Instance for `Eff`?

[natefaubion]

Is there a reason for there to not be an instance for Eff proper? I know in JS its "technically" not parallel. But if Eff was implemented on a true multi-threaded platform like the JVM, it definitely would be possible. Also I'd argue that a MonadPar instance for Eff is no different than MonadPar for Aff where you liftEff everything.

[paf31]

I think it's a reasonable law that par f x y be observationally identical to par (flip f) y x and that seems false for Eff.

[natefaubion]

That law does not necessarily hold for Aff.

[natefaubion]

In fact it doesn't hold for anything that supports synchronicity on platforms without multithreading.

[paf31]

Is the suggested instance par = liftA2?

[natefaubion]

The suggested instance is implemented via the FFI, such that a platform that has true multithreading could have a better implementation.

[natefaubion]

But yes, for the JS platform, that's what it would be.

[JordanMartinez]

I'm guessing this issue applies to Effect now rather than Eff.