non-behavior DOM about agda-frp-js HOT 4 CLOSED

Just a placeholder response for now, I'm packing to head off to PoPL. Glad to see that someone is going over my code!

Yes, the current API is certainly not expressive enough. Join shouldn't be a problem (indeed I think the JS code is already there). Non-behaviour DOM nodes is a bit trickier, as you need to make sure that a DOM(t) node is only constructed at time t. At the moment, the invariants that are being maintained by the JS FRP implementation are, er, underspecified. It would be nice to make these invariants more precise, to know what extensions to the existing API are considered safe.

from agda-frp-js.

Awesome, thanks I started to look at implementing join today. Hopefully I'll update this post with a link to the fork when I have something (mostly I need to read over / understand your signal js runtime.)

I updated the gist to compile with an example postulated join. Is a join at the level of Behaviors like that ok, or should it be specialized to "Dom w" rather than arbitrary reactive type "A"?

from agda-frp-js.

Yes, I was expecting join to have the type ⟦ Beh (Beh A) ⇒ Beh A ⟧. As you may detect, you are about to enter an extremely untested bit of the JS runtime.

from agda-frp-js.

Discussion has moved to #3.

from agda-frp-js.