Re: window state stuff

[Havoc Pennington <hp redhat com>]

Darin Adler <darin eazel com> writes: 
> As part of a recent bug fix, jsh just added code to the Nautilus
> window_present function that moves it on to the current virtual desktop. I
> haven't looked at the code myself so I don't know the details, but I figure
> this might be a bit more controversial than the "classic" contents of the
> function, so might not make it in.
> 
> On the other hand, maybe "/* FIXME add desktop-moving stuff from WM spec */"
> is about this very issue?
> 

Hmm, a relevant quote from a mail on the sawfish list John pointed me
to:

  |
  |-When the _NET_ACTIVE_WINDOW client message is sent to the root
  | window, Sawfish fails to switch to the workspace of the specified
  | window if it's not on the current workspace; it merely removes the
  | focus from the current window.  (I was under the impression that the
  | purpose of this message was to allow the WM policy to prevail as to
  | whether the window was moved to the current workspace or the
  | workspace was changed to that of the window, and hence I should not
  | be trying to do one or the other myself before sending this message.
  | Am I mistaken?)

  there's a comment in the source file reminding me to implement this

That makes it seem to me that if I am calling gdk_window_focus() which
uses _NET_ACTIVE_WINDOW, it would be wrong to also fool around with
desktops, since the WM is responsible for handling that.

Havoc