Re: Defining secure-enough origins.

On Thu, Aug 28, 2014 at 6:14 PM, Jeffrey Yasskin <jyasskin@google.com> wrote:
> Since an origin is just (uri-scheme, uri-host, uri-port)--effectively a
> string--but insecurity and authentication in MIX change based on
> whether "the user agent discovers only after performing a
> TLS-handshake that the TLS-protection offered is either weak or
> deprecated", I'm not sure it's appropriate to talk about authenticated
> or insecure "origins". I think it's the _resource_ that becomes
> insecure if it turns out to have been transferred over a TLS-deficient
> connection.

Origins are very much objects. They are either a tuple (sometimes
comparisons are just done on the scheme), an identifier, or a pointer
to another origin (called alias at the moment). Some engines might
also have the certificate pinned to the origin so for TLS-origins an
additional comparison is made other than scheme/host/port.


> The "authenticated environment" term is nice, because it's easy to get
> to an environment from any IDL description.

>From the last time Ian weighed in it seems we need to move from
environment to global object. But I haven't had the time to dig in
that again.


-- 
http://annevankesteren.nl/

Received on Sunday, 31 August 2014 17:02:36 UTC