Re: GdkScreen size-changed and gdk_monitor_get_workarea problem

[Bastien Nocera <hadess hadess net>]

On Thu, 2018-11-08 at 12:34 +0100, Sebastian Geiger (Lanoxx) wrote:
> Dear Bastien,
>
> thanks for the reply. I will attach a sample next time.
>
> My App runs with the X11 backend.
>
> I was able to resolve the issue yesterday, and found that it was not 
> directly a Problem of GTK+.
>
> The problem is, that the window manager computes the workarea 
> asynchronosly, presumably its handled in a g_add_idle callback.
> When I tried to use get_geometry instead of get_workarea I was
> always 
> getting the correct values, but I specifically needed the workarea
> so 
> that was not helping.
>
> My solution was then to register a filter function with 
> gdk_window_add_filter which listens for _NET_WORKAREA property
> changes 
> and then queue a g_add_idle callback which handles the change.
>
> I tested the behavior with mutter and metacity and both had the same 
> problem.
>
> I am not sure if this is by design, or a bug. However, I would like
> to 
> suggest that we add a small note to the documentation of the 
> monitor-changed signal and the size-change signal, stating that
> values reported by gdk_monitor_get_workarea might arrive delayed.
> Alternatively, fixing mutter and metacity would be my preferred
> solution 
> but would probably be more work and I have not the necessary
> knowledge 
> about mutter code to fix this myself.

I'd start with filing a bug against mutter (I'm not sure metacity is
still maintained), and see whether the GDK API documentation should be
amended after discussing it there.

Unless you're writing a panel of sorts, I don't think end-user
applications should use this API in any case, with the window manager
taking care of the window layout.

Cheers