On Wed, Feb 8, 2012 at 2:33 PM, Roman Cheplyaka wrote:

[1 of 1] Text.Regex.Applicative.Interface(.......jhci: sigFromClass: Control.Applicative.Applicative [(Control.Applicative.pure, forall f0 a . Control.Applicative.Applicative f0 => a -> f0 a),(Control.Applicative.<*>, forall f0 a b . Control.Applicative.Applicative f0 => f0 (a -> b) -> f0 a -> f0 b)] Control.Applicative.<*

Could it be improved?

ooh. bad error message. Yeah, I'll add a check in the renamer for attempting to create instances with functions that don't exist, that will produce a much better message and not inhibit further error checking.

Also, I'd like to update the 'applicative' package to match the current 'base', but before I spend time on that I'd like to check if there are no objections to merging the changes into the jhc repository.

Sure, if you can update it, that would be cool. You may want to see if you can just use the external build system such as is done for bytestring, you would just need to create a lib/ext/applicative.yaml and associated patch if it doesn't compile as is from hackage. Though, I don't mind integrating applicative directly as at some point I'd like to have the standard Monad derive from those classes as would make sense, since Monad has special syntax it would mean that the package would have to be included as part of jhc. But making it as compatible as can be would be good. john

-- Roman I. Cheplyaka :: http://ro-che.info/

_______________________________________________ jhc mailing list jhc@haskell.org http://www.haskell.org/mailman/listinfo/jhc