ghc.git
6 years agoFix accidentally commited debug form of FIND_LLVM_PROG.
David Terei [Mon, 4 Mar 2013 01:59:42 +0000 (17:59 -0800)] 
Fix accidentally commited debug form of FIND_LLVM_PROG.

6 years agoMerge branch 'master' of darcs.haskell.org:/srv/darcs//ghc
Ian Lynagh [Mon, 4 Mar 2013 00:40:38 +0000 (00:40 +0000)] 
Merge branch 'master' of darcs.haskell.org:/srv/darcs//ghc

6 years agoComments and type signatures only
Simon Peyton Jones [Sun, 3 Mar 2013 23:04:35 +0000 (23:04 +0000)] 
Comments and type signatures only

6 years agoTreat equalities with incompatible kinds as "irreducible" constraints
Simon Peyton Jones [Sun, 3 Mar 2013 23:04:09 +0000 (23:04 +0000)] 
Treat equalities with incompatible kinds as "irreducible" constraints

Originally we had the invariant that CTyEqCan and CFunEqCan have LHS
and RHS with compatible kinds.  This is important because if they have
different kinds, then a substitution using the CTyEqCan can give rise
to an ill-kinded type, which in turn makes typeKind crash, and this
led to Trac #7696.  (The possibility of this happening really only
occurred when we introduced kind polymorphism.)

I thought at first this was going to be really awkward to solve, but
happily it turned out to be easy.  We already have CIrredEvCan
constraints, which are "stuck"; we can't use them and we can't solve
them.  Yet. After some substitution from solving other constraints we
may be able to make progress.

So for equality constraints where the LHS and RHS don't have compatible kinds
(although perhaps not YET compatible, eg k and *, just needing to
unify k := *), we now generate a CIrredEvCan, plus the necessary kind
equality constraint.

This entailed some refactoring of course, but only in TcCanonical.  In
particular, the emitKindConstraint code has gone, in favour of a kind
check in canEqLeaf.  See Note [Equalities with incompatible kinds] in
TcCanonical, and Note [CIrredEvCan constraints] in TcRnTypes

6 years agoMake sure that Constraint is unrelated to other kinds in tcIsSubKind
Simon Peyton Jones [Sun, 3 Mar 2013 22:49:23 +0000 (22:49 +0000)] 
Make sure that Constraint is unrelated to other kinds in tcIsSubKind

This was causing the bug reported in Trac #7697

6 years agoAutomatically add the $(exeext) to program names
Ian Lynagh [Sun, 3 Mar 2013 22:49:41 +0000 (22:49 +0000)] 
Automatically add the $(exeext) to program names

We now define _PROGNAME, and _PROG is automatically defined with
$(exeext). This will shortly automatically use the right exeext
depending on what stage it is being compiled with (exeext may be
different for different stages when cross-compiling).

6 years agoUse the automatic variables for touchy too
Ian Lynagh [Sun, 3 Mar 2013 22:37:57 +0000 (22:37 +0000)] 
Use the automatic variables for touchy too

6 years agoAutomatically define _INPLACE variables for perl programs too
Ian Lynagh [Sun, 3 Mar 2013 22:29:57 +0000 (22:29 +0000)] 
Automatically define _INPLACE variables for perl programs too

and use them for split

6 years agoinstallation fixes
Ian Lynagh [Sun, 3 Mar 2013 21:26:25 +0000 (21:26 +0000)] 
installation fixes

6 years agoRemove the _INSTALL_IN support in the build system
Ian Lynagh [Sun, 3 Mar 2013 21:09:58 +0000 (21:09 +0000)] 
Remove the _INSTALL_IN support in the build system

It doesn't seem to do anything that _INSTALL and _INSTALL_INPLACE
can't do.

