ghc.git
8 years agoShow LLVM mangler and code gen passes at v2
David Terei [Tue, 22 Nov 2011 20:30:32 +0000 (12:30 -0800)] 
Show LLVM mangler and code gen passes at v2

8 years agoFormatting fix.
David Terei [Tue, 22 Nov 2011 20:23:43 +0000 (12:23 -0800)] 
Formatting fix.

8 years agoMerge branch 'master' of http://darcs.haskell.org/ghc
Ian Lynagh [Tue, 22 Nov 2011 20:55:38 +0000 (20:55 +0000)] 
Merge branch 'master' of darcs.haskell.org/ghc

8 years agoFix comments.
Michal Terepeta [Mon, 21 Nov 2011 20:36:17 +0000 (21:36 +0100)] 
Fix comments.

8 years agoDrop ".exe" exetention from eventlog file name
Duncan Coutts [Mon, 21 Nov 2011 11:32:15 +0000 (11:32 +0000)] 
Drop ".exe" exetention from eventlog file name

Fixes ticket #5472

8 years agoTabs -> Spaces
David Terei [Tue, 22 Nov 2011 20:06:37 +0000 (12:06 -0800)] 
Tabs -> Spaces

8 years agoFormatting fix
David Terei [Tue, 22 Nov 2011 20:06:20 +0000 (12:06 -0800)] 
Formatting fix

8 years agoExplicitly handle unsupported Cmm prim ops.
David Terei [Tue, 22 Nov 2011 20:05:52 +0000 (12:05 -0800)] 
Explicitly handle unsupported Cmm prim ops.

8 years agoFormatting fixes
David Terei [Tue, 22 Nov 2011 19:27:25 +0000 (11:27 -0800)] 
Formatting fixes

8 years agoSpecify unsupported ops, don't just use a catch all
David Terei [Tue, 22 Nov 2011 19:24:16 +0000 (11:24 -0800)] 
Specify unsupported ops, don't just use a catch all

Better to specifically list the unsupported cases in code
than to have a catch all that panics. The later method hides
problems when new constructors are added such as the recent
additions to the supported Cmm prim ops that weren't ported
to the C backend since no one noticed.

8 years agoRemove registerised code for dead architectures: mips, ia64, alpha,
David Terei [Mon, 21 Nov 2011 20:05:18 +0000 (12:05 -0800)] 
Remove registerised code for dead architectures: mips, ia64, alpha,
hppa1, m68k

8 years agoFix missing primop in PprC
David Terei [Sun, 20 Nov 2011 01:38:18 +0000 (17:38 -0800)] 
Fix missing primop in PprC

8 years agoFormatting of PprC
David Terei [Sun, 20 Nov 2011 01:38:04 +0000 (17:38 -0800)] 
Formatting of PprC

8 years agoTabs -> Spaces
David Terei [Sat, 19 Nov 2011 19:36:21 +0000 (11:36 -0800)] 
Tabs -> Spaces

8 years agoTabs -> Spaces
David Terei [Sat, 19 Nov 2011 06:41:16 +0000 (22:41 -0800)] 
Tabs -> Spaces

8 years agoTabs -> Spaces
David Terei [Sat, 19 Nov 2011 04:09:52 +0000 (20:09 -0800)] 
Tabs -> Spaces

8 years agoRemove some old comments about the mangler
David Terei [Sat, 19 Nov 2011 03:47:24 +0000 (19:47 -0800)] 
Remove some old comments about the mangler

8 years agoFix warnings in ByteCodeItbls
David Terei [Sat, 19 Nov 2011 00:39:24 +0000 (16:39 -0800)] 
Fix warnings in ByteCodeItbls

8 years agoTabs -> Spaces + cleaning
David Terei [Fri, 18 Nov 2011 23:44:14 +0000 (15:44 -0800)] 
Tabs -> Spaces + cleaning

8 years agoEnable '-split-objs' with llvm backend
David Terei [Fri, 18 Nov 2011 22:42:47 +0000 (14:42 -0800)] 
Enable '-split-objs' with llvm backend

8 years agoBetter document the driver pipeline.
David Terei [Fri, 18 Nov 2011 22:31:00 +0000 (14:31 -0800)] 
Better document the driver pipeline.

8 years agoTweaks to safe haskell documentation.
David Terei [Tue, 22 Nov 2011 17:59:47 +0000 (09:59 -0800)] 
Tweaks to safe haskell documentation.

