ignore-this: bfa4480dd239dda2a02ac391b6a9219c
934]
935[rts_isProfiled should be a visible API (fixes T2615(dyn))
936Simon Marlow <marlowsd@gmail.com>**20100915083941
937 Ignore-this: b8ac09bb9d1a929bf45c6122f8485561
938]
939[Fix the "lost due to fragmentation" calculation
940Simon Marlow <marlowsd@gmail.com>**20100914145945
941 Ignore-this: cdffcc9f3061c3a33da5171be111fc43
942 It was counting the space used by block descriptors as "lost"
943]
944[fix +RTS -S output: use peak_mblocks_allocated, now that mblocks can be freed
945Simon Marlow <marlowsd@gmail.com>**20100914135030
946 Ignore-this: 65d21e5f86d3ab6ab4d6c255180b6968
947]
948[Fix egregious bug in deeplyInstantiate
949simonpj@microsoft.com**20100915070325
950 Ignore-this: 22ede973038877af2673339aaf5de6cf
951 
952 This resulted in an infinite loop in applyTypeToArgs, in syb
953]
954[Improve HsSyn pretty printing
955simonpj@microsoft.com**20100915070255
956 Ignore-this: 7c8e2d86a482453c7e69e22bc31cb03f
957]
958[Remove (most of) the FiniteMap wrapper
959Ian Lynagh <igloo@earth.li>**20100914201703
960 We still have
961     insertList, insertListWith, deleteList
962 which aren't in Data.Map, and
963     foldRightWithKey
964 which works around the fold(r)WithKey addition and deprecation.
965]
966[Improve ASSERT
967simonpj@microsoft.com**20100914113900
968 Ignore-this: dbc0363be5924f543316e77f7d18dd77
969]
970[Comment on what an "enumeration" type is
971simonpj@microsoft.com**20100914113850
972 Ignore-this: c09c8591e3140f305d55fbf945adbf95
973]
974[Make absent-arg wrappers work for unlifted types (fix Trac #4306)
975simonpj@microsoft.com**20100914113827
976 Ignore-this: 1945e56779329e8b79780403710aba98
977 
978 Previously we were simply passing arguments of unlifted
979 type to a wrapper, even if they were absent, which was
980 stupid.
981 
982 See Note [Absent error Id] in WwLib.
983]
984[Comments only
985simonpj@microsoft.com**20100914113641
986 Ignore-this: 3191ce856c9b5d9700cedc9b149b8097
987]
988[Move error-ids to MkCore (from PrelRules)
989simonpj@microsoft.com**20100914113635
990 Ignore-this: c3d820db62ba6139dd7c96bf97e51bb5
991 
992 and adjust imports accordingly
993]
994[More wibbles to deriving error messages
995simonpj@microsoft.com**20100914113523
996 Ignore-this: bd2df662644961138fa209aec843a2aa
997]
998[Fix getThreadCPUTime()
999Simon Marlow <marlowsd@gmail.com>**20100913153838
1000 Ignore-this: 950e048a5724086534b74c609c7d5ed
1001 ever since the patch "Check with sysconf _POSIX_THREAD_CPUTIME", it
1002 has been returning incorrect results, because the sysconf variable to
1003 check should have been _SC_THREAD_CPUTIME, not _POSIX_THREAD_CPUTIME.
1004]
1005[filter out the gcc-lib directory from the rts package's library-dirs
1006Simon Marlow <marlowsd@gmail.com>**20100913101259
1007 Ignore-this: 46dc1dccbfee8a65f9243e125eee117f
1008 fixes problems when building with GHC 6.10 on Windows
1009]
1010[Don't include GC time in heap profiles (#4225)
1011Simon Marlow <marlowsd@gmail.com>**20100913133852
1012 Ignore-this: 68ac48b004b311384b5996c6b33ba5cc
1013]
1014[Use clock_gettime (if available) to measure the process CPU time
1015Simon Marlow <marlowsd@gmail.com>**20100913133818
1016 Ignore-this: 8c9300df9b929bfc1db4713c9b6065b3
1017 This is much more accurate than getrusage, which was giving misleading
1018 results when trying to time very quick operations like a minor GC.
1019]
1020[make stg_arg_bitmaps public, and available via the GHCi linker (#3672)
1021Simon Marlow <marlowsd@gmail.com>**20100913105235
1022 Ignore-this: e18efd0bd77c521e5530fb59e93b5a42
1023]
1024[fix typo
1025Simon Marlow <marlowsd@gmail.com>**20100913105100
1026 Ignore-this: 6049eea21208864203b2d79db2edd143
1027]
1028[Update release notes and docs with LLVM info.
1029David Terei <davidterei@gmail.com>**20100914072135
1030 Ignore-this: 5b3d0e5c9d5da98ed6ae9c2e8e1f6f30
1031]
1032[Remove defaultExtensionFlags
1033Ian Lynagh <igloo@earth.li>**20100913165949
1034 The default should do into languageExtensions instead
1035]
1036[Improve crash message
1037simonpj@microsoft.com**20100913170407
1038 Ignore-this: 5c26a9979f18be8cd12cea823c9f4b5a
1039]
1040[Fix Trac #4302, plus a little refactoring
1041simonpj@microsoft.com**20100913170355
1042 Ignore-this: cf6886b587aa0e8d723362183625d946
1043]
1044[Fix build with 6.10
1045Ian Lynagh <igloo@earth.li>**20100913160048]
1046[Haddock fixes
1047simonpj@microsoft.com**20100913120510
1048 Ignore-this: f3157d6969f10d4cbd593000a477138b
1049]
1050[Remove two old junk files
1051simonpj@microsoft.com**20100913103426
1052 Ignore-this: ed7af5ef1b9592178909a8d876345302
1053]
1054[Super-monster patch implementing the new typechecker -- at last
1055simonpj@microsoft.com**20100913095048
1056 Ignore-this: 14d14a1e4d7a414f5ae8d9d89d1c6a4b
1057 
1058 This major patch implements the new OutsideIn constraint solving
1059 algorithm in the typecheker, following our JFP paper "Modular type
1060 inference with local assumptions". 
1061 
1062 Done with major help from Dimitrios Vytiniotis and Brent Yorgey.
1063 
1064]
1065[Fix simplifier statistics
1066simonpj@microsoft.com**20100909085441
1067 Ignore-this: 48e383655aafc912dea15c4d94382863
1068]
1069[Trace output
1070simonpj@microsoft.com**20100908170056
1071 Ignore-this: 4b67fa4b310fbf0a16b852686d2d3294
1072]
1073[Better debug output
1074simonpj@microsoft.com**20100908170047
1075 Ignore-this: 410cef00616dda7c0c162f65216e8ca3
1076]
1077[Add Outputable instance for OccEncl
1078simonpj@microsoft.com**20100908150510
1079 Ignore-this: 6362ef9028287d84f070eaf8963c1bfc
1080]
1081[Better simplifier counting
1082simonpj@microsoft.com**20100907214840
1083 Ignore-this: 9d4722703f8f47447e86a28c8c50e0ea
1084]
1085[Put liftStringName into the known-key names
1086simonpj@microsoft.com**20100906112415
1087 Ignore-this: 287064d14ff484da1a6dea6924bc9235
1088]
1089[Deprecate NoRelaxedPolyRec
1090simonpj@microsoft.com**20100903234519
1091 Ignore-this: 607217e77f6bc1b91bf57dfd8dd2b967
1092]
1093[Buglet in Core Lint
1094simonpj@microsoft.com**20100903234457
1095 Ignore-this: 277535d51b396d3b4b0265a0939c2d4
1096]
1097[Give seqId the right type
1098simonpj@microsoft.com**20100903093556
1099 Ignore-this: d1fc7a73dea160614a8d4ddc930f99cd
1100]
1101[Remove dead code
1102simonpj@microsoft.com**20100903093548
1103 Ignore-this: 92cc3f7651445aa349ee7f114d3ec758
1104]
1105[Comments and layout
1106simonpj@microsoft.com**20100903093502
1107 Ignore-this: 9987d1409e654992c1cb1be35cb87728
1108]
1109[Remove checkFreeness; no longer needed
1110simonpj@microsoft.com**20100902233227
1111 Ignore-this: c96a12ac9794290aa30402317d88c095
1112]
1113[Assert
1114simonpj@microsoft.com**20100902073642
1115 Ignore-this: 4be1ab2f6096665ae5ec7fdd1f025a67
1116]
1117[Add aserts
1118simonpj@microsoft.com**20100902073211
1119 Ignore-this: e1409441217fd070c5a7f9ee4cca99ab
1120]
1121[Wibbles
1122simonpj@microsoft.com**20100831113540
1123 Ignore-this: 903811ab493a7b560a62eb86fcf3ee25
1124]
1125[Wibble to allow phantom types in Enum
1126simonpj@microsoft.com**20100825112711
1127 Ignore-this: fdef1c50d92b4a3d46bbe4cbfd8a83ea
1128]
1129[Add HsCoreTy to HsType
1130simonpj@microsoft.com**20100824141845
1131 Ignore-this: 4ca742b099f9cc90af3167f1012dbba6
1132 
1133 The main thing here is to allow us to provide type
1134 signatures for 'deriving' bindings without pain.
1135]
1136[Comments
1137simonpj@microsoft.com**20100823223654
1138 Ignore-this: dd412a55839430c436902d8699d6900b
1139]
1140[Wibbles to error message
1141simonpj@microsoft.com**20100823163308
1142 Ignore-this: 4d6cd8e613762dca8135c2e3b09264ec
1143]
1144[Correct type signatures
1145simonpj@microsoft.com**20100823153045
1146 Ignore-this: 42942309221a443258246098f9c0a13b
1147]
1148[Add missing signatures
1149simonpj@microsoft.com**20100823112413
1150 Ignore-this: 8ee1ce40456306de469938c02df4fed5
1151]
1152[Add type signatures in "deriving" bindings
1153simonpj@microsoft.com**20100820234230
1154 Ignore-this: 4726b28968cf65ec16cb65b7e0e7303e
1155]
1156[Minor
1157dimitris@microsoft.com**20100820131021]
1158[Be a bit less aggressive in mark-many inside a cast
1159simonpj@microsoft.com**20100819104804
1160 Ignore-this: 3fd48fe7647ec7a58c2032cd86ca4d4f
1161]
1162[Wibble
1163simonpj@microsoft.com**20100818185738
1164 Ignore-this: d5c939311377c0d0c7244aa339193315
1165]
1166[Pretty printing change
1167simonpj@microsoft.com**20100818065436
1168 Ignore-this: 4f7e70976dbe52f95effb3e634dfef5d
1169]
1170[Remember to zonk FlatSkols!
1171simonpj@microsoft.com**20100811143555
1172 Ignore-this: 84f7f9dbda97f561a918c69308ddef9a
1173]
1174[De-polymorphise
1175simonpj@microsoft.com**20100730151217
1176 Ignore-this: a9304487b983e517a9083fd697f77576
1177]
1178[Work around missing type signature in Happy
1179simonpj@microsoft.com**20100730122405
1180 Ignore-this: 7f241a655d93c5ad7763a7ffe8db0c7a
1181 
1182 Happy generates
1183 
1184       notHappyAtAll = error "Blah"
1185 
1186 without a type signature, and currently the new
1187 typechecker doesn't generalise it.  This patch
1188 says "no monomorphism restriction" which makes it
1189 generalise again.
1190 
1191 Better would be to add a type sig to Happy's template
1192]
1193[Add two local type signatures
1194simonpj@microsoft.com**20100729152611
1195 Ignore-this: afa99bcc515469aa0990d44d8c18a9e6
1196]
1197[Second test from Simon's laptop
1198simonpj@microsoft.com**20100729091703
1199 Ignore-this: 4dc64cadae314a5a1b05cc5326918a83
1200]
1201[Test commit from Simon's laptop
1202simonpj@microsoft.com**20100729091344
1203 Ignore-this: 109eff835cc19e9f93799d12f09b0ba7
1204]
1205[Add OutsideIn flag
1206simonpj@microsoft.com**20100728075525
1207 Ignore-this: 69c2f5c3a15fa653f6da80598aa8d74d
1208]
1209[Layout only
1210simonpj@microsoft.com**20100727141539
1211 Ignore-this: 1a58a8fe80ba8bced18ae81a2efb9495
1212]
1213[Improvement to SimplUtils.mkLam
1214simonpj@microsoft.com**20100727131659
1215 Ignore-this: 739beaefa79baa7e0ebeb5b2b6d1ea91
1216]
1217[Give the correct kind to unsafeCoerce#
1218simonpj@microsoft.com**20100727131538
1219 Ignore-this: 6b787de3b398c6d7a61fa04fccd15fd6
1220]
1221[Suppress warnings about recursive INLINE in output of desugarer
1222simonpj@microsoft.com**20100727094549
1223 Ignore-this: a361f7238c0fcba526d46326722c42e
1224]
1225[Rename CorePrep.tryEtaReduce to tryEtaReducePrep
1226simonpj@microsoft.com**20100726231253
1227 Ignore-this: 4375ddace205745244ba224ae012252
1228 
1229 This avoids the name clash with the similar but
1230 not identical CoreUtils.tryEtaReduce
1231]
1232[Add a trace message
1233simonpj@microsoft.com**20100719211111
1234 Ignore-this: b5daebe46e50c8cf28cc693f84bbf099
1235]
1236[Don't use RelaxedPolyRec in the compiler; it's built in now
1237simonpj@microsoft.com**20100719170441
1238 Ignore-this: a2e4489cdf63478e46282a421ee7aec3
1239]
1240[Remove duplicated #defines for FreeBSD
1241Matthias Kilian <kili@outback.escape.de>**20100912181518
1242 Ignore-this: d16214fef8635c7c9ef4edec4e8e7896
1243]
1244[Don't fail with absolute silence
1245Matthias Kilian <kili@outback.escape.de>**20100912150506
1246 Ignore-this: 479e2321f39b263fa2d9f80491e5e9f7
1247]
1248[Add a release note: "-dynload wrapper" removed
1249Ian Lynagh <igloo@earth.li>**20100911195809]
1250[put back the conversion of warn-lazy-unlifted-bindings into an error until 7.2
1251Ian Lynagh <igloo@earth.li>**20100911193434
1252 I think we'll currently still have too many people with old versions of
1253 alex/happy to want to make this an error now.
1254]
1255[6.14 -> 7.0
1256Ian Lynagh <igloo@earth.li>**20100911192837]
1257[Add a couple more release notes
1258Ian Lynagh <igloo@earth.li>**20100911162059]
1259[Document -dsuppress-module-prefixes
1260Ian Lynagh <igloo@earth.li>**20100911162005]
1261[Enable -fregs-graph with -O2; fixes #2790
1262Ian Lynagh <igloo@earth.li>**20100910191301]
1263[Remove unused code
1264Ian Lynagh <igloo@earth.li>**20100909170207]
1265[Fix warnings
1266Ian Lynagh <igloo@earth.li>**20100909154348]
1267[Fix warnings
1268Ian Lynagh <igloo@earth.li>**20100909150957]
1269[Remove context completion
1270lykahb@gmail.com**20100901160153
1271 Ignore-this: dc61b259dcb7063f0c76f56788b5d2af
1272 Now completion suggests to remove only modules added to context before.
1273]
1274[avoid Foreign.unsafePerformIO
1275Ross Paterson <ross@soi.city.ac.uk>**20100909125531
1276 Ignore-this: 5cabeae4cffec8fc17ef7c0cabbea22a
1277]
1278[updates to the release notes
1279Simon Marlow <marlowsd@gmail.com>**20100909111450
1280 Ignore-this: a4d25ad8815c305b7e0f21fd4f6ee37b
1281]
1282[newAlignedPinnedByteArray#: avoid allocating an extra word sometimes
1283Simon Marlow <marlowsd@gmail.com>**20100909110805
1284 Ignore-this: 996a3c0460068ab2835b4920905b3e75
1285]
1286[Finish breaking up vectoriser utils
1287benl@ouroborus.net**20100909061311
1288 Ignore-this: 217fe1d58a3e8bb13200bcb81353a416
1289]
1290[Move VectType module to Vectorise tree
1291benl@ouroborus.net**20100909042451
1292 Ignore-this: 5af8cf394d4835911259ca3ffb6774c5
1293]
1294[Sort all the PADict/PData/PRDict/PRepr stuff into their own modules
1295benl@ouroborus.net**20100909035147
1296 Ignore-this: 53436329773347cad793adbd83e90a9e
1297]
1298[Break out Repr and PADict stuff for vectorisation of ADTs to their own modules
1299benl@ouroborus.net**20100909025759
1300 Ignore-this: d2b7d2f79332eda13416449742f7cf1c
1301]
1302[Break out conversion functions to own module
1303benl@ouroborus.net**20100909023332
1304 Ignore-this: 613f2666b6ca7f2f8876fcc1e4a59593
1305]
1306[Comments and formatting only
1307benl@ouroborus.net**20100909022117
1308 Ignore-this: c8e30139d730669e5db44f0ef491a588
1309]
1310[Remove "-dynload wrapper"; fixes trac #4275
1311Ian Lynagh <igloo@earth.li>**20100908213251]
1312[Don't set visibility on Windows
1313Ian Lynagh <igloo@earth.li>**20100905122442
1314 With gcc 4.5.0-1, using visibility hidden gives:
1315     error: visibility attribute not supported in this configuration; ignored
1316]
1317[Fix warnings on Windows
1318Ian Lynagh <igloo@earth.li>**20100905111201
1319 Ignore-this: c5cce63bb1e0c7a27271bed78d25fbc5
1320]
1321[Fix gcc wrapper for new mingw binaries
1322Ian Lynagh <igloo@earth.li>**20100905001807
1323 Ignore-this: f6acc8c911055ffce632bac138ccc939
1324]
1325[Don't pass our gcc options to stage0 ghc's gcc; they may not be suitable
1326Ian Lynagh <igloo@earth.li>**20100905103129]
1327[Update intree-mingw creation
1328Ian Lynagh <igloo@earth.li>**20100904225559]
1329[Update commands to build in-tree mingw
1330Ian Lynagh <igloo@earth.li>**20100904215112]
1331[Break out hoisting utils into their own module
1332benl@ouroborus.net**20100908074102
1333 Ignore-this: e3ba4ed0252a2def1ed88a9e14c58fea
1334]
1335[Break out closure utils into own module
1336benl@ouroborus.net**20100908072040
1337 Ignore-this: 216172b046ff101cf31a1753667a5383
1338]
1339[Move VectVar module to Vectorise tree
1340benl@ouroborus.net**20100908065904
1341 Ignore-this: 1fba5333d29927dba4275381e1a7f315
1342]
1343[Break out vectorisation of expressions into own module
1344benl@ouroborus.net**20100908065128
1345 Ignore-this: 6a952b80fb024b5291f166477eb1976
1346]
1347[Break out TyCon classifier into own module
1348benl@ouroborus.net**20100908063111
1349 Ignore-this: da754c4ef6960b4e152ea1bf8c04ab6f
1350]
1351[Break out vectorisation of TyConDecls into own module
1352benl@ouroborus.net**20100908052004
1353 Ignore-this: c0ab4fb2a05ca396efe348b384db1ebf
1354]
1355[Break out type vectorisation into own module
1356benl@ouroborus.net**20100907110311
1357 Ignore-this: 67bd70a21d16468daf68dd3ec1ff7d62
1358]
1359[Tidy up the ArchHasAdjustorSupport definition
1360Ian Lynagh <igloo@earth.li>**20100904144234]
1361[ppc: switch handling of 'foreign import wrapper' (FIW) to libffi
1362Sergei Trofimovich <slyfox@community.haskell.org>**20100829192859
1363 Ignore-this: 662ea926681ebea0759e2a04a38e82b7
1364 
1365 Joseph Jezak reported darcs-2.4.4 SIGSEGV in interactive mode in ghc-6.12.3.
1366 So I've concluded ppc also has rotten native adjustor. I don't have hardware
1367 to verify the patch (ticket #3516 should help to test it), but I think it will
1368 help (as similar patch helped for ia64 and ppc64).
