Re: [rtcweb] Bridged line appearance? (Re: Usecase & architecture: Browser application with separate webserver & voipserver)

[Tim Panton <[e-mail]>]

On 7 Sep 2011, at 15:59, Cullen Jennings wrote:

> 
> On Sep 7, 2011, at 8:16 AM, Matthew Kaufman wrote:
> 
>> On 9/7/2011 4:10 AM, Harald Alvestrand wrote:
>>> 
>>> Since more people than I may be confused, I'm forking a subthread...
>>> 
>>> what is a bridged line appearance, and why is it hard to do in SIP?
>>> 
>>> My terminology cache is totally blank.
>>> 
>> 
>> Key system (and home phone) emulation requires two things that are hard to do (because there's no final specification, not because it is impossible) (and because there's 3+ alternatives, and almost no phones implement more than one of them) in SIP:
>> 
>> 1. Shared line appearance
>> 
>> This is where you can be on a call on one handset, see that the line is in use at the other handset. Place the call on hold at the first, pick it up at the second.
>> 
>> In business PBX cases, this is used for executive/assistant cases. In key system emulation it is used for "Bob, pick up the call holding on line 3". And for home phones it is the usual "put the phone on hold in the kitchen, pick it up in the den." (My wife wanted this functionality, so I had to add proprietary support for shared line appearance on my home PBX.)
>> 
>> 2. Bridged line appearance
>> 
>> This is where you can be on a call on one handset, see that line is in use at the other handset, and pick up the second handset to join the call.
>> 
>> In the business PBX and key system case this is sometimes used for supervisors to join a call to help, but the real common case is the home phones... "kids, go pick up the extension in the living room and talk to grandma."
>> 
>> With POTS, this works by simply paralleling two handsets on the same copper pair. For SIP it requires everything shared line appearance does *plus* automatic barge-in + conference on pickup.
>> 
>> The BLISS WG has been working, for a long time, on taking the various interim proposals and creating a standard out of them, but we're still not there... and yet without it, there's no way to use phones from two different vendors to accomplish this.
>> 
>> Whereas with WebRTC implemented *without SIP*, it is fairly easy to build a web app that runs on multiple browsers from multiple vendors and implements this.
> 
> I've sort of been humoring you but I'm going to push this one a bit. I think you are wrong. 
> 
> Give me one reasons why it would be any harder to do if the browser did SIP than if it did the type of API you are talking about. I think you are just wrong here. The web app in the JS and in the web server would still just use SIP to set up media the way it wanted - just like you would do with the other API. No one is proposing we use the BLISS shared line appearance or anything like that - we are proposing that when A wants to say what codecs to use for sending media to B, use that part of SIP. 
> 
> (oh, and you could replace SIP with the word Jingle in the above paragraph and it would equally true)

But try that paragraph with BRI or 'RFC 5456' or 'Freeswitch's XML socket' or ASCF's ICE over HTTPS  and it sounds a bit weird. Why ? 
what is the difference between being locked into one of them vs SIP?

Arguably SIP comes with more expectations and baggage than any of the above and there is the risk. Folks will embed a full SIP stack
before we know it - and the mere existence of a market in SBCs shows why that would be 'a bad thing' (tm)

Tim. (speaking for himself).