[OpenID] Bug in OpenID RP implementations

Fri Jan 2 00:36:39 UTC 2009

Hi Peter,

I (mostly) share your idealistic view of myopenid.com.  I use it as a frame
of reference often.  However, I don't agree with all of their decisions (AX
type URIs and non-https enforcement among them).

It seems to me that http*s*://homepw.myopenid.com/ results in
myopenid.comreferring to an HTTPS OP endpoint, and the
http://homepw.myopenid.com/ referring to an HTTP OP endpoint.  Fair enough.
If you start your discovery securely it makes sense to continue with secure
authentication.  And if your discovery was insecure, well then the
authentication isn't secure whether it uses HTTPS or not since the discovery
could have been hijacked and authentication rerouted to another address.

Ideally, myopenid.com and all other OPs would redirect identity page
requests that come in on HTTP to HTTPS so that all claimed identifiers and
authentication would occur over HTTPS to provide higher security to users.

Regarding whether mere users have the choice of HTTPS being used for auth,
they *may*.  In your myopenid.com claimed id example the user has no control
since myopenid.com hosts the claimed id page.  However, if you take
http://blog.nerdbank.net, which is one of my claimed identifiers which
delegate to myopenid.com (and others via XRDS), since I am the one to write
the delegation tags I get to decide whether to use an OP's HTTPS or HTTP
endpoint.  So yes, the user may have a choice about using a HTTP(S) OP
endpoint, but they likely will not have that choice if they use an OP-hosted
identity page.

--
Andrew Arnott
"I [may] not agree with what you have to say, but I'll defend to the death
your right to say it." - Voltaire

On Thu, Jan 1, 2009 at 4:51 PM, Peter Williams <pwilliams at rapattoni.com>wrote:

>  Yes SP=RP. Remind me to use RP only, here, and keep SP for OAUTH (since
> OAUTH and openid seem to have a potential marriage ahead).
>
>
>
> I'm now logged into my classical OP site (homepw.myopenid.com). I can't
> see how to control "my" metadata to allow "authentication" only over https.
>
>
>
> Should I infer in the movement that "really" OPs decide whether the
>  https-class openid they provision will or will not "authenticate" over
> https – and not mere users?
>
>
>
> For me, myopenid OP is/was the "gold standard" of openid OP implementations
> – an expression of the core use cases and the movement's management/control
> goals in their most primitive form. If I keep that belief, evidently
> users/subscribers are not "really" supposed to deciding whether https must
> be used during authentication. RPs may decide so. OPs may decide so. But
> users cannot.
>
>
>
> *From:* Andrew Arnott [mailto:andrewarnott at gmail.com]
> *Sent:* Thursday, January 01, 2009 3:27 PM
> *To:* Peter Williams
> *Cc:* Eddy Nigg (StartCom Ltd.); OpenID List
>
> *Subject:* Re: [OpenID] Bug in OpenID RP implementations
>
>
>
> SP?  Do you mean RP?  SP is an acronym that applies to