Haskell: Network conduit "callback" questions

Question

I'm using network-conduit and runTCPServer to power my stranded server.

In this case:

-- | Helper which represents a conduit chain for each client connection
serverApp :: Application SessionIO
serverApp appdata = do
    -- blah blah blah initializer code
    appSource appdata $$ decoder =$= protocol =$= encoder =$ appSink appdata

...

-- | Handles an input stream of 'Packet' objects and dispatches them
protocol :: Conduit Packet SessionIO Packet
protocol = undefined

How often does protocol get executed? If I wanted something to be timely executed every so often on a client socket would it be safe to put its logic in protocol, or is it only called each time data actually reaches it?

I would appreciate if you explained the way the conduit "callbacks" work. :)

Regards.

Answer 1

The Conduit Overview over at FP Complete explains the flow control of Conduit: specifically that conduits only go up the chain when data is needed. So if you're relying on time-based side-effects, relying on a conduit being called is not the way to do that.

This does depend a lot on your use case and the structure of your conduit chain, however, because you could put delays into a Conduit to rate-limit it or something similar.