ghc-blaze-html ghc-citeproc-hs ghc-hamlet ghc-json ghc-ltk ghc-MonadCatchIO-mtl ghc-pandoc-types ghc-texmath
I *think* these are false positives or possibly a yum corner-case dependency issue.
Is it perhaps because of prepending "0:" epoch before the hashes?
At least I am not able to reproduce: all the packages install fine for me.
Jens
ghc-blaze-html ghc-citeproc-hs ghc-hamlet ghc-json ghc-ltk ghc-MonadCatchIO-mtl ghc-pandoc-types ghc-texmath
Doh, ok I finally get that this is a multilib issue!
Basically ghc is not multilib in Fedora but because of the way we compose trees the presence of ghc-*-devel packages induces both archs to be installed on x86_64. However the ghc-*-prof packages (which are now being merged into and obsoleted by newer ghc-*-devel) are not multilib, and hence this new problem.
We should really just get rid of ghc-*.i686 from x86_64 trees, IMO, or failing that force repoclosure by including both archs of ghc-*-prof too. (Note that F16 no longer has any ghc-*-prof packages.)
Jens
haskell-devel@lists.fedoraproject.org