1369]
1370[Binary no longer has unusable UNPACK pragmas, so no need to turn of -Werror
1371Ian Lynagh <igloo@earth.li>**20100904133339]
1372[Don't haddock packages that we aren't going to install
1373Ian Lynagh <igloo@earth.li>**20100903231921]
1374[Give haddock per-package source entity paths; fixes #3810
1375Ian Lynagh <igloo@earth.li>**20100903221335]
1376[update for containers-0.4
1377Simon Marlow <marlowsd@gmail.com>**20100903105131
1378 Ignore-this: 556eac0e4926c9b8af6b66d7b069302c
1379]
1380[Fix for nursery resizing: the first block's back pointer should be NULL
1381Simon Marlow <marlowsd@gmail.com>**20100827102818
1382 Ignore-this: fb68938e3f1e291e3c9e5e8047f9dcd2
1383 I'm not sure if this could lead to a crash or not, but it upsets +RTS -DS
1384 Might be related to #4265
1385]
1386[Add some -no-user-package-conf flags
1387Ian Lynagh <igloo@earth.li>**20100902224726
1388 Stops user-installed packages breaking the build
1389]
1390[Fix warnings: Remove unused imports
1391Ian Lynagh <igloo@earth.li>**20100902204342]
1392[Finish breaking up VectBuiltIn and VectMonad, and add comments
1393benl@ouroborus.net**20100831100724
1394 Ignore-this: 65604f3d22d03433abc12f10be40050d
1395]
1396[Fix warnings
1397benl@ouroborus.net**20100830083746
1398 Ignore-this: 2a0e000985f694582a6f9a9261ff2739
1399]
1400[Break up vectoriser builtins module
1401benl@ouroborus.net**20100830070900
1402 Ignore-this: b86bd36a7875abdcf16763902ba2e637
1403]
1404[Move VectCore to Vectorise tree
1405benl@ouroborus.net**20100830053415
1406 Ignore-this: d5763ca6424285b39a58c7792f4a84a1
1407]
1408[Split out vectoriser environments into own module
1409benl@ouroborus.net**20100830050252
1410 Ignore-this: 5319111c74831394d2c29b9aedf5a766
1411]
1412[Comments and formatting to vectoriser, and split out varish stuff into own module
1413benl@ouroborus.net**20100830042722
1414 Ignore-this: d3f0c98ed8124dd0fca9a2ccea3e15fd
1415]
1416[Fix warnings
1417benl@ouroborus.net**20100830040340
1418 Ignore-this: d6cfad803ad4617e7fdaa62e4a895282
1419]
1420[Fix warning about multiply exported name
1421benl@ouroborus.net**20100830035243
1422 Ignore-this: 27ce2c1d22d9f99929d16a426343044e
1423]
1424[Vectorisation of method types
1425benl@ouroborus.net**20100830032941
1426 Ignore-this: 75614571d5c246a4906edb3b39ab1e0b
1427]
1428[Comments and formatting to vectoriser
1429benl@ouroborus.net**20100830032516
1430 Ignore-this: fe665b77108501c7960d858be3290761
1431]
1432[Implement -dsuppress-module-prefixes
1433benl@ouroborus.net**20100830032428
1434 Ignore-this: 2bb8bad9c60ef9044132bba118010687
1435]
1436[Whitespace only
1437benl@ouroborus.net**20100527045629
1438 Ignore-this: 4c160dfa77727e659817b6af9c84684a
1439]
1440[Disambiguate a function name
1441Ian Lynagh <igloo@earth.li>**20100828225827]
1442[users_guide.xml is now generated
1443Ian Lynagh <igloo@earth.li>**20100828225751]
1444[Pass more -pgm flags in the ghc wrapper; fixes #3863
1445Ian Lynagh <igloo@earth.li>**20100827204537]
1446[Add a new-IO manager release note
1447Ian Lynagh <igloo@earth.li>**20100827171616]
1448[Merge a duplicate release note
1449Ian Lynagh <igloo@earth.li>**20100827171511]
1450[Typo, spotted by Johan Tibell
1451Ian Lynagh <igloo@earth.li>**20100827153914]
1452[First pass at 6.14.1 release notes
1453Ian Lynagh <igloo@earth.li>**20100826220811]
1454[Fix typo
1455Ian Lynagh <igloo@earth.li>**20100824201330]
1456[FIX BUILD: add rts_isProfiled to the symbol table
1457Simon Marlow <marlowsd@gmail.com>**20100826094319
1458 Ignore-this: 9536ddb0a94721c8dec03a2a981cfa83
1459]
1460[Fix the DPH package cleaning/profiled mess even more (the build was broken)
1461Simon Marlow <marlowsd@gmail.com>**20100826084436
1462 Ignore-this: 49d7e4db2fb53b856c213c74c8969d82
1463]
1464[Remove the debugging memory allocator - valgrind does a better job
1465Simon Marlow <marlowsd@gmail.com>**20100824113537
1466 Ignore-this: a3731a83dc18b0fd0de49452e695a7ca
1467 
1468 I got fed up with the constant bogus output from the debugging memory
1469 allocator in RtsUtils.c.  One problem is that we allocate memory in
1470 constructors that then isn't tracked, because the debugging allocator
1471 hasn't been initialised yet.
1472 
1473 The bigger problem is that for a given piece of leaking memory it's
1474 impossible to find out where it was allocated; however Valgrind gives
1475 output like this:
1476 
1477 ==6967== 8 bytes in 1 blocks are still reachable in loss record 1 of 7
1478 ==6967==    at 0x4C284A8: malloc (vg_replace_malloc.c:236)
1479 ==6967==    by 0x4C28522: realloc (vg_replace_malloc.c:525)
1480 ==6967==    by 0x6745E9: stgReallocBytes (RtsUtils.c:213)
1481 ==6967==    by 0x68D812: setHeapAlloced (MBlock.c:91)
1482 ==6967==    by 0x68D8E2: markHeapAlloced (MBlock.c:116)
1483 ==6967==    by 0x68DB56: getMBlocks (MBlock.c:240)
1484 ==6967==    by 0x684F55: alloc_mega_group (BlockAlloc.c:305)
1485 ==6967==    by 0x6850C8: allocGroup (BlockAlloc.c:358)
1486 ==6967==    by 0x69484F: allocNursery (Storage.c:390)
1487 ==6967==    by 0x694ABD: allocNurseries (Storage.c:436)
1488 ==6967==    by 0x6944F2: initStorage (Storage.c:217)
1489 ==6967==    by 0x673E3C: hs_init (RtsStartup.c:160)
1490 
1491 which tells us exactly what the leaking bit of memory is.  So I don't
1492 think we need our own debugging allocator.
1493]
1494[free the entries in the thread label table on exit
1495Simon Marlow <marlowsd@gmail.com>**20100824112606
1496 Ignore-this: c9d577c06548cda80791e590e40d35b3
1497]
1498[Panic in the right way
1499simonpj@microsoft.com**20100825091614
1500 Ignore-this: e6ea4f6dfd2aea088828ea7a945ddd5f
1501]
1502[Fix the DPH/profiled make thing (again)
1503simonpj@microsoft.com**20100825091602
1504 Ignore-this: bc58fa48034ac40cf7be4170958ea29e
1505]
1506[Don't test for gcc flags before we've found gcc
1507Ian Lynagh <igloo@earth.li>**20100824131401]
1508[Change how the dblatex/lndir problem is worked around
1509Ian Lynagh <igloo@earth.li>**20100824130938
1510 Hack: dblatex normalises the name of the main input file using
1511 os.path.realpath, which means that if we're in a linked build tree,
1512 it find the real source files rather than the symlinks in our link
1513 tree. This is fine for the static sources, but it means it can't
1514 find the generated sources.
1515 
1516 We therefore also generate the main input file, so that it really
1517 is in the link tree, and thus dblatex can find everything.
1518]
1519[Clean the generated userguide sources
1520Ian Lynagh <igloo@earth.li>**20100824105827
1521 Ignore-this: 39b4f9702c688c053ed3273b20969597
1522]
1523[DPH should not even be built if GhcProfiled
1524simonpj@microsoft.com**20100823133439
1525 Ignore-this: 62acbf83de5b70ff6d27ab38ce9218ae
1526 
1527 It's not just when cleaning!
1528]
1529[The templateHaskellOk check should only run in stage2
1530simonpj@microsoft.com**20100823133353
1531 Ignore-this: f6dc9292923a1ca201953c5f58c0af3c
1532 
1533 Because rtsIsProfiled is only available in stage2
1534]
1535[Add a couple of missing tests for EAGER_BLACKHOLE
1536Simon Marlow <marlowsd@gmail.com>**20100823104654
1537 Ignore-this: 70c981b86370b0c7564b29b057650897
1538 This was leading to looping and excessive allocation, when the
1539 computation should have just blocked on the black hole. 
1540 
1541 Reported by Christian Höner zu Siederdissen <choener@tbi.univie.ac.at>
1542 on glasgow-haskell-users.
1543]
1544[Don't check for swept blocks in -DS.
1545Marco Túlio Gontijo e Silva <marcot@marcot.eti.br>**20100718225526
1546 Ignore-this: ad5dcf3c247bc19fbef5122c1142f3b2
1547 
1548 The checkHeap function assumed the allocated part of the block contained only
1549 alive objects and slops.  This was not true for blocks that are collected using
1550 mark sweep.  The code in this patch skip the test for this kind of blocks.
1551]
1552[Fix "darcs get"
1553Ian Lynagh <igloo@earth.li>**20100822183542]
1554[Add "darcs-all upstreampull"
1555Ian Lynagh <igloo@earth.li>**20100822163419
1556 This pulls from the upstream repos, for those packages which have
1557 upstreams
1558]
1559[Generate the bit in the user guide where we say what -fglasgow-exts does
1560Ian Lynagh <igloo@earth.li>**20100822155514
1561 Stops the docs going out of sync with the code.
1562]
1563[Factor out the packages file parsing in darcs-all
1564Ian Lynagh <igloo@earth.li>**20100822154813]
1565[Document --supported-extensions
1566Ian Lynagh <igloo@earth.li>**20100822134530]
1567[fix extraction of command stack of arguments of arrow "forms" (fixes #4236)
1568Ross Paterson <ross@soi.city.ac.uk>**20100822090022
1569 Ignore-this: a93db04ec4f20540642a19cdc67d1666
1570 
1571 The command stack was being extracted (by unscramble) with the outermost
1572 type first, contrary to the comment on the function.
1573]
1574[minor fix to comment
1575Ross Paterson <ross@soi.city.ac.uk>**20100822085838
1576 Ignore-this: 8d203ba2600eaf4cf21b043dcfa96cdc
1577]
1578[Add the RTS library path to the library search path
1579Ian Lynagh <igloo@earth.li>**20100820155523
1580 In case the RTS is being explicitly linked in. For #3807.
1581]
1582[Remove some duplication of C flags
1583Ian Lynagh <igloo@earth.li>**20100819233743
1584 We now use the CONF_CC_OPTS_STAGEn C flags in machdepCCOpts, rather than
1585 repeating them there.
1586]
1587[Set -fno-stack-protector in CONF_CC_OPTS_STAGE* rathre than extra-gcc-opts
1588Ian Lynagh <igloo@earth.li>**20100819233031
1589 The latter is only used when compiling .hc files, whereas we need it for
1590 .c files too.
1591]
1592[Give clearer errors for bad input in the packages file; suggested by pejo
1593Ian Lynagh <igloo@earth.li>**20100819232420]
1594[Set -march=i686 on OS X x86 in the configure variables
1595Ian Lynagh <igloo@earth.li>**20100819230939
1596 We used to set it only in machdepCCOpts, so this is more consistent
1597]
1598[Give each stage its own Config.hs
1599Ian Lynagh <igloo@earth.li>**20100819224709
1600 This also means the file is generated in a dist directory, not a
1601 source directory.
1602]
1603[Fix cleaning when GhcProfiled = YES
1604Ian Lynagh <igloo@earth.li>**20100819131346
1605 We need to include the DPH cleaning rules, even though we don't build DPH
1606 when GhcProfiled = YES.
1607]
1608[stgReallocBytes(DEBUG): don't fail when the ptr passed in is NULL
1609Simon Marlow <marlowsd@gmail.com>**20100817150836
1610 Ignore-this: 4b5063e65e01399f64a33f0d0555ff38
1611]
1612[Use make-command in rules/bindist.mk
1613Ian Lynagh <igloo@earth.li>**20100818191243
1614 Rather than it having its own specialised version
1615]
1616[Use make-command when installing packages
1617Ian Lynagh <igloo@earth.li>**20100818190600]
1618[Add _DATA_FILES to package-data.mk files
1619Ian Lynagh <igloo@earth.li>**20100818185801]
1620[Add a "make-command" utility Makefile function
1621Ian Lynagh <igloo@earth.li>**20100818183055]
1622[LLVM: Nicer format for lack of shared lib warning
1623David Terei <davidterei@gmail.com>**20100817145207
1624 Ignore-this: 753d45762601d87761614937a1bb6716
1625]
1626[fix FP_CHECK_ALIGNMENT for autoconf 2.66 (fixes #4252)
1627Ross Paterson <ross@soi.city.ac.uk>**20100816142442
1628 Ignore-this: cd784b8888d32b3b2cc2cc0969ec40f
1629 
1630 Recent versions of AS_LITERAL_IF don't like *'s.  Fix from
1631 
1632 http://blog.gmane.org/gmane.comp.sysutils.autoconf.general/month=20100701
1633]
1634[Refactor the command-line argument parsing (again)
1635simonpj@microsoft.com**20100816074453
1636 Ignore-this: 26dc9e37a88660a887a2e316ed7a9803
1637 
1638 This change allows the client of CmdLineParser a bit more flexibility,
1639 by giving him an arbitrary computation (not just a deprecation
1640 message) for each flag. 
1641 
1642 There are several clients, so there are lots of boilerplate changes.
1643 
1644 Immediate motivation: if RTS is not profiled, we want to make
1645 Template Haskell illegal.  That wasn't with the old setup.
1646]
1647[Add upstream repo to the packages file
1648Ian Lynagh <igloo@earth.li>**20100815154741]
1649[Make the "tag" column of the packages file always present
1650Ian Lynagh <igloo@earth.li>**20100815151657
1651 It makes the parsing simpler if we always have the same number of columns
1652]
1653[Disable object splitting on OSX; works around #4013
1654Ian Lynagh <igloo@earth.li>**20100815134759]
1655[Return memory to the OS; trac #698
1656Ian Lynagh <igloo@earth.li>**20100813170402]
1657[Reduce the xargs -s value we use on Windows
1658Ian Lynagh <igloo@earth.li>**20100812223721
1659 With 30000 I was getting:
1660     xargs: value for -s option should be < 28153
1661]
1662[LLVM: Enable shared lib support on Linux x64
1663David Terei <davidterei@gmail.com>**20100813191534
1664 Ignore-this: 642ed37af38e5f17d419bf4f09332671
1665]
1666[Re-do the arity calculation mechanism again (fix Trac #3959)
1667simonpj@microsoft.com**20100813161151
1668 Ignore-this: d4a2aa48150b503b20c25351a79decfb
1669 
1670 After rumination, yet again, on the subject of arity calculation,
1671 I have redone what an ArityType is (it's purely internal to the
1672 CoreArity module), and documented it better.  The result should
1673 fix #3959, and I hope the related #3961, #3983.
1674 
1675 There is lots of new documentation: in particular
1676  * Note [ArityType] 
1677    describes the new datatype for arity info
1678 
1679  * Note [State hack and bottoming functions]
1680    says how bottoming functions are dealt with, particularly
1681    covering catch# and Trac #3959
1682 
1683 I also found I had to be careful not to eta-expand single-method
1684 class constructors; see Note [Newtype classes and eta expansion].
1685 I think this part could be done better, but it works ok.
1686]
1687[Comments only
1688simonpj@microsoft.com**20100813161019
1689 Ignore-this: baf68300d8bc630bf0b7ab27647b33a0
1690]
1691[Modify FloatOut to fix Trac #4237
1692simonpj@microsoft.com**20100813163120
1693 Ignore-this: ffc8d00d4b7f0a8a785fcef312900413
1694 
1695 The problem was that a strict binding was getting floated
1696 out into a letrec. This only happened when profiling was
1697 on.  It exposed a fragility in the floating strategy.  This
1698 patch makes it more robust.  See
1699       Note [Avoiding unnecessary floating]
1700]
1701[Fix egregious bug in SetLevels.notWorthFloating
1702simonpj@microsoft.com**20100813161429
1703 Ignore-this: d22865f48d417e6a6b732de3dfba378f
1704 
1705 This bug just led to stupid code, which would
1706 later be optimised away, but better not to generate
1707 stupid code in the first place.
1708]
1709[Delete GhcLibProfiled
1710simonpj@microsoft.com**20100813140152
1711 Ignore-this: 2e1a3f677308be726bd022f45e2fd856
1712 
1713 Simon M and I looked at this, and we think GhcLibProfiled is
1714 (a) not needed (b) confusing.
1715 
1716 Ian should review.
1717 
1718 Really, if GhcProfiled is on we should also
1719 check that 'p' is in the GhcLibWays
1720]
1721[Do not build DPH when GhcProfiled (fixes #4172)
1722simonpj@microsoft.com**20100813140021
1723 Ignore-this: 9e20181643b456e841f845ae0cab0a9a
1724 
1725 Reason: DPH uses Template Haskell and TH doesn't work
1726 in a profiled compiler
1727]
1728[Fix Trac #4220
1729simonpj@microsoft.com**20100812131319
1730 Ignore-this: 33141cfd81627592150a9e5973411ff8
1731 
1732 For deriving Functor, Foldable, Traversable with empty
1733 data cons I just generate a null equation
1734    f _ = error "urk"
1735 
1736 There are probably more lurking (eg Enum) but this will do for now.
1737]
1738[Improve the Specialiser, fixing Trac #4203
1739simonpj@microsoft.com**20100812131133
1740 Ignore-this: 482afbf75165e24a80527a6e52080c07
1741 
1742 Simply fixing #4203 is a tiny fix: in case alterantives we should
1743 do dumpUDs *including* the case binder. 
1744 
1745 But I realised that we can do better and wasted far too much time
1746 implementing the idea.  It's described in
1747     Note [Floating dictionaries out of cases]
1748]
1749[Comments
1750simonpj@microsoft.com**20100812101456
1751 Ignore-this: 6362fe887d25688c12ef2c3cf5554ce4
1752]
1753[Comments only
1754simonpj@microsoft.com**20100812101439
1755 Ignore-this: 7ed2f5fc08811cbe9958c2309a9ed1fa
1756]
1757[Fix bug in linting of shadowed case-alternative binders
1758simonpj@microsoft.com**20100812101413
1759 Ignore-this: 9212a5e2c03421749f5935b3944ecf53
1760]
1761[Comments and spacing only
1762simonpj@microsoft.com**20100812101347
1763 Ignore-this: ed59a7dae7decb24470709dc1c118dbb
1764]
1765[Add more info to more parse error messages (#3811)
1766Ian Lynagh <igloo@earth.li>**20100809233108]
1767[Run finalizers *after* updating the stable pointer table (#4221)
1768Simon Marlow <marlowsd@gmail.com>**20100810133739
1769 Ignore-this: b0462f80dd64eac71e599d8a9f6dd665
1770 Silly bug really, we were running the C finalizers while the StablePtr
1771 table was still in a partially-updated state during GC, but finalizers
1772 are allowed to call freeStablePtr() (via hs_free_fun_ptr(), for
1773 example), and chaos ensues.
1774]
1775[Do the dependency-omitting for 'make 1' in a slightly different way
1776Simon Marlow <marlowsd@gmail.com>**20100810093446
1777 Ignore-this: af15edd3a1492cbd93111316b57e02e4
1778 
1779 I encountered a couple of things that broke after Ian's previous
1780 patch: one was my nightly build scripts that use 'make stage=2' at the
1781 top level, and the other is 'make fast' in libraries/base, which uses
1782 'stage=0' to avoid building any compilers.
1783 
1784 So my version of this patch is more direct: it just turns off the
1785 appropriate dependencies using a variable set by 'make 1', 'make 2',
1786 etc.
1787]
1788[Integrate new I/O manager, with signal support
1789Johan Tibell <johan.tibell@gmail.com>**20100724102355
1790 Ignore-this: eb092857a2a1b0ca966649caffe7ac2b
1791]
1792[Add DoAndIfThenElse support
1793Ian Lynagh <igloo@earth.li>**20100808194625]
1794[Make another parse error more informative
1795Ian Lynagh <igloo@earth.li>**20100808193340]
1796[Make a parse error say what it is failing to parse; part of #3811
1797Ian Lynagh <igloo@earth.li>**20100808155732]
1798[Send ghc progress output to stdout; fixes #3636
1799Ian Lynagh <igloo@earth.li>**20100808142542]
1800[Fix the HsColour test in the build system
1801Ian Lynagh <igloo@earth.li>**20100805155319
1802 Ignore-this: ba2752b04801a253e891b31e1914485d
1803]
1804[Fix the -lm configure test; fixes #4155
1805Ian Lynagh <igloo@earth.li>**20100805142508
1806 Ignore-this: 358b8b1074d2d22fb8d362ea6d8b80d6
1807]
1808[Don't restrict filenames in line pragmas to printable characters; fixes #4207
1809Ian Lynagh <igloo@earth.li>**20100805135011
1810 Ignore-this: e3d32312127165e40e6eaa919193d60b
1811 "printable" is ASCII-only, whereas in other locales we can get things like
1812 # 1 "<línea-de-orden>"
1813]
1814[Ensure extension flags are flattened in the Cmm phase
1815Ian Lynagh <igloo@earth.li>**20100805133614
1816 If we start with a .cmmcpp file then they don't get flattened in
1817 the CmmCpp phase, as we don't run that phase.
1818]
1819[Add "cmmcpp" as a Haskellish source suffix
1820Ian Lynagh <igloo@earth.li>**20100805132555]
1821[On amd64/OSX we don't need to be given memory in the first 31bits
1822Ian Lynagh <igloo@earth.li>**20100805120600
1823 Ignore-this: 42eb64e25ad4b66ae022884305e0297b
1824 as PIC is always on
1825]
1826[NCG: Don't worry about trying to re-freeze missing coalescences
1827benl@ouroborus.net**20100702053319
1828 Ignore-this: ea05cbee19b6c5c410db41292cbb64b0
1829]
1830[Make -rtsopts more flexible
1831Ian Lynagh <igloo@earth.li>**20100805011137
1832 The default is a new "some" state, which allows only known-safe flags
1833 that we want on by default. Currently this is only "--info".
1834]
1835[Test for (fd < 0) before trying to FD_SET it
1836Ian Lynagh <igloo@earth.li>**20100804173636]
1837[Remove "On by default" comments in DynFlags
1838Ian Lynagh <igloo@earth.li>**20100802110803
1839 Ignore-this: 2a51055277b5ce9f0e98e1438b212027
1840 These make less sense now we support multiple languges. The
1841 "languageExtensions" function gives the defaults.