6 years agoUse unlit_INPLACE rather than UNLIT in the build system
Ian Lynagh [Sun, 3 Mar 2013 20:27:51 +0000 (20:27 +0000)] 
Use unlit_INPLACE rather than UNLIT in the build system

6 years agoUse more of the automatically generated INPLACE variables
Ian Lynagh [Sun, 3 Mar 2013 20:20:06 +0000 (20:20 +0000)] 
Use more of the automatically generated INPLACE variables

6 years agoDefine $(prog)_INPLACE variables automatically, and start using them
Ian Lynagh [Sun, 3 Mar 2013 18:49:23 +0000 (18:49 +0000)] 
Define $(prog)_INPLACE variables automatically, and start using them

6 years agoWhitespace only
Ian Lynagh [Sun, 3 Mar 2013 18:32:10 +0000 (18:32 +0000)] 
Whitespace only

6 years agoExport dopt_set, dopt_unset
Ian Lynagh [Sun, 3 Mar 2013 17:35:02 +0000 (17:35 +0000)] 
Export dopt_set, dopt_unset

We don't need them, but GHC API users might.

6 years agoRemove some unused build system variables
Ian Lynagh [Sun, 3 Mar 2013 17:33:11 +0000 (17:33 +0000)] 
Remove some unused build system variables

6 years agoFix missing backticks and parentheses in error messages; fixes #7734
Ian Lynagh [Sun, 3 Mar 2013 16:30:23 +0000 (16:30 +0000)] 
Fix missing backticks and parentheses in error messages; fixes #7734

Patch from monoidal

6 years agoAdd missing parenthesis
Ian Lynagh [Sun, 3 Mar 2013 13:43:49 +0000 (13:43 +0000)] 
Add missing parenthesis

6 years agoFix build when dph is not in the tree; fixes #7733
Ian Lynagh [Sun, 3 Mar 2013 13:41:46 +0000 (13:41 +0000)] 
Fix build when dph is not in the tree; fixes #7733

Patch from Stephen Blackheath.

6 years agoAdd PACKAGES_STAGE0 to build-dirs too; fixes #7700
Ian Lynagh [Sun, 3 Mar 2013 13:35:41 +0000 (13:35 +0000)] 
Add PACKAGES_STAGE0 to build-dirs too; fixes #7700

This is sometimes needed when cross-compiling, as some packages may be
built in stage 0 but not stage 1.

In order to make everything work out, this also removes the requirement
that the build-dirs are in dependency order

6 years agoTypofix.
Edward Z. Yang [Sun, 3 Mar 2013 00:13:04 +0000 (16:13 -0800)] 
Typofix.

Signed-off-by: Edward Z. Yang <ezyang@mit.edu>
6 years agoUpdate source pointer.
Edward Z. Yang [Sun, 3 Mar 2013 00:05:00 +0000 (16:05 -0800)] 
Update source pointer.

Signed-off-by: Edward Z. Yang <ezyang@mit.edu>
6 years agoMake codeOutput and friends return the filename that they have created
Ian Lynagh [Sat, 2 Mar 2013 23:49:41 +0000 (23:49 +0000)] 
Make codeOutput and friends return the filename that they have created

6 years agoMake codeOutput panic if hscTarget is HscInterpreted
Ian Lynagh [Sat, 2 Mar 2013 21:30:29 +0000 (21:30 +0000)] 
Make codeOutput panic if hscTarget is HscInterpreted

Doesn't look like it causes any failures in the full test suite, so
presumably it's fine!

6 years agoSimplify the definition of getOutputFilename
Ian Lynagh [Sat, 2 Mar 2013 19:16:38 +0000 (19:16 +0000)] 
Simplify the definition of getOutputFilename

6 years agoDon't redundantly set hscTarget to the same value it already was
Ian Lynagh [Sat, 2 Mar 2013 17:39:43 +0000 (17:39 +0000)] 
Don't redundantly set hscTarget to the same value it already was

6 years agoAdd iOS specific module layout entry to LLVM codegen; fixes #7721
Ian Lynagh [Sat, 2 Mar 2013 15:12:03 +0000 (15:12 +0000)] 
Add iOS specific module layout entry to LLVM codegen; fixes #7721

Patch from Stephen Blackheath

6 years agoAdd OSiOS, and define and use platformUsesFrameworks; part of #7720
Ian Lynagh [Sat, 2 Mar 2013 13:57:10 +0000 (13:57 +0000)] 
Add OSiOS, and define and use platformUsesFrameworks; part of #7720

6 years agoios fix from Stephen Blackheath; part of #7707
Ian Lynagh [Sat, 2 Mar 2013 13:15:21 +0000 (13:15 +0000)] 
ios fix from Stephen Blackheath; part of #7707

Apparently ios doesn't understand .type %function.

