[asterisk-dev] Via/maddr: (was Re: [Code Review] Properly routeresponses according to the Via headers in the request)

[Watkins, Bradley]

>-----Original Message-----
>From: asterisk-dev-bounces at lists.digium.com 
>[mailto:asterisk-dev-bounces at lists.digium.com] On Behalf Of 
>Olle E. Johansson
>Sent: Monday, January 03, 2011 4:09 AM
>To: Asterisk Developers Mailing List
>Cc: Olle E Johansson
>Subject: Re: [asterisk-dev] Via/maddr: (was Re: [Code Review] 
>Properly routeresponses according to the Via headers in the request)
>
>
>3 jan 2011 kl. 09.51 skrev Klaus Darilion:
>
>>> I have no examples of maddr, but here's the one from the RFC:
>>> Via: SIP / 2.0 / UDP first.example.com: 4000;ttl=16;maddr=224.2.0.1 
>>> ;branch=z9hG4bKa7c6a8dlze.1
>>> 
>>> Since this has a ttl value, the maddr is supposed to be multicast.
>> 
>> Btw: Has anybody ever seen usage of maddr or can describe to 
>me what it is useful for? I always thought I'm quite familiar 
>with SIP, but still I do not understand why maddr is needed at all.
>> 
>> 
>In the bug tracker, you can find examples of a broken Nortel 
>system that doesn't send a valid hostname or IP in the via 
>header, but use maddr to define a port. To me, that seems 
>broken but valid. 
>
>I had a discussion with Iñaki about maddr in via headers and 
>we suggested that someone should write an RFC called "via 
>maddr considered harmful". That why I want to be able to 
>disable support of it. Using multicast is very uncommon and I 
>don't want other systems to tell me where to send responses. 
>
>/O
>

Indeed I am forced to interoperate with exactly such a Nortel system (a CS1000, if anybody cares), and quite frankly it doesn't work with vanilla Asterisk.  I have to hack around it using Kamailio (which I use anyway, but still...).

They pretty much use maddr for any time they didn't feel like properly implementing something in SIP. :/

- Brad