Re: gdbus function call on interface NULL does not work

[Simon McVittie <simon mcvittie collabora co uk>]

On 27/08/13 14:43, Colin Walters wrote:
> On Tue, 2013-08-27 at 11:01 +0100, Simon McVittie wrote:
>
> > but that doesn't seem very well thought-out, and I'd be tempted to
> > delete it from the spec.
>
> I think the idea with this, though of course Havoc would know better,
> was to enable dbus-python users to just do dbus.Object() without
> dbus.Interface().

That might have been the reason, although it has to be said that
dbus-python isn't actually particularly Pythonic ("explicit is better
than implicit" and "refuse the temptation to guess", after all).

Thiago suggested on https://bugs.freedesktop.org/show_bug.cgi?id=68597
that it might also have been carried over from DCOP (which had
interfaces, but they were optional and in practice never used).

> Although maybe that was before introspection was added and dbus-python
> parsed it; nowadays it could block on receiving the introspection and
> pick from that on the client side which interface to use.

The fact that dbus-python parses introspection continues to annoy me: it
violates "don't be remotely-crashable", which is one of my design rules
for D-Bus. (If the service changes its types, you get some sort of
strange TypeError, or even silent misinterpretation of the result,
rather than a D-Bus error that you need to be prepared to handle anyway.)

    S