Re: [pkix] fyi: Sovereign Keys: an EFF proposal for more secure TLS

[Adam Langley <[e-mail]>]

On Mon, Dec 19, 2011 at 10:10 AM, Rob Stradling
<[e-mail]> wrote:
> But as for discouraging incorrectly encoded certificates: if just one major TLS
> Client rejected Adam's "terrible thing" signature algorithm parameters hack,
> then I'd imagine this would provide sufficient incentive for Adam to abandon the
> idea.

That's pretty much true - our tolerance for breaking things is pretty limited.

As a browser, we're in largely the same situation as the CAs ("there's
no point in taking a hard line because users will just move to a
different browser.") None the less, we do try to win small victories
where we can, and we take a fair amount of flack for it. (We banned
EDH parameters < 512 bits rather than work around it. We're pushing
1/n-1 record splitting before Firefox and IE, and thus taking a lot of
the heat. We're banning MD2/4/5 in certificate chains. We're banning
RSA keys < 1024 bits.)

But a way to include extra data in a certificate chain is valuable,
and not just for us (i.e.
https://github.com/moxie0/Convergence/wiki/TACK) We'll take it if we
can, even if it means having to pinch our noses.


Cheers

AGL