On Jan 9, 2007, at 22:35 , Claus Reinke wrote:
- The existing module used undecidable instances, a Haskell extension which is supported by GHC but not by Hugs.
??
as far as i recall, Hugs was the first Haskell implementation to bite the bullet and accept undecidable instances as a design option (although its predecessor apparently also had the undecidability feature).
http://cvs.haskell.org/Hugs/pages/users_guide/class-extensions.html
are you thinking of interactions with other extensions, perhaps?
Sorry, my bad, that was confused. Some earlier comments from this thread to add to the confusion: Sven Panne wrote:
Effectively removing Network.CGI from the Hugs community will probably break more programs, so I'll vote for non-overlapping instances, too.
Ian Lynagh wrote:
Hugs compatibility is also important, though, and undecidable instances are marked "probably no" for Haskell'.
The bottom line is that the instance: instance (MonadTrans t, MonadCGI m, Monad (t m)) => MonadCGI (t m) where needs both overlapping and undecidable instances, doesn't work in Hugs, and probably won't be Haskell'-compatible. However, the problem should be solved now, since that instance is no longer in the cgi package. /Bjorn