Re: [RAI] Device control using SIP

["Shekh-Yusef, Rifaat (Rifaat)" <[e-mail]>]

> If you *want* to define a new method and headers and such, then you could argue PUBLISH is wrong...

:)
Just as a reminder, take a look at the following link:
https://datatracker.ietf.org/doc/draft-yusef-dispatch-action-ref/
This was our first proposal, to use REFER; we introduced INVOKE after the discussion in SPLICES meeting and the concerns around overloading the REFER method, which is already heavily overloaded.

No one of the authors of these two draft really "*want* to define a new method".
What we really want is to define a proper mechanism for invoking actions on a UA.

Regards,
 Rifaat


> -----Original Message-----
> From: Hadriel Kaplan [mailto:[e-mail]]
> Sent: Friday, June 24, 2011 11:48 AM
> To: Shekh-Yusef, Rifaat (Rifaat)
> Cc: Gonzalo Camarillo; [e-mail]
> Subject: Re: [RAI] Device control using SIP
> 
> 
> On Jun 23, 2011, at 10:50 PM, Shekh-Yusef, Rifaat (Rifaat) wrote:
> 
> > The Controller can learn the state of the Controlled device using other
> means.
> > For example, a proxy application that is in the signaling path is aware of
> the state of the Controlled device, so it does not need to subscribe to
> anything.
> 
> I was only pointing out one can use SUB/NOT to learn it if one needs to learn
> it.  If you don't need to learn anything before issuing a command, that's
> fine. (personally I wouldn't rely on a Proxy to know the state of a UA by the
> messages that Proxy happens to forward, but to each his own)
> 
> 
> > Also, I thought that PUBLISH is used to publish state changes, not
> manipulate the state of a dialog.
> > Your proposal seems to overload and change the meaning of the PUBLISH
> request.
> 
> No it doesn't overload it.  It uses it for an application no one has used it
> for before, maybe, but SIP methods are not one-method-per-use-case.
> 
> I think one could argue it either way.  If you *want* to define a new method
> and headers and such, then you could argue PUBLISH is wrong... but if you want
> to not have to change SIP just to do this thing then I think it's not too hard
> to describe how PUBLISH is right.  This is all just legal academic posturing
> anyway; I think it's pretty clear there's nothing actually broken with using a
> PUBLISH to do this in practice. (ie, it won't screw up SIP the protocol, UAs,
> proxies, b2buas, deployments, etc.)
> 
> -hadriel