Clarify the relation of focus and propagate_to_hidden

[xarvic]

As i understand it, hidden in propagate_to_hidden() means the widget, is inside the widget tree, but can't be seen. Not because it is overlaid by another component, but because it doesn't have a position inside the current window (like a widget inside a hidden Tab). On the other hand, a component in a scroll would not be "hidden" according to my understanding, even if you can't see it right now.
Therefore a component can participate in focus if and only if it is not hidden.
Which means for one thing that my implementation for the Lifecycle::BuildFocusChain in #1632 is wrong since it includes all hidden tabs. But also, to fix this we would probably need a context method children_state_changed() which would request a rebuild of the focus_chain (for an example when selecting another Tab). Calling children changed should be enough.

Another thing we should think about is what happens to focused widgets which become hidden.

[cmyr]

Your understanding is correct; a 'hidden' widget is a widget that is part of the widget tree but is not currently a functional part of the UI, generally because it is part of a switch/either/tab that is not visible.

I also agree that a widget that is in a scroll view and is not onscreen is not "hidden".

[xarvic]

Ok, then i will create a PR with the changes. I would like to test in BuildFocusChain if the focused widget is still a part of the non hidden tree, send ResignFocus otherwise. Is that ok? I also would like to change Lifecycle::should_propagate_to_hidden() -> bool to Lifecycle::propagate_to_hidden() -> Option<Self>. Then we could change the focus event into a focus event with only the resign part. Therefore even if someone tries to focus a hidden widget he won't succeed, but were still able send the unfocus to the specific widget.

[cmyr]

Ok, then i will create a PR with the changes. I would like to test in BuildFocusChain if the focused widget is still a part of the non hidden tree, send ResignFocus otherwise. Is that ok? I also would like to change Lifecycle::should_propagate_to_hidden() -> bool to Lifecycle::propagate_to_hidden() -> Option<Self>. Then we could change the focus event into a focus event with only the resign part. Therefore even if someone tries to focus a hidden widget he won't succeed, but were still able send the unfocus to the specific widget.

The first part sounds good. The second part is... maybe okay? I'm always a little scared of complicating the logic more than possible. Thinking this through, I'm not sure if it really matters if a hidden widget gets the 'resign' message, assuming it gets some message when it becomes hidden? (does this happen? 😬)

[xarvic]

I don't think so. Since hidden means the parent widget decided at some point not to show the widget anymore, the widget itself doesn't know that it's hidden. If a textbox inside a tab widget is focused and you select another tab, druid thinks the textbox is still focused until you click another one.

[xarvic]

I think we can close this now