1842]
1843[Fix build: Add newline to end of ghc-pkg/Main.hs
1844Ian Lynagh <igloo@earth.li>**20100801183206]
1845[Add a versions haddock binary for Windows
1846Ian Lynagh <igloo@earth.li>**20100801180917]
1847[ghc-pkg: don't fail, if a file is already removed
1848ich@christoph-bauer.net**20100725162606
1849 Ignore-this: 5501d6812c31f4da525c7fb24f6dcaed
1850]
1851[Remove push-all from file list in boot script (push-all no longer exists)
1852Ian Lynagh <igloo@earth.li>**20100801121841
1853 Ignore-this: eec130f06610d8728a57626682860a1a
1854]
1855[Add error checking to boot-pkgs script
1856Ian Lynagh <igloo@earth.li>**20100801121432
1857 Ignore-this: 8afd6663db443c774bad45d75bbfe950
1858]
1859[Add more error checking to the boot script
1860Ian Lynagh <igloo@earth.li>**20100801113628]
1861[Remove libHSrtsmain.a before creating it
1862Ian Lynagh <igloo@earth.li>**20100801005432
1863 Otherwise it isn't updated properly if rts/Main.c changes
1864]
1865[Expose the functions haddock needs even when haddock is disabled; #3558
1866Ian Lynagh <igloo@earth.li>**20100731115506]
1867[Always haddock by default
1868Ian Lynagh <igloo@earth.li>**20100730235001
1869 Revert this patch:
1870     Matthias Kilian <kili@outback.escape.de>**20090920181319
1871     Don't build haddock if HADDOC_DOCS = NO, and disable HADDOC_DOCS
1872         if GhcWithInterpreter = NO
1873     Haddock uses TcRnDriver.tcRnGetInfo, which is only available if
1874     GHCI is built. Set HADDOC_DOCS to NO if GhcWithInterpreter is NO,
1875     and disable the haddock build if HADDOC_DOCS = NO.
1876]
1877[Add a debugTrace for the phases that we run
1878Ian Lynagh <igloo@earth.li>**20100729201503]
1879[* Add StringPrimL as a constructor for Template Haskell (Trac #4168)
1880simonpj@microsoft.com**20100730131922
1881 Ignore-this: 520d0a0a14b499b299e8b2be8d148ff0
1882   
1883 There are already constructors for IntPrim, FloatPrim etc,
1884 so this makes it more uniform.
1885   
1886 There's a corresponding patch for the TH library
1887]
1888[Add thread affinity support for FreeBSD
1889Gabor Pali <pgj@FreeBSD.org>**20100720001409
1890 Ignore-this: 6c117b8219bfb45445089e82ee470410
1891 - Implement missing functions for setting thread affinity and getting real
1892   number of processors.
1893 - It is available starting from 7.1-RELEASE, which includes a native support
1894   for managing CPU sets.
1895 - Add __BSD_VISIBLE, since it is required for certain types to be visible in
1896   addition to POSIX & C99.
1897]
1898[Disable symbol visibility pragmas for FreeBSD
1899Ian Lynagh <igloo@earth.li>**20100729012507
1900 Do not use GCC pragmas for controlling visibility, because it causes
1901 "undefined reference" errors at link-time.  The true reasons are
1902 unknown, however FreeBSD 8.x includes GCC 4.2.1 in the base system,
1903 which might be buggy.
1904]
1905[Fix numeric escape sequences parsing
1906Anton Nikishaev <anton.nik@gmail.com>**20100721194208
1907 Ignore-this: dd71935b1866b5624f7975c45ad519a1
1908 This fixes trac bug #1344
1909]
1910[Explicitly give the right path to perl when making the OS X installer; #4183
1911Ian Lynagh <igloo@earth.li>**20100728163030]
1912[Set -fno-stack-protector in extra-gcc-opts; fixes #4206
1913Ian Lynagh <igloo@earth.li>**20100728161957
1914 We were using it only when building the RTS, and only on certain
1915 platforms. However, some versions of OS X need the flag, while others
1916 don't support it, so we now test for it properly.
1917]
1918[Make PersistentLinkerState fields strict; fixes #4208
1919Ian Lynagh <igloo@earth.li>**20100727201911
1920 Ignore-this: fc5cfba48cd16624f6bb15a7a03a3b4
1921 We modify fields a lot, so we retain the old value if they aren't forced.
1922]
1923[Don't rebuild dependency files unnecessarily when doing "make 1" etc
1924Ian Lynagh <igloo@earth.li>**20100726211512
1925 Ignore-this: d91a729e5113aa964cc67768e92e57ef
1926]
1927[LLVM: If user specifies optlo, don't use '-O' levels
1928David Terei <davidterei@gmail.com>**20100726105650
1929 Ignore-this: e05e103b09d1de937540ffad7983f88e
1930]
1931[Flatten flags for ghci's :show
1932Ian Lynagh <igloo@earth.li>**20100725135320]
1933[Add support for Haskell98 and Haskell2010 "languages"
1934Ian Lynagh <igloo@earth.li>**20100724230121]
1935[Rename "language" varibles etc to "extension", and add --supported-extensions
1936Ian Lynagh <igloo@earth.li>**20100724223624]
1937[Separate language option handling into 2 phases
1938Ian Lynagh <igloo@earth.li>**20100724212013
1939 We now first collect the option instructions (from the commandline,
1940 from pragmas in source files, etc), and then later flatten them into
1941 the list of enabled options. This will enable us to use different
1942 standards (H98, H2010, etc) as a base upon which to apply the
1943 instructions, when we don't know what the base will be when we start
1944 collecting instructions.
1945]
1946[Separate the language flags from the other DynFlag's
1947Ian Lynagh <igloo@earth.li>**20100724133103
1948 Ignore-this: 47bb8d42e621e47016b66c7472bd6cb5
1949]
1950[Set stage-specific CC/LD opts in the bindist configure.ac
1951Ian Lynagh <igloo@earth.li>**20100724113748
1952 Ignore-this: f06926d185a35ddd05490ca4a257e992
1953]
1954[Use different CC/LD options for different stages
1955Ian Lynagh <igloo@earth.li>**20100723223059]
1956[Add some error belchs to the linker, when we find bad magic numbers
1957Ian Lynagh <igloo@earth.li>**20100723200822]
1958[Add some more linker debugging prints
1959Ian Lynagh <igloo@earth.li>**20100723180237]
1960[When (un)loading an object fails, say which object in teh panic
1961Ian Lynagh <igloo@earth.li>**20100723162649]
1962[Add a release note: GHCi import syntax
1963Ian Lynagh <igloo@earth.li>**20100721193647]
1964[Deprecate NewQualifiedOperators extension (rejected by H')
1965Ian Lynagh <igloo@earth.li>**20100719150909
1966 Ignore-this: 6e7e3bedc5360c5975f73497b3e6cba5
1967]
1968[LLVM: Allow optlc and optlo to override default params for these systools
1969David Terei <davidterei@gmail.com>**20100722181631
1970 Ignore-this: e60af7941996f7170fb3bfb02a002082
1971]
1972[LLVM: Code and speed improvement to dominateAllocs pass.
1973David Terei <davidterei@gmail.com>**20100721143654
1974 Ignore-this: 9fb7058c8a2afc005521298c7b8d0036
1975]
1976[Comments only
1977simonpj@microsoft.com**20100721144257
1978 Ignore-this: b3091ddcd1df271eb85fe90978ab7adc
1979]
1980[Fix inlining for default methods
1981simonpj@microsoft.com**20100721144248
1982 Ignore-this: 61a11a8f741f775000c6318aae4b3191
1983 
1984 This was discombobulated by a patch a week ago;
1985 now fixed, quite straightforwardly.  See
1986 Note [Default methods and instances]
1987]
1988[Allow reification of existentials and GADTs
1989simonpj@microsoft.com**20100721090437
1990 Ignore-this: 20f1ccd336cc25aff4d4d67a9ac2211a
1991 
1992 It turns out that TH.Syntax is rich enough to express even GADTs,
1993 provided we express them in equality-predicate form.  So for
1994 example
1995 
1996   data T a where
1997      MkT1 :: a -> T [a]
1998      MkT2 :: T Int
1999 
2000 will appear in TH syntax like this
2001 
2002   data T a = forall b. (a ~ [b]) => MkT1 b
2003            | (a ~ Int) => MkT2
2004 
2005 While I was at it I also improved the reification of types,
2006 so that we use TH.TupleT and TH.ListT when we can.
2007]
2008[add numSparks# primop (#4167)
2009Simon Marlow <marlowsd@gmail.com>**20100720153746
2010 Ignore-this: f3f925e7de28f3f895213aefbdbe0b0f
2011]
2012[LLVM: Decrease max opt level used under OSX to avoid bug
2013David Terei <davidterei@gmail.com>**20100720160938
2014 Ignore-this: 34b0b3550f00b27b00ad92f8232745e5
2015 
2016 Currently, many programs compiled with GHC at -O2 and LLVM
2017 set to -O3 will segfault (only under OSX). Until this issue
2018 is fixed I have simply 'solved' the segfault by lowering
2019 the max opt level for LLVM used to -O2 under OSX.
2020 
2021 All these recent changes to OSX should mean its finally as
2022 stable as Linux and Windows.
2023]
2024[LLVM: Fix OSX to work again with TNTC disabled.
2025David Terei <davidterei@gmail.com>**20100720160845
2026 Ignore-this: 8dc98139cfa536b2a64aa364d040b581
2027]
2028[LLVM: Fix printing of local vars so LLVM works with -fnew-codegen
2029David Terei <davidterei@gmail.com>**20100720160302
2030 Ignore-this: d883c433dfaed67921a8c5360e1f9f6a
2031]
2032[Use a separate mutex to protect all_tasks, avoiding a lock-order-reversal
2033Simon Marlow <marlowsd@gmail.com>**20100716150832
2034 Ignore-this: ffbdb4ee502e0f724d57acb9bfbe9d92
2035 In GHC 6.12.x I found a rare deadlock caused by this
2036 lock-order-reversal:
2037 
2038 AQ cap->lock
2039   startWorkerTask
2040     newTask
2041       AQ sched_mutex
2042 
2043 scheduleCheckBlackHoles
2044   AQ sched_mutex
2045    unblockOne_
2046     wakeupThreadOnCapabilty
2047       AQ cap->lock
2048 
2049 so sched_mutex and cap->lock are taken in a different order in two
2050 places.
2051 
2052 This doesn't happen in the HEAD because we don't have
2053 scheduleCheckBlackHoles, but I thought it would be prudent to make
2054 this less likely to happen in the future by using a different mutex in
2055 newTask.  We can clearly see that the all_tasks mutex cannot be
2056 involved in a deadlock, becasue we never call anything else while
2057 holding it.
2058]
2059['make fast' in a package does not build any compilers
2060Simon Marlow <marlowsd@gmail.com>**20100715125904
2061 Ignore-this: f27e70faf3944831dad16e89a4e273da
2062]
2063[LLVM: Fix up botched last commit
2064David Terei <davidterei@gmail.com>**20100719104823
2065 Ignore-this: a32e0f6a38cb9e02527eb8ca69b3eb59
2066]
2067[LLVM: Fix warning introduce in last commit.
2068David Terei <davidterei@gmail.com>**20100719103411
2069 Ignore-this: e9c92a9402aff50d60ab26e6ad441bfc
2070]
2071[LLVM: Use mangler to fix up stack alignment issues on OSX
2072David Terei <davidterei@gmail.com>**20100718231000
2073 Ignore-this: 9f6e8cb855269cb3a5ac1a23480d0e71
2074]
2075[Fix #4195 (isGadtSyntaxTyCon returns opposite result)
2076illissius@gmail.com**20100715134134
2077 Ignore-this: a90403f893030432b5c15d743647f350
2078]
2079[Update to time 1.2.0.3
2080Ian Lynagh <igloo@earth.li>**20100717181810
2081 Ignore-this: 1ccb4801a73f399e6718ce556543ede1
2082]
2083[Reorder RTS --info output
2084Ian Lynagh <igloo@earth.li>**20100717162356]
2085[Fix unreg prof build: Define CCS_SYSTEM in stg/MiscClosures.h
2086Ian Lynagh <igloo@earth.li>**20100717142832
2087 Ignore-this: 9675f3f51b6dac40483155344e7f45b6
2088]
2089[Make mkDerivedConstants as a stage 1 program
2090Ian Lynagh <igloo@earth.li>**20100717000827
2091 Ignore-this: 5357403461b209b8606f1d33defb51cf
2092 This way it gets the defines for the right platform when cross-compiling
2093]
2094[Don't generate Haskell dependencies if we don't have any Haskell sources
2095Ian Lynagh <igloo@earth.li>**20100717000800
2096 Ignore-this: 454abd0358f535b7e789327125c9206c
2097]
2098[Link programs that have no Haskell objects with gcc rather than ghc
2099Ian Lynagh <igloo@earth.li>**20100716235303
2100 Ignore-this: f65588b69675edea616cc434e769b0a4
2101]
2102[Use gcc to build C programs for stages >= 1
2103Ian Lynagh <igloo@earth.li>**20100716223703
2104 Ignore-this: 9f843a4e17285cda582117504707f9e7
2105]
2106[Add platform info to "ghc --info" output
2107Ian Lynagh <igloo@earth.li>**20100716141953]
2108[Tidy up Config.hs generation
2109Ian Lynagh <igloo@earth.li>**20100716140630]
2110[Fix HC porting test in makefiles
2111Ian Lynagh <igloo@earth.li>**20100716010808
2112 Ignore-this: 6052c1dd022a6108ab2236a299ee1d84
2113 Now that we are trying to support cross compilation, we can't use
2114     "$(TARGETPLATFORM)" != "$(HOSTPLATFORM)"
2115 as a test for HC-porting.
2116]
2117[Change a BUILD var to a HOST var
2118Ian Lynagh <igloo@earth.li>**20100716002558]
2119[Remove an unnecessary #include
2120Ian Lynagh <igloo@earth.li>**20100715233930
2121 Ignore-this: dcede249de6be7e3c9305c9279c2ca07
2122]
2123[Split up some make commands, so that errors aren't overlooked
2124Ian Lynagh <igloo@earth.li>**20100715152237
2125 Ignore-this: fb69b0a25d9ca71dae5e75d38db675cd
2126 When we ask make to run "a | b", if a fails then the pipeline might
2127 still exit successfuly.
2128]
2129[Remove an unnecessary #include
2130Ian Lynagh <igloo@earth.li>**20100715143000
2131 Ignore-this: 4e098cac5dda2dd595ca0a0f5121853c
2132]
2133[Simplify some more CPP __GLASGOW_HASKELL__ tests
2134Ian Lynagh <igloo@earth.li>**20100715142500]
2135[Remove some code only used with GHC 6.11.*
2136Ian Lynagh <igloo@earth.li>**20100715141720]
2137[__GLASGOW_HASKELL__ >= 609 is now always true
2138Ian Lynagh <igloo@earth.li>**20100715141544]
2139[Correct the values in ghc_boot_platform.h
2140Ian Lynagh <igloo@earth.li>**20100714223717
2141 Ignore-this: 4c99116f7ac73fadbd6d16807f57a693
2142]
2143[Change some TARGET checks to HOST checks
2144Ian Lynagh <igloo@earth.li>**20100714184715]
2145[LLVM: Add inline assembly to binding.
2146David Terei <davidterei@gmail.com>**20100714152530
2147 Ignore-this: 72a7b5460c128ed511e8901e5889fe2b
2148]
2149[LLVM: Fix mistype in last commit which broke TNTC under win/linux.
2150David Terei <davidterei@gmail.com>**20100714153339
2151 Ignore-this: 302d7957e3dded80368ebade5312ab35
2152]
2153[Remove unnecessary #include
2154Ian Lynagh <igloo@earth.li>**20100713153704
2155 Ignore-this: c37d3127b1dc68f59270c07173994c28
2156]
2157[Change some TARGET tests to HOST tests in the RTS
2158Ian Lynagh <igloo@earth.li>**20100713141034
2159 Which was being used seemed to be random
2160]
2161[LLVM: Add in new LLVM mangler for implementing TNTC on OSX
2162David Terei <davidterei@gmail.com>**20100713183243
2163 Ignore-this: 394fb74d7f9657d8b454bd0148d24bf7
2164]
2165[Refactor where an error message is generated
2166simonpj@microsoft.com**20100713115733
2167 Ignore-this: f94467856238586fcbbe48537141cf78
2168]
2169[Comments only
2170simonpj@microsoft.com**20100713115703
2171 Ignore-this: 5815442c4e69b9ec331b34242a596253
2172]
2173[Comments on data type families
2174simonpj@microsoft.com**20100713115640
2175 Ignore-this: 90a333bb7f7d64a49fb7dd180d893f6b
2176]
2177[Fix Trac #T4136: take care with nullary symbol constructors
2178simonpj@microsoft.com**20100707135945
2179 Ignore-this: 2a717a24fefcd593ea41c23dad351db0
2180 
2181 When a nullary constructor is a symbol eg (:=:) we need
2182 to take care.  Annoying.
2183]
2184[Fix Trac #4127 (and hence #4173)
2185simonpj@microsoft.com**20100707123125
2186 Ignore-this: 98bb6d0f7182b59f8c93596c61f9785d
2187 
2188 The change involves a little refactoring, so that the default
2189 method Ids are brought into scope earlier, before the value
2190 declarations are compiled.  (Since a value decl may contain
2191 an instance decl in a quote.)
2192 
2193 See Note [Default method Ids and Template Haskell] in
2194 TcTyClsDcls.
2195]
2196[Fix second bug in Trac #4127
2197simonpj@microsoft.com**20100701140124
2198 Ignore-this: c8d1cc27364fe9ee5a52acb1ecb5cdd9
2199 
2200 This bug concerned the awkward shadowing we do for
2201 Template Haskell declaration brackets.  Lots of
2202 comments in
2203 
2204   Note [Top-level Names in Template Haskell decl quotes]
2205]
2206[ia64: switch handling of 'foreign import wrapper' (FIW) to libffi
2207Sergei Trofimovich <slyfox@community.haskell.org>**20100709213922
2208 Ignore-this: fd07687e0089aebabf62de85d2be693
2209 
2210 I tried to build darcs-2.4.4 with ghc-6.12.3 and got coredumps when darcs is used
2211 in interactive mode. I tried test from ticket #3516 and found out FIW code is broken.
2212 Instead of fixing it I just switched to libffi. Result built successfully, passed
2213 'foreign import wrapper' test from ticket #3516 and builds working darcs.
2214]
2215[* storage manager: preserve upper address bits on 64bit machines (thanks to zygoloid)
2216Sergei Trofimovich <slyfox@community.haskell.org>**20100709115917
2217 Ignore-this: 9f1958a19992091ddc2761c389ade940
2218 
2219 Patch does not touch amd64 as it's address lengts is 48 bits at most, so amd64 is unaffected.
2220 
2221 the issue: during ia64 ghc bootstrap (both 6.10.4 and 6.12.3) I
2222 got the failure on stage2 phase:
2223     "inplace/bin/ghc-stage2"   -H32m -O -H64m -O0 -w ...
2224     ghc-stage2: internal error: evacuate: strange closure type 15
2225         (GHC version 6.12.3 for ia64_unknown_linux)
2226         Please report this as a GHC bug:  http://www.haskell.org/ghc/reportabug
2227     make[1]: *** [libraries/dph/dph-base/dist-install/build/Data/Array/Parallel/Base/Hyperstrict.o] Aborted
2228 
2229 gdb backtrace (break on 'barf'):
2230 Breakpoint 1 at 0x400000000469ec31: file rts/RtsMessages.c, line 39.
2231 (gdb) run -B/var/tmp/portage/dev-lang/ghc-6.12.3/work/ghc-6.12.3/inplace/bin --info
2232 Starting program: /var/tmp/portage/dev-lang/ghc-6.12.3/work/ghc-6.12.3/inplace/lib/ghc-stage2 -B/var/tmp/portage/dev-lang/ghc-6.12.3/work/ghc-6.12.3/inplace/bin --info
2233 [Thread debugging using libthread_db enabled]
2234 
2235 Breakpoint 1, barf (s=0x40000000047915b0 "evacuate: strange closure type %d") at rts/RtsMessages.c:39
2236 39        va_start(ap,s);
2237 (gdb) bt
2238 #0  barf (s=0x40000000047915b0 "evacuate: strange closure type %d") at rts/RtsMessages.c:39
2239 #1  0x400000000474a1e0 in evacuate (p=0x6000000000147958) at rts/sm/Evac.c:756
2240 #2  0x40000000046d68c0 in scavenge_srt (srt=0x6000000000147958, srt_bitmap=7) at rts/sm/Scav.c:348
2241 ...
2242 
2243 > 16:52:53 < zygoloid> slyfox: i'm no ghc expert but it looks like HEAP_ALLOCED_GC(q)
2244 >                      is returning true for a FUN_STATIC closure
2245 > 17:18:43 < zygoloid> try: p HEAP_ALLOCED_miss((unsigned long)(*p) >> 20, *p)
2246 > 17:19:12 < slyfox> (gdb) p HEAP_ALLOCED_miss((unsigned long)(*p) >> 20, *p)
2247 > 17:19:12 < slyfox> $1 = 0
2248 > 17:19:40 < zygoloid> i /think/ that means the mblock_cache is broken
2249 > 17:22:45 < zygoloid> i can't help further. however i am suspicious that you seem to have pointers with similar-looking low 33
2250 >                      bits and different high 4 bits, and it looks like such pointers get put into the same bucket in
2251 >                      mblock_cache.
