From: Hugh O. B. <hb...@re...> - 2008-12-18 21:22:07
|
On Thu, Dec 18, 2008 at 08:17:55PM +0000, Daniel P. Berrange wrote: > On Thu, Dec 18, 2008 at 10:08:25AM -0600, Federico Mena Quintero wrote: > > On Wed, 2008-12-17 at 19:51 -0600, Anthony Liguori wrote: > > > > [backing store] > > > > > > Is this just double buffering in GTK parlance? If you avoid disabling > > > double buffering, does that help? > > > > Nope, it's different. You can ask the X server to use backing store if > > possible --- if the server can save the contents of a window, it can > > avoid sending Expose events when the window is unobscured. That is, > > nothing goes over the wire in the ideal case. > > > > [This is similar to requesting "save-unders" for temporary windows, but > > it saves your actual contents, not the underlying contents.] > > Ok, I see how this can make a big difference for slow X connections. > Obviously it'll have a bit of a memory overhead in the X server > to maintain the backing store. > > ACk to the patch in principle, once its has been #ifdef'd to not hardcode > the use of GTK X backend on all platforms. > > Daniel Glad to see this going in, it will be very heplful on the oVirt cloud setup (slow console connections will be the rule rather than the exception). --Hugh |