Re: Integrating tools with Cabal

* Mikhail Glushenkov [2012-11-09 19:36:58+0100]

Hi,

On Fri, Nov 9, 2012 at 6:25 PM, Roman Cheplyaka wrote:

...

searching for ld in path. found ld at /usr/bin/ld ("/home/feuerbach/bin/gen-iface",["-c","/tmp/17368.c","-o","/tmp/17368.o"]) ("/usr/bin/ld",["-x","-r","/tmp/17368.o","-o","/tmp/17369.o"]) /usr/bin/ld returned ExitFailure 1 with error message: /tmp/17368.o: file not recognized: File truncated Setup: /tmp/17369.o: does not exist

I looked into this - this is caused by configureLd' in Distribution.Simple.GHC. This function tries to determine whether ld supports the -x flag by compiling a short C program ("int foo() {}") with ghc (apparently ghc supports .c files as input).

Since ghc just invokes gcc when given .c files, and you're trying to mimic ghc, I think it makes sense to support this behaviour. If you think that this is ugly, you can put all such hacky code into a separate driver program, and leave your analyser unmodified.

At this point I'd rather work on a separate module in spirit of Distribution.Simple.*HC and hope that Cabal devs wouldn't mind to accept it. This will be essentially Cabal support for haskell-suite[1]. [1]: http://www.nbroberg.se/haskell-suite/ Of course, fixing #57 would be ideal, but given the complexity of existing modules I guess we're going to live with this for quite some time. Roman