dbus & threading (fix) ...
michael.meeks at novell.com
Mon Mar 8 02:56:50 PST 2010
On Fri, 2010-03-05 at 16:33 -0500, Havoc Pennington wrote:
> I had a similar suggestion (I think?) in the last paragraph here:
Yep - glad we're on the same page :-)
> I might almost suggest that switching libdbus to that model is easier
> than trying to understand whatever libdbus is doing now.
Sure - on the other hand, we have a GNOME release coming up really soon
now (weeks away), and an Evolution (using the best dbus available) that
has hideous blocking problems without my (pretty trivial) patch.
Is there any chance we can get a new release of dbus-1 with the fix, so
that we can depend on in the next two weeks ? The sad irony currently is
that HEAD Evo now uses IMAPX giving it a nice interactivity boost,
modulo the potential 25 second hangs from moving e-d-s from CORBA to
dbus ;-) [ the right direction overall of course with this fix ].
On Mon, 2010-03-08 at 11:03 +0100, Alexander Larsson wrote:
> I agree. As long as we do the switch automatically when the first
> non-main-thread operation is called. Always having the extra thread
> for simple one-thread apps is pretty lame.
Quite; that is the ORBit2 model; it seems to work reasonably. Having
said that - IMHO, writing a simple new socket layer for GDBus such that
we don't have to link to libdbus, that doesn't have to lug around the
OOM safety, and hyper-sensitive message validation logic might be a
better approach, and doing the marshalling in glib directly could be
michael.meeks at novell.com <><, Pseudo Engineer, itinerant idiot
More information about the dbus