Re: [TLS] Call for Consensus on removal of renegotiation

[Martin Thomson <[e-mail]>]

On 25 June 2014 13:03, Joseph Salowey (jsalowey) <[e-mail]> wrote:
> 1.  In favor of removing renegotiation
> 2.  In favor of removing renegotiation with the addition of rekey facility
> 3.  Not in favor of removing renegotiation

2 > 1 >> 3 > 4

>From a browser perspective, 1 would be OK: we simply don't have
connections that live that long.  But when you have long-lived
connections, rekeying is highly valuable.  It's also clear to me that
some users want to rekey more often than that even.

I don't think that there is any value in keeping renegotiation, but
having artificial constraints on what can change.  I think that it's
significantly easier to reason about *anything* where the things that
are immutable simply cannot change.

For my #4.... I'll note that the options don't include anything like
the poor-mans renegotiation: end the TLS session and make a new one on
the same underlying transport.  Is that intentional?  In case you
decide to poll that option, I think that it's strictly worse than
renegotiation due to the 'dead air' problem.