On the disappearance of CLDouble
Manuel, other interested c2hs people, GHC 6.12 removes the CLDouble type from the FFI libs. Of course this means c2hs currently does not compile. The question is what level we should handle this at. Do we also stop recognising and translating long double in C code? Perhaps we should recognise it but fail with an error message to the effect that there is no corresponding Haskell type. Suggestions welcome. Duncan
According to the GHC trac, it appears that CLDouble is slated for
re-inclusion after a proper implementation. The FFI requires it, so
they'll hopefully progress relatively quickly.
Given that, I think it would be best to avoid drastic changes in c2hs
to support this (hopefully) temporary situation. I wouldn't mind a
fail with error as suggested.
John
On Thu, Dec 10, 2009 at 1:05 AM, Duncan Coutts
Manuel, other interested c2hs people,
GHC 6.12 removes the CLDouble type from the FFI libs. Of course this means c2hs currently does not compile. The question is what level we should handle this at. Do we also stop recognising and translating long double in C code? Perhaps we should recognise it but fail with an error message to the effect that there is no corresponding Haskell type.
Suggestions welcome.
Duncan
_______________________________________________ C2hs mailing list C2hs@haskell.org http://www.haskell.org/mailman/listinfo/c2hs
Duncan Coutts:
GHC 6.12 removes the CLDouble type from the FFI libs. Of course this means c2hs currently does not compile. The question is what level we should handle this at. Do we also stop recognising and translating long double in C code? Perhaps we should recognise it but fail with an error message to the effect that there is no corresponding Haskell type.
Yes, failing with a helpful error message seems best, especially as support for the type should be back at some point. Manuel
participants (3)
-
Duncan Coutts -
John Lato -
Manuel M T Chakravarty