2252 ...
2253 > 17:36:16 < zygoloid> slyfox: try changing the definition of MbcCacheLine to StgWord64, see if that helps
2254 > 17:36:31 < zygoloid> that's in includes/rts/storage/MBlock.h
2255 And it helped!
2256]
2257[Fixing link failure of compiler on ia64 ('-Wl,' prefixed value passed directly to ld)
2258Sergei Trofimovich <slyfox@community.haskell.org>**20100708180943
2259 Ignore-this: ced99785e1f870ee97e5bec658e2504f
2260 
2261     /usr/bin/ld -Wl,--relax -r -o dist-stage1/build/HSghc-6.10.4.o \
2262                                   dist-stage1/build/BasicTypes.o dist-stage1/build/DataCon.o ...
2263     /usr/bin/ld: unrecognized option '-Wl,--relax'
2264 
2265 If we just drop '-Wl,' part it will not help as '-r' and '--relax' are incompatible.
2266 
2267 Looks like '-Wl,--relax' was skipped by earlier binutils' ld as unknown option.
2268 Removing ia64 specific path.
2269]
2270[LLVM: Allow getelementptr to use LlvmVar for indexes.
2271David Terei <davidterei@gmail.com>**20100712152529
2272 Ignore-this: 9e158d9b89a86bca8abf11d082328278
2273]
2274[Move all the warning workarounds to one place
2275Ian Lynagh <igloo@earth.li>**20100710161723]
2276[xhtml is now warning-free
2277Ian Lynagh <igloo@earth.li>**20100710144635]
2278[Move a bit of build system code
2279Ian Lynagh <igloo@earth.li>**20100709224534]
2280[adapt to the new async exceptions API
2281Simon Marlow <marlowsd@gmail.com>**20100709125238
2282 Ignore-this: 55d845e40b9daed3575c1479d8dda1d5
2283]
2284[quiet some new spewage
2285Simon Marlow <marlowsd@gmail.com>**20100709091521
2286 Ignore-this: de7f91976bbc9789e6fd7091f05c25c0
2287]
2288[New asynchronous exception control API (ghc parts)
2289Simon Marlow <marlowsd@gmail.com>**20100708144851
2290 Ignore-this: 56320c5fc61ae3602d586609387aae22
2291 
2292 As discussed on the libraries/haskell-cafe mailing lists
2293   http://www.haskell.org/pipermail/libraries/2010-April/013420.html
2294 
2295 This is a replacement for block/unblock in the asychronous exceptions
2296 API to fix a problem whereby a function could unblock asynchronous
2297 exceptions even if called within a blocked context.
2298 
2299 The new terminology is "mask" rather than "block" (to avoid confusion
2300 due to overloaded meanings of the latter).
2301 
2302 In GHC, we changed the names of some primops:
2303 
2304   blockAsyncExceptions#   -> maskAsyncExceptions#
2305   unblockAsyncExceptions# -> unmaskAsyncExceptions#
2306   asyncExceptionsBlocked# -> getMaskingState#
2307 
2308 and added one new primop:
2309 
2310   maskUninterruptible#
2311 
2312 See the accompanying patch to libraries/base for the API changes.
2313]
2314[remove outdated comment
2315Simon Marlow <marlowsd@gmail.com>**20100708100840
2316 Ignore-this: afb2e9f6fe1f1acda51b0cbdf2637176
2317]
2318[remove 'mode: xml' emacs settings (#2208)
2319Simon Marlow <marlowsd@gmail.com>**20100708100817
2320 Ignore-this: 3a8d997fb90e01ca88dc47fb95feeba0
2321]
2322[typo in comment
2323Simon Marlow <marlowsd@gmail.com>**20100616111359
2324 Ignore-this: d3ef9288d6d6b9ab3bacbe09e0d9801c
2325]
2326[Win32 getProcessElapsedTime: use a higher-resolution time source
2327Simon Marlow <marlowsd@gmail.com>**20100708093223
2328 Ignore-this: 821989d4ff7ff2bff40cee71a881521c
2329 QueryPerformanceCounter() on Windows gives much better resolution than
2330 GetSystemTimeAsFileTime().
2331]
2332[alpha: switch handling of 'foreign import wrapper' (FIW) to libffi
2333Sergei Trofimovich <slyfox@community.haskell.org>**20100708065318
2334 Ignore-this: ddee15876737a6aa7f6dabc8ff79ce0d
2335 
2336 I tried to build ghc-6.12.3 and found out FIW part of code
2337 does not compile anymore. It uses absent functions under #ifdef.
2338 Instead of fixing it I just switched to libffi. Result built successfully
2339 and passed 'foreign import wrapper' test I wrote for trac ticket #3516.
2340 
2341 I didn't try to build -HEAD yet, but this patch only removes code, so
2342 it should not make -HEAD worse.
2343]
2344[Reorder the CPP flags so -optP can override the platform defines
2345Ian Lynagh <igloo@earth.li>**20100708203523]
2346[Add docs for DatatypeContexts extension
2347Ian Lynagh <igloo@earth.li>**20100707230907
2348 Ignore-this: 8158f03b35a2d7442a75fe85d6f1b1c7
2349]
2350[Make datatype contexts an extension (on by default) (DatatypeContexts)
2351Ian Lynagh <igloo@earth.li>**20100707212529
2352 Ignore-this: 6885ff510a0060610eeeba65122caef5
2353]
2354[LLVM: Fix various typos in comments
2355David Terei <davidterei@gmail.com>**20100707220448
2356 Ignore-this: 1ba3e722f150492da2f9d485c5795e80
2357]
2358[Handle haddock headers when looking for LANGUAGE/OPTIONS_GHC pragmas
2359Ian Lynagh <igloo@earth.li>**20100707120423
2360 Ignore-this: a75aa67690284a6cee3e62c943d4fd01
2361]
2362[Make pragState call mkPState, rather than duplicating everything
2363Ian Lynagh <igloo@earth.li>**20100706173007
2364 Ignore-this: 61fe24b99dbe7a42efff1a9dd703a75c
2365 This also means that extsBitmap gets set, whereas is was just being set
2366 to 0 before.
2367]
2368[LLVM: Add alias type defenitions to LlvmModule.
2369David Terei <davidterei@gmail.com>**20100707142053
2370 Ignore-this: eee6ad5385563ccf08e664d2634a03f2
2371]
2372[LLVM: Use packed structure type instead of structure type
2373David Terei <davidterei@gmail.com>**20100707120320
2374 Ignore-this: a06e0359d182291b81cae56993ca385e
2375 
2376 The regular structure type adds padding to conform to the platform ABI,
2377 which causes problems with structures storing doubles under windows since
2378 we don't conform to the platform ABI there. So we use packed structures
2379 instead now that don't do any padding.
2380]
2381[Make mkPState and pragState take their arguments in the same order
2382Ian Lynagh <igloo@earth.li>**20100706172611]
2383[Remove an out-of-date comment
2384Ian Lynagh <igloo@earth.li>**20100706172217
2385 Ignore-this: 710ebd7d2dc01c1b0f1e58a5b6f85701
2386]
2387[LLVM: Stop llvm saving stg caller-save regs across C calls
2388David Terei <davidterei@gmail.com>**20100705162629
2389 Ignore-this: 28b4877b31b9358e682e38fc54b88658
2390 
2391 This is already handled by the Cmm code generator so LLVM is simply
2392 duplicating work. LLVM also doesn't know which ones are actually live
2393 so saves them all which causes a fair performance overhead for C calls
2394 on x64. We stop llvm saving them across the call by storing undef to
2395 them just before the call.
2396]
2397[LLVM: Add in literal undefined value to binding
2398David Terei <davidterei@gmail.com>**20100705161544
2399 Ignore-this: 95d8361b11584aaeec44c30e76916470
2400]
2401[LLVM: Add a literal NULL value to binding
2402David Terei <davidterei@gmail.com>**20100705161308
2403 Ignore-this: 9507b4b12c1157498704a9d1e5860f12
2404 
2405 Patch from Erik de Castro Lopo <erikd@mega-nerd.com>.
2406]
2407[refactor import declaration support (#2362)
2408Simon Marlow <marlowsd@gmail.com>**20100705104557
2409 Ignore-this: ee034ac377078a7e92bfada1907c86a0
2410]
2411[Disable dynamic linking optimisations on OS X
2412Simon Marlow <marlowsd@gmail.com>**20100705103014
2413 Ignore-this: b04420d3705c51112797758d17b2e40c
2414 To improve performance of the RTS when dynamically linked on x86, I
2415 previously disabled -fPIC for certain critical modules (the GC, and a
2416 few others).  However, build reports suggest that the dynamic linker
2417 on OS X doesn't like this, so I'm disabling this optimsation on that
2418 platform.
2419]
2420[trac #2362 (full import syntax in ghci)
2421amsay@amsay.net**20100625032632
2422 Ignore-this: a9d0859d84956beb74e27b797431bf9c
2423 'import' syntax is seperate from ':module' syntax
2424]
2425[Simplify ghc-pkg's Cabal dependencies
2426Ian Lynagh <igloo@earth.li>**20100704184155
2427 We no longer support building with a compiler that doesn't come with
2428 base 4.
2429]
2430[Use Cabal to configure the dist-install ghc-pkg; fixes trac #4156
2431Ian Lynagh <igloo@earth.li>**20100704132612]
2432[Remove dead code (standalone deriving flag no longer used in parser)
2433Ian Lynagh <igloo@earth.li>**20100701162058]
2434[LLVM: Use the inbounds keyword for getelementptr instructions.
2435David Terei <davidterei@gmail.com>**20100702160511
2436 Ignore-this: 3708e658a4c82b78b1402393f4405541
2437]
2438[threadPaused: fix pointer arithmetic
2439Simon Marlow <marlowsd@gmail.com>**20100701085046
2440 Ignore-this: b78210e5d978f18ffd235f1c78a55a23
2441 Noticed by Henrique Ferreiro <hferreiro@udc.es>, thanks!
2442]
2443[LLVM: Change more operations to use getelementptr
2444David Terei <davidterei@gmail.com>**20100701161856
2445 Ignore-this: fb24eb124e203f50680c6fec3ff9fe7d
2446]
2447[Add the haskell2010 package
2448Simon Marlow <marlowsd@gmail.com>**20100630125532
2449 Ignore-this: e9b011313f283a8ff2fcda7d029a01f
2450]
2451[LLVM: Use getelementptr instruction for a lot of situations
2452David Terei <davidterei@gmail.com>**20100630181157
2453 Ignore-this: 34d314dd8dffad9bdcffdc525261a49d
2454 
2455 LLVM supports creating pointers in two ways, firstly through
2456 pointer arithmetic (by casting between pointers and ints)
2457 and secondly using the getelementptr instruction. The second way
2458 is preferable as it gives LLVM more information to work with.
2459 
2460 This patch changes a lot of pointer related code from the first
2461 method to the getelementptr method.
2462]
2463[remove out of date comments; point to the wiki
2464Simon Marlow <marlowsd@gmail.com>**20100625100313
2465 Ignore-this: 95f363a373534b9471b1818102ec592d
2466]
2467[NCG: allocatableRegs is only giving us 8 SSE regs to allocate to
2468benl@ouroborus.net**20100629054321
2469 Ignore-this: b3e0fa0b4ce988a0258dc12261989ee0
2470]
2471[LLVM: Use intrinsic functions for pow, sqrt, sin, cos
2472David Terei <davidterei@gmail.com>**20100628182949
2473 Ignore-this: 98a0befaca3fe2b36d710d8ff9f062c4
2474 
2475 Instead of calling the C library for these Cmm functions
2476 we use intrinsic functions provided by llvm. LLVM will
2477 then either create a compile time constant if possible, or
2478 use a cpu instruction or as a last resort call the C
2479 library.
2480]
2481[LLVM: Fix test '2047' under linux-x64
2482David Terei <davidterei@gmail.com>**20100628165256
2483 Ignore-this: 41735d4f431a430db636621650ccd71e
2484]
2485[LLVM: Fix test 'ffi005' under linux-x64
2486David Terei <davidterei@gmail.com>**20100628155355
2487 Ignore-this: 841f3142c63cc898ac4c3f89698a837e
2488]
2489[LLVM: Update to use new fp ops introduced in 2.7
2490David Terei <davidterei@gmail.com>**20100628144037
2491 Ignore-this: 5dd2e5964e3c039d297ed586841e706b
2492]
2493[Add noalias and nocapture attributes to pointer stg registers
2494David Terei <davidterei@gmail.com>**20100628115120
2495 Ignore-this: 492a1e723cb3a62498d240d7de92dd7
2496 
2497 At the moment this gives a very slight performance boost of around 1 - 2%.
2498 Future changes to the generated code though so that pointers are kept as
2499 pointers more often instead of being cast to integer types straight away
2500 should hopefully improve the benefit this brings.
2501 
2502]
2503[during shutdown, only free the heap if we waited for foreign calls to exit
2504Simon Marlow <marlowsd@gmail.com>**20100628090536
2505 Ignore-this: d545384a4f641d701455d08ef1217479
2506]
2507[Fix typo in -ddump-pass's document.
2508shelarcy <shelarcy@gmail.com>**20100620070759
2509 Ignore-this: f4f1ddb53f147949e948147d89190c37
2510]
2511[Add #undefs for posix source symbols when including papi.h
2512dmp@rice.edu**20100624163514
2513 Ignore-this: 8a1cba21b880d12a75a75f7e96882053
2514 
2515 Validation fails when validating with PAPI support (i.e. GhcRtsWithPapi  = YES
2516 in validate.mk).  The problem is that the posix symbols are defined by a header
2517 included from papi.h. Compilation then fails because these symbols are
2518 redefined in PosixSource.h.
2519 
2520 This patch adds an undefine for the posix symbols after including papi.h and
2521 before including PosixSource.h. The #undefines are localized to Papi.c since
2522 that is the only case where they are getting defined twice.
2523]
2524[Use machdepCCOpts in runPhase_MoveBinary; fixes trac #3952
2525Ian Lynagh <igloo@earth.li>**20100625220953]
2526[LLVM: Fix bug with calling tail with empty list
2527David Terei <davidterei@gmail.com>**20100625115729
2528 Ignore-this: 46b4b32c8d92372a2d49794a96fe1613
2529]
2530[Fix warnings
2531benl@ouroborus.net**20100624091339
2532 Ignore-this: 5ba4bbd6abb9c9d1fb8c5d21ab73f218
2533]
2534[NCG: Comments and formatting only
2535benl@ouroborus.net**20100624083121
2536 Ignore-this: 86002e72c30d06bcc876d8c49f4caa5a
2537]
2538[NCG: Do the actual reversing of SCCs
2539benl@ouroborus.net**20100624082717
2540 Ignore-this: 12d2027ea118e751fbb48b27126150ef
2541]
2542[NCG: Fix dumping of graphs in regalloc stats for graph allocator
2543benl@ouroborus.net**20100624082625
2544 Ignore-this: 2b971bc9e0318099a9afb0e0db135730
2545]
2546[NCG: Reverse SCCs after each round in the graph allocator
2547benl@ouroborus.net**20100624082437
2548 Ignore-this: f0152e4039d6f16f7b5a99b286538116
2549]
2550[NCG: Don't actually complain on unreachable code blocks
2551benl@ouroborus.net**20100624081445
2552 Ignore-this: e7335ae6120917cb858c38c7c6da8e24
2553]
2554[NCG: Do explicit check for precondition of computeLiveness
2555benl@ouroborus.net**20100624080747
2556 Ignore-this: e7053c4e5e4c3c746b5ebf016913424a
2557 
2558  computeLiveness requires the SCCs of blocks to be in reverse dependent
2559  order, and if they're not it was silently giving bad liveness info,
2560  yielding a bad allocation.
2561 
2562  Now it complains, loudly.
2563]
2564[NCG: Fix off-by-one error in realRegSqueeze
2565benl@ouroborus.net**20100623095813
2566 Ignore-this: ab0698686d4c250da8e207f734f8252d
2567]
2568[NCG: Handle stripping of liveness info from procs with no blocks (like stg_split_marker)
2569benl@ouroborus.net**20100623091209
2570 Ignore-this: c0319b6cc62ec713afe4eb03790406e3
2571]
2572[NCG: Emit a warning on unreachable code block instead of panicing
2573benl@ouroborus.net**20100623085002
2574 Ignore-this: d20314b79e3c31e764ed4cd97290c696
2575]
2576[NCG: Remember to keep the entry block first when erasing liveness info
2577Ben.Lippmeier@anu.edu.au**20090917104429
2578 Ignore-this: 1b0c1df19d622858d50ffb6a01f2cef0
2579]
2580[NCG: Refactor representation of code with liveness info
2581Ben.Lippmeier@anu.edu.au**20090917090730
2582 Ignore-this: 2aebb3b02ebd92e547c5abad9feb0f0d
2583 
2584  * I've pushed the SPILL and RELOAD instrs down into the
2585    LiveInstr type to make them easier to work with.
2586 
2587  * When the graph allocator does a spill cycle it now just
2588    re-annotates the LiveCmmTops instead of converting them
2589    to NatCmmTops and back.
2590 
2591  * This saves working out the SCCS again, and avoids rewriting
2592    the SPILL and RELOAD meta instructions into real machine
2593    instructions.
2594]
2595[NCG: Add sanity checking to linear allocator
2596Ben.Lippmeier@anu.edu.au**20090917090335
2597 Ignore-this: 5a442be8b5087d04bc8b58dffa9ea080
2598 If there are are unreachable basic blocks in the native code then the
2599 linear allocator might loop. Detect this case and panic instead.
2600]
2601[NCG: Refactor LiveCmmTop to hold a list of SCCs instead of abusing ListGraph
2602Ben.Lippmeier@anu.edu.au**20090917060332
2603 Ignore-this: 3fec8d69ed0f760e53a202f873d5d9cb
2604]
2605[NCG: Allow the liveness map in a LiveInfo to be Nothing
2606Ben.Lippmeier@anu.edu.au**20090917043937
2607 Ignore-this: 5f82422d54d1b0ffc0589eb7e82fb7a4
2608]
2609[NCG: Also show the result of applying coalesings with -ddump-asm-regalloc-stages
2610Ben.Lippmeier.anu.edu.au**20090917034427
2611 Ignore-this: 76bd6d5ca43adb2167cb25832cbaa80b
2612]
2613[Fix panic when running "ghc -H"; trac #3364
2614Ian Lynagh <igloo@earth.li>**20100624234011
2615 The problem is that showing SDoc's looks at the static flags global
2616 variables, but those are panics while we are parsing the static flags.
2617 We work around this by explicitly using a fixed prettyprinter style.
2618]
2619[Allow for stg registers to have pointer type in llvm BE.
2620David Terei <davidterei@gmail.com>**20100621175839
2621 Ignore-this: fc09b1a8314aef0bde945c77af1124fb
2622 
2623 Before all the stg registers were simply a bit type or
2624 floating point type but now they can be declared to have
2625 a pointer type to one of these. This will allow various
2626 optimisations in the future in llvm since the type is
2627 more accurate.
2628]
2629[Add support for parameter attributes to the llvm BE binding
2630David Terei <davidterei@gmail.com>**20100624111744
2631 Ignore-this: 77f3c0c7bf8f81c4a154dc835ae7bcba
2632 
2633 These allow annotations of the code produced by the backend
2634 which should bring some perforamnce gains. At the moment
2635 the attributes aren't being used though.
2636]
2637[Cast some more nats to StgWord to be on the safe side
2638Simon Marlow <marlowsd@gmail.com>**20100624105700
2639 Ignore-this: e6176683856f9872fdeb2358bb065bb8
2640 And add a comment about the dangers of int overflow
2641]
2642[comments only
2643Simon Marlow <marlowsd@gmail.com>**20100624105105
2644 Ignore-this: fc8f762f4c3a5ffca2f8da2bc63ac2a4
2645]
2646[Fix an arithmetic overflow bug causing crashes with multi-GB heaps
2647Simon Marlow <marlowsd@gmail.com>**20100624104654
2648 Ignore-this: 67210755aa098740ff5230347be0fd5d
2649]
2650[Add support for collecting PAPI native events
2651dmp@rice.edu**20100622195953
2652 Ignore-this: 7269f9c4dfb2912a024eb632200fcd1
2653 
2654 This patch extends the PAPI support in the RTS to allow collection of native
2655 events. PAPI can collect data for native events that are exposed by the
2656 hardware beyond the PAPI present events. The native events supported on your
2657 hardware can found by using the papi_native_avail tool.
2658 
2659 The RTS already allows users to specify PAPI preset events from the command
2660 line. This patch extends that support to allow users to specify native events.
2661 The changes needed are:
2662 
2663 1) New option (#) for the RTS PAPI flag for native events. For example, to
2664    collect the native event 0x40000000, use ./a.out +RTS -a#0x40000000 -sstderr
2665 
2666 2) Update the PAPI_FLAGS struct to store whether the user specified event is a
2667    papi preset or a native event
2668 
2669 3) Update init_countable_events function to add the native events after parsing
2670    the event code and decoding the name using PAPI_event_code_to_name
2671 
2672]
2673[Don't warn about unused bindings with parents in .hs-boot files; trac #3449
2674Ian Lynagh <igloo@earth.li>**20100624110351]
2675[fix the home_imps filter to allow for 'import "this" <module>'
2676Simon Marlow <marlowsd@gmail.com>**20100621125535
2677 Ignore-this: da4e605b0513afc32a4e7caa921a2c76
2678 In the PackageImports extension, import "this" means "import from the
2679 current package".
