[PATCH 2/3] shell: Add implementation of fullscreen.

[Juan Zhao]

  Thanks a lot for review!
Still have one question:
On 02/27/2012 04:02 AM, Kristian Hoegsberg wrote:
>> +	if (es->output) {
> This is never true since reset_shell_surface_type() unmaps it.  But
> even when that is fixed, we can't change the surface position, size or
> stacking until we receive the fullscreen buffer from the client.  If
> we do, we're very likely to see a frame or two with the surface
> jumping or moving before we get the real fullscreen buffer.  All we
> should do here is save the fullscreen parameters and send the
> configure.  Once we get the new buffer in configure, we can do all the
> changes (create black surface, raise, move, add transform etc).
One concern here:
The case is like this: if the client called set_fullscreen, but not 
response that configure event.
No new buffer is attached after that.
In this condition, the surface type is fullscreen and we won't paint it 
as fullscreen.

So we add one condition here to figure out such case, the surface is 
already mapped and assigned an output.

But if we assume the client must attach a new buffer after it called 
set_fullscreen, this action is useless.


Thanks a lot again,
Juan