TypelevelAs should add `cats.syntax` import

[NeQuissimus]

Version 0.1.5

import cats.effect.Resource

val foo: Resource[A, B] = ???
val x = foo.map(_ => ())

The rule TypelevelAs.as will complain and suggest to rewrite the map line to .void. Resource does not have a void method :) The check should probably be limited to types that provide the necessary functions.

[armanbilge]

It does, if you do import cats.syntax.all._ :) I think there's a way for scalafix to add imports if necessary.

[NeQuissimus]

Oh, that is possible. I wonder if there is a good way to make that part of the message in such a situation maybe?

My actual issue turned out to be that I had multiple Functor instances (from Temporal and Sync) and the void syntax could not decide which one to pick :)

[armanbilge]

Here's how an http4s scalafix adds an import if needed. https://github.com/typelevel/typelevel-scalafix/blob/8e246387dbed1e772933d2f757bc7609bc909861/modules/http4s/rules/src/main/scala/org/typelevel/fix/Http4sLiteralsSyntax.scala#L41-L43

[NeQuissimus]

Let's close this issue because there is no issue :)

[armanbilge]

Actually there is an issue, the import is not being added :)

[NeQuissimus]

Right but you'd have to make the rule a lot more specific then, would you not? What would happen to a foo.map(_ => ()) where foo does not have a void in cats.syntax?

[armanbilge]

Good point. So even after adding the import, this Scalafix could still be suggesting broken code. Seems to me like another reason to keep it open.

[NeQuissimus]

👍 I agree with that. This almost needs something like "do we have an implicit Functor? OK, just add the syntax" and otherwise bail out.

[DavidGregory084]

The (WIP) port of the RemoveInstancesImport rule does something similar to that actually: https://github.com/typelevel/typelevel-scalafix/pull/31/files#diff-b2f050feb2124095e5835138e89f5f116e4781fc5aeb560ea4415dcf7f17f926R46