*** Bug 107301 has been marked as a duplicate of this bug. ***

Possible causes to investigate:
 - do we do initial window focus during placement and skip 
   placement if a position is set?
 - is it just that with mouse focus the window isn't coming 
   up under the pointer?

Created attachment 35193 [details] [review]
Windows that manually position themselves should be focused too

The code looks (roughly) like this:

  if (did_placement)
    if (takes_focus_on_map)
     meta_window_focus (window,
	 meta_display_get_current_time_roundtrip(window->display));

where did_placement means "did metacity do the placement?".  Since the
application did the placement, this variable is false and thus everything else
is ignored.  I remember staring at the "if (did_placement)" for quite a while
when I was working on focus stealing prevention and wondering why in the world
it would possibly be there.  I couldn't find a single good reason for it, but
on the other hand I didn't have a good reason to remove it, so I didn't.  But
now we do have a good reason to get rid of it...

Comment on attachment 35193 [details] [review]
Windows that manually position themselves should be focused too

I think the idea of did_placement was to focus new windows, but not focus old
windows that are being re-shown.
In other words, to avoid focusing in all cases from inside implement_showing()

The trivial alternate patch would add a "shown for first time" flag to
MetaWindow, but perhaps we can think of something more elegant, I don't know.

Created attachment 35248 [details] [review]
Add showing_for_first_time flag

I tried looking through the various flags we already had in window.h to see if
we had anything that would provide the information we wanted.  I couldn't see
anything, so I just went with the showing_for_first_time flag as you suggested.

Comment on attachment 35248 [details] [review]
Add showing_for_first_time flag

Simple enough, thanks.

committed.