Introduction to D-BUS

Havoc Pennington hp at redhat.com
Mon Aug 28 06:56:10 PDT 2006


Jeroen T. Vermeulen wrote:
> 4. If it's true that a client can create a proxy on an object behind an
> un-activated service's connection (as Raphaël's Activation Tutorial says),
> why does the D-BUS Tutorial's section on the Python binding say that one
> of the two ways of listening for a signal, which uses a proxy, requires
> the service to be active first?

I don't know much about the python bindings, but presumably the python 
signal implementation wants to bind to the unique name or something like 
that.

> 5. Again, if it's true that a service doesn't get activated until a method
> on one of its objects is invoked, how does the caller get to the right
> interface on that object?

The assumption is that the local and remote apps both have foreknowledge 
of the interface via some means, e.g. there's documentation of the 
interface shared by both app authors, or whatever. Because D-Bus isn't 
"statically typesafe" there's no need to have the interface definition, 
as long as you know the method name and signature.

Havoc



More information about the dbus mailing list