Good point. It’s certainly easier to change in that direction though I’m sure some weirdo will get a regression because they do something important in a catch block. ![]()
That works if you’re actively looking for xyz but it’s more of a hazzle if you don’t know the exact name and just browse through the list of set methods.
This is an interesting twist that I didn’t think of. Would be convenient to significantly reduce the amount of new API we would have to introduce. The big drawback is that there’s no direct way of hooking up with the component’s attach and detach events which means that you would have to remember to manually close the registration to avoid leaking memory in case the signal instance lives longer than the component instance. We could make it work with the help of yet another ThreadLocal but I’m not sure we want to go that route.