What design do you think would be better?

Both Datastar and HTMX have the same issue: they want to pretend to be HTML. So they force themselves into writing several DSLs like this to pretend that this is still "just HTML".

Since they have a full "Datastar expression" language, I'd just expand that. Then you wouldn't need these weird constructs:

   <button data-on-click__window__debounce.500ms.leading="$foo = ''"></button>

Well they are both "just HTML", Datastar being better about being spec compliant using the dataset attributes and all.

By expanding the "Datastar expression" language, do you mean a compiled language or moving all the expressions into JS files or something else?

> By expanding the "Datastar expression" language, do you mean a compiled language or moving all the expressions into JS files or something else?

Doesn't matter. They literally already have a Javascript-like DSL: https://data-star.dev/guide/datastar_expressions

Datastar is spec compliant HTML. Please point me to the code that's not HTML spec compliant

this is how signals are described in the functional reactive programming world. here is the paper by the guy http://conal.net/papers/push-pull-frp/push-pull-frp.pdf