6 years agoUse 'r11' rather than 'fp' on Arm; part of #7707
Ian Lynagh [Sat, 2 Mar 2013 13:05:05 +0000 (13:05 +0000)] 
Use 'r11' rather than 'fp' on Arm; part of #7707

They're both the same register, and Linux seems happy with both,
but ios only accepts r11.

6 years agoUse .globl rather than .global on arm; part of #7707
Ian Lynagh [Sat, 2 Mar 2013 12:45:26 +0000 (12:45 +0000)] 
Use .globl rather than .global on arm; part of #7707

Apparently ios only understands .globl, but Linux appears to understand
both.

6 years agoGive hsc2hs different options in different stages; fixes #7705
Ian Lynagh [Sat, 2 Mar 2013 12:22:20 +0000 (12:22 +0000)] 
Give hsc2hs different options in different stages; fixes #7705

Based on patch from Stephen Blackheath.

6 years agoRemove some more unused build system code
Ian Lynagh [Sat, 2 Mar 2013 12:12:26 +0000 (12:12 +0000)] 
Remove some more unused build system code

6 years agoRemove the unused REAL_SHELL variable in the build system
Ian Lynagh [Sat, 2 Mar 2013 02:04:09 +0000 (02:04 +0000)] 
Remove the unused REAL_SHELL variable in the build system

6 years agoFollow Windows -> Windows_Host change in config.mk.in
Ian Lynagh [Sat, 2 Mar 2013 02:03:34 +0000 (02:03 +0000)] 
Follow Windows -> Windows_Host change in config.mk.in

6 years agoRemove some more unused build system variables
Ian Lynagh [Sat, 2 Mar 2013 01:26:22 +0000 (01:26 +0000)] 
Remove some more unused build system variables

6 years agoRemove a couple more single-use variables
Ian Lynagh [Sat, 2 Mar 2013 01:25:21 +0000 (01:25 +0000)] 
Remove a couple more single-use variables

6 years agoRemove unused values from Config.hs
Ian Lynagh [Sat, 2 Mar 2013 01:23:45 +0000 (01:23 +0000)] 
Remove unused values from Config.hs

6 years agoRemove some single-use build system variables
Ian Lynagh [Sat, 2 Mar 2013 01:20:14 +0000 (01:20 +0000)] 
Remove some single-use build system variables

6 years agosmall build system tweak
Ian Lynagh [Sat, 2 Mar 2013 01:11:58 +0000 (01:11 +0000)] 
small build system tweak

6 years agoLATEX_DOCS is no longer used
Ian Lynagh [Sat, 2 Mar 2013 00:05:22 +0000 (00:05 +0000)] 
LATEX_DOCS is no longer used

6 years agoRemove some more unused build system variables
Ian Lynagh [Fri, 1 Mar 2013 23:45:30 +0000 (23:45 +0000)] 
Remove some more unused build system variables

6 years agoMake ghc.mk a little prettier
Ian Lynagh [Fri, 1 Mar 2013 23:41:14 +0000 (23:41 +0000)] 
Make ghc.mk a little prettier

6 years agoFix haddocking after the build system changes
Ian Lynagh [Fri, 1 Mar 2013 22:43:30 +0000 (22:43 +0000)] 
Fix haddocking after the build system changes

6 years agoFix when terminfo is built; fixes #7700
Ian Lynagh [Fri, 1 Mar 2013 22:29:19 +0000 (22:29 +0000)] 
Fix when terminfo is built; fixes #7700

We now use the 'host' to determine whether stage0 builds terminfo,
and 'target' to determine whether stage1 does.

