[Pkg-haskell-maintainers] Bug#752460: haskell-nettle: FTBFS for mips/mipsel due to lack of GHCi support

[Colin Watson]

On Mon, Jun 23, 2014 at 05:48:00PM +0000, Dejan Latinovic wrote:
> While trying to build haskell-nettle on mips/mipsel architecture,
> build fails on testing with an error:
> 
> > Building nettle-0.1.0...
> > Preprocessing library nettle-0.1.0...
> > [ 1 of 11] Compiling Crypto.Nettle.Hash.Types ( src/Crypto/Nettle/Hash/Types.hs, dist-ghc/build/Crypto/Nettle/Hash/Types.o )
> > [ 2 of 11] Compiling Crypto.Nettle.KeyedHash ( src/Crypto/Nettle/KeyedHash.hs, dist-ghc/build/Crypto/Nettle/KeyedHash.o )
> > [ 3 of 11] Compiling Crypto.Nettle.HMAC ( src/Crypto/Nettle/HMAC.hs, dist-ghc/build/Crypto/Nettle/HMAC.o )
> > [ 4 of 11] Compiling Nettle.Utils     ( src/Nettle/Utils.hs, dist-ghc/build/Nettle/Utils.o )
> > [ 5 of 11] Compiling Crypto.Nettle.Hash.ForeignImports ( dist-ghc/build/Crypto/Nettle/Hash/ForeignImports.hs, dist-ghc/build/Crypto/Nettle/Hash/ForeignImports.o )
> > ghc: panic! (the 'impossible' happened)
> >   (GHC version 7.6.3 for mipsel-unknown-linux):
> > 	Cant do annotations without GHCi
> >     {src/Crypto/Nettle/Hash/ForeignImports.hsc:138:16-44}
> >     "HLint: ignore Use camelCase"

See https://github.com/stbuehler/haskell-nettle/pull/4.

I wonder if it would be worth backporting the GHC patch referenced from
that bug, rather than continuing to apply workarounds in various modules
that we're just going to have to undo later?

-- 
Colin Watson                                       [cjwatson at debian.org]