8 years agouType_defer: only call mkErrInfo if -ddump-tc-trace is on (#5631)
Simon Marlow [Tue, 22 Nov 2011 14:09:05 +0000 (14:09 +0000)] 
uType_defer: only call mkErrInfo if -ddump-tc-trace is on (#5631)

Profiling shows we're wasting a lot of time calling mkErrInfo and
tidying the context, work that is wasted if we're not tracing.

There is more to this compilation performance regression than this,
but this is one bug.

8 years agomerge
Simon Marlow [Tue, 22 Nov 2011 11:26:28 +0000 (11:26 +0000)] 
merge

8 years agofix warning
Simon Marlow [Mon, 21 Nov 2011 16:01:34 +0000 (16:01 +0000)] 
fix warning

8 years agoDo not wrap Coercions with ticks (fixes GADT3(profasm))
Simon Marlow [Mon, 21 Nov 2011 14:52:06 +0000 (14:52 +0000)] 
Do not wrap Coercions with ticks (fixes GADT3(profasm))

8 years agoDon't add the -x flag to ld when joining object files
Simon Marlow [Mon, 21 Nov 2011 13:27:15 +0000 (13:27 +0000)] 
Don't add the -x flag to ld when joining object files

8 years agoMerge /Users/benl/devel/ghc/ghc-head-devel
Ben Lippmeier [Tue, 22 Nov 2011 07:33:07 +0000 (18:33 +1100)] 
Merge /Users/benl/devel/ghc/ghc-head-devel

8 years agoUpdate Safe Haskell documentation for new design in 7.4
David Terei [Tue, 22 Nov 2011 07:26:44 +0000 (23:26 -0800)] 
Update Safe Haskell documentation for new design in 7.4

8 years agovectoriser: fix pvoids problem when vectorising enumerations
Ben Lippmeier [Tue, 22 Nov 2011 07:16:39 +0000 (18:16 +1100)] 
vectoriser: fix pvoids problem when vectorising enumerations

Also give a civilised error message when the data type to vectorised contains no data.

8 years agoAdd doc for -msse4.2 flag.
David Terei [Tue, 22 Nov 2011 00:07:56 +0000 (16:07 -0800)] 
Add doc for -msse4.2 flag.

8 years agoImprove documentation on backends
David Terei [Mon, 21 Nov 2011 23:57:42 +0000 (15:57 -0800)] 
Improve documentation on backends

8 years agoTurn on LLVM backend if 'keep-llvm-files' flag specified.
David Terei [Mon, 21 Nov 2011 22:34:04 +0000 (14:34 -0800)] 
Turn on LLVM backend if 'keep-llvm-files' flag specified.

8 years agoImprove code generator documentation.
David Terei [Mon, 21 Nov 2011 19:39:09 +0000 (11:39 -0800)] 
Improve code generator documentation.

8 years agoReplace EkCtxt by an SDoc
Jose Pedro Magalhaes [Mon, 21 Nov 2011 11:18:16 +0000 (11:18 +0000)] 
Replace EkCtxt by an SDoc

We were never really manipulating the EkCtxt after we had built it,
so it's simpler to just pass the final SDoc.

8 years agoRename ? to OpenKind and ?? to ArgKind
Jose Pedro Magalhaes [Mon, 21 Nov 2011 10:47:44 +0000 (10:47 +0000)] 
Rename ? to OpenKind and ?? to ArgKind

The previous names were not informative at all, and now we have
named kinds like Constraint and datatype promotion to kind, so
we might as well name these too.

I tried to update some comments to the new names, but certainly
many references to the old names remain.

8 years agoFix bug in the handling of TSOs in the compacting GC (#5644)
Simon Marlow [Fri, 18 Nov 2011 17:05:44 +0000 (17:05 +0000)] 
Fix bug in the handling of TSOs in the compacting GC (#5644)

8 years agoFix #5636: Use clang as assembler on OSX when LLVM >= 3.0
David Terei [Mon, 21 Nov 2011 11:01:32 +0000 (03:01 -0800)] 
Fix #5636: Use clang as assembler on OSX when LLVM >= 3.0

LLVM doesn't support the OS X system assembler anymore so we must use
their assembler through clang. Also improved error messages when various
LLVM tools can't be run.