We don't build it if the platform is either Windows or ios.

6 years agoRename $(Windows) to $(Windows_Host)
Ian Lynagh [Fri, 1 Mar 2013 21:46:02 +0000 (21:46 +0000)] 
Rename $(Windows) to $(Windows_Host)

6 years agoRemove some unused variables
Ian Lynagh [Fri, 1 Mar 2013 21:31:29 +0000 (21:31 +0000)] 
Remove some unused variables

6 years agoChange how the build system handles packages
Ian Lynagh [Fri, 1 Mar 2013 16:34:33 +0000 (16:34 +0000)] 
Change how the build system handles packages

This makes the build system a little simpler, and in particular
will make it easier to handle the changes needed for cross-compilation.

6 years agoComment the fix to Trac #7702
Simon Peyton Jones [Fri, 1 Mar 2013 17:55:07 +0000 (17:55 +0000)] 
Comment the fix to Trac #7702

6 years agoMerge branch 'master' of darcs.haskell.org:/home/darcs/ghc
Simon Peyton Jones [Fri, 1 Mar 2013 17:52:54 +0000 (17:52 +0000)] 
Merge branch 'master' of darcs.haskell.org:/home/darcs/ghc

6 years agoUpdate to binary 0.7.0.0
Ian Lynagh [Fri, 1 Mar 2013 13:59:21 +0000 (13:59 +0000)] 
Update to binary 0.7.0.0

6 years agoFix memory leak in CoreWriter (Trac #7702)
Andrew Farmer [Mon, 18 Feb 2013 09:32:35 +0000 (03:32 -0600)] 
Fix memory leak in CoreWriter (Trac #7702)

6 years agoMake the #includes a little more correct in OSThreads.c
Ian Lynagh [Fri, 1 Mar 2013 00:52:33 +0000 (00:52 +0000)] 
Make the #includes a little more correct in OSThreads.c

6 years agoFix sys/cpuset.h on FreeBSD 9.1; patch from kazu-yamamoto; fixes #7708
Ian Lynagh [Fri, 1 Mar 2013 00:50:47 +0000 (00:50 +0000)] 
Fix sys/cpuset.h on FreeBSD 9.1; patch from kazu-yamamoto; fixes #7708

6 years agoSmall refactoring: Move the end-of-pipeline move into pipeLoop
Ian Lynagh [Thu, 28 Feb 2013 13:09:24 +0000 (13:09 +0000)] 
Small refactoring: Move the end-of-pipeline move into pipeLoop

6 years agoSmall refactoring: Don't pass redundant stop_phase to runPipeline
Ian Lynagh [Thu, 28 Feb 2013 13:00:12 +0000 (13:00 +0000)] 
Small refactoring: Don't pass redundant stop_phase to runPipeline

6 years agoFix some m4/shell syntax in aclocal.m4, spotted by Jan Stolarek
Ian Lynagh [Thu, 28 Feb 2013 12:52:59 +0000 (12:52 +0000)] 
Fix some m4/shell syntax in aclocal.m4, spotted by Jan Stolarek

[ ] is m4 syntax, so we need to use 'test foo' rather than '[ foo ]'.

6 years agoSmall refactoring: Don't pass a redundant output_spec around
Ian Lynagh [Thu, 28 Feb 2013 02:40:21 +0000 (02:40 +0000)] 
Small refactoring: Don't pass a redundant output_spec around

6 years agoInline get_output_fn
Ian Lynagh [Thu, 28 Feb 2013 02:38:37 +0000 (02:38 +0000)] 
Inline get_output_fn

It's now only used once

6 years agoSmall refactoring
Ian Lynagh [Thu, 28 Feb 2013 02:21:19 +0000 (02:21 +0000)] 
Small refactoring

pipeLoop now returns the final dflags, rather than having
runPipeline' extract it from the end state.

