preventDefault about purescript-halogen

[paf31]

I think it is.

EventHandler (Aff eff input) is inhabited by preventDefault $> x. All we need is for x to not call the callback, but I'm not sure if that's possible. @jdegoes ?

[jdegoes]

Aff eff input is a computation which must produce an input or end in error, there are no other options. It's exactly like Eff eff input except asynchronous. However, one use forkAff if one wishes to fork the computation, and then const-map it to whatever. That should be enough to build this, right? Also Handler is becoming a monad transformer, isn't it?

[paf31]

I think @cryogenian wants to not produce any input to the state machine at all. I think handler' will let you do that. The underlying representation of a handler certainly allows it.

[jdegoes]

Perhaps a combinator, then, which operates on EventHandler a, goes to EventHandler b for all b, by mapping the Maybe i to Nothing?

[paf31]

It's not EventHandler, it's the attribute itself:

https://github.com/slamdata/purescript-halogen/blob/master/src/Halogen/HTML/Attributes.purs#L147

But yes, a combinator would probably do the trick. Rolling the Maybe into the new EventHandlerT would probably be a good thing.

[paf31]

@cryogenian Now you can use cancel if you don't want to generate any input event.

@puffnfresh You mentioned using Free which I think was a good idea, but I stuck with WriterT because I now need to turn this into EventHandlerT (transformer) and we don't have a free monad transformer yet in purescript-free.

[puffnfresh]

@paf31 that's fine.

We were not able to write a stack-safe FreeT in Scala due to the JVM. I wonder if JS will have the same problem...

[paf31]

Can't you play the same trick with Gosub? Interpreting the monad should (and I'm guessing obviously) be possible safely with tailrec.