2680]
2681[Use the standard C wrapper code for the ghc-$version.exe wrapper
2682Ian Lynagh <igloo@earth.li>**20100622202859
2683 Ignore-this: 60cd3e6db3afb63e6ba9e2db3b033580
2684]
2685[Don't rely on "-packagefoo" working; use "-package foo" instead
2686Ian Lynagh <igloo@earth.li>**20100622202547]
2687[Remove unnecessary C #includes
2688Ian Lynagh <igloo@earth.li>**20100622172919]
2689[Make the ghci.exe wrapper call the right ghc.exe
2690Ian Lynagh <igloo@earth.li>**20100622172247]
2691[More updates to datalayout description in llvm BE
2692David Terei <davidterei@gmail.com>**20100622165339
2693 Ignore-this: b0c604fe7673b0aa7c7064694d574437
2694]
2695[Remove LlvmAs phase as the llvm opt tool now handles this phase
2696David Terei <davidterei@gmail.com>**20100622144044
2697 Ignore-this: b9fd8f959702b6af014e2fa654bede3
2698 
2699 This phase originally invoked the llvm-as tool that turns a textual
2700 llvm assembly file into a bit code file for the rest of llvm to deal
2701 with. Now the llvm opt tool can do this itself, so we don't need to
2702 use llvm-as anymore.
2703]
2704[Update datalayout info in llvm BE
2705David Terei <davidterei@gmail.com>**20100622123457
2706 Ignore-this: 89b043d211225dcd819f30549afe1840
2707]
2708[Fix handling of float literals in llvm BE
2709David Terei <davidterei@gmail.com>**20100622121642
2710 Ignore-this: a3b5f382ad4b5a426ad4b581664506fa
2711]
2712[Declare some top level globals to be constant when appropriate
2713David Terei <davidterei@gmail.com>**20100621174954
2714 Ignore-this: 44832f65550d4f995d11c01cc1affef5
2715 
2716 This involved removing the old constant handling mechanism
2717 which was fairly hard to use. Now being constant or not is
2718 simply a property of a global variable instead of a separate
2719 type.
2720]
2721[Reduce the number of passes over the cmm in llvm BE
2722David Terei <davidterei@gmail.com>**20100621125220
2723 Ignore-this: cb2f4e54e8d0f982d5087fbeee35c73c
2724]
2725[Fix negate op not working for -0 in llvm backend
2726David Terei <davidterei@gmail.com>**20100621123606
2727 Ignore-this: c5d76e5cffa781fed074137851b1347f
2728]
2729[ROLLBACK: picCCOpts: -dynamic should not entail -optc-fPIC
2730Simon Marlow <marlowsd@gmail.com>**20100621100409
2731 Ignore-this: f2fac7df33d3919199befc59bd455414
2732 and add a comment to explain why it was wrong.  This fixes the dyn
2733 test failures that sprang up recently.
2734]
2735[Check files are really created in libffi
2736Ian Lynagh <igloo@earth.li>**20100620163724
2737 when we think that the libffi build creates them, so they just depend
2738 on the libffi build stamp.
2739]
2740[Improve the missing-import-list warning
2741Ian Lynagh <igloo@earth.li>**20100620124320
2742 Ignore-this: 551e5fdf2dfb56b49d249e0cebaa6115
2743]
2744[Tweak missing-import-list warning
2745Ian Lynagh <igloo@earth.li>**20100620122622
2746 Ignore-this: 360cdf59ae13d66ded181129325506c4
2747]
2748[trac #1789 (warnings for missing import lists)
2749amsay@amsay.net**20100618150649
2750 Ignore-this: b0b0b1e048fbca0817c1e6fade1153fa
2751]
2752[Refix docs for sizeofByteArray#/sizeofMutableByteArray# (#3800)
2753Ian Lynagh <igloo@earth.li>**20100620103749]
2754[Remove some old commented out code
2755Ian Lynagh <igloo@earth.li>**20100620000459]
2756[SET_ARR_HDR's last argument is now a number of bytes, rather than words
2757Ian Lynagh <igloo@earth.li>**20100619235214
2758 This avoids unnecessary work and potential loss of information
2759]
2760[Replace an (incorrect) bytes-to-words calculation with ROUNDUP_BYTES_TO_WDS
2761Ian Lynagh <igloo@earth.li>**20100619234310]
2762[FIX #38000 Store StgArrWords payload size in bytes
2763Antoine Latter <aslatter@gmail.com>**20100101183346
2764 Ignore-this: 7bf3ab4fc080c46311fc10b179361bb6
2765]
2766[Add win32 datalayout support to llvm backend
2767David Terei <davidterei@gmail.com>**20100618131733
2768 Ignore-this: 4b7bffaa8ef38c628ab852c1a6c1c009
2769]
2770[Remove unused 'ddump-opt-llvm' flag
2771David Terei <davidterei@gmail.com>**20100618101237
2772 Ignore-this: f78467496d986897e49d82646ee2907e
2773]
2774[generate "movl lbl(%reg1), %reg2" instructions, better codegen for -fPIC
2775Simon Marlow <marlowsd@gmail.com>**20100618082258
2776 Ignore-this: a25567ebff9f575303ddc8f2deafebbf
2777]
2778[joinToTargets: fix a case of panic "handleComponent cyclic"
2779Simon Marlow <marlowsd@gmail.com>**20100618082147
2780 Ignore-this: 765baeefbb5a41724004acd92405cecc
2781]
2782[comment typo
2783Simon Marlow <marlowsd@gmail.com>**20100618082102
2784 Ignore-this: e495610b7dd5ec30b02938638b56cb7
2785]
2786[Add support of TNTC to llvm backend
2787David Terei <davidterei@gmail.com>**20100618093205
2788 Ignore-this: 2c27d21668374a5b0d5e844882c69439
2789 
2790 We do this through a gnu as feature called subsections,
2791 where you can put data/code into a numbered subsection
2792 and those subsections will be joined together in descending
2793 order by gas at compile time.
2794]
2795[Don't automatically insert a -fvia-C flag in an unregisterised compiler
2796Ian Lynagh <igloo@earth.li>**20100617190901
2797 Ignore-this: eb25a9a338fade9e17c153da7c5f27e9
2798 The default object mode is already HscC, so it's unnecessary, and
2799 -fvia-C generates a deprecated flag warning now.
2800]
2801[In PosixSource.h, conditionally define things based on platform
2802Ian Lynagh <igloo@earth.li>**20100617174912
2803 This may not be ideal, but it should get GHC building on all platforms
2804 again.
2805]
2806[disable -fPIC for the GC for performance reasons
2807Simon Marlow <marlowsd@gmail.com>**20100617140025
2808 Ignore-this: c7c152bbff71ef7891eaee8ff39fc281
2809 see comment for details
2810]
2811[picCCOpts: -dynamic should not entail -optc-fPIC
2812Simon Marlow <marlowsd@gmail.com>**20100617115259
2813 Ignore-this: d71e42bd56e4bd107d2c431b801855e5
2814]
2815[Make getAllocations() visible
2816Simon Marlow <marlowsd@gmail.com>**20100617113259
2817 Ignore-this: 1b7fb38a01358c0acbe8987df07d23f2
2818]
2819[Fix the symbol visibility pragmas
2820Simon Marlow <marlowsd@gmail.com>**20100617105758
2821 Ignore-this: 76552500865473a1dbebbc1cc2def9f0
2822]
2823[pick up changes to $(GhcStage1HcOpts) without re-configuring the ghc package
2824Simon Marlow <marlowsd@gmail.com>**20100616124718
2825 Ignore-this: afb56d5560c813051285607fefb15493
2826]
2827[Fix bindisttest Makefile
2828Ian Lynagh <igloo@earth.li>**20100616205611
2829 Ignore-this: 39cd352152422f378572fc3859c5a377
2830]
2831[Remove some more unused make variables
2832Ian Lynagh <igloo@earth.li>**20100616180519]
2833[Convert some more variable names to FOO_CMD, for consistency
2834Ian Lynagh <igloo@earth.li>**20100616175916]
2835[Rename some variables from FOO to FOO_CMD
2836Ian Lynagh <igloo@earth.li>**20100616161108
2837 This fixes a problem with commands like gzip, where if $GZIP is exported
2838 in the environment, then when make runs a command it'll put the Makefile
2839 variable's value in the environment. But gzip treats $GZIP as arguments
2840 for itself, so when we run gzip it thinks we're giving it "gzip" as an
2841 argument.
2842]
2843[Make the "show" target work anywhere in the build tree
2844Ian Lynagh <igloo@earth.li>**20100616122910
2845 Ignore-this: 299d40cbe16112accd9f14e56fa12158
2846]
2847[Change ghc-pwd's license to a string Cabal recognises
2848Ian Lynagh <igloo@earth.li>**20100615204015
2849 Ignore-this: c935b6ad7f605aab0168997a90b40fc6
2850]
2851[fix warning
2852Simon Marlow <marlowsd@gmail.com>**20100604205933
2853 Ignore-this: 2aaa4ed6a8b9ae1e39adc4696aaf14a3
2854]
2855[--install-signal-handles=no does not affect the timer signal (#1908)
2856Simon Marlow <marlowsd@gmail.com>**20100527214627
2857 Ignore-this: b0c51f1abdb159dc360662485095a11a
2858]
2859[Small optimisation: allocate nursery blocks contiguously
2860Simon Marlow <marlowsd@gmail.com>**20100509194928
2861 Ignore-this: e650e99e9ea9493d2efb245d565beef4
2862 This lets automatic prefetching work better, for a tiny performance boost
2863]
2864[fix -fforce-recomp setting: module is PrimOp, not PrimOps
2865Simon Marlow <marlowsd@gmail.com>**20100507084507
2866 Ignore-this: f76e0d9b643682ec0e8fb7d91afdea68
2867]
2868[it should be an error to use relative directories (#4134)
2869Simon Marlow <marlowsd@gmail.com>**20100615151740
2870 Ignore-this: 2068021701832e018ca41b22877921d5
2871]
2872[missing include-dirs or library-dirs is only a warning now (#4104)
2873Simon Marlow <marlowsd@gmail.com>**20100615151702
2874 Ignore-this: e3114123cef147bbd28ccb64581a1afb
2875]
2876[fix #3822: desugaring case command in arrow notation
2877Ross Paterson <ross@soi.city.ac.uk>**20100615225110
2878 Ignore-this: 477d6c460b4174b94b4cd113fa5b9d19
2879 
2880 Get the set of free variables from the generated case expression:
2881 includes variables in the guards and decls that were missed before,
2882 and is also a bit simpler.
2883]
2884[Deprecate the -fvia-C flag; trac #3232
2885Ian Lynagh <igloo@earth.li>**20100615151836
2886 Ignore-this: c2452b2648bf7e44546465c1b964fce
2887]
2888[Avoid using the new ~~ perl operator in the mangler
2889Ian Lynagh <igloo@earth.li>**20100615151236
2890 Ignore-this: 709a7ba4e514b1596841b3ba7e5c6cc
2891]
2892[stmAddInvariantToCheck: add missing init of invariant->lock (#4057)
2893Simon Marlow <marlowsd@gmail.com>**20100615123643
2894 Ignore-this: 3b132547fa934cecf71a846db2a5f70e
2895]
2896[Add new LLVM code generator to GHC. (Version 2)
2897David Terei <davidterei@gmail.com>**20100615094714
2898 Ignore-this: 4dd2fe5854b64a3f0339d484fd5c238
2899 
2900 This was done as part of an honours thesis at UNSW, the paper describing the
2901 work and results can be found at:
2902 
2903 http://www.cse.unsw.edu.au/~pls/thesis/davidt-thesis.pdf
2904 
2905 A Homepage for the backend can be found at:
2906 
2907 http://hackage.haskell.org/trac/ghc/wiki/Commentary/Compiler/Backends/LLVM
2908 
2909 Quick summary of performance is that for the 'nofib' benchmark suite, runtimes
2910 are within 5% slower than the NCG and generally better than the C code
2911 generator.  For some code though, such as the DPH projects benchmark, the LLVM
2912 code generator outperforms the NCG and C code generator by about a 25%
2913 reduction in run times.
2914 
2915]
2916[Fix Trac #4127: build GlobalRdrEnv in GHCi correctly
2917simonpj@microsoft.com**20100615070626
2918 Ignore-this: d907e3bfa7882878cea0af172aaf6e84
2919 
2920 GHCi was building its GlobalRdrEnv wrongly, so that the
2921 gre_par field was bogus.  That in turn fooled the renamer.
2922 The fix is easy: use the right function!  Namely, call
2923 RnNames.gresFromAvail rather than availsToNameSet.
2924]
2925[Comments, and improvement to pretty-printing of HsGroup
2926simonpj@microsoft.com**20100615070409
2927 Ignore-this: ec8358f2485370b20226a97ec84e9024
2928]
2929[Don't reverse bindings in rnMethodBinds (fix Trac #4126)
2930simonpj@microsoft.com**20100614163935
2931 Ignore-this: a6ffbb5af6f51b142ed0aeae8ee5e3a9
2932]
2933[Fix Trac #4120: generate a proper coercion when unifying forall types
2934simonpj@microsoft.com**20100614134311
2935 Ignore-this: 601592bb505305f1954cbe730f168da4
2936 
2937 This was just a blatant omission, which hasn't come up before.
2938 Easily fixed, happily.
2939]
2940[Use mkFunTy to ensure that invariants are respected
2941simonpj@microsoft.com**20100614134159
2942 Ignore-this: 67dcada7a4e8d9927581cd77af71b6f
2943]
2944[Remove redundant debug code
2945simonpj@microsoft.com**20100601154151
2946 Ignore-this: e6ff11c04c631cf6aac73788cbcf02b5
2947]
2948[Fix Trac #4099: better error message for type functions
2949simonpj@microsoft.com**20100531140413
2950 Ignore-this: 3f53ca98cf770577818b9c0937482577
2951 
2952 Now we only want about "T is a type function and might not be
2953 injective" when matchin (T x) against (T y), which is the case
2954 that is really confusing.
2955]
2956[Gruesome fix in CorePrep to fix embarassing Trac #4121
2957simonpj@microsoft.com**20100614132726
2958 Ignore-this: fe82d15474afaac3e6133adfd7a7e055
2959 
2960 This is a long-lurking bug that has been flushed into
2961 the open by other arity-related changes.  There's a
2962 long comment
2963 
2964      Note [CafInfo and floating]
2965 
2966 to explain. 
2967 
2968 I really hate the contortions we have to do through to keep correct
2969 CafRef information on top-level binders.  The Right Thing, I believe,
2970 is to compute CAF and arity information later, and merge it into the
2971 interface-file information when the latter is generated.
2972 
2973 But for now, this hackily fixes the problem.
2974]
2975[Fix a bug in CorePrep that meant output invariants not satisfied
2976simonpj@microsoft.com**20100531150013
2977 Ignore-this: d34eb36d8877d3caf1cf2b20de426abd
2978 
2979 In cpePair I did things in the wrong order so that something that
2980 should have been a CprRhs wasn't.  Result: a crash in CoreToStg.
2981 Fix is easy, and I added more informative type signatures too.
2982]
2983[Robustify the treatement of DFunUnfolding
2984simonpj@microsoft.com**20100531145332
2985 Ignore-this: 8f5506ada4d89f6ab8ad1e8c3ffb09ba
2986 
2987 See Note [DFun unfoldings] in CoreSyn.  The issue here is that
2988 you can't tell how many dictionary arguments a DFun needs just
2989 from looking at the Arity of the DFun Id: if the dictionary is
2990 represented by a newtype the arity might include the dictionary
2991 and value arguments of the (single) method.
2992 
2993 So we need to record the number of arguments need by the DFun
2994 in the DFunUnfolding itself.  Details in
2995    Note [DFun unfoldings] in CoreSyn
2996]
2997[Fix spelling in comment
2998simonpj@microsoft.com**20100614132259
2999 Ignore-this: bbf0d55f2e5f10ef9c74592c12f9201c
3000]
3001[Update docs on view patterns
3002simonpj@microsoft.com**20100614074801
3003 Ignore-this: 8617b9078800d4942d71f142a5b6c831
3004]
3005[Fix printing of splices; part of #4124
3006Ian Lynagh <igloo@earth.li>**20100613154838
3007 Just putting parens around non-atomic expressions isn't sufficient
3008 for splices, as only the $x and $(e) forms are valid input.
3009]
3010[In ghci, catch IO exceptions when calling canonicalizePath
3011Ian Lynagh <igloo@earth.li>**20100613134627
3012 We now get an exception if the path doesn't exist
3013]
3014[Whitespace only
3015Ian Lynagh <igloo@earth.li>**20100612213119]
3016[Whitespace only
3017Ian Lynagh <igloo@earth.li>**20100612165450]
3018[Update ghci example output in user guide; patch from YitzGale in #4111
3019Ian Lynagh <igloo@earth.li>**20100612162250]
3020[Fix #4131 missing UNTAG_CLOSURE in messageBlackHole()
3021benl@ouroborus.net**20100611044614]
3022[messageBlackHole: fix deadlock bug caused by a missing 'volatile'
3023Simon Marlow <marlowsd@gmail.com>**20100610080636
3024 Ignore-this: 3cda3054bb45408aa9bd2d794b69c938
3025]
3026[Pass --no-tmp-comp-dir to Haddock (see comment)
3027Simon Marlow <marlowsd@gmail.com>**20100604083214
3028 Ignore-this: bfa4d74038637bd149f4d878b4eb8a87
3029]
3030[Track changes to DPH libs
3031Roman Leshchinskiy <rl@cse.unsw.edu.au>**20100607052903
3032 Ignore-this: 4dbc3f8418af3e74b3fc4f9a9dfe7764
3033]
3034[Track changes to DPH libs
3035Roman Leshchinskiy <rl@cse.unsw.edu.au>**20100607012642
3036 Ignore-this: 5d4e498171a3c57ab02621bfaea82cff
3037]
3038[In ghc-pkg, send warnings to stderr
3039Ian Lynagh <igloo@earth.li>**20100606161726
3040 Ignore-this: 56927d13b5e1c1ce2752734f0f9b665b
3041]
3042[Re-add newlines to enable layout for multi-line input.
3043Ian Lynagh <igloo@earth.li>**20100602180737
3044 Patch from Adam Vogt <vogt.adam@gmail.com>
3045 Partial fix for #3984
3046]
3047[Don't use unnecessary parens when printing types (Fix Trac 4107)
3048simonpj@microsoft.com**20100604110143
3049 Ignore-this: a833714ab13013c4345b222f4e87db1d
3050 
3051    f :: Eq a => a -> a
3052 rather than
3053    f :: (Eq a) => a -> a
3054]
3055[Track DPH library changes
3056Roman Leshchinskiy <rl@cse.unsw.edu.au>**20100604005728
3057 Ignore-this: 32bc2fbea6ad975e89545d4c42fd7c30
3058]
3059[fix --source-entity option passed to Haddock: we needed to escape a #
3060Simon Marlow <marlowsd@gmail.com>**20100603125459
3061 Ignore-this: d52ae6188b510c482bcebb23f0e553ae
3062]
3063[__stg_EAGER_BLACKHOLE_INFO -> __stg_EAGER_BLACKHOLE_info (#4106)
3064Simon Marlow <marlowsd@gmail.com>**20100602091419
3065 Ignore-this: 293315ac8f86fd366b8d61992ecc7961
3066]
3067[Add xhtml package (a new dependency of Haddock; not installed/shipped)
3068Simon Marlow <marlowsd@gmail.com>**20100602090101
3069 Ignore-this: af0ac8b91abe98f7fdb624ea0a4dee20
3070]
3071[Use UserInterrupt rather than our own Interrupted exception (#4100)
3072Simon Marlow <marlowsd@gmail.com>**20100602082345
3073 Ignore-this: 1909acf2f452593138b9f85024711714
3074]
3075[Add the global package DB to ghc --info (#4103)
3076Simon Marlow <marlowsd@gmail.com>**20100602082233
3077 Ignore-this: fd5c0e207e70eb0f62606c45dc5b8124
3078]
3079[rts/sm/GC.c: resize_generations(): Remove unneeded check of number of generations.
3080Marco Túlio Gontijo e Silva <marcot@debian.org>**20100528115612
3081 Ignore-this: 6f1bea62917c01c7adac636146132c97
3082 
3083 This "if" is inside another "if" which checks for RtsFlags.GcFlags.generations
3084 > 1, so testing this again is redundant, assuming the number of generations
3085 won't change during program execution.
3086]
3087[rts/sm/BlockAlloc.c: Small comment correction.
3088Marco Túlio Gontijo e Silva <marcot@debian.org>**20100526205839
3089 Ignore-this: bd2fcd4597cc872d80b0e2eeb1c3998a
3090]
3091[rts/sm/GC.c: Annotate constants.
3092Marco Túlio Gontijo e Silva <marcot@debian.org>**20100526205707
3093 Ignore-this: f232edb89383564d759ed890a18f602f
3094]
3095[includes/rts/storage/GC.h: generation_: n_words: Improve comment.
