[regression] Function type attributes are generated as member function?

[kevinresol]

[regression] Function type attributes are generated as member function?

[back2dos]

Ok, there's a workaround for now. The main idea was not to cause rerendering if callbacks change.

The null error can easily be avoided. But if we want someFunctionAttribute == null to work, that's a different story. We could make it so that one can write hasSomeFunctionAttribute instead. Or we burry this optimization

[back2dos]

Hmm, I think I like the optimization that's currently disabled and I'm thinking to reenable it in a few cases.

For starters, if the function is optional and has a non-null default, because then we know it'll never be null so the wrapping is safe.

Beyond that, I would also propose that when the attribute is mandatory and not Null it's meaningful to assume that null is invalid, especially since passing null as attribute in HXX probably doesn't work anyway due to type inference. So for mandatory non-Null function attributes, I'm inclined to enable this. We could make an explicit null check and throw an error that states more explicitly what's wrong, e.g. invalid null passed to onPress attribute of <Main /> or something.

[kevinresol]

We can have a try