- From: Michal Zalewski <lcamtuf@coredump.cx>
- Date: Tue, 4 Nov 2014 15:07:43 -0800
- To: Tanvi Vyas <tanvi@mozilla.com>
- Cc: Chris Palmer <palmer@google.com>, Joel Weinberger <jww@chromium.org>, Frederik Braun <fbraun@mozilla.com>, Pete Freitag <pete@foundeo.com>, "public-webappsec@w3.org" <public-webappsec@w3.org>
For what it's worth, for all the attention that HTTPS is getting of recent (for good reasons), I suspect that an opportunistic compromise of a JS-hosting CDN is a more endemic and pervasive risk to most users than the possibility of active MITM on unencrypted wifi or so. While I'm strongly in support of pushing high-profile sites over to HTTPS (*koff* Amazon *koff*)... unless we're completely writing HTTP off, it's probably reasonable to allow the "bad CDN" problem to be addressed by SRI independently of deploying HTTPS everywhere. /mz On Tue, Nov 4, 2014 at 2:41 PM, Tanvi Vyas <tanvi@mozilla.com> wrote: > SRI on an non-secure origin does not protect the user from an active network > attacker that is MITM'ing their connections and changing their content. But > SRI on a non-secure origin does protect the user who is not subject to an > active attack and who is trying to source javascript from a third party that > has been compromised. For me, that's a good enough reason to allow SRI on > non-secure origins. > > Most users aren't going to understand SRI at all and it's not a user facing > feature, so we aren't giving them any new or misleading security guarantees. > There may be other new features that we want to restrict to HTTPS as a stick > to get developers to migrate to secure origins, but features like SRI that > can only help protect users shouldn't be put in that bucket. > > > On 11/3/14 4:28 PM, Chris Palmer wrote: >> >> On Mon, Nov 3, 2014 at 4:02 PM, Joel Weinberger <jww@chromium.org> wrote: >> >>>> Although it would be desirable for every site to use HTTPS, >>>> I don't think that SRI is the right way of promoting this. >>> >>> This isn't a matter of promoting HTTPS; it's a matter of suggesting to >>> users >>> and developers that they're getting a security property that they're >>> simply >>> not getting. >> >> Exactly. >> >> I'm increasingly concerned about this idea, popping up in several >> contexts now, that there can be any security at all without at least >> strong server authentication, data integrity, and data >> confidentiality. Like, we need those as the baseline minimum, so that >> we can *begin* to think about the next problems, like metadata >> confidentiality. >> >> Putzing around in the margins like this (Better Than Nothingism) is >> not going to help users. >> > >
Received on Tuesday, 4 November 2014 23:08:30 UTC