reflex-dom VS reactive-banana

dyn allows “using” a dynamic widget. The returned event activates each time the dynamic changes.

reflex-dom

As I grew up on a Macintosh, I was always interested in graphical user interfaces. In my free time, I'm currently working on refreshing my reactive-banana library so that it will hopefully work well with the upcoming JavaScript backend. (The library relies on a few non-trivial garbage collector features, which were not implemented in GHCJS a decade ago. Essentially, I ended up with a program that gave different results, depending on whether it was compiled with GHC native or with GHCJS. 😅)

I am building a GUI-application in GTK and try to use reactive-banana as the style of FRP seems a lot better to work with in Haskell. The basis works: receiving events from the GTK-side lets me properly update the internal state. The problem however is when I want to create my own events and hook them onto the rest of the event-network I get a deadlock. I already sent an issue on Github, https://github.com/HeinrichApfelmus/reactive-banana/issues/216 , but looking at the commits it seems it is not that actively maintained anymore. I was able to find some kind of work-around for a basic use-case, but it is not enough when dealing with an event-stream that will fire multiple events. I created the following test-code to check where the problem lies (a bit simplified from the one on Github): {-# LANGUAGE RecursiveDo, LambdaCase #-} module Main where import Control.Monad.IO.Class (liftIO) import qualified Reactive.Banana as B import qualified Reactive.Banana.Frameworks as BF

There are some examples that also use event-switching, like https://github.com/HeinrichApfelmus/reactive-banana/blob/master/reactive-banana-wx/src/BarTab.hs , which also seems to use execute. They use <$ though instead of <$>: