Concurrency in Haskell

NOTE: You may reference Control.Currency for a more up to date insight.

In this post I am exploring something called forkIO. Here is the context:

We are creating a simple backend server. This server has a SQLite db instance hosted.

What we like to do here is host the server, doing two things: - Listening for queries via finger, - Allowing updates & injections of user data to the db instance

As a result we would need to fork the main process, to allow us to continue listening(first process) while we inject the user data into the db (second process).

Do let me know if I get any of this wrong :P

Material referenced: http://www.haskell.org/ghc/docs/papers/concurrent-haskell.ps.gz

Takeaways: - impractical to do IO with side-effecting functions: - sub-expr eval is context specific due to non-strictness of haskell - MutVars are references to a mutable location in the state - state-transformer semantics for I/O are unsatisfactory because we do not want the side-effects to be updated simultaneously but rather incrementally - Concurrent Haskell introduces: - processes && process init - atomically mutable state: Support inter process comms && coop - concurrency is non-deterministic, - we may have 2 processes writing to the file, of which the outcome is unknown - forkIO requires inter-process syncing: Evaluation of thunks can be synced