On 06/10/2009 02:37 AM, Bill Crawford wrote: > Eric Anholt wrote: >> On Wed, 2009-05-20 at 10:59 +0100, Luke Benstead wrote: >>> Hi all, >>> >>> I recently had a discussion with a Wine developer and I mentioned that >>> I had on several occasions been left at a low resolution when a >>> fullscreen application crashes. He said that Wine could perhaps code >>> in a fix which guaranteed the resolution was reset but he wasn't sure >>> where this problem lied, or more specifically which project should fix >>> it, as native games also suffer the same problem. >>> >>> Incidentally, the previous day I had been in a PC store and started up >>> SuperTux on a netbook in the shop, I came back a few minutes later to >>> find the desktop at a low resolution, the game had obviously crashed >>> (desperate not to give a bad impression of Linux to passers by I reset >>> the netbook ;) ) so I know it's quite a common issue. >>> >>> After more discussion with the Wine dev I learned/realized that >>> Windows generally doesn't suffer a similar problem, because setting >>> the CDS_FULLSCREEN flag in the DEVMODE structure (passed to >>> ChangeDisplaySettings) indicates a temporary resolution change. If the >>> program crashes, or is ALT+TABbed away from, Windows (usually) resets >>> the resolution back to native. >>> >>> After some discussion on some forums I found this bug report: >>> https://bugs.freedesktop.org/show_bug.cgi?id=14255 >>> >>> I really just interested to find out from you guys where you think the >>> problem lies? Is it something that the X.org devs can fix? Or is it >>> something that should be fixed higher up? Is a "temp flag" perhaps >>> something that can be passed to xrandr when a resolution is set? >> I'd love to see an extension to RandR to handle this. It's been a >> serious problem of X's modesetting APIs forever. > > In particular, the notifications of a change need to include the flag so > that window managers can avoid totally rearranging desktop panels and > windows to fit the tiny resolution :o) as the mess left afterwards can > take a while to clear up if like me you retentively arrange everything. > Perhaps, even, the temporary flag would cause *no* notification to be > sent, but then you'd have to detect when that client lost focus and > change resolution, rather than rely on the WM to do it.
Couldn't we just do a resolution change on map/unmap? Or is that not sufficient? ~ C. _______________________________________________ xorg-devel mailing list [email protected] http://lists.x.org/mailman/listinfo/xorg-devel