8 years agoSimplify a regexp and improve a couple of comments
Ian Lynagh [Sun, 20 Nov 2011 18:23:37 +0000 (18:23 +0000)] 
Simplify a regexp and improve a couple of comments

8 years agoAdd branch command to sync-all
Max Bolingbroke [Sat, 19 Nov 2011 11:27:38 +0000 (11:27 +0000)] 
Add branch command to sync-all

8 years agoImprove the way we call "rm" in the build system; fixes trac #4916
Ian Lynagh [Sat, 19 Nov 2011 01:29:05 +0000 (01:29 +0000)] 
Improve the way we call "rm" in the build system; fixes trac #4916

We avoid calling "rm -rf" with no file arguments; this fixes cleaning
on Solaris, where that fails.

We also check for suspicious arguments: anything containing "..",
starting "/", or containing a "*" (you need to call $(wildcard ...)
yourself now if you really want globbing). This should make things
a little safer.

8 years agoRemove a redundant tabs warning kludge
Ian Lynagh [Fri, 18 Nov 2011 22:22:44 +0000 (22:22 +0000)] 
Remove a redundant tabs warning kludge

The tabs have already disappeared

8 years agoMerge branch 'master' of http://darcs.haskell.org/ghc
Ian Lynagh [Fri, 18 Nov 2011 19:07:33 +0000 (19:07 +0000)] 
Merge branch 'master' of darcs.haskell.org/ghc

8 years agoUpdate for a mutable localeEncoding
Max Bolingbroke [Fri, 18 Nov 2011 17:49:04 +0000 (17:49 +0000)] 
Update for a mutable localeEncoding

8 years agoMerge branch 'master' of http://darcs.haskell.org/ghc
Ian Lynagh [Fri, 18 Nov 2011 16:42:11 +0000 (16:42 +0000)] 
Merge branch 'master' of darcs.haskell.org/ghc

8 years agofix new warnings with gcc 4.6
Simon Marlow [Fri, 18 Nov 2011 14:59:49 +0000 (14:59 +0000)] 
fix new warnings with gcc 4.6

8 years agohscParse: do not record a dependency on non-existent source files
Simon Marlow [Fri, 18 Nov 2011 14:59:37 +0000 (14:59 +0000)] 
hscParse: do not record a dependency on non-existent source files

8 years agoCreate the directory for dump files to be put in
Ian Lynagh [Fri, 18 Nov 2011 13:48:33 +0000 (13:48 +0000)] 
Create the directory for dump files to be put in

Now that we don't necessarily put them in the same place as the source,
the directory may not exist.

8 years agoAdd -dumpdir flag; fixes trac #5198
Ian Lynagh [Fri, 18 Nov 2011 02:06:57 +0000 (02:06 +0000)] 
Add -dumpdir flag; fixes trac #5198

-ddump-* files are put in the given directory.

-outputdir also sets the dumpdir.

8 years agoTrack #included files for recompilation checking (#4900, #3589)
Simon Marlow [Fri, 18 Nov 2011 12:46:01 +0000 (12:46 +0000)] 
Track #included files for recompilation checking (#4900, #3589)

This was pretty straightforward: collect the filenames in the lexer,
and add them in to the tcg_dependent_files list that the typechecker
collects.

Note that we still don't get #included files in the ghc -M output.
Since we don't normally lex the whole file in ghc -M, this same
mechanism can't be used directly.

8 years agoThe "Linking" message should go through compilationProgressMessage
Simon Marlow [Fri, 18 Nov 2011 12:42:32 +0000 (12:42 +0000)] 
The "Linking" message should go through compilationProgressMessage

Currently it is appearing on stderr, whereas the rest of the messages
from --make go to stdout, this change fixes that.

8 years agoTweak to pretty printing for UsageFile
Simon Marlow [Fri, 18 Nov 2011 12:41:41 +0000 (12:41 +0000)] 
Tweak to pretty printing for UsageFile

8 years agoFix the vectorisation of workers of data constructors
Manuel M T Chakravarty [Fri, 18 Nov 2011 04:05:01 +0000 (15:05 +1100)] 
Fix the vectorisation of workers of data constructors

8 years agoBetter documentation for stack alignment design
David Terei [Thu, 17 Nov 2011 23:38:02 +0000 (15:38 -0800)] 
Better documentation for stack alignment design

8 years agoTabs -> Spaces + formatting fixes
David Terei [Thu, 17 Nov 2011 23:35:33 +0000 (15:35 -0800)] 
Tabs -> Spaces + formatting fixes

