On March 27th, 2012, 5:21 p.m., Martin Gräßlin wrote:

kwin/geometry.cpp (Diff revision 1)
QRect Workspace::clientArea(clientAreaOption opt, int screen, int desktop) const
292	if (is_multihead)	298	return QRect(0, 0, displayWidth(), displayHeight());
293	return Kephal::ScreenUtils::screenGeometry(screen_number);
294	else
295	return Kephal::ScreenUtils::desktopGeometry();

after thinking about it I could not see any difference any more. In both cases the complete display is returned.

On March 27th, 2012, 7:39 p.m., Thomas Lübking wrote:

Pot. related to how many screens kephal sees in the multihead (maybe even xinerama instead xrandr?) case.
If it sees the "other" screen as well, desktopGeometry() would largely differ from screenGeomtry() but latter be equal to display(Width|Height)()

So by dropping kephal, this pitfall could have implicitly be avoided.
-> Did you test on a multihead config (i'm tonight not in the mood to break my setup ;-)

of course I did not test on multihead. But my old laptop has a xorg.conf prepared for multihead and I could run a recompile there.

- Martin

On March 27th, 2012, 5:19 p.m., Martin Gräßlin wrote:

Kephal has turned into not being more than a wrapper around QDesktopWidget and does not even provide syntax sugar.

currently running kwin without kephal on a multiscreen setup - so far everything's fine.