On Wed, Feb 4, 2009 at 1:13 PM, Colin Walters <walters@xxxxxxxxxx> wrote: > On Wed, Feb 4, 2009 at 11:37 AM, Olivier Galibert <galibert@xxxxxxxxx> wrote: >> On Wed, Feb 04, 2009 at 01:42:42PM +0000, Richard Hughes wrote: >>> Well, sending the data does cause a wakeup, yes. The interface stops >>> polling if there's no client connected. >> >> If I understood correctly, the problem was more about the "wideness" >> of the wakeup, i.e. everything connected to the bus was waking up, >> unswapping, etc rather than just what was interested by the message. > > Almost certainly what you're thinking of (or have heard about) is the > NameOwnerChanged problem: > http://bugs.freedesktop.org/show_bug.cgi?id=14183 > > It's definitely something we should fix (and the system was designed > to support avoid wakeups), but it's certainly not as bad as every > process waking up on every message; just when something joins or > leaves the bus. And, just for anyone keeping score at home about DBus and power consumption, a patch exists now in that bug. -- fedora-devel-list mailing list fedora-devel-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/fedora-devel-list
