Re: [TLS] Renego Indication RI patch interaction with TLS major version interop

[Marsh Ray <[e-mail]>]

> I just mean that the article is misleading in that it implies that the vast
> majority of servers are implementing the renegotiation protection draft
> incorrectly, when really they are implementing something else wrongly that
> doesn't have a *practical* effect on renegotiation protection.

I agree. Still it seems weird though that patching for RI would
introduce new breakage systematically across multiple implementations or
distributions, which is why I brought it to this list. No evidence of a
spec bug yet it seems.

Given the adoption rate (0 to 12% increase in time period of graph) I
would have thought the "all servers" (red) line would have shown a
little of the wild variation from the "renego patched" (blue) line.
Perhaps it's not all servers, but non-patched servers on the red line.

> If the test
> fails for 0x03FF then, yes, there may be something to worry about. But I
> think we would all be quite relieved if supporting versions >= 0x0400 was
> one of the top 10 (or 100) most pressing issues regarding TLS.

Still, we need to be aware of it. As we've seen, this class of issue can
just as easily be a problem with clients advertising version 1.1. Thank
you Yngve and Opera for compiling and publishing this info!

> Regarding NSS on servers: I think it will become increasingly common due to
> initiatives like the following
> http://fedoraproject.org/wiki/FedoraCryptoConsolidation
> http://en.opensuse.org/SharedCertStore

Those are interesting projects. Though I'm not sure I'm ready to unify
my web browser and SSH server under a single trust model yet. :-)

- Marsh