Macos fires WindowEvent::Focused(false) on WinitWindow creation

[TimonPost]

In this commit: b457329#diff-4803bb2335a17d700335730b9f62828e001faa602b9b894dff572422c737b78aR480 the author fires the Focused(false) event on macos window creation. It seems to be inconsistent with the other platforms.

In my case on macos I receive first Focused(true) and then Focused(false) when creating a window. The Focused(true) is caused by window_did_become_key and then Focused(false) is caused by the window init function queuing it when it is created.

On other platforms it seems like the focused boolean is automatically set to false upon creation without firing an event. This seems to create divergent behavior on different platforms as macos has an additional event now.

For now my fix is the following:

 WindowEvent::Focused(_focused) => {
        self.focused = winit.has_focus();
}

Sending an event on window creation creates unexpected behavior. Evenwhile my window is focused winit just informed me it isn't.

[TimonPost]

cc: @amrbashir @kchibisov

[kchibisov]

In my case on macos I receive first Focused(true) and then Focused(false) when creating a window. The Focused(true) is caused by window_did_become_key and then Focused(false) is caused by the window init function queuing it when it is created.

So do you have Focused(false) for focused window?

[TimonPost]

Yes! When i print the output of Window::has_focus and the event WindowEvent::Focused(x) boolean it shows they are different. The event tells me my window is not focused while the function tells me it is focused. I think its a macos only problem as we did not get windows bugs reported.

[kchibisov]

@TimonPost could you verify whether my patch changes anything for you? Because I can't repro the original issue you're having in the first place to begin with.

[TimonPost]

Will give it a try soon. Yes I found it hard to replicate the same behavior in an standalone app. We do have quite some complex setup of our window logic so many moving parts.

[kchibisov]

Yeah, that would be nice, since my testing target is usually alacritty, and it works fine with the current logic we have.

[TimonPost]

I will still confirm the changes just need to get to work where I have a MacBook.

[kchibisov]

Well, it was already confirmed, but yeah, sure.