6 years agoTypo
Gabor Greif [Wed, 27 Feb 2013 23:55:26 +0000 (00:55 +0100)] 
Typo

6 years agoMinor doc typofix.
Edward Z. Yang [Wed, 27 Feb 2013 22:22:55 +0000 (14:22 -0800)] 
Minor doc typofix.

Signed-off-by: Edward Z. Yang <ezyang@mit.edu>
6 years agoClarify comment
Gabor Greif [Tue, 26 Feb 2013 21:42:50 +0000 (22:42 +0100)] 
Clarify comment

6 years agoUpdate submodule
Gabor Greif [Tue, 26 Feb 2013 21:42:19 +0000 (22:42 +0100)] 
Update submodule

6 years agoFix a TODO in the compiler
Ian Lynagh [Tue, 26 Feb 2013 18:57:39 +0000 (18:57 +0000)] 
Fix a TODO in the compiler

AnnProvenance now has Functor, Foldable, Traversable instances.

6 years agoFix parsing of pragmas containing unicode characters; fixes #7671
Ian Lynagh [Tue, 26 Feb 2013 01:27:43 +0000 (01:27 +0000)] 
Fix parsing of pragmas containing unicode characters; fixes #7671

6 years agoOnly emit %write_barrier primitive for THREADED_RTS
Gabor Greif [Mon, 25 Feb 2013 22:03:23 +0000 (23:03 +0100)] 
Only emit %write_barrier primitive for THREADED_RTS

6 years agoChange how unboxed tuples are lexed; fixes #7627
Ian Lynagh [Mon, 25 Feb 2013 19:02:57 +0000 (19:02 +0000)] 
Change how unboxed tuples are lexed; fixes #7627

(# is now always a lexeme, even if followed by a symbol.

6 years agoSplit SettingsCCompilerFlags into non-link and link portions
Gabor Greif [Mon, 25 Feb 2013 17:53:31 +0000 (18:53 +0100)] 
Split SettingsCCompilerFlags into non-link and link portions

This fixes certain older GCCs which do not accept link options when assembling or compiling:

  ppc_85xx-gcc: --hash-size=31: linker input file unused because linking not done
  ppc_85xx-gcc: --reduce-memory-overheads: linker input file unused because linking not done

and diagnose this to stderr.

6 years agoFix autoconf code to find LLVM tools.
Geoffrey Mainland [Mon, 25 Feb 2013 08:06:46 +0000 (08:06 +0000)] 
Fix autoconf code to find LLVM tools.

The loop exit condition was testing ${LLC} instead of $1, which was
incorrect. While I'm here, quote the path being tested since it may contain
spaces (e.g. on Windows), and don't search paths that don't exist, which
eliminates un-useful error messages from find.

6 years agoFix the compiler plugin example; patch from edsko; fixes #7682
Ian Lynagh [Sun, 24 Feb 2013 16:20:04 +0000 (16:20 +0000)] 
Fix the compiler plugin example; patch from edsko; fixes #7682

6 years agoUse unicode quote characters in error messages etc; fixes #2507
Ian Lynagh [Sun, 24 Feb 2013 00:26:07 +0000 (00:26 +0000)] 
Use unicode quote characters in error messages etc; fixes #2507

We only use the unicode characters if the locale supports them.

6 years agox86: promote arguments to C functions according to the ABI (#7383)
Simon Marlow [Wed, 20 Feb 2013 11:43:33 +0000 (11:43 +0000)] 
x86: promote arguments to C functions according to the ABI (#7383)

I don't think the x86-64 version is quite right, but this ought to be
enough to pass cgrun071.

This code is terrible and needs a complete refactor.  There's a lot of
duplication, and we ought to be specifying the ABI in a much more
abstract way (like LLVM).

6 years agoTurn -dynamic-too off in the build system for now
Ian Lynagh [Sat, 23 Feb 2013 13:16:01 +0000 (13:16 +0000)] 
Turn -dynamic-too off in the build system for now

6 years agoTweak the sed command so that it works on OSX
Ian Lynagh [Sat, 23 Feb 2013 02:33:20 +0000 (02:33 +0000)] 
Tweak the sed command so that it works on OSX

Using { } with sed on OS X requires using newlines rather than
semicolons. I've rewritten the sed so that no { } are necessary.

