On Thu, Oct 04, 2012 at 01:23:42PM +0100, Ian Lynagh wrote: > On Mon, Oct 01, 2012 at 06:29:13PM +0200, Simon Hengel wrote: > > On Mon, Oct 01, 2012 at 04:05:24PM +0200, Simon Hengel wrote: > > > On Sat, Sep 01, 2012 at 02:09:28PM +0100, Ian Lynagh wrote: > > > > On Fri, Aug 31, 2012 at 05:58:13PM +0200, Simon Hengel wrote: > > > > > > > > > > The library component of Haddock is still not build with GHC. Would > > > > > it > > > > > be sane to build/include the library component of Haddock with GHC, so > > > > > that there is only one "build flavor" of Haddock? > > > > > > > > I suspect the current setup predates the ability for executables to > > > > depend on libraries in the same Cabal package. I'd suggest that the > > > > haddock executable should be changed to depend on the haddock library, > > > > and as far as I can see there is no reason why that should cause > > > > problems in the GHC build. > > > > > > Ok, I just changed that [1] (on branch ghc-7.6). Now the executable > > > depends on the library. I'm validating it right now; if there are no > > > issues I'll push it. > > > > After making an other minor modification to the cabal file [1] it fails > > with: > > > > ghc-cabal: XXX ghc-cabal can't handle more than one buildinfo yet > > > > Would this require modifications to ghc-cabal? > > Ah, hmm. > > I think it would be easiest to use 2 dist directories: In dist-lib, > build the library only, and in dist-exe build the executable only.
I now only depend on the library when we are not in-ghc-tree. It's still an improvement for two reasons: * All modules are now compiled, even in-ghc-tree, so that the library (hopefully) won't break again. * The build is much faster if we are not in-ghc-tree So I think this is properly addressed. Cheers, Simon _______________________________________________ Cvs-ghc mailing list Cvs-ghc@haskell.org http://www.haskell.org/mailman/listinfo/cvs-ghc
