ListT done right in transformers

[sjshuck]

[original issue 88 by hecate a.k.a. @Kleidukos]

Since ListT has been properly deprecated now, it would be nice to have a proper implementation in transformers nowadays. Gabriella Gonzalez's list-tranformer package¹ would be a suitable replacement.

What do you think?

¹ https://hackage.haskell.org/package/list-transformer

[sjshuck]

[original comment by @sjshuck]

Friendly bump

[Bodigrim]

There is newtype ListT = ListT (m (Maybe (a, ListT m a))) hidden inside of another GHC boot library, directory. Even while it's probably wise for transformers and directory not to depend on each other, I think it would be helpful to define it in the same way (and not how list-transformer does it).

[sjshuck]

That seems fine.

One unanswered question here is whether or not to CPS. I haven't looked into ListT at all versus things like ConduitT that have a separate o and a; I wonder whether codensity considerations apply here. Because as encoded above, the m (... ListT m a) does seem to be "capturing the new state". Intuitively not CPS-ing ListT is correct, but I want to double-check, with an eye to projected usage patterns.