6 years agoAdd another chapter to the hi-rule novel
Ian Lynagh [Fri, 22 Feb 2013 23:47:26 +0000 (23:47 +0000)] 
Add another chapter to the hi-rule novel

6 years agoMake the build system use -dynamic-too if both way v and dyn are being built
Ian Lynagh [Fri, 22 Feb 2013 23:36:53 +0000 (23:36 +0000)] 
Make the build system use -dynamic-too if both way v and dyn are being built

6 years agoAdd separate rules for all .hi files, rather than using %.hi style
Ian Lynagh [Fri, 22 Feb 2013 22:18:50 +0000 (22:18 +0000)] 
Add separate rules for all .hi files, rather than using %.hi style

If a file is created by a %.hi rule, and the actual filename isn't
mentioned in the makefiles, then make will treat it as an 'intermediate
file' and delete it when it is finished.

We'd been lucky so far that .hi files weren't actually being built due
to our rules (but rather, as side-effects of the .o rules). However,
when using -dynamic-too to build, we had a rule
    $1/$2/build/%.$$(dyn_osuf): $1/$2/build/%.$$(v_hisuf)
which meant that building a .dyn_o could cause the rule for the
corresponding .hi to be used, and the .hi may then be deleted later on.
This was exacerbated by a bug in GNU make 3.81 which caused make to
enter an infinite loop if running in parallel mode:
    http://lists.gnu.org/archive/html/bug-make/2013-02/msg00020.html

Adding
    .SECONDARY:
would stop make from deleting the intermediate files. However, this
caused make to take a pathologically long time (it appeared to be
live-locked for 2 hours before I killed it) with our build system.

This patch instead creates lines like
    $(eval $(call hi-rule,libraries/base/dist-install/build/Unsafe/Coerce.dyn_hi libraries/base/dist-install/build/Unsafe/Coerce.hi  : %hi: %o  libraries/base/Unsafe/Coerce.hs))
in the .depend files, which results in a rule like
    libraries/base/dist-install/build/Unsafe/Coerce.dyn_hi libraries/base/dist-install/build/Unsafe/Coerce.hi  : %hi: %o  libraries/base/Unsafe/Coerce.hs ;
which, as the files are now all named in the makefiles, means they are
no longer intermediate files so do not get deleted.

6 years agoFix displayed string
Gabor Greif [Fri, 22 Feb 2013 17:13:13 +0000 (18:13 +0100)] 
Fix displayed string

6 years agoAlways quote arguments to removeFiles (fixes Windows install).
Geoffrey Mainland [Fri, 22 Feb 2013 15:26:55 +0000 (15:26 +0000)] 
Always quote arguments to removeFiles (fixes Windows install).

Thanks to Daniel Pratt <colorblinddad@gmail.com> for pointing out the failure
and fix.

6 years agoMerge branch 'master' of darcs.haskell.org:/srv/darcs//ghc
Ian Lynagh [Thu, 21 Feb 2013 16:49:39 +0000 (16:49 +0000)] 
Merge branch 'master' of darcs.haskell.org:/srv/darcs//ghc

6 years agoBuild fixes following build system refactoring
Ian Lynagh [Thu, 21 Feb 2013 15:58:53 +0000 (15:58 +0000)] 
Build fixes following build system refactoring

6 years agoSeparate the non-way-specific hs-suffix rules
Ian Lynagh [Thu, 21 Feb 2013 15:32:46 +0000 (15:32 +0000)] 
Separate the non-way-specific hs-suffix rules

This means we don't define them multiple times

6 years agoSplit non-way-specific variables off from distdir-way-opts into distdir-opts
Ian Lynagh [Thu, 21 Feb 2013 15:05:22 +0000 (15:05 +0000)] 
Split non-way-specific variables off from distdir-way-opts into distdir-opts

The hsc2hs, alex and happy options variables are now also
non-way-specific, as the files are shared between all ways.

