D-Bus and click-to-dial

[Simon McVittie]

On 13/07/15 07:24, Daniel Pocock wrote:
> I'm just wondering if anyone can share any comments on the suitability
> of D-Bus for click-to-dial and other computer-telephony-integration
> (CTI), and whether such work is already under way?

The fact that only one thing can own each well-known name (i.e. D-Bus
activation cannot reliably choose between several implementations with
the same "priority" in the search path) makes D-Bus less than ideally
suited for this, IMO.

> - somebody clicks a phone number in one application and it sends a D-Bus
> message.  The message is received by a softphone or some daemon passes
> the number to a desk phone

There's a standard for tel: URIs, and there's a freedesktop.org protocol
for dispatching URIs to URI-handlers (it's something like "look up the
MIME-type handler for x-scheme-handler/tel"). That might be a better
solution for this side of things than using D-Bus activation, because
the MIME-handler spec understands concepts like "choose one of these
three potential handlers" and "prefer this handler while running KDE".

> - an incoming call arrives on a softphone or a desk phone and a signal
> is sent to all interested applications

URI-handlers are not sufficient for this side, and D-Bus does seem a
better fit.

Telepathy was meant to be a standard API for this sort of thing,
although development has stalled. telepathy-snom did this for a
particular brand of hardware desk-phone, but if I remember correctly, it
never made it past the proof-of-concept stage.

(I used to work on Telepathy; I don't any more.)

-- 
Simon McVittie
Collabora Ltd. <http://www.collabora.com/>