[Standards-JIG] Phone service discovery

Jean-Louis Seguineau jean-louis.seguineau at laposte.net
Wed May 10 06:01:13 UTC 2006


Nolan, I'm sorry to say I disagree with your otherwise perfectly valid
concern. From JEP-30, a client application should query the 'phone service'
JID to get the relevant disco#info. In the reply it should receive
<identity/> elements qualified with category/type, and some <feature/>
elements. I believe the addressing scheme support must be at the feature
level, not the category level. In addition, the support for a particular
addressing scheme should be expressed in a different namespace than Jingle. 
Addressing is a generic concern in XMPP. Being able to discover a particular
addressing scheme support spans beyond Jingle and telephone numbers.
Consider also that you will never reach the PSTN directly, because Jingle IS
a VoIP implementation. And we have not yet defined how an XMPP phone IRI
should look like. But this belongs to another thread...   

In using category/type, we would actually be adding an 'a-priori' semantic
to a service type with regard to addressing scheme. Following your proposal,
we would assume the ONLY way to interact with a PSTN service is by directly
providing a phone number. But you may find services willing to expose PSTN
through some address translation of their own. You will also find services
acting as front end to PBXs that will accept user's JIDs/IRI AND phone
numbers, or either JIDs OR phone number. There could be endless
combinations. 

I may want to create a service to sends SMS and place calls to users' mobile
phones without ever exposing the users' numbers for privacy reasons. In such
service, I will never need a phone pad. The service will be in the 'voice'
category/type. And the addressing feature will indicate JID only support.

Deciding if you show a phone pad on a client is the result of combining many
different features, such as phone number support, in band DTMF, and others.
We must not create artificial limitations in the protocol because of
'legacy' usage or perception. 

Jean-Louis

-----Original Message-----
Message: 4
Date: Tue, 09 May 2006 16:31:56 -0500
From: Nolan Eakins <sneakin at semanticgap.com>
Subject: Re: [Standards-JIG] Phone service discovery
To: Jabber protocol discussion list <standards-jig at jabber.org>
Message-ID: <44610A4C.4010708 at semanticgap.com>
Content-Type: text/plain; charset="iso-8859-1"

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Jean-Louis Seguineau wrote:
> Thanks, this is exactly the point; a user does not care how the call will
be
> handled. The user just need to know which JID is providing the voice
> service. It does NOT matter if the service provides a gateway to SIP,
PSTN,
> Skype or whatever. It could even be a direct integration with a PBX.

It does matter since the addressing schemes for each is different. For
POTS you're essentially limited to numbers whereas the actual VoIP apps
use user names possibly scoped by a domain. Differentiating could allow
unruly client developers to include a dialing pad or at least a place to
punch in phone numbers.

- - Nolan







More information about the Standards mailing list