It isn't clear to me that arrowized FRP is incremental. I think some of the less pure FRPs are (like Flapjax) given that they do a topological sort of the signal graph to repair it (if they were going to redo everything, the sort wouldn't be necessary).
AFRP is a good example here. AFRP semantics are easily stated in a recomputational way. It makes it necessary to talk about causal AFRP and bounded history AFRP which are nice terms to think about a computation (if sort of obvious). Then, the efficient implementations (of causal AFRP) themselves are incremental for efficiency.
data (i ~> o) = A (i -> (o, i ~> o))
data (i ~> o) = forall s. A (i -> s -> (o, s))
in each case, the inputs are consumed sequentially, the outputs returned immediately, and the local state updated for the next time around.If you're referring to the ability for a new event to update the signal network only partially (in Elliott's terminology, "push" semantics) then there's Amsden's TimeFlies library.
I find that to be completely desirable. That behavior is exactly what I want... simply done more efficiently [0].
[0] Also modulo a whole bunch of local state. AFRP handles this well by having a good notion of what it means to "switch an arrow in" but it's been a challenge for applicative/monadic FRP. Rx programming (e.g. "not FRP") tends to solve this problem by ignoring its existence and just littering local state everywhere.