What happens now after "delimited continuations" is merged to GHC?

Just to provide a picture for the general landscape, eff belongs to a family of effect libraries that uses a technique called “evidence-passing” to avoid the high performance cost brought by free monads, mtl typeclasses and the likes. They don’t necessarily have delimited continuations but all have very good performance compared to say, polysemy or unspecialized fused-effects:

  • effectful doesn’t have delimited continuation but has generally very good support for IO.
  • cleff is a library that I wrote that is intended to be beginner-friendly, it also doesn’t have delimited continuation.
  • eveff and mpeff are proof-of-concept libraries of the paper that first formally described evidence-passing

I personally think that evidence-passing is the future of Haskell effect libraries, because it’s easy to understand, has low performance overhead and works very well with other techniques (including either delimited continuation or MonadUnliftIO, it’s delimited continuation and MonadUnliftIO that don’t work well with each other).

4 Likes