[fdo] Seeking clarification on Window Properties

[Christian Schmidt]

Hi,

I am trapped in a discussion that is thrown back and forth between two 
groups of developers, and would hope to get a verdict from freedesktop.org.

The bug, simplified, is this:

1. A window manager creates a _NET_WM_STATE property with 
_NET_WM_STATE_HIDDEN atom to tell an application window that it is no 
longer shown, and sends a notification.
2. The application window receives the notification and state change, 
and stops redrawing.
3. The window manager decides that the application is visible again, 
deletes the complete _NET_WM_STATE property (as _NET_WM_STATE_HIDDEN was 
the only atom in it), and sends a notification.
4. The application queries the property and receives something that is 
not actually a property, and treats this as an error, keeping its frozen 
state.

The application developers claim that the WM must keep the property, 
even if it is empty.
The WM developers claim that a non-existing property should be treated 
the same as an empty property, even if it existed at some time in the past.

I found no part of the specification that enforces persistence of a 
property, or one that defines the identity between an empty but 
existing, and a non-existing property.

Can you please clarify on this?

Best regards,
Christian