3096Marco Túlio Gontijo e Silva <marcot@debian.org>**20100526204615
3097 Ignore-this: f5d5feefa8f7b552303978f1804fea23
3098]
3099[Add PPC_RELOC_LOCAL_SECTDIFF support; patch from PHO in #3654
3100Ian Lynagh <igloo@earth.li>**20100601204211
3101 Ignore-this: 51293b7041cdce3ce7619ef11cf7ceb
3102]
3103[powerpc-apple-darwin now supports shared libs
3104Ian Lynagh <igloo@earth.li>**20100601173325]
3105[Vectoriser: only treat a function as scalar if it actually computes something
3106Roman Leshchinskiy <rl@cse.unsw.edu.au>**20100601045630
3107 Ignore-this: e5d99a6ddb62052e3520094a5af47552
3108]
3109[Add a release notes file for 6.14.1
3110Ian Lynagh <igloo@earth.li>**20100530171117
3111 Ignore-this: 1941e6d3d1f4051b69ca2f17a1cf84d6
3112]
3113[Check dblatex actually creates the files we tell it to
3114Ian Lynagh <igloo@earth.li>**20100530171043
3115 Ignore-this: ccc72caea2313be05cbac59bb54c0603
3116 If it fails, it still exits successfully.
3117]
3118[Add darwin to the list of OSes for which we use mmap
3119Ian Lynagh <igloo@earth.li>**20100529145016
3120 Ignore-this: a86d12a3334aaaafc86f7af9dbb0a7ae
3121 Patch from Barney Stratford
3122]
3123[Simplify the CPP logic in rts/Linker.c
3124Ian Lynagh <igloo@earth.li>**20100529144929
3125 Ignore-this: 1288f5b752cc1ab8b1c90cfd0ecfdf68
3126]
3127[Fix validate on OS X
3128Ian Lynagh <igloo@earth.li>**20100529154726]
3129[OS X x86_64 fix from Barney Stratford
3130Ian Lynagh <igloo@earth.li>**20100529122440]
3131[OS X 64 installer fixes from Barney Stratford
3132Ian Lynagh <igloo@earth.li>**20100528234935]
3133[fix warning
3134Simon Marlow <marlowsd@gmail.com>**20100525155812
3135 Ignore-this: f34eee3fe3d89579fd8d381c91ced750
3136]
3137[Fix doc bugs (#4071)
3138Simon Marlow <marlowsd@gmail.com>**20100525155728
3139 Ignore-this: aa25be196de567de360075022a1942f7
3140]
3141[Make sparks into weak pointers (#2185)
3142Simon Marlow <marlowsd@gmail.com>**20100525150435
3143 Ignore-this: feea0bb5006007b82c932bc3006124d7
3144 The new strategies library (parallel-2.0+, preferably 2.2+) is now
3145 required for parallel programming, otherwise parallelism will be lost.
3146]
3147[If you say 'make' or 'make stage=2' here, pretend we're in the ghc dir
3148Simon Marlow <marlowsd@gmail.com>**20100525085301
3149 Ignore-this: 78b740337aa460915c812cbbcdae5321
3150]
3151[Another attempt to get these #defines right
3152Simon Marlow <marlowsd@gmail.com>**20100525154313
3153 Ignore-this: 460ca0c47d81cd25eae6542114f67899
3154 Apparently on Solaris it is an error to omit _ISOC99_SOURCE when using
3155 _POSIX_C_SOURCE==200112L.
3156]
3157[Add configure flags for the location of GMP includes/library; fixes #4022
3158Ian Lynagh <igloo@earth.li>**20100525221616
3159 Ignore-this: fc3060caf995d07274ec975eeefbdf3e
3160]
3161[Refactor pretty printing of TyThings to fix Trac #4015
3162simonpj@microsoft.com**20100525153126
3163 Ignore-this: 8f15053b7554f62caa84201d2e4976d2
3164]
3165[When haddocking, we need the dependencies to have been built
3166Ian Lynagh <igloo@earth.li>**20100525145830
3167 as haddock loads the .hi files with the GHC API.
3168]
3169[Fix profiling output; spotted by jlouis
3170Ian Lynagh <igloo@earth.li>**20100525111217
3171 We were outputing the number of words allocated in a column titled "bytes".
3172]
3173[Improve printing of TyThings; fixes Trac #4087
3174simonpj@microsoft.com**20100525114045
3175 Ignore-this: da2a757a533454bba80b9b77cc5a771
3176]
3177[Spelling in comments
3178simonpj@microsoft.com**20100525114001
3179 Ignore-this: 270f3da655e526cf04e27db7a01e29c0
3180]
3181[Refactor (again) the handling of default methods
3182simonpj@microsoft.com**20100525113910
3183 Ignore-this: 6686f6cdb878d57abf6b49fec64fcbb1
3184 
3185 This patch fixes Trac #4056, by
3186 
3187  a) tidying up the treatment of default method names
3188  b) removing the 'module' argument to newTopSrcBinder
3189 
3190 The details aren't that interesting, but the result
3191 is much tidier. The original bug was a 'nameModule' panic,
3192 caused by trying to find the module of a top-level name.
3193 But TH quotes generate Internal top-level names that don't
3194 have a module, and that is generally a good thing. 
3195 
3196 Fixing that in turn led to the default-method refactoring,
3197 which also makes the Name for a default method be handled
3198 in the same way as other derived names, generated in BuildTyCl
3199 via a call newImplicitBinder.  Hurrah.
3200]
3201[Don't do SpecConstr on NOINLINE things (Trac #4064)
3202simonpj@microsoft.com**20100525112807
3203 Ignore-this: 452be0a2cef0042fb67275c2827b5f72
3204 
3205 Since the RULE from specialising gets the same Activation as
3206 the inlining for the Id itself there's no point in specialising
3207 a NOINLINE thing, because the rule will be permanently switched
3208 off.
3209 
3210 See Note [Transfer activation] in SpecConstr
3211 and Note [Auto-specialisation and RULES] in Specialise.
3212]
3213[Change our #defines to work on FreeBSD too
3214Simon Marlow <marlowsd@gmail.com>**20100524105828
3215 Ignore-this: b23ede46211e67859206c0ec57d6a86f
3216 With glibc, things like _POSIX_C_SOURCE and _ISOC99_SOURCE are
3217 additive, but on FreeBSD they are mutually exclusive.  However, it
3218 turns out we only need to define _POSIX_C_SOURCE and _XOPEN_SOURCE to
3219 get all the C99 stuff we need too, so there's no need for any #ifdefs.
3220 
3221 Submitted by: Gabor PALI <pgj@FreeBSD.org>
3222]
3223[Add a missing UNTAG_CLOSURE, causing bus errors on Sparc
3224Simon Marlow <marlowsd@gmail.com>**20100524105547
3225 Ignore-this: a590b5391d6f05d50c8c088456c3c166
3226 We just about got away with this on x86 which isn't
3227 alignment-sensitive.  The result of the memory load is compared
3228 against a few different values, but there is a fallback case that
3229 happened to be the right thing when the pointer was tagged.  A good
3230 bug to find, nonetheless.
3231]
3232[Add wiki links
3233Simon Marlow <marlowsd@gmail.com>**20100520095953
3234 Ignore-this: c22f126cde166e6207922b2eb51d29e3
3235]
3236[the 'stage=0' trick to disable all compiler builds stopped working; fix it
3237Simon Marlow <marlowsd@gmail.com>**20100520104455
3238 Ignore-this: bb6fae9056471612c8dbf06916188c33
3239]
3240[Comments and formatting only
3241benl@ouroborus.net**20100524014021
3242 Ignore-this: 64579c38154728b632e358bec751cc0b
3243]
3244[Core prettyprinter fixes. Patch from Tim Chevalier. Fixes #4085
3245Ian Lynagh <igloo@earth.li>**20100522225048]
3246[Fix the RTS debug_p build
3247Ian Lynagh <igloo@earth.li>**20100522163127]
3248[Unset $CFLAGS for "GNU non-executable stack" configure test; fixes #3889
3249Ian Lynagh <igloo@earth.li>**20100521165005
3250 With gcc 4.4 we get
3251     Error: can't resolve `.note.GNU-stack' {.note.GNU-stack section} - `.Ltext0' {.text section}
3252 when running gcc with the -g flag. To work around this we unset
3253 CFLAGS when running the test.
3254]
3255[Don't run "set -o igncr" before configuring libffi
3256Ian Lynagh <igloo@earth.li>**20100520162918
3257 Ignore-this: 489fa94df23f2adf4ff63c8ede2c0794
3258 It used to make the build work on cygwin, but now it breaks it instead:
3259     config.status: creating include/Makefile
3260     gawk: ./confLqjohp/subs.awk:1: BEGIN {\r
3261     gawk: ./confLqjohp/subs.awk:1: ^ backslash not last character on line
3262     config.status: error: could not create include/Makefile
3263     make[2]: *** [libffi/stamp.ffi.configure-shared] Error 1
3264     make[1]: *** [all] Error 2
3265]
3266[Stop passing -Wl,-macosx_version_min to gcc
3267Ian Lynagh <igloo@earth.li>**20100520154003
3268 Fixes a build failure on OS X 10.6. When linking
3269     rts/dist/build/libHSrts-ghc6.13.20100519.dylib
3270 we got
3271     ld: symbol dyld_stub_binding_helper not defined (usually in crt1.o/dylib1.o/bundle1.o)
3272     collect2: ld returned 1 exit status
3273]
3274[Fix build on FreeBSD; patch from Gabor PALI
3275Ian Lynagh <igloo@earth.li>**20100519140552]
3276[Fix package shadowing order (#4072)
3277Simon Marlow <marlowsd@gmail.com>**20100519104617
3278 Ignore-this: 26ea5e4bb5dff18618b807a54c7d6ebb
3279 
3280 Later packages are supposed to shadow earlier ones in the stack,
3281 unless the ordering is overriden with -package-id flags.
3282 Unfortunately an earlier fix for something else had sorted the list of
3283 packages so that it was in lexicographic order by installedPackageId,
3284 and sadly our test (cabal/shadow) didn't pick this up because the
3285 lexicographic ordering happened to work for the test.  I've now fixed
3286 the test so it tries both orderings.
3287]
3288[Set more env variables when configuring libffi
3289Ian Lynagh <igloo@earth.li>**20100518185014
3290 We now tell it where to find ld, nm and ar
3291]
3292[Set the location of ar to be the in-tree ar on Windows
3293Ian Lynagh <igloo@earth.li>**20100518181556]
3294[Change another / to </> to avoid building paths containing \/
3295Ian Lynagh <igloo@earth.li>**20100518172015
3296 This will hopefully fix #2889.
3297]
3298[Fix #4074 (I hope).
3299Simon Marlow <marlowsd@gmail.com>**20100518113214
3300 Ignore-this: 73cd70f5bc6f5add5247b61985c03fc1
3301 
3302 1. allow multiple threads to call startTimer()/stopTimer() pairs
3303 2. disable the timer around fork() in forkProcess()
3304 
3305 A corresponding change to the process package is required.
3306]
3307[we don't have a gcc-lib in LIB_DIR any more
3308Simon Marlow <marlowsd@gmail.com>**20100401102351
3309 Ignore-this: f41acd2d8f8e6763aa8bd57a0b44a7e4
3310]
3311[In validate, use gmake if available; based on a patch from Gabor PALI
3312Ian Lynagh <igloo@earth.li>**20100517200654]
3313[Remove duplicate "./configure --help" output; fixes #4075
3314Ian Lynagh <igloo@earth.li>**20100516141206]
3315[Update various 'sh boot's to 'perl boot'
3316Ian Lynagh <igloo@earth.li>**20100516122609
3317 Spotted by Marco Túlio Gontijo e Silva
3318]
3319[add missing initialisation for eventBufMutex
3320Simon Marlow <marlowsd@gmail.com>**20100514094943
3321 Ignore-this: 7f75594a8cb54fbec5aebd46bb959f45
3322]
3323[Undo part of #4003 patch
3324Simon Marlow <marlowsd@gmail.com>**20100513142017
3325 Ignore-this: cb65db86a38a7e5ccee9f779e489d104
3326 We still need the workaround for when compiling HEAD with 6.12.2
3327 
3328]
3329[fix !TABLES_NEXT_TO_CODE
3330Simon Marlow <marlowsd@gmail.com>**20100510151934
3331 Ignore-this: fccb859b114bef1c3122c98e60af51
3332]
3333[looksLikeModuleName: allow apostrophe in module names (#4051)
3334Simon Marlow <marlowsd@gmail.com>**20100510094741
3335 Ignore-this: df9348f3ba90608bec57257b47672985
3336]
3337[add the proper library dependencies for GhcProfiled=YES
3338Simon Marlow <marlowsd@gmail.com>**20100506122118
3339 Ignore-this: 6236993aa308ab5b5e1e5ea5f65982a
3340]
3341[Fix Trac #4003: fix the knot-tying in checkHiBootIface
3342simonpj@microsoft.com**20100511075026
3343 Ignore-this: a9ce2a318386fdc8782848df84592002
3344 
3345 I had incorrectly "optimised" checkHiBootIface so that it forgot
3346 to update the "knot-tied" type environment.
3347 
3348 This patch fixes the HEAD
3349]
3350[Re-engineer the derived Ord instance generation code (fix Trac #4019)
3351simonpj@microsoft.com**20100510133333
3352 Ignore-this: 8fe46e4dad27fbee211a7928acf372c2
3353   
3354 As well as fixing #4019, I rejigged the way that Ord instances are
3355 generated, which should make them faster in general.  See the
3356 Note [Generating Ord instances].
3357 
3358 I tried to measure the performance difference from this change, but
3359 the #4019 fix only removes one conditional branch per iteration, and
3360 I couldn't measure a consistent improvement.  But still, tihs is
3361 better than before.
3362]
3363[Make arity of INLINE things consistent
3364simonpj@microsoft.com**20100510133005
3365 Ignore-this: 15e7abf803d1dcb3f4ca760d2d939d0d
3366 
3367 We eta-expand things with INLINE pragmas;
3368 see Note [Eta-expanding INLINE things].
3369 
3370 But I eta-expanded it the wrong amount when the function
3371 was overloaded.  Ooops.
3372]
3373[Compacting GC fix, we forgot to thread the new bq field of StgTSO.
3374Simon Marlow <marlowsd@gmail.com>**20100510082325
3375 Ignore-this: a079c8446e2ad53efff6fd95d0f3ac80
3376]
3377[Add version constraints for the boot packages; fixes trac #3852
3378Ian Lynagh <igloo@earth.li>**20100509175051
3379 When using the bootstrapping compiler, we now explicitly constrain
3380 the version of boot packages (Cabal, extensible-exceptions, etc) to the
3381 in-tree version, so that the build system is less fragile should the
3382 user have a newer version installed for the bootstrapping compiler.
3383]
3384[Don't include inter-package dependencies when compiling with stage 0; #4031
3385Ian Lynagh <igloo@earth.li>**20100509130511
3386 This fixes a problem when building with GHC 6.12 on Windows, where
3387 dependencies on stage 0 (bootstrapping compiler) packages have absolute
3388 paths c:/ghc/..., and make gets confused by the colon.
3389]
3390[Add a ghc.mk for bindisttest/
3391Ian Lynagh <igloo@earth.li>**20100508223911]
3392[Move some make variables around so they are available when cleaning
3393Ian Lynagh <igloo@earth.li>**20100508212405]
3394[Optimise checkremove a bit
3395Ian Lynagh <igloo@earth.li>**20100508202006]
3396[Improve the bindisttest Makefile
3397Ian Lynagh <igloo@earth.li>**20100508195450]
3398[Add tools to test that cleaning works properly
3399Ian Lynagh <igloo@earth.li>**20100508194105]
3400[Tweak the ghc-pkg finding code
3401Ian Lynagh <igloo@earth.li>**20100508125815
3402 It now understand the ghc-stage[123] names we use in-tree, and it won't
3403 go looking for any old ghc-pkg if it can't find the one that matches
3404 ghc.
3405]
3406[Add a way to show what cleaning would be done, without actually doing it
3407Ian Lynagh <igloo@earth.li>**20100508122438]
3408[Tidy up the "rm" flags in the build system
3409Ian Lynagh <igloo@earth.li>**20100508115745]
3410[Correct install-name for dynamic Darwin rts
3411pho@cielonegro.org**20100508151155
3412 Ignore-this: 6d31716c8c113dcb46e9cb925c4201df
3413]
3414[PIC support for PowerPC
3415pho@cielonegro.org**20100508143900
3416 Ignore-this: 3673859a305398c4acae3f4d7c997615
3417 
3418 PPC.CodeGen.getRegister was not properly handling PicBaseReg.
3419 It seems working with this patch, but I'm not sure this change is correct.
3420]
3421[Fix makefile loop (#4050)
3422pho@cielonegro.org**20100507140707
3423 Ignore-this: 3a1cb13d0600977e74d17ac26cbef83d
3424 
3425 The libtool creates "libffi.dylib" and "libffi.5.dylib" but not "libffi.5.0.9.dylib". Having it in libffi_DYNAMIC_LIBS causes an infinite makefile loop.
3426]
3427[Fix crash in nested callbacks (#4038)
3428Simon Marlow <marlowsd@gmail.com>**20100507093222
3429 Ignore-this: cade85e361534ce711865a4820276388
3430 Broken by "Split part of the Task struct into a separate struct
3431 InCall".
3432]
3433[Add $(GhcDynamic) knob, set to YES to get stage2 linked with -dynamic
3434Simon Marlow <marlowsd@gmail.com>**20100428205241
3435 Ignore-this: 1db8bccf92099785ecac39aebd27c92d
3436 Default currently NO.
3437 
3438 Validate passed with GhcDynamic=YES on x86/Linux here.
3439 
3440 The compiler is currently slower on x86 when linked -dynamic,
3441 because the GC inner loop has been adversely affected by -fPIC, I'm
3442 looking into how to fix it.
3443]
3444[omit "dyn" from the way appended to the __stginit label
3445Simon Marlow <marlowsd@gmail.com>**20100428204914
3446 Ignore-this: 14183f3defa9f2bde68fda6729b740bc
3447 When GHCi is linked dynamically, we still want to be able to load
3448 non-dynamic object files.
3449]
3450[improvements to findPtr(), a neat hack for browsing the heap in gdb
3451Simon Marlow <marlowsd@gmail.com>**20100506115427
3452 Ignore-this: ac57785bb3e13b97a5945f753f068738
3453]
3454[Fix +RTS -G1
3455Simon Marlow <marlowsd@gmail.com>**20100506110739
3456 Ignore-this: 86a5de39a94d3331a4ee1213f82be497
3457]
3458[Enable the "redundant specialise pragmas" warning; fixes trac #3855
3459Ian Lynagh <igloo@earth.li>**20100506175351]
3460[Find the correct external ids when there's a wrapper
3461simonpj@microsoft.com**20100506164135
3462 Ignore-this: 636266407b174b05b2b8646cc73062c0
3463 
3464 We were failing to externalise the wrapper id for a function
3465 that had one.
3466]
3467[Add a comment about pattern coercions
3468simonpj@microsoft.com**20100506164027
3469 Ignore-this: 17428089f3df439f65d892e23e8ed61a
3470]
3471[Comments only
3472simonpj@microsoft.com**20100506163829
3473 Ignore-this: 169167b6463873ab173cc5750c5be469
3474]
3475[Make a missing name in mkUsageInfo into a panic
3476simonpj@microsoft.com**20100506163813
3477 Ignore-this: b82ff1b8bf89f74f146db7cb5cc4c4d7
3478 
3479 We really want to know about this!
3480]
3481[Refactoring of hsXxxBinders
3482simonpj@microsoft.com**20100506163737
3483 Ignore-this: 97c6667625262b160f9746f7bea1c980
3484 
3485 This patch moves various functions that extract the binders
3486 from a HsTyClDecl, HsForeignDecl etc into HsUtils, and gives
3487 them consistent names.
3488]
3489[Fix Trac #3966: warn about useless UNPACK pragmas
3490simonpj@microsoft.com**20100506163337
3491 Ignore-this: 5beb24b686eda6113b614dfac8490df1
3492 
3493 Warning about useless UNPACK pragmas wasn't as easy as I thought.
3494 I did quite a bit of refactoring, which improved the code by refining
3495 the types somewhat.  In particular notice that in DataCon, we have
3496 
3497     dcStrictMarks   :: [HsBang]
3498     dcRepStrictness :: [StrictnessMarks]
3499 
3500 The former relates to the *source-code* annotation, the latter to
3501 GHC's representation choice.
3502]
3503[Make tcg_dus behave more sanely; fixes a mkUsageInfo panic
3504simonpj@microsoft.com**20100506162719
3505 Ignore-this: d000bca15b0e127e297378ded1bfb81b
3506 
3507 The tcg_dus field used to contain *uses* of type and class decls,
3508 but not *defs*.  That was inconsistent, and it really went wrong
3509 for Template Haskell bracket.  What happened was that
3510  foo = [d| data A = A
3511                   f :: A -> A
3512                   f x = x |]
3513 would find a "use" of A when processing the top level of the module,
3514 which in turn led to a mkUsageInfo panic in MkIface.  The cause was
3515 the fact that the tcg_dus for the nested quote didn't have defs for
3516 A.
3517]
3518[Add a HsExplicitFlag to SpliceDecl, to improve Trac #4042
3519simonpj@microsoft.com**20100506161523
3520 Ignore-this: e4e563bac2fd831cc9e94612f5b4fa9d
3521 
3522 The issue here is that
3523 
3524     g :: A -> A
3525     f
3526     data A = A
3527 
3528 is treated as if you'd written $(f); that is the call of
3529 f is a top-level Template Haskell splice.  This patch
3530 makes sure that we *first* check the -XTemplateHaskellFlag
3531 and bleat about a parse error if it's off.  Othewise we
3532 get strange seeing "A is out of scope" errors.
