Can't use eq if order is also imported

[joprice]

Is this expected?

import cats._
import cats.std.all._
import cats.syntax.eq._
import cats.syntax.order._

class Test {
  1 === 2
}

// [error] ... value === is not a member of Int
// [error]   1 === 2

[joprice]

Ok this makes sense, since OrderSyntax extends PartialOrderSyntax extends EqSyntax.

Is there a library-level trick to avoid this ambiguity? Moving the order import into another enclosing scope, such as within the Test class, raises it's priority, so that's a usercode-level workaround..

[ceedubs]

I don't know of any way that we can work around this within cats, but I think that we should be sure to address the conflicting import problem in any import guide that may be created as part of #1026.

[diesalbla]

@kailuowang It has been three years since this issue was opened. Looking at it for a minute, I can see only two ways forward:

One way is to modify the traits OrderSyntax and PartialOrderSyntax, so that they no longer inherit from EqSyntax or each other. Each one would only provide the specific methods and operators introduced by their respective type-class, and not those from the other ones. This is similar to how import cats.syntax.applicative._ does not bring .map or .void, and one has to use cats.syntax.functor._. This way would be both binary and source-breaking.

The other way is to leave the cats code as is, and close this issue.

[kailuowang]

The fix requires breaking changes. I would vote for closing this issue. @LukaJCB what do you think?

[tpolecat]

Agree re: closing

[diesalbla]

Pertinent but not main topic: Are there any plans for a cats 2.0 release, at some point, one in which breaking changes such as this one could be added?

[kailuowang]

The roadmap for 2019 is here https://github.com/typelevel/cats/blob/master/ROADMAP_2019.md
For breaking changes like the fix you suggested we'll have to wait for 3.0, which is likely to be post 2020. It's not ideal to have inconsistencies, but we have to evaluate the benefit of fixing it against the cost of the breaking change for users.

[kailuowang]

sorry accidentally closed it, I meant to wait on hearing luka.

[LukaJCB]

I also agree to closing this issue, we can discuss refinement of the roadmap in the appropriate issue :)