6 years agoRename hs-suffix-rules-srcdir to hs-suffix-way-rules-srcdir
Ian Lynagh [Thu, 21 Feb 2013 14:40:14 +0000 (14:40 +0000)] 
Rename hs-suffix-rules-srcdir to hs-suffix-way-rules-srcdir

6 years agoRename hs-suffix-rules to hs-suffix-way-rules
Ian Lynagh [Thu, 21 Feb 2013 14:31:34 +0000 (14:31 +0000)] 
Rename hs-suffix-rules to hs-suffix-way-rules

6 years agoSmall build system refactoring; no functional changes
Ian Lynagh [Thu, 21 Feb 2013 14:17:51 +0000 (14:17 +0000)] 
Small build system refactoring; no functional changes

6 years agoFix #7710: do not AutoDeriveTypeable for synonyms and type families
Jose Pedro Magalhaes [Thu, 21 Feb 2013 09:33:31 +0000 (09:33 +0000)] 
Fix #7710: do not AutoDeriveTypeable for synonyms and type families

6 years agoMerge branch 'master' of darcs.haskell.org:/srv/darcs//ghc
Ian Lynagh [Thu, 21 Feb 2013 01:31:50 +0000 (01:31 +0000)] 
Merge branch 'master' of darcs.haskell.org:/srv/darcs//ghc

6 years agoSome -dynamic-too fixes
Ian Lynagh [Wed, 20 Feb 2013 14:08:39 +0000 (14:08 +0000)] 
Some -dynamic-too fixes

6 years agoPackage 'binary' is not a boot package any more
Gabor Greif [Wed, 20 Feb 2013 21:27:10 +0000 (22:27 +0100)] 
Package 'binary' is not a boot package any more

6 years agoUpdate submodules
Gabor Greif [Wed, 20 Feb 2013 20:46:50 +0000 (21:46 +0100)] 
Update submodules

6 years agoBetter handling of find llvm tools. Use IFS as opposed to more hacky tr
David Terei [Wed, 20 Feb 2013 12:05:50 +0000 (04:05 -0800)] 
Better handling of find llvm tools. Use IFS as opposed to more hacky tr
approach. This way can handle spaces in paths.

6 years agoFix bug in setNumCapabilities
Simon Marlow [Wed, 20 Feb 2013 10:36:25 +0000 (10:36 +0000)] 
Fix bug in setNumCapabilities

We were changing n_capabilities after we had released the
Capabilities, which lead to a range of interesting crashes.  This
should fix test failures in setnumcapabilities001.

6 years agoRemove printExceptionAndWarnings (deprecated since 7.2)
Ian Lynagh [Wed, 20 Feb 2013 01:05:58 +0000 (01:05 +0000)] 
Remove printExceptionAndWarnings (deprecated since 7.2)

6 years agoAdd a "deprecated in" pragma
Ian Lynagh [Wed, 20 Feb 2013 01:03:20 +0000 (01:03 +0000)] 
Add a "deprecated in" pragma

6 years agoMerge branch 'master' of darcs.haskell.org:/srv/darcs//ghc
Ian Lynagh [Tue, 19 Feb 2013 17:58:01 +0000 (17:58 +0000)] 
Merge branch 'master' of darcs.haskell.org:/srv/darcs//ghc

6 years agoRemove gblock and gunblock
Ian Lynagh [Tue, 19 Feb 2013 17:07:16 +0000 (17:07 +0000)] 
Remove gblock and gunblock

6 years agoFix segfault in retainer profiling when using multiple cores (#5909)
Simon Marlow [Tue, 19 Feb 2013 09:58:31 +0000 (09:58 +0000)] 
Fix segfault in retainer profiling when using multiple cores (#5909)

Thanks to @akio on the ticket for the diagnosis and the patch.  I
modified the comments a bit.

6 years agofix some dependencies
Simon Marlow [Mon, 18 Feb 2013 14:33:02 +0000 (14:33 +0000)] 
fix some dependencies