From: Rik F. <fa...@va...> - 2000-11-06 21:55:18
|
On Mon 6 Nov 2000 17:58:13 +0100, Andreas Ehliar <eh...@ly...> wrote: > On Mon, Nov 06, 2000 at 10:41:06AM -0500, Rik Faith wrote: > > The server will remain in fullscreen mode until: > > 1) another drawable is used as a parameter to glXMakeCurrent, > > 2) the fullscreen drawable is destroyed, > > 3) the client dies. > > snip > > > Some clients may not behave well if full screen mode is entered > > automatically. Because of this, the full screen feature will only > > be enabled if LIBGL_DRI_AUTOFULLSCREEN is set to "enable", "true", > > "t", "on", "yes", "y", or "1". This will give the user full control > > over whether or not full screen mode is used. (The automatic > > fullscreen mode is most useful for clients that open a single > > fullscreen window (e.g., Q3A), and will not work well if the client > > wants to open regular X windows (e.g., for pull down menus, etc.).) > > Does "will not work well" involve sluggish response or complete > and utter failure? Closer to utter failure -- an application may bring up a menu in a window, but you won't see it, so you'll never be able to select something from it. For applications like that, don't set LIBGL_DRI_AUTOFULLSCREEN -- wait instead for the "real" fullscreen mode that will be a GLX extension. > Would it be impractical to exit fullscreen mode if another window > suddenly obscures part of the fullscreen window. Yes. > > VT switching will be prevented by holding the vtSema and via the use > > of the new xf86EnableVTSwitch call, which will allow enabling and > > disabling VT switching in a clean manner (this builds on work that > > was put into XFree86 for *BSD). > > Will it still be possible to recover from a broken client? If the client > is sent a SIGSTOP for example? Or if it is spinning around in a loop without > any exit in sight? For now, you'd have to kill the client in some out-of-band fashion. |