Greetings all:

SIPIT 12 was held in Stockholm, Sweden Feb 24-28.
Hotsip hosted the event and did a phenomenal job.
We had global attendance. There were over 100 attendees
from over 40 organizations. I estimate there were nearly
100 discrete implementations represented.

We used the new domain "sipit.net" for the first time
at this event, serving primary DNS from Hotsip provided
servers at the event. Each team had its own subdomain and
was able to edit its own zone file directly. Teams were
much more willing and able to test SIP using domain names
instead of numeric IP addresses. Several teams were making
frequent adjustments to their SRV records, testing proper
resolution and routing. We will continue to provide this
infrustructure component at future events.

The common theme that emerged at this event was one of working
towards richer content and features. Testers were attempting
more intricate media constructions (both multimedia and 
alternative codecs on single streams), richer presence state,
and better security features.

There was a sharp increase in the number of implementations
supporting TLS and SIPS. Startup stuttered as folks wrestled
with setting up trust relationships. Most implementations had
been testing with self-signed certificates and required prior
knowledge of peer's certificates before accepting connections
based on them. There was some confusion around which fields
to use inside a certificate for an identifying key. We were able
to work past those and get a large spiral to function using
hop-by-hop TLS indicated by SRV. We identified a concern with
respect to record-routing when using TLS, similar to the old
question about record-routing when using non-default transports.
How can you be sure that the record-routing elements on either
side of you will accept each other's credentials? This will
require some list discussion. For now, the safe answer is to
always record-route if you are using TLS.

A smaller set of implementations supported the sips: URI scheme.
Most handled the extra requirements it imposed correctly. A few
couldn't be prevented from stepping down to sip:. 

We had two implementations of S/MIME that successfully exchanged
S/MIME encrypted messages. They also exercised the 493 response
correctly during initial testing.

It's possible that every implementation of sigcomp on the planet
was represented at the event. These teams tested extensively,
finding a few specification clarifications in the process.

Simple testing showed strong interoperability of the sip-events
infrastructure, but reinforced the need for all implementations
to adopt the baseline pidf document format and follow the 
guidelines therein for extending it. Several teams were attempting
to provide richer functionality through non-interworking extensions.

On the floor, the most frequent issues I heard being discussed
revolved around non-trivial use of SDP (things like specifying
different ptimes for each codec listed on a given m line).

I asked at the event for feedback on whether the twice per year
format was working or if the participants would prefer to return
to more frequent events. I only got one response, in favor of
twice a year.

I had several people ask that the multiparty tests not be
separated from the main testfloor (they were held in a
separate room in Stockholm). I also would prefer they be
in the same room at future events. I didn't receive many
specification interpretations this time, and I suspect its
partly because it was inconvenient to bring the questions to
me.
 
If you attended the event, and have comments about what was
effective and what wasn't, please send them to this list or
directly to me. The next event will be hosted by Mitel in
Canada in August.

RjS

_______________________________________________
Sip-implementors mailing list
[EMAIL PROTECTED]
http://lists.cs.columbia.edu/mailman/listinfo/sip-implementors

Reply via email to