Re: Generalize filterM to Applicative

On 14-12-30 12:34 PM, David Feuer wrote:

I realized just now that we can actually make the type a little bit more general still, interpreting `filterM` for lists as being an applicative filter *producing* lists:

filterM :: (Applicative f, Foldable t) => (a -> f Bool) -> t a -> f [a]

I'm +1 on the more limited generalization proposal, but not on this one. I've been feeling for a while that we need a subclass of Traversable to generalize functions like filter, mapMaybe, and partition (potentially span, splitAt and others as well). If you call this missing class Filterable, the properly-generalized signature would be

filterM :: (Applicative f, Filterable t) => (a -> f Bool) -> t a -> f (t a)

On Mon, Dec 29, 2014 at 3:17 PM, Edward Kmett wrote:

...

I think this is the interpretation I think we're probably best left with.

I agree that names matter and that anachronistic labels confuse, but I think it is the lesser of evils to widen the definition of the 'M' suffix than it is to double up on almost all the names taken in the environment and force the entire community to go through a positively gigantic deprecation cycle.

-Edward

On Mon, Dec 29, 2014 at 1:31 PM, Andreas Abel wrote:

...

Or you can widen the interpretation of suffix ...M as "effectful", which could be a monadic or applicative effect.

On 29.12.2014 16:50, Kim-Ee Yeoh wrote:

...

On Mon, Dec 29, 2014 at 7:58 PM, Edward Kmett mailto:ekmett@gmail.com> wrote:

Is it "madness" to want to avoid namespace proliferation and maximize the usefulness of an existing combinator now that the constraints that forged it have changed to allow it to be slightly more permissive?

Madness is such strong language for this august list.

May I speak on behalf of haskell newcomers for a time?

Haskell places such an emphasis on uniformity and regularity. Functions with names that end with M once meant they were monadic variants of those that don't. That's no longer uniformly the case, because of the FAM restructuring.

The names of functions matter.

Anachronistic labels confuse.

Leaving filterM with a type signature of Applicative cannot be the long-term solution.

-- Kim-Ee

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

-- Andreas Abel <>< Du bist der geliebte Mensch.

Department of Computer Science and Engineering Chalmers and Gothenburg University, Sweden

andreas.abel@gu.se http://www2.tcs.ifi.lmu.de/~abel/

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

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

-- Mario Blazevic mblazevic@stilo.com Stilo International This message, including any attachments, is for the sole use of the intended recipient(s) and may contain confidential and privileged information. Any unauthorized review, use, disclosure, copying, or distribution is strictly prohibited. If you are not the intended recipient(s) please contact the sender by reply email and destroy all copies of the original message and any attachments.