3533]
3534[Change an assert to a warn
3535simonpj@microsoft.com**20100506161111
3536 Ignore-this: 739a4fb4c7940376b0f2c8ad52a1966c
3537 
3538 This is in the constraint simplifier which I'm about
3539 to rewrite, so I'm hoping the assert isn't fatal!
3540]
3541[Tidy up debug print a little
3542simonpj@microsoft.com**20100506161027
3543 Ignore-this: bd5492878e06bee1cddcbb3fc4df66d8
3544]
3545[Remove useless UNPACK pragmas
3546simonpj@microsoft.com**20100506161012
3547 Ignore-this: 3e5ab1a7cf58107034412a798bc214e5
3548]
3549[Add WARNM2 macro, plus some refactoring
3550simonpj@microsoft.com**20100506160808
3551 Ignore-this: 2ab4f1f0b5d94be683036e77aec09255
3552]
3553[Use -Wwarn for the binary package, becuase it has redundant UNPACK pragmas
3554simonpj@microsoft.com**20100506160750
3555 Ignore-this: cf0d3a11473e28bfce9602e716e69a5f
3556]
3557[Fix Trac #3966: warn about unused UNPACK pragmas
3558simonpj@microsoft.com**20100409201812
3559 Ignore-this: c96412596b39c918b5fb9b3c39ce2119
3560]
3561[Fix Trac #3953: fail earlier when using a bogus quasiquoter
3562simonpj@microsoft.com**20100409201748
3563 Ignore-this: ef48e39aa932caed538643985234f043
3564]
3565[Fix Trac #3965: tighten conditions when deriving Data
3566simonpj@microsoft.com**20100409184420
3567 Ignore-this: 96f7d7d2da11565d26b465d7d0497ac9
3568 
3569 It's tricky to set up the context for a Data instance.  I got it wrong
3570 once, and fixed it -- hence the "extra_constraints" in
3571 TcDeriv.inferConstraints. 
3572 
3573 But it still wasn't right!  The tricky bit is that dataCast1 is only
3574 generated when T :: *->*, and dataCast2 when T :: *->*->*. (See
3575 the code in TcGenDeriv for dataCastX.
3576]
3577[Fix Trac #3964: view patterns in DsArrows
3578simonpj@microsoft.com**20100409165557
3579 Ignore-this: d823c182831d5e2e592e995b16180e2f
3580 
3581 Just a missing case; I've eliminated the catch-all so
3582 that we get a warning next time we extend HsPat
3583]
3584[Fix Trac #3955: renamer and type variables
3585simonpj@microsoft.com**20100409163710
3586 Ignore-this: bd5ec64d76c0f583bf5f224792bf294c
3587 
3588 The renamer wasn't computing the free variables of a type declaration
3589 properly.  This patch refactors a bit, and makes it more robust,
3590 fixing #3955 and several other closely-related bugs.  (We were
3591 omitting some free variables and that could just possibly lead to a
3592 usage-version tracking error.
3593]
3594[Layout only
3595simonpj@microsoft.com**20100409163506
3596 Ignore-this: 1f14990b5aa0b9821b84452fb34e9f41
3597]
3598[Give a better deprecated message for INCLUDE pragmas; fixes #3933
3599Ian Lynagh <igloo@earth.li>**20100506130910
3600 We now have a DeprecatedFullText constructor, so we can override the
3601 "-#include is deprecated: " part of the warning.
3602]
3603[De-haddock a comment that confuses haddock
3604Ian Lynagh <igloo@earth.li>**20100506123607]
3605[Fix comment to not confuse haddock
3606Ian Lynagh <igloo@earth.li>**20100506113642]
3607[Detect EOF when trying to parse a string in hp2ps
3608Ian Lynagh <igloo@earth.li>**20100506000830]
3609[Make the demand analyser sdd demands for strict constructors
3610simonpj@microsoft.com**20100505200936
3611 Ignore-this: eb32632adbc354eb7a5cf884c263e0d3
3612 
3613 This opportunity was spotted by Roman, and is documented in
3614 Note [Add demands for strict constructors] in DmdAnal.
3615]
3616[Fix interaction of exprIsCheap and the lone-variable inlining check
3617simonpj@microsoft.com**20100505200723
3618 Ignore-this: f3cb65085c5673a99153d5d7b6559ab1
3619 
3620 See Note [Interaction of exprIsCheap and lone variables] in CoreUnfold
3621 
3622 This buglet meant that a nullary definition with an INLINE pragma
3623 counter-intuitively didn't get inlined at all.  Roman identified
3624 the bug.
3625]
3626[Matching cases in SpecConstr and Rules
3627simonpj@microsoft.com**20100505200543
3628 Ignore-this: f5c28c780fbf8badce84c6fdc9aa1779
3629 
3630 This patch has zero effect.  It includes comments,
3631 a bit of refactoring, and a tiny bit of commment-out
3632 code go implement the "matching cases" idea below.
3633 
3634 In the end I've left it disabled because while I think
3635 it does no harm I don't think it'll do any good either.
3636 But I didn't want to lose the idea totally. There's
3637 a thread called "Storable and constant memory" on
3638 the libraries@haskell.org list (Apr 2010) about it.
3639 
3640 Note [Matching cases]
3641 ~~~~~~~~~~~~~~~~~~~~~
3642 {- NOTE: This idea is currently disabled.  It really only works if
3643          the primops involved are OkForSpeculation, and, since
3644         they have side effects readIntOfAddr and touch are not.
3645         Maybe we'll get back to this later .  -}
3646   
3647 Consider
3648    f (case readIntOffAddr# p# i# realWorld# of { (# s#, n# #) ->
3649       case touch# fp s# of { _ ->
3650       I# n# } } )
3651 This happened in a tight loop generated by stream fusion that
3652 Roman encountered.  We'd like to treat this just like the let
3653 case, because the primops concerned are ok-for-speculation.
3654 That is, we'd like to behave as if it had been
3655    case readIntOffAddr# p# i# realWorld# of { (# s#, n# #) ->
3656    case touch# fp s# of { _ ->
3657    f (I# n# } } )
3658]
3659[Comments only
3660simonpj@microsoft.com**20100504163629
3661 Ignore-this: 3be12df04714aa820bce706b5dc8a9cb
3662]
3663[Comments only
3664simonpj@microsoft.com**20100504163529
3665 Ignore-this: 791e2fd39c7d880ce1dc80ebdf3a5398
3666]
3667[Comments only
3668simonpj@microsoft.com**20100504163457
3669 Ignore-this: f19e9ffeb3d65770b1595bca5f97a59d
3670]
3671[Comments only (about type families)
3672simonpj@microsoft.com**20100417145032
3673 Ignore-this: dd39425ef2155d52dbf55a4d5fd97cb8
3674]
3675[Fix hp2ps when the .hp file has large string literals
3676Ian Lynagh <igloo@earth.li>**20100505191921]
3677[In build system, call package-config after including package data
3678Ian Lynagh <igloo@earth.li>**20100504225035
3679 Otherwise the $1_$2_HC_OPTS variable gets clobbered.
3680]
3681[runghc: flush stdout/stderr on an exception (#3890)
3682Simon Marlow <marlowsd@gmail.com>**20100505133848
3683 Ignore-this: 224c1898cec64cb1c94e0d7033e7590e
3684]
3685[Remove the Unicode alternative for ".." (#3894)
3686Simon Marlow <marlowsd@gmail.com>**20100505121202
3687 Ignore-this: 2452cd67281667106f9169747b6d784f
3688]
3689[tidyup; no functional changes
3690Simon Marlow <marlowsd@gmail.com>**20100505115015
3691 Ignore-this: d0787e5cdeef1dee628682fa0a46019
3692]
3693[Make the running_finalizers flag task-local
3694Simon Marlow <marlowsd@gmail.com>**20100505114947
3695 Ignore-this: 345925d00f1dca203941b3c5d84c90e1
3696 Fixes a bug reported by Lennart Augustsson, whereby we could get an
3697 incorrect error from the RTS about re-entry from a finalizer,
3698]
3699[add a MAYBE_GC() in killThread#, fixes throwto003(threaded2) looping
3700Simon Marlow <marlowsd@gmail.com>**20100505114746
3701 Ignore-this: efea04991d6feed04683a42232fc85da
3702]
3703[Allow filepath-1.2.*
3704Simon Marlow <marlowsd@gmail.com>**20100505101139
3705 Ignore-this: 1b5580cd9cd041ec48f40cd37603326a
3706]
3707[BlockedOnMsgThrowTo is possible in resurrectThreads (#4030)
3708Simon Marlow <marlowsd@gmail.com>**20100505094534
3709 Ignore-this: ac24a22f95ffeaf480187a1620fdddb2
3710]
3711[Don't raise a throwTo when the target is masking and BlockedOnBlackHole
3712Simon Marlow <marlowsd@gmail.com>**20100505094506
3713 Ignore-this: 302616931f61667030d77ddfbb02374e
3714]
3715[Fix build with GHC 6.10
3716Ian Lynagh <igloo@earth.li>**20100504180302
3717 In GHC 6.10, intersectionWith is (a -> b -> a) instead of (a -> b -> c),
3718 so we need to jump through some hoops to get the more general type.
3719]
3720[The libffi patches are no longer needed
3721Ian Lynagh <igloo@earth.li>**20100504171603]
3722[Use the in-tree windres; fixes trac #4032
3723Ian Lynagh <igloo@earth.li>**20100504170941]
3724[Print unfoldings on lambda-bound variables
3725Simon PJ <simonpj@microsoft.com>**20100503181822
3726 Ignore-this: 2fd5a7502cc6273d96258e0914f0f8cd
3727 
3728 ...in the unusual case where they have one;
3729 see Note [Case binders and join points] in Simplify.lhs
3730]
3731[Replace FiniteMap and UniqFM with counterparts from containers.
3732Milan Straka <fox@ucw.cz>**20100503171315
3733 Ignore-this: a021972239163dbf728284b19928cebb
3734 
3735 The original interfaces are kept. There is small performance improvement:
3736 - when compiling for five nofib, we get following speedups:
3737     Average                -----           -2.5%
3738     Average                -----           -0.6%
3739     Average                -----           -0.5%
3740     Average                -----           -5.5%
3741     Average                -----          -10.3%
3742 - when compiling HPC ten times, we get:
3743     switches                          oldmaps   newmaps
3744     -O -fasm                          117.402s  116.081s (98.87%)
3745     -O -fasm -fregs-graph             119.993s  118.735s (98.95%)
3746     -O -fasm -fregs-iterative         120.191s  118.607s (98.68%)
3747]
3748[Make the demand analyser take account of lambda-bound unfoldings
3749Simon PJ <simonpj@microsoft.com>**20100503151630
3750 Ignore-this: 2ee8e27d4df2debfc79e6b8a17c32bc1
3751 
3752 This is a long-standing lurking bug. See Note [Lamba-bound unfoldings]
3753 in DmdAnal.
3754 
3755 I'm still not really happy with this lambda-bound-unfolding stuff.
3756]
3757[Fix dynamic libs on OS X, and enable them by default
3758Ian Lynagh <igloo@earth.li>**20100503150302]
3759[Switch back to using bytestring from the darcs repo; partially fixes #3855
3760Ian Lynagh <igloo@earth.li>**20100502113458]
3761[Fix some cpp warnings when building on FreeBSD; patch from Gabor PALI
3762Ian Lynagh <igloo@earth.li>**20100428150700]
3763[Fix "make 2"
3764Ian Lynagh <igloo@earth.li>**20100427162212
3765 The new Makefile logic was enabling the stage 1 rules when stage=2,
3766 so "make 2" was rebuilding stage 1.
3767]
3768[Inplace programs depend on their shell wrappers
3769Ian Lynagh <igloo@earth.li>**20100427160038]
3770[--make is now the default (#3515), and -fno-code works with --make (#3783)
3771Simon Marlow <marlowsd@gmail.com>**20100427122851
3772 Ignore-this: 33330474fa4703f32bf9997462b4bf3c
3773 If the command line contains any Haskell source files, then we behave
3774 as if --make had been given.
3775 
3776 The meaning of the -c flag has changed (back): -c now selects one-shot
3777 compilation, but stops before linking.  However, to retain backwards
3778 compatibility, -c is still allowed with --make, and means the same as
3779 --make -no-link.  The -no-link flag has been un-deprecated.
3780 
3781 -fno-code is now allowed with --make (#3783); the fact that it was
3782 disabled before was largely accidental, it seems.  We also had some
3783 regressions in this area: it seems that -fno-code was causing a .hc
3784 file to be emitted in certain cases.  I've tidied up the code, there
3785 was no need for -fno-code to be a "mode" flag, as far as I can tell.
3786 
3787 -fno-code does not emit interface files, nor does it do recompilation
3788 checking, as suggested in #3783.  This would make Haddock emit
3789 interface files, for example, and I'm fairly sure we don't want to do
3790 that.  Compiling with -fno-code is pretty quick anyway, perhaps we can
3791 get away without recompilation checking.
3792]
3793[remove duplicate docs for -e in --help output (#4010)
3794Simon Marlow <marlowsd@gmail.com>**20100426140642
3795 Ignore-this: 187ff893ba8ffa0ec127867a7590e38d
3796]
3797[workaround for #4003, fixes HEAD build with 6.12.2
3798Simon Marlow <marlowsd@gmail.com>**20100426103428
3799 Ignore-this: c4bc445dc8052d4e6efef3f1daf63562
3800]
3801[Make sure all the clean rules are always included
3802Ian Lynagh <igloo@earth.li>**20100424181823
3803 In particular, this fixes a problem where stage3 bits weren't being cleaned
3804]
3805[Correct the name of the amd64/FreeBSD platform in PlatformSupportsSharedLibs
3806Ian Lynagh <igloo@earth.li>**20100424132830
3807 We weren't getting sharedlibs on amd64/FreeBSD because of this
3808]
3809[Include DPH docs in bindists
3810Ian Lynagh <igloo@earth.li>**20100424123101]
3811[reinstate eta-expansion during SimplGently, to fix inlining of sequence_
3812Simon Marlow <marlowsd@gmail.com>**20100423124853
3813 Ignore-this: 4fa0fd5bafe0d6b58fc81076f50d5f8d
3814]
3815[fix 64-bit value for W_SHIFT, which thankfully appears to be not used
3816Simon Marlow <marlowsd@gmail.com>**20100422213605
3817 Ignore-this: 525c062d2456c224ec8d0e083edd3b55
3818]
3819[Add missing constant folding and optimisation for unsigned division
3820Simon Marlow <marlowsd@gmail.com>**20100422213443
3821 Ignore-this: fb10d1cda0852fab0cbcb47247498fb3
3822 Noticed by Denys Rtveliashvili <rtvd@mac.com>, see #4004
3823]
3824[Fix the GHC API link in the main doc index.html
3825Ian Lynagh <igloo@earth.li>**20100422213226]
3826[Give the right exit code in darcs-all
3827Ian Lynagh <igloo@earth.li>**20100421171339
3828 Our END block was calling system, which alters $?. So now we save and
3829 restore it.
3830]
3831[Use StgWord64 instead of ullong
3832Ian Lynagh <igloo@earth.li>**20100421162336
3833 This patch also fixes ullong_format_string (renamed to showStgWord64)
3834 so that it works with values outside the 32bit range (trac #3979), and
3835 simplifies the without-commas case.
3836]
3837[Implement try10Times in Makefile
3838Ian Lynagh <igloo@earth.li>**20100420165909
3839 Avoid using seq, as FreeBSD has jot instead.
3840]
3841[Fix crash in non-threaded RTS on Windows
3842Simon Marlow <marlowsd@gmail.com>**20100420122125
3843 Ignore-this: 28b0255a914a8955dce02d89a7dfaca
3844 The tso->block_info field is now overwritten by pushOnRunQueue(), but
3845 stg_block_async_info was assuming that it still held a pointer to the
3846 StgAsyncIOResult.  We must therefore save this value somewhere safe
3847 before putting the TSO on the run queue.
3848]
3849[Expand the scope of the event_buf_mutex to cover io_manager_event
3850Simon Marlow <marlowsd@gmail.com>**20100420122026
3851 Ignore-this: 185a6d84f7d4a35997f10803f6dacef1
3852 I once saw a failure that I think was due to a race on
3853 io_manager_event, this should fix it.
3854]
3855[Flags -auto and -auto-all operate only on functions not marked INLINE.
3856Milan Straka <fox@ucw.cz>**20100331191050
3857 Ignore-this: 3b63580cfcb3c33d62ad697c36d94d05
3858]
3859[Spelling correction for LANGUAGE pragmas
3860Max Bolingbroke <batterseapower@hotmail.com>**20100413192825
3861 Ignore-this: 311b51ba8d43f6c7fd32f48db9a88dee
3862]
3863[Update the user guide so it talks about the newer "do rec" notation everywhere
3864Ian Lynagh <igloo@earth.li>**20100416205416
3865 Some of the problems highlighted in trac #3968.
3866]
3867[Fix typo
3868Ian Lynagh <igloo@earth.li>**20100416205412]
3869[Fix Trac #3950: unifying types of different kinds
3870simonpj@microsoft.com**20100412151845
3871 Ignore-this: d145b9de5ced136ef2c39f3ea4a04f4a
3872 
3873 I was assuming that the unifer only unified types of the
3874 same kind, but now we can "defer" unsolved constraints that
3875 invariant no longer holds.  Or at least is's more complicated
3876 to ensure. 
3877 
3878 This patch takes the path of not assuming the invariant, which
3879 is simpler and more robust.  See
3880 Note [Mismatched type lists and application decomposition]
3881]
3882[Fix Trac #3943: incorrect unused-variable warning
3883simonpj@microsoft.com**20100412151630
3884 Ignore-this: 52459f2b8b02c3cb120abe674dc9a060
3885 
3886 In fixing this I did the usual little bit of refactoring
3887]
3888[Convert boot and boot-pkgs to perl
3889Ian Lynagh <igloo@earth.li>**20100415143919
3890 This stops us having to worry about sh/sed/... portability.
3891]
3892[Use $(MAKE), not make, when recursively calling make
3893Ian Lynagh <igloo@earth.li>**20100415121453]
3894[Remove the ghc_ge_609 makefile variables
3895Ian Lynagh <igloo@earth.li>**20100412235658
3896 They are now guaranteed to be YES
3897]
3898[Increase the minimum version number required to 6.10 in configure.ac
3899Ian Lynagh <igloo@earth.li>**20100412235313]
3900[The bootstrapping compiler is now required to be > 609
3901Ian Lynagh <igloo@earth.li>**20100409161046]
3902[Handle IND_STATIC in isRetainer
3903Ian Lynagh <igloo@earth.li>**20100409104207
3904 IND_STATIC used to be an error, but at the moment it can happen
3905 as isAlive doesn't look through IND_STATIC as it ignores static
3906 closures. See trac #3956 for a program that hit this error.
3907]
3908[Add Data and Typeable instances to HsSyn
3909David Waern <david.waern@gmail.com>**20100330011020
3910 Ignore-this: c3f2717207b15539fea267c36b686e6a
3911 
3912 The instances (and deriving declarations) have been taken from the ghc-syb
3913 package.
3914]
3915[Fix for derefing ThreadRelocated TSOs in MVar operations
3916Simon Marlow <marlowsd@gmail.com>**20100407092824
3917 Ignore-this: 94dd7c68a6094eda667e2375921a8b78
3918]
3919[sanity check fix
3920Simon Marlow <marlowsd@gmail.com>**20100407092746
3921 Ignore-this: 9c18cd5f5393e5049015ca52e62a1269
3922]
3923[get the reg liveness right in the putMVar# heap check
3924Simon Marlow <marlowsd@gmail.com>**20100407092724
3925 Ignore-this: b1ba07a59ecfae00e9a1f8391741abc
3926]
3927[initialise the headers of MSG_BLACKHOLE objects properly
3928Simon Marlow <marlowsd@gmail.com>**20100407081712
3929 Ignore-this: 183dcd0ca6a395d08db2be12b02bdd79
3930]
3931[initialise the headers of MVAR_TSO_QUEUE objects properly
3932Simon Marlow <marlowsd@gmail.com>**20100407081514
3933 Ignore-this: 4b4a2f30cf2fb69ca4128c41744687bb
3934]
3935[undo debugging code
3936Simon Marlow <marlowsd@gmail.com>**20100406142740
3937 Ignore-this: 323c2248f817b6717c19180482fc4b00
3938]
3939[putMVar#: fix reg liveness in the heap check
3940Simon Marlow <marlowsd@gmail.com>**20100406135832
3941 Ignore-this: cddd2c7807ac7612c9b2c4c0d384d284
3942]
3943[account for the new BLACKHOLEs in the GHCi debugger
3944Simon Marlow <marlowsd@gmail.com>**20100406133406
3945 Ignore-this: 4d4aeb4bbada3f50dc1fb0123f565e8f
3946]
3947[don't forget to deRefTSO() in tryWakeupThread()
3948Simon Marlow <marlowsd@gmail.com>**20100406130411
3949 Ignore-this: 171d57c4f8653835dec0b69f9be9881c
3950]
3951[Fix bug in popRunQueue
3952Simon Marlow <marlowsd@gmail.com>**20100406091453
3953 Ignore-this: 9d3cec8f18f5c5cbd51751797386eb6f
3954]
3955[fix bug in migrateThread()
3956Simon Marlow <marlowsd@gmail.com>**20100401105840
3957 Ignore-this: 299bcf0d1ea0f8865f3e845eb93d2ad3
3958]
3959[Remove the IND_OLDGEN and IND_OLDGEN_PERM closure types
3960Simon Marlow <marlowsd@gmail.com>**20100401093519
3961 Ignore-this: 95f2480c8a45139835eaf5610217780b
3962 These are no longer used: once upon a time they used to have different
3963 layout from IND and IND_PERM respectively, but that is no longer the
3964 case since we changed the remembered set to be an array of addresses
3965 instead of a linked list of closures.