8 years agoMerge branch 'master' of http://darcs.haskell.org/ghc
Simon Peyton Jones [Thu, 17 Nov 2011 17:45:22 +0000 (17:45 +0000)] 
Merge branch 'master' of darcs.haskell.org/ghc

8 years agoFix panic (missing keepWanteds) in simplifyApprox
Dimitrios Vytiniotis [Thu, 17 Nov 2011 16:39:54 +0000 (16:39 +0000)] 
Fix panic (missing keepWanteds) in simplifyApprox

8 years agoGhcProfiled: don't automatically add -auto-all
Simon Marlow [Thu, 17 Nov 2011 11:59:10 +0000 (11:59 +0000)] 
GhcProfiled: don't automatically add -auto-all

8 years agoAdd some missing top-level SCCs
Simon Marlow [Thu, 17 Nov 2011 11:58:39 +0000 (11:58 +0000)] 
Add some missing top-level SCCs

8 years agoAdd a getStablePtr for flushStdHandles_closure
Simon Marlow [Thu, 17 Nov 2011 13:17:53 +0000 (13:17 +0000)] 
Add a getStablePtr for flushStdHandles_closure

8 years agoMerge branch 'master' of http://darcs.haskell.org/ghc
Simon Peyton Jones [Thu, 17 Nov 2011 16:02:23 +0000 (16:02 +0000)] 
Merge branch 'master' of darcs.haskell.org/ghc

8 years agoMake DsMeta work right on tuples (cf Trac #5597)
Simon Peyton Jones [Thu, 17 Nov 2011 13:14:14 +0000 (13:14 +0000)] 
Make DsMeta work right on tuples (cf Trac #5597)

8 years agoRemove a quadratic complexity blow-up in coercionKind
Simon Peyton Jones [Thu, 17 Nov 2011 13:13:15 +0000 (13:13 +0000)] 
Remove a quadratic complexity blow-up in coercionKind
thereby fixing Trac #5631.

See Note [Nested InstCos] in Coercion

8 years agoWarning police
Manuel M T Chakravarty [Thu, 17 Nov 2011 12:14:11 +0000 (23:14 +1100)] 
Warning police

8 years agoSpecial case dictionary abstraction and application during vectorisation
Manuel M T Chakravarty [Thu, 17 Nov 2011 11:33:49 +0000 (22:33 +1100)] 
Special case dictionary abstraction and application during vectorisation

8 years agoTabs -> spaces & other white space
Manuel M T Chakravarty [Tue, 15 Nov 2011 08:32:31 +0000 (19:32 +1100)] 
Tabs -> spaces & other white space

8 years agoFix #4211: No need to fixup stack using mangler on OSX
David Terei [Wed, 16 Nov 2011 03:21:34 +0000 (19:21 -0800)] 
Fix #4211: No need to fixup stack using mangler on OSX

We now manage the stack correctly on both x86 and i386, keeping
the stack align at (16n bytes - word size) on function entry
and at (16n bytes) on function calls. This gives us compatability
with LLVM and GCC.

8 years agoRemoving the default grouping clause from the SQL-like comprehension notation ;
George Giorgidze [Wed, 2 Nov 2011 22:43:36 +0000 (23:43 +0100)] 
Removing the default grouping clause from the SQL-like comprehension notation ;

8 years agoMerge /Users/benl/devel/ghc/ghc-head-devel
Ben Lippmeier [Thu, 17 Nov 2011 02:31:37 +0000 (13:31 +1100)] 
Merge /Users/benl/devel/ghc/ghc-head-devel

8 years agovectoriser: Use Sels2 type for vector of selectors in PDatas Sum2 instance
Ben Lippmeier [Thu, 17 Nov 2011 02:30:58 +0000 (13:30 +1100)] 
vectoriser: Use Sels2 type for vector of selectors in PDatas Sum2 instance

8 years agovectoriser: mkCast args the other way around, and fix warning
Ben Lippmeier [Thu, 17 Nov 2011 02:30:02 +0000 (13:30 +1100)] 
vectoriser: mkCast args the other way around, and fix warning

8 years agovectoriser: follow constraint solver changes in vectoriser
Ben Lippmeier [Thu, 17 Nov 2011 01:35:35 +0000 (12:35 +1100)] 
vectoriser: follow constraint solver changes in vectoriser

8 years agoMerge /Users/benl/devel/ghc/ghc-head-devel
Ben Lippmeier [Thu, 17 Nov 2011 01:34:57 +0000 (12:34 +1100)] 
Merge /Users/benl/devel/ghc/ghc-head-devel

