Amount of D-Bus messages with Telepathy

Alban Crequy alban.crequy at collabora.co.uk
Wed Jan 26 09:27:47 PST 2011


Le Wed, 26 Jan 2011 16:49:30 +0000,
Simon McVittie <simon.mcvittie at collabora.co.uk> a écrit :

> On Wed, 26 Jan 2011 at 17:35:22 +0100, Thiago Macieira wrote:
> > Em quarta-feira, 26 de janeiro de 2011, às 16:06:39, Simon McVittie escreveu:
> > > (Telepathy doesn't block on D-Bus calls, apart from a few calls to the bus
> > > daemon.)
> > 
> > You mean like the 1091 AddMatch calls, which are 59% of the calls made? :-)
> 
> So you'd think, but in fact dbus-glib just fires them off with the ignore-reply
> flag and hopes they'll work; it's not as if we'd be able to recover from them
> failing, so that seems fair really.
> 
> (Alban, were you using the dbus-glib patch from
> https://bugs.freedesktop.org/show_bug.cgi?id=23846 to only bind to individual
> signals, or the traditional one-match-per-interface behaviour? 59% does seem
> like a lot.)

I did my tests without the patch.

About half of Telepathy's match rules are NameOwnerChanged rules on Gabble: there
is 1 such rule per DBusGProxy, even if most proxies are on Gabble's D-Bus objects.

The patch seems to do 2 different things:
1. bind on individual signals (controversial change)
2. avoid duplicate DBusGProxy's "NameOwnerChanged" match rules on the same
connection (fd.o #23846 Comment #5)

I don't remember the details of the patch but it should be possible to fix only
the point 2 and consider the point 1 later if we reach a consensus.

-- 
Alban


More information about the dbus mailing list