Hm. I hadn't considered the liftM2 version. It would need Monoid a instead of Monoid r, and it would behave differently. The version I suggested is motivated by a library for a Fran-like functional event-based programming that I'm preparing to release, based on Cont. In the monoid def I gave corresponds, mempty is the never-occurring event, and mappend is a sort of union of two events.
I sure don't know how to resolve these situations of more than one credible instance. I'm seeing more & more of them.
Thanks for the pointer on library submissions.
I'm with you about liftM0 (and liftA0).
Cheers, - Conal
On 9/8/07, Stefan O'Rear <stefanor@cox.net> wrote:
On Sat, Sep 08, 2007 at 05:26:09PM -0700, Conal Elliott wrote:
> I'd like to see the following addition to Control.Monad.Cont in mtl:
>
> instance Monoid r => Monoid (Cont r a) where
> mempty = Cont mempty
> m `mappend` m' = Cont (runCont m `mappend` runCont m')
>
> What's the process for suggesting, discussing, and making such changes?
http://haskell.org/haskellwiki/Library_submissions
(My 2 cents: Why not mempty = return mempty; mappend = liftM2 mappend ?
Instances are best if unambiguous.)
(Why is there no liftM0 ?)
Stefan
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
iD8DBQFG40G1FBz7OZ2P+dIRAiIgAJ0bhthqmBvcpoqLnNF+RcbccUBKYACeO4QO
sJ9sUvnkvkfZsgIBbgq6r9U=
=cCFe
-----END PGP SIGNATURE-----