[Patches] Various things related to building .app bundles

[Dan Knapp]

9 patches for repository http://darcs.haskell.org/cabal:

Fri May 27 19:25:44 EDT 2011  Dan Knapp <dankna at gmail.com>
  * Pursuant to Cabal ticket #850.  Fix for "C" headers specified with
includes: which are actually ObjectiveC headers (or some other
dialect).  Works by asking the C compiler which source files include
which headers, then inferring the dialects of those source files from
their filenames, combining the resulting dialects for each header with
a monoid, and mapping the final dialect back onto a filename extension
for the dummy file to be used in the foreign-library tests.

Sat May 28 08:15:56 EDT 2011  Dan Knapp <dankna at gmail.com>
  * One-line fix to the prior patch for ticket #850; I had omitted a
case in my declaration for the Monoid CDialect instance.

Sat May 28 08:30:22 EDT 2011  Dan Knapp <dankna at gmail.com>
  * Pursuant to ticket #504, expand the check for a known extension on
the file provided to main-is: to also allow known C-language
extensions.  This uses my Distribution.Simple.CCompiler from the
ticket #850 patch so that the information on known extensions isn't
duplicated.  The ticket #504 discussion suggests that ghc needs
-no-hs-main in this situation, but that appears to no longer be the
case; also see discussion on GHC ticket #2459.  There might still be
an issue if the entry point is WinMain() instead of main(), but I'm
not really set up to test that.

Sat May 28 12:19:21 EDT 2011  Dan Knapp <dankna at gmail.com>
  * Change to my earlier patch for ticket #504, executables with a C
main function.  I swear this worked for me in my test case, but I
found that it doesn't work now, and I can't see how it ever did.  The
issue is that, when looking for the source file, the new C extensions
weren't known so it didn't come up.  I added knowledge about them by
making Distribution.Simple.PreProcess depend on
Distribution.Simple.CCompiler.

Sat May 28 20:56:14 EDT 2011  Dan Knapp <dankna at gmail.com>
  * Enhancement as requested by my ticket #851: a c-source-dirs field
in BuildInfo.  Implemented for configure, build, and sdist, which
should be everything.

Sat May 28 22:06:06 EDT 2011  Dan Knapp <dankna at gmail.com>
  * My fix for #851 actually broke my earlier fix for #850; this patch
corrects things so they play nice together.

Sun May 29 04:55:44 EDT 2011  Dan Knapp <dankna at gmail.com>
  * Fix the regression described in #852 and caused by my patch for #850.

Sun May 29 21:23:50 EDT 2011  Dan Knapp <dankna at gmail.com>
  * Changes to handle building C source after Haskell source when
making mixed-language executables, pursuant to ticket #852, so that C
source can include *_stub.h files which are generated by the Haskell
compilation.  (This should already work in the library case.)  Since
ghc --make can't by itself infer the modules that need to be built
when it's passed a .c file, we need to tell it.

Tue May 31 04:43:24 EDT 2011  Dan Knapp <dankna at gmail.com>
  * Fix for ticket #853, passing the appropriate ObjC-GC flag to ghc
and gcc when it's needed, based on a new BuildInfo field objc-gc: with
the values Disabled (passes nothing), Optional, and Mandatory.


-- 
Dan Knapp
"An infallible method of conciliating a tiger is to allow oneself to
be devoured." (Konrad Adenauer)
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Patches
Type: application/octet-stream
Size: 102148 bytes
Desc: not available
URL: <http://www.haskell.org/pipermail/cabal-devel/attachments/20110602/6b6a5acd/attachment-0001.obj>