Re: Proposal: A pinning mechanism for CSP?

Yea, that is one of the problems with real-world pinning deployment today.
If I get something wrong I might lock out users for the max-age they were
previously pinned to. Safely being able reset that is compelling. :)

Aloha Mike,
--
Jim Manico
@Manicode
(808) 652-3805

On Jan 23, 2015, at 9:44 AM, Mike West <mkwst@google.com> wrote:

On Fri, Jan 23, 2015 at 6:38 PM, Jim Manico <jim.manico@owasp.org> wrote:

> All I'm saying is that if pinning config can be set via a manifest like
> structure vs headers, I'd suggest that headers take precedence.
>

The csp-pinning proposal (and PKP (and HSTS)) ignores any pin for a host
that already has a pin. First one in wins, which is good for security, as
it can't be maliciously overridden. Of course, it's bad for flexibility for
the same reasons. *shrug* That's a trade off I think we should allow
developers to make.

Actually, there might be some subtlety here (can't HSTS/PKP turn itself off
with a 0 max-age? Chris? Ryan? I didn't see that logic in a quick skim of
the RFCs)


>
> If I'm way off base or being disruptive, let me know off-list and I'll go
> back to lurking and popcorn.
>

Not at all! Feedback/questions are totally welcome!

-mike