On Tue, Nov 22, 2016 at 09:17:32AM +0000, Baldur Blöndal wrote:

1) is good, I assume 3) will happen once the dust settles on AMP

Yes, there's a lot of conditional CPP in there already, but this doesn't seem severe enough to justify more.

2016-11-22 8:56 GMT+00:00 Edward Kmett :

They could be CPP'd around on newer versions, sure, at the cost of a fair bit of noise in the source code. Presumably it was considered simpler to just have one type signature. I've CC'd Ross Paterson, the maintainer of transformers. Ultimately it comes down to a 3 way decision between 1.) cleaner code by keeping the status quo, or 2.) cleaner haddocks by changing the code as requested here conditionally with CPP, or 3.) loss of backwards compatibility by just changing the type completely.

I don't particularly care which of the first two options are taken -- and the first even has the benefit of requiring zero labor -- but I think it'd be a bad idea to lose backwards compatibility by taking the third.

-Edward

On Mon, Nov 21, 2016 at 5:52 PM, Baldur Blöndal wrote:

Does this belong here? Applicative [1] and Alternative [2] instances of StateT have redundant constraints

> instance (Functor m, Monad m) => Applicative (StateT s m) > instance (Functor m, MonadPlus m) => Alternative (StateT s m)

The Functor constraint may be removed

[1] https://hackage.haskell.org/package/transformers-0.5.2.0/docs/src/ Control.Monad.Trans.State.Lazy.html#line-201 [2] https://hackage.haskell.org/package/transformers-0.5.2.0/docs/src/ Control.Monad.Trans.State.Lazy.html#line-210

_______________________________________________ Libraries mailing list Libraries@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/libraries