Yep, that's right. There seem to be a few factors involved: - Most of that discussion happened before a full description of transducers was up anywhere, so a l

Yep, that's right. There seem to be a few factors involved: - Most of that discussion happened before a full description of transducers was up anywhere, so a lot of these details are just completely missing. - Haskell's type system is not expressive enough to capture certain interesting properties, like linearity. (To be clear, it's quite possible to translate any arbitrary transducer into Haskell -- but the type system will not prevent all the mistakes it is possible to make.) - There are different classes of transducers that Haskell people actively would not want to have identical types. (ie. stateful vs. stateless, early-terminating vs. unbounded) Many of the discussions revolve only one or a few of these classes -- typically the stateless unbounded version, which turns out to have a particularly simple encoding, but it definitely varies.