On 31 Dec 2015, at 00:59, Brian E Carpenter
<brian(_dot_)e(_dot_)carpenter(_at_)gmail(_dot_)com> wrote:
On 31/12/2015 09:26, Olle E. Johansson wrote:
...
So we have bugs related to IPv6, we have a group of authors and the
documents still are not going anywhere because a lack of interest.
Interest is usually generated by unhappy users, and there isn't quite enough
of
the world depending on IPv6 yet to create the necessary unhappiness. I'm not
pessimistic though: we just have to be patient, and eventually the
pointy-haired
bosses will start to insist on fixing IPv6 bugs.
And when they do, I want the IETF to support the efforts and put energy into it.
We need IPv6 and dual stack to work properly in all applications.
I don’t want the efforts of these pointy-haired-boss-supported engineers to be
met
like our work was handled. I know there’s a lot of more sexy stuff than fixing
bugs in
old protocols, but I am personally surprised that it has required so much time
and effort
to fix small issues. Considering we found the issues at SIPit almost five years
ago and
started to make noises about it, I’m disappointed that developers still haven’t
got the
needed information to code things right.
We have found issues surrounding TLS use in SIP at SIPit events, but the
interest
in fixing that seems even smaller. Based on my experience of the IPv6 work, I
am not
sure I have the resources and energy to even start working with it funded by
myself.
And it isn't just the IETF. Just to take a random example that bit me
recently,
why doesn't the Python socket module define IPPROTO_IPV6? A tiny thing that is
easy to work around, but I gather it's been like that for years.
Ouch.
Documents (in response to Jari’s question)
----------------------------------------------------------
Here’s one (now expired) draft we put a lot of efforts into and it got stuck
somewhere
in dispatch or sipcore - the chairs couldn’t make up their mind what to do with
it.
I personally still feel that developers can get a lot of help by a document
like this.
This document was more of a foundation for the other work, an informational
document
based on experiences we found at SIPit and other tests.
https://tools.ietf.org/html/draft-klatsky-dispatch-ipv6-impact-ipv4-03
The other draft got dispatched to the SIPcore wg, has gotten a review after a
long time. We
started working on an update based on the review in time for an IETF meeting
(don’t remember which one) but then the WG meeting got cancelled and I think we
lost energy
there. Yes, we should pursue the work at some point.
https://datatracker.ietf.org/doc/draft-johansson-sip-dual-stack/
The third one was never written, but discussed and planned. That’s the one
where we discuss happy eyeballs for SIP transports. Basically “avoid UDP,
it can’t easily be solved unless you use DNS records in a clever way."
/O