rhine-0.1.0.0: src/FRP/Rhine/SF.hs
{-# LANGUAGE GADTs #-}
{-# LANGUAGE RankNTypes #-}
module FRP.Rhine.SF where
-- rhine
import FRP.Rhine.Clock
import FRP.Rhine.ResamplingBuffer
import FRP.Rhine.Schedule
import FRP.Rhine.SyncSF
{- | 'SF' is an abbreviation for "signal function".
It represents a side-effectful asynchronous /__s__ignal __f__unction/, or signal network,
where input, data processing (including side effects) and output
need not happen at the same time.
The type parameters are:
* 'm': The monad in which side effects take place.
* 'cl': The clock of the whole signal network.
It may be sequentially or parallely composed from other clocks.
* 'a': The input type. Input arrives at the rate @Leftmost cl@.
* 'b': The output type. Output arrives at the rate @Rightmost cl@.
-}
data SF m cl a b where
-- | A synchronous monadic stream function is the basic building block.
-- For such an 'SF', data enters and leaves the system at the same rate as it is processed.
Synchronous
:: ( cl ~ Leftmost cl, cl ~ Rightmost cl)
=> SyncSF m cl a b
-> SF m cl a b
-- | Two 'SF's may be sequentially composed if there is a matching 'ResamplingBuffer' between them.
Sequential
:: ( Clock m clab, Clock m clcd
, TimeDomainOf clab ~ TimeDomainOf clcd
, TimeDomainOf clab ~ TimeDomainOf (Rightmost clab)
, TimeDomainOf clcd ~ TimeDomainOf (Leftmost clcd)
)
=> SF m clab a b
-> ResamplingBuffer m (Rightmost clab) (Leftmost clcd) b c
-> SF m clcd c d
-> SF m (SequentialClock m clab clcd) a d
-- | Two 'SF's with the same input and output data may be parallely composed.
Parallel
:: ( Clock m cl1, Clock m cl2
, TimeDomainOf cl1 ~ TimeDomainOf (Rightmost cl1)
, TimeDomainOf cl2 ~ TimeDomainOf (Rightmost cl2)
, TimeDomainOf cl1 ~ TimeDomainOf cl2
, TimeDomainOf cl1 ~ TimeDomainOf (Leftmost cl1)
, TimeDomainOf cl2 ~ TimeDomainOf (Leftmost cl2)
)
=> SF m cl1 a b
-> SF m cl2 a b
-> SF m (ParallelClock m cl1 cl2) a b