Re: [Haskell-cafe] Haskell-Cafe Digest, Vol 217, Issue 16

On 19 Sep 2021, at 8:24 pm, Richard Eisenberg wrote:

This was all clear enough. What I didn't understand is why there was a typeclass to capture this or why we needed other monads -- until I associated "monad" with "programmable semicolon". At that point, I realized that the idea is very powerful.

Thus, e.g., the Hasql library (and of course also STM before that) leverage the ability to define a specialised Monad in which only specific "impure" actions are available to be able to "retry" a (Postgres Database in the case of Hasql) transaction, without worrying about side-effects persisting from the initial attempt. The Conduit and Streaming ecosystems are also testaments to the effectiveness of the abstraction. And then there's always State, which e.g. makes it possible to define mapAccumL and mapAccumR for all Traversable structures, or more generally thread an evolving through a sequence pure computations. Thus while there's reportedly been a shift lately from multi-layered Monad Transformer stacks to RIO or algebraic effect systems, Monads remain powerful and useful tools. I no longer remember exactly when they "clicked" for me, perhaps there wasn't any "Aha" moment. Just growing familiarity through multiple exposures to the many use-cases. Still haven't used "Cont" at all, so don't have any intuition for whether/when/how I might put it to good use... If someone wrote a good overview for that module, I'd be happy to learn more. -- Viktor.