Conflicts:
compiler/vectorise/Vectorise/Type/PRepr.hs

8 years agoRemove executable mode from some files
David Terei [Wed, 16 Nov 2011 18:17:23 +0000 (10:17 -0800)] 
Remove executable mode from some files

8 years agoTabs -> Spaces + formatting fixes
David Terei [Wed, 16 Nov 2011 21:37:34 +0000 (13:37 -0800)] 
Tabs -> Spaces + formatting fixes

8 years agoAdd Safe Haskell '-fwarn-safe', '-fwarn-unsafe', '-fno-safe-infer' flags
David Terei [Wed, 16 Nov 2011 21:22:27 +0000 (13:22 -0800)] 
Add Safe Haskell '-fwarn-safe', '-fwarn-unsafe', '-fno-safe-infer' flags

8 years agoFix mistake in user guide
David Terei [Tue, 15 Nov 2011 22:02:31 +0000 (14:02 -0800)] 
Fix mistake in user guide

8 years agoMove mkPiTypes back to Type, rename mkForAllArrowKinds to mkPiKinds
Jose Pedro Magalhaes [Wed, 16 Nov 2011 18:08:32 +0000 (18:08 +0000)] 
Move mkPiTypes back to Type, rename mkForAllArrowKinds to mkPiKinds

8 years agoChanges to the kind checker
Jose Pedro Magalhaes [Wed, 16 Nov 2011 16:27:17 +0000 (16:27 +0000)] 
Changes to the kind checker

We now always check against an expected kind. When we really don't know what
kind to expect, we match against a new meta kind variable.

Also, we are more explicit about tuple sorts:
  HsUnboxedTuple                  -> Produced by the parser
  HsBoxedTuple                    -> Certainly a boxed tuple
  HsConstraintTuple               -> Certainly a constraint tuple
  HsBoxedOrConstraintTuple        -> Could be a boxed or a constraint
                                  tuple. Produced by the parser only,
                                  disappears after type checking

8 years agoBetter documentation for AnyK
Jose Pedro Magalhaes [Tue, 15 Nov 2011 15:20:01 +0000 (15:20 +0000)] 
Better documentation for AnyK

8 years agoMinor fix, and transform a failure into a warning
Jose Pedro Magalhaes [Tue, 15 Nov 2011 10:07:10 +0000 (10:07 +0000)] 
Minor fix, and transform a failure into a warning

This warning has to be addressed later, though. It is only
relevant when using -XPolyKinds.

8 years agoBetter error messages when we know the expected kind in tcHsSigType
Jose Pedro Magalhaes [Mon, 14 Nov 2011 09:50:01 +0000 (09:50 +0000)] 
Better error messages when we know the expected kind in tcHsSigType

8 years agoAdd dependencies on all .hs-boot TyCons in rnTyClDecls
Jose Pedro Magalhaes [Mon, 14 Nov 2011 14:00:29 +0000 (14:00 +0000)] 
Add dependencies on all .hs-boot TyCons in rnTyClDecls

8 years agoMerge branch 'master' of http://darcs.haskell.org/ghc
Dimitrios Vytiniotis [Wed, 16 Nov 2011 17:40:43 +0000 (17:40 +0000)] 
Merge branch 'master' of darcs.haskell.org/ghc

8 years agoGHC gets a new constraint solver. More efficient and smaller in size.
Dimitrios Vytiniotis [Wed, 16 Nov 2011 16:12:48 +0000 (16:12 +0000)] 
GHC gets a new constraint solver. More efficient and smaller in size.

8 years agoGive a warning when -rtsopts/-with-rtsopts are used with -no-hs-main
Simon Marlow [Wed, 16 Nov 2011 15:51:44 +0000 (15:51 +0000)] 
Give a warning when -rtsopts/-with-rtsopts are used with -no-hs-main

8 years agoFix trashing of the masking state in STM (#5238)
Simon Marlow [Wed, 16 Nov 2011 14:21:49 +0000 (14:21 +0000)] 
Fix trashing of the masking state in STM (#5238)

8 years agoDoc changes following changes to the way -rtsopts works
Simon Marlow [Wed, 16 Nov 2011 15:40:40 +0000 (15:40 +0000)] 
Doc changes following changes to the way -rtsopts works

Mainly, -rtsopts doesn't work with -no-hs-main, and you have to do
something in your main() to get the effect of -rtsopts (and
-with-rtsopts).

