- From: Martin Thomson <martin.thomson@gmail.com>
- Date: Mon, 7 Mar 2016 14:27:10 +1100
- To: Mark Nottingham <mnot@mnot.net>
- Cc: Stephen Farrell <stephen.farrell@cs.tcd.ie>, The IESG <iesg@ietf.org>, Mike Bishop <michael.bishop@microsoft.com>, HTTP WG <ietf-http-wg@w3.org>
On 7 March 2016 at 13:19, Mark Nottingham <mnot@mnot.net> wrote: > It's just saying that clients can and use additional means to validate certificates; i.e., they're not obligated to accept a cert if it passes the 2818 checks. In practice, browsers do pinning checks, blacklist checks, revocation checks [1], CT signature checks, user override checks, and probably things that I'm not aware of. The intent was to avoid limiting validation behaviour. My initial reaction was that this wasn't interoperable. I still think we could do better, but don't want to burden this effort unreasonably, defining what it means to validate a certificate turns out to be hard. [1] Hah, had you going there, we don't. Well... unless there is a must-pin policy.
Received on Monday, 7 March 2016 03:27:42 UTC