3966]
3967[Change the representation of the MVar blocked queue
3968Simon Marlow <marlowsd@gmail.com>**20100401091605
3969 Ignore-this: 20a35bfabacef2674df362905d7834fa
3970 
3971 The list of threads blocked on an MVar is now represented as a list of
3972 separately allocated objects rather than being linked through the TSOs
3973 themselves.  This lets us remove a TSO from the list in O(1) time
3974 rather than O(n) time, by marking the list object.  Removing this
3975 linear component fixes some pathalogical performance cases where many
3976 threads were blocked on an MVar and became unreachable simultaneously
3977 (nofib/smp/threads007), or when sending an asynchronous exception to a
3978 TSO in a long list of thread blocked on an MVar.
3979 
3980 MVar performance has actually improved by a few percent as a result of
3981 this change, slightly to my surprise.
3982 
3983 This is the final cleanup in the sequence, which let me remove the old
3984 way of waking up threads (unblockOne(), MSG_WAKEUP) in favour of the
3985 new way (tryWakeupThread and MSG_TRY_WAKEUP, which is idempotent).  It
3986 is now the case that only the Capability that owns a TSO may modify
3987 its state (well, almost), and this simplifies various things.  More of
3988 the RTS is based on message-passing between Capabilities now.
3989]
3990[eliminate some duplication with a bit of CPP
3991Simon Marlow <marlowsd@gmail.com>**20100330154355
3992 Ignore-this: 838f7d341f096ca14c86ab9c81193e36
3993]
3994[Make ioManagerDie() idempotent
3995Simon Marlow <marlowsd@gmail.com>**20100401100705
3996 Ignore-this: a5996b43cdb2e2d72e6e971d7ea925fb
3997 Avoids screeds of "event buffer overflowed; event dropped" in
3998 conc059(threaded1).
3999]
4000[Move a thread to the front of the run queue when another thread blocks on it
4001Simon Marlow <marlowsd@gmail.com>**20100329144521
4002 Ignore-this: c518ff0d41154680edc811d891826a29
4003 This fixes #3838, and was made possible by the new BLACKHOLE
4004 infrastructure.  To allow reording of the run queue I had to make it
4005 doubly-linked, which entails some extra trickiness with regard to
4006 GC write barriers and suchlike.
4007]
4008[remove non-existent MUT_CONS symbols
4009Simon Marlow <marlowsd@gmail.com>**20100330152600
4010 Ignore-this: 885628257a9d03f2ece2a754d993014a
4011]
4012[change throwTo to use tryWakeupThread rather than unblockOne
4013Simon Marlow <marlowsd@gmail.com>**20100329144613
4014 Ignore-this: 10ad4965e6c940db71253f1c72218bbb
4015]
4016[tiny GC optimisation
4017Simon Marlow <marlowsd@gmail.com>**20100329144551
4018 Ignore-this: 9e095b9b73fff0aae726f9937846ba92
4019]
4020[New implementation of BLACKHOLEs
4021Simon Marlow <marlowsd@gmail.com>**20100329144456
4022 Ignore-this: 96cd26793b4e6ab9ddd0d59aae5c2f1d
4023 
4024 This replaces the global blackhole_queue with a clever scheme that
4025 enables us to queue up blocked threads on the closure that they are
4026 blocked on, while still avoiding atomic instructions in the common
4027 case.
4028 
4029 Advantages:
4030 
4031  - gets rid of a locked global data structure and some tricky GC code
4032    (replacing it with some per-thread data structures and different
4033    tricky GC code :)
4034 
4035  - wakeups are more prompt: parallel/concurrent performance should
4036    benefit.  I haven't seen anything dramatic in the parallel
4037    benchmarks so far, but a couple of threading benchmarks do improve
4038    a bit.
4039 
4040  - waking up a thread blocked on a blackhole is now O(1) (e.g. if
4041    it is the target of throwTo).
4042 
4043  - less sharing and better separation of Capabilities: communication
4044    is done with messages, the data structures are strictly owned by a
4045    Capability and cannot be modified except by sending messages.
4046 
4047  - this change will utlimately enable us to do more intelligent
4048    scheduling when threads block on each other.  This is what started
4049    off the whole thing, but it isn't done yet (#3838).
4050 
4051 I'll be documenting all this on the wiki in due course.
4052 
4053]
4054[Fix warnings (allow pushOnRunQueue() to not be inlined)
4055Simon Marlow <marlowsd@gmail.com>**20100401114559
4056 Ignore-this: f40bfbfad70a5165a946d11371605b7d
4057]
4058[remove out of date comment
4059Simon Marlow <marlowsd@gmail.com>**20100401105853
4060 Ignore-this: 26af88dd418ee0bcda7223b3b7e4e8d2
4061]
4062[tidy up spacing in stderr traces
4063Simon Marlow <marlowsd@gmail.com>**20100326163122
4064 Ignore-this: 16558b0433a274be217d4bf39aa4946
4065]
4066[Fix an assertion that was not safe when running in parallel
4067Simon Marlow <marlowsd@gmail.com>**20100325143656
4068 Ignore-this: cad08fb8900eb3a475547af0189fcc47
4069]
4070[Never jump directly to a thunk's entry code, even if it is single-entry
4071Simon Marlow <marlowsd@gmail.com>**20100325114847
4072 Ignore-this: 938da172c06a97762ef605c8fccfedf1
4073 I don't think this fixes any bugs as we don't have single-entry thunks
4074 at the moment, but it could cause problems for parallel execution if
4075 we ever did re-introduce update avoidance.
4076]
4077[Rename forgotten -dverbose-simpl to -dverbose-core2core in the docs.
4078Milan Straka <fox@ucw.cz>**20100331153626
4079 Ignore-this: 2da58477fb96e1cfb80f37dddd7c422c
4080]
4081[Add -pa and -V to the documentation of time profiling options.
4082Milan Straka <fox@ucw.cz>**20100329191121
4083 Ignore-this: be74d216481ec5a19e5f40f85e6e3d65
4084]
4085[Keep gcc 4.5 happy
4086Simon Marlow <marlowsd@gmail.com>**20100330120425
4087 Ignore-this: 7811878cc2bd1ce9cfbb5bf102fe3454
4088]
4089[Fix warning compiling Linker.c for PPC Mac
4090naur@post11.tele.dk**20100403182355
4091 Ignore-this: e2d2448770c9714ce17dd6cf3e297063
4092 The warning message eliminated is:
4093 > rts/Linker.c:4756:0:
4094 >      warning: nested extern declaration of 'symbolsWithoutUnderscore'
4095]
4096[Fix error compiling AsmCodeGen.lhs for PPC Mac (mkRtsCodeLabel)
4097naur@post11.tele.dk**20100403181656
4098 Ignore-this: deb7524ea7852a15a2ac0849c8c82f74
4099 The error messages eliminated are:
4100 > compiler/nativeGen/AsmCodeGen.lhs:875:31:
4101 >     Not in scope: `mkRtsCodeLabel'
4102 > compiler/nativeGen/AsmCodeGen.lhs:879:31:
4103 >     Not in scope: `mkRtsCodeLabel'
4104 > compiler/nativeGen/AsmCodeGen.lhs:883:31:
4105 >     Not in scope: `mkRtsCodeLabel'
4106]
4107[Fix error compiling AsmCodeGen.lhs for PPC Mac (DestBlockId)
4108naur@post11.tele.dk**20100403180643
4109 Ignore-this: 71e833e94ed8371b2ffabc2cf80bf585
4110 The error message eliminated is:
4111 > compiler/nativeGen/AsmCodeGen.lhs:637:16:
4112 >     Not in scope: data constructor `DestBlockId'
4113]
4114[Fix boot-pkgs's sed usage to work with Solaris's sed
4115Ian Lynagh <igloo@earth.li>**20100401153441]
4116[Pass "-i org.haskell.GHC" to packagemaker when building the OS X installer
4117Ian Lynagh <igloo@earth.li>**20100331144707
4118 This seems to fix this failure:
4119 [...]
4120 ** BUILD SUCCEEDED **
4121 rm -f -f GHC-system.pmdoc/*-contents.xml
4122 /Developer/usr/bin/packagemaker -v --doc GHC-system.pmdoc\
4123              -o /Users/ian/to_release/ghc-6.12.1.20100330/GHC-6.12.1.20100330-i386.pkg
4124 2010-03-31 15:08:15.695 packagemaker[13909:807] Setting to : 0 (null)
4125 2010-03-31 15:08:15.709 packagemaker[13909:807] Setting to : 0 org.haskell.glasgowHaskellCompiler.ghc.pkg
4126 2010-03-31 15:08:15.739 packagemaker[13909:807] relocate: (null) 0
4127 2010-03-31 15:08:15.740 packagemaker[13909:807] *** Terminating app due to uncaught exception 'NSInvalidArgumentException', reason: '*** -[NSXMLDocument initWithXMLString:options:error:]: nil argument'
4128 2010-03-31 15:08:15.741 packagemaker[13909:807] Stack: (
4129     2511962091,
4130     2447007291,
4131     2511961547,
4132     2511961610,
4133     2432803204,
4134     453371,
4135     447720,
4136     436209,
4137     435510,
4138     9986,
4139     9918
4140 )
4141 make[1]: *** [framework-pkg] Trace/BPT trap
4142 make: *** [framework-pkg] Error 2
4143]
4144[Use machdepCCOpts when compiling the file to toggle -(no-)rtsopts
4145Ian Lynagh <igloo@earth.li>**20100331161302
4146 Should fix toggling on OS X "Snow Leopard". Diagnosed by Roman Leshchinskiy.
4147]
4148[Avoid a non-portable use of tar reported by Roman Leshchinskiy
4149Ian Lynagh <igloo@earth.li>**20100330145802]
4150[Don't install EXTRA_PACKAGES by default
4151Simon Marlow <marlowsd@gmail.com>**20100330142714
4152 Ignore-this: d4cc8f87a6de8d9d1d6dc9b77130b3
4153]
4154[fix a non-portable printf format
4155Simon Marlow <marlowsd@gmail.com>**20100330134437
4156 Ignore-this: d41c23c54ec29654cb2049de1e588570
4157]
4158[avoid single quote in #error
4159Simon Marlow <marlowsd@gmail.com>**20100330120346
4160 Ignore-this: 663f39e7a27fead2f648fbf22d345bb4
4161]
4162[use FMT_Word64 instead of locally-defined version
4163Simon Marlow <marlowsd@gmail.com>**20100330114650
4164 Ignore-this: 82697b8095dffb3a8e196c687006ece0
4165]
4166[remove old/unused DotnetSupport and GhcLibsWithUnix
4167Simon Marlow <marlowsd@gmail.com>**20100330123732
4168 Ignore-this: c68814868b3671abdc369105bbeafe6c
4169]
4170[fix return type cast in f.i.wrapper when using libffi (#3516)
4171Simon Marlow <marlowsd@gmail.com>**20100329154220
4172 Ignore-this: f898eb8c9ae2ca2009e539735b92c438
4173 
4174 Original fix submitted by
4175   Sergei Trofimovich <slyfox@community.haskell.org>
4176 modified by me:
4177  - exclude 64-bit types
4178  - compare uniques, not strings
4179  - #include "ffi.h" is conditional
4180]
4181[libffi: install 'ffitarget.h' header as sole 'ffi.h' is unusable
4182Simon Marlow <marlowsd@gmail.com>**20100329135734
4183 Ignore-this: f9b555ea289d8df1aa22cb6faa219a39
4184 Submitted by: Sergei Trofimovich <slyfox@community.haskell.org>
4185 Re-recorded against HEAD.
4186]
4187[avoid a fork deadlock (see comments)
4188Simon Marlow <marlowsd@gmail.com>**20100329132329
4189 Ignore-this: 3377f88b83bb3b21e42d7fc5f0d866f
4190]
4191[tidy up the end of the all_tasks list after forking
4192Simon Marlow <marlowsd@gmail.com>**20100329132253
4193 Ignore-this: 819d679875be5f344e816210274d1c29
4194]
4195[Add a 'setKeepCAFs' external function (#3900)
4196Simon Marlow <marlowsd@gmail.com>**20100329110036
4197 Ignore-this: ec532a18cad4259a09847b0b9ae2e1d2
4198]
4199[Explicitly check whether ar supports the @file syntax
4200Ian Lynagh <igloo@earth.li>**20100329123325
4201 rather than assuming that all GNU ar's do.
4202 Apparently OpenBSD's older version doesn't.
4203]
4204[Fix the format specifier for Int64/Word64 on Windows
4205Ian Lynagh <igloo@earth.li>**20100327182126
4206 mingw doesn't understand %llu/%lld - it treats them as 32-bit rather
4207 than 64-bit. We use %I64u/%I64d instead.
4208]
4209[Fix the ghci startmenu item
4210Ian Lynagh <igloo@earth.li>**20100326235934
4211 I'm not sure what changed, but it now doesn't work for me without
4212 the "Start in" field being set.
4213]
4214[Fix paths to docs in "Start Menu" entries in Windows installer; fixes #3847
4215Ian Lynagh <igloo@earth.li>**20100326155917]
4216[Add a licence file for the Windows installer to use
4217Ian Lynagh <igloo@earth.li>**20100326155130]
4218[Add gcc-g++ to the inplace mingw installation; fixes #3893
4219Ian Lynagh <igloo@earth.li>**20100326154714]
4220[Add the licence file to the Windows installer. Fixes #3934
4221Ian Lynagh <igloo@earth.li>**20100326152449]
4222[Quote the paths to alex and happy in configure
4223Ian Lynagh <igloo@earth.li>**20100325143449
4224 Ignore-this: d6d6e1a250f88985bbeea760e63a79db
4225]
4226[Use </> rather than ++ "/"
4227Ian Lynagh <igloo@earth.li>**20100325133237
4228 This stops us generating paths like
4229     c:\foo\/ghc460_0/ghc460_0.o
4230 which windres doesn't understand.
4231]
4232[Append $(exeext) to utils/ghc-pkg_dist_PROG
4233Ian Lynagh <igloo@earth.li>**20100324233447
4234 Fixes bindist creation
4235]
4236[A sanity check
4237Simon Marlow <marlowsd@gmail.com>**20100325110500
4238 Ignore-this: 3b3b76d898c822456857e506b7531e65
4239]
4240[do_checks: do not set HpAlloc if the stack check fails
4241Simon Marlow <marlowsd@gmail.com>**20100325110328
4242 Ignore-this: 899ac8c29ca975d03952dbf4608d758
4243 
4244 This fixes a very rare heap corruption bug, whereby
4245 
4246  - a context switch is requested, which sets HpLim to zero
4247    (contextSwitchCapability(), called by the timer signal or
4248    another Capability).
4249 
4250  - simultaneously a stack check fails, in a code fragment that has
4251    both a stack and a heap check.
4252 
4253 The RTS then assumes that a heap-check failure has occurred and
4254 subtracts HpAlloc from Hp, although in fact it was a stack-check
4255 failure and retreating Hp will overwrite valid heap objects.  The bug
4256 is that HpAlloc should only be set when Hp has been incremented by the
4257 heap check.  See comments in rts/HeapStackCheck.cmm for more details.
4258 
4259 This bug is probably incredibly rare in practice, but I happened to be
4260 working on a test that triggers it reliably:
4261 concurrent/should_run/throwto001, compiled with -O -threaded, args 30
4262 300 +RTS -N2, run repeatedly in a loop.
4263]
4264[comments and formatting only
4265Simon Marlow <marlowsd@gmail.com>**20100325104617
4266 Ignore-this: c0a211e15b5953bb4a84771bcddd1d06
4267]
4268[Change how perl scripts get installed; partially fixes #3863
4269Ian Lynagh <igloo@earth.li>**20100324171422
4270 We now regenerate them when installing, which means the path for perl
4271 doesn't get baked in
4272]
4273[Pass the location of gcc in the ghc wrapper script; partially fixes #3863
4274Ian Lynagh <igloo@earth.li>**20100324171408
4275 This means we don't rely on baking a path to gcc into the executable
4276]
4277[Quote the ar path in configure
4278Ian Lynagh <igloo@earth.li>**20100324162043]
4279[Remove unused cUSER_WAY_NAMES cUSER_WAY_OPTS
4280Ian Lynagh <igloo@earth.li>**20100324145048]
4281[Remove unused cCONTEXT_DIFF
4282Ian Lynagh <igloo@earth.li>**20100324145013]
4283[Remove unused cEnableWin32DLLs
4284Ian Lynagh <igloo@earth.li>**20100324144841]
4285[Remove unused cGHC_CP
4286Ian Lynagh <igloo@earth.li>**20100324144656]
4287[Fix the build for non-GNU-ar
4288Ian Lynagh <igloo@earth.li>**20100324132907]
4289[Tweak the Makefile code for making .a libs; fixes trac #3642
4290Ian Lynagh <igloo@earth.li>**20100323221325
4291 The main change is that, rather than using "xargs ar" we now put
4292 all the filenames into a file, and do "ar @file". This means that
4293 ar adds all the files at once, which works around a problem where
4294 files with the same basename in a later invocation were overwriting
4295 the existing file in the .a archive.
4296]
4297[Enable shared libraries on Windows; fixes trac #3879
4298Ian Lynagh <igloo@earth.li>**20100320231414
4299 Ignore-this: c93b35ec5b7a7fa6ddb286d17a616216
4300]
4301[Add the external core PDF to the new build system
4302Ian Lynagh <igloo@earth.li>**20100321161909]
4303[Allow specifying $threads directly when validating
4304Ian Lynagh <igloo@earth.li>**20100321112835]
4305[Remove LazyUniqFM; fixes trac #3880
4306Ian Lynagh <igloo@earth.li>**20100320213837]
4307[UNDO: slight improvement to scavenging ...
4308Simon Marlow <marlowsd@gmail.com>**20100319153413
4309 Ignore-this: f0ab581c07361f7b57eae02dd6ec893c
4310 
4311 Accidnetally pushed this patch which, while it validates, isn't
4312 correct.
4313 
4314 rolling back:
4315 
4316 Fri Mar 19 11:21:27 GMT 2010  Simon Marlow <marlowsd@gmail.com>
4317   * slight improvement to scavenging of update frames when a collision has occurred
4318 
4319     M ./rts/sm/Scav.c -19 +15
4320]
4321[slight improvement to scavenging of update frames when a collision has occurred
4322Simon Marlow <marlowsd@gmail.com>**20100319112127
4323 Ignore-this: 6de2bb9614978975f17764a0f259d9bf
4324]
4325[Don't install the utf8-string package
4326Ian Lynagh <igloo@earth.li>**20100317212709]
4327[Don't use -Bsymbolic when linking the RTS
4328Ian Lynagh <igloo@earth.li>**20100316233357
4329 This makes the RTS hooks work when doing dynamic linking
4330]
4331[Fix Trac #3920: Template Haskell kinds
4332simonpj@microsoft.com**20100317123519
4333 Ignore-this: 426cac7920446e04f3cc30bd1d9f76e2
4334 
4335 Fix two places where we were doing foldl instead of foldr
4336 after decomposing a Kind.  Strange that the same bug appears
4337 in two quite different places!
4338]
4339[copy_tag_nolock(): fix write ordering and add a write_barrier()
4340Simon Marlow <marlowsd@gmail.com>**20100316143103
4341 Ignore-this: ab7ca42904f59a0381ca24f3eb38d314
4342 
4343 Fixes a rare crash in the parallel GC.
4344 
4345 If we copy a closure non-atomically during GC, as we do for all
4346 immutable values, then before writing the forwarding pointer we better
4347 make sure that the closure itself is visible to other threads that
4348 might follow the forwarding pointer.  I imagine this doesn't happen
4349 very often, but I just found one case of it: in scavenge_stack, the
4350 RET_FUN case, after evacuating ret_fun->fun we then follow it and look
4351 up the info pointer.
4352]
4353[Add sliceP mapping to vectoriser builtins
4354benl@ouroborus.net**20100316060517
4355 Ignore-this: 54c3cafff584006b6fbfd98124330aa3
4356]
4357[Comments only
4358benl@ouroborus.net**20100311064518
4359 Ignore-this: d7dc718cc437d62aa5b1b673059a9b22
4360]
4361[TAG 2010-03-16
4362Ian Lynagh <igloo@earth.li>**20100316005137
4363 Ignore-this: 234e3bc29e2f26cc59d7b03d780cc352
4364]
4365Patch bundle hash:
436629abe184123b10bc6e6d2a24494077c20fd7a2fe
4367-----BEGIN PGP SIGNATURE-----
4368Version: GnuPG v1.4.10 (Darwin)
4369
4370iD8DBQFMoe73FOecpxqG73IRAl6qAKCJEXtEX0grR1MoR4eGkZT9ZZGq7wCcDKv1
4371OLF2qE6zbZfnaZCEXoZ2xoE=
4372=cTAa
4373-----END PGP SIGNATURE-----