8 years agoMake the --fast option to validate faster, and add --normal
Simon Marlow [Wed, 16 Nov 2011 13:00:55 +0000 (13:00 +0000)] 
Make the --fast option to validate faster, and add --normal

The --fast option now disables the following:
  - dynamic libs
  - bindist and bindisttest

Which knocks several minutes off validate for me, but it's still over
30 minutes using 5 cores on 64-bit Linux.

Usual caveats apply: if you're using --fast, then make sure you aren't
doing anything that might destabilise dynamic libs or binary dists.

8 years agoCreate parent directories when touching the object file (#5584)
Simon Marlow [Wed, 16 Nov 2011 10:27:02 +0000 (10:27 +0000)] 
Create parent directories when touching the object file (#5584)

8 years agoGenerate the C main() function when linking a binary (fixes #5373)
Simon Marlow [Tue, 15 Nov 2011 15:43:28 +0000 (15:43 +0000)] 
Generate the C main() function when linking a binary (fixes #5373)

Rather than have main() be statically compiled as part of the RTS, we
now generate it into the tiny C file that we compile when linking a
binary.

The main motivation is that we want to pass the settings for the
-rtsotps and -with-rtsopts flags into the RTS, rather than relying on
fragile linking semantics to override the defaults, which don't work
with DLLs on Windows (#5373).  In order to do this, we need to extend
the API for initialising the RTS, so now we have:

void hs_init_ghc (int *argc, char **argv[],   // program arguments
                  RtsConfig rts_config);      // RTS configuration

hs_init_ghc() can optionally be used instead of hs_init(), and allows
passing in configuration options for the RTS.  RtsConfig is a struct,
which currently has two fields:

typedef struct {
    RtsOptsEnabledEnum rts_opts_enabled;
    const char *rts_opts;
} RtsConfig;

but might have more in the future.  There is a default value for the
struct, defaultRtsConfig, the idea being that you start with this and
override individual fields as necessary.

In fact, main() was in a separate static library, libHSrtsmain.a.
That's now gone.

8 years agoAdd -fpedantic-bottoms, and document it
Simon Peyton Jones [Wed, 16 Nov 2011 14:03:30 +0000 (14:03 +0000)] 
Add -fpedantic-bottoms, and document it

I did a bit of refactoring (of course) at the same time.
See the discussion in Trac #5587.  Most of the real change
is in CoreArity.

8 years agoFormatting fix
Simon Peyton Jones [Wed, 16 Nov 2011 12:29:08 +0000 (12:29 +0000)] 
Formatting fix

8 years agoFix CaseIdentity optimisaion
Simon Peyton Jones [Wed, 16 Nov 2011 10:37:47 +0000 (10:37 +0000)] 
Fix CaseIdentity optimisaion

In fixing one bug I'd introduced another;
   case x of { T -> T; F -> F }
wasn't getting optmised!  Trivial to fix.

8 years agoFurther wibbles to calcUnfoldingGuidance, with documentation of same
Simon Peyton Jones [Wed, 16 Nov 2011 10:36:50 +0000 (10:36 +0000)] 
Further wibbles to calcUnfoldingGuidance, with documentation of same

I'd gotten into a state in which top-level x = y bindings weren't
getting inlined!

8 years agoReplace unused variable with "_"
Simon Peyton Jones [Tue, 15 Nov 2011 17:52:22 +0000 (17:52 +0000)] 
Replace unused variable with "_"

8 years agoMerge branch 'master' of http://darcs.haskell.org/ghc
Simon Peyton Jones [Tue, 15 Nov 2011 17:19:13 +0000 (17:19 +0000)] 
Merge branch 'master' of darcs.haskell.org/ghc

8 years agoImprove documentation of SPECIALISE pragma
Simon Peyton Jones [Tue, 15 Nov 2011 17:18:02 +0000 (17:18 +0000)] 
Improve documentation of SPECIALISE pragma

8 years agoFix Trac #5628: equality on data types with no constructors
Simon Peyton Jones [Tue, 15 Nov 2011 17:17:28 +0000 (17:17 +0000)] 
Fix Trac #5628: equality on data types with no constructors

8 years agofurther fixes to the #5505 fix.
Simon Marlow [Tue, 15 Nov 2011 15:44:53 +0000 (15:44 +0000)] 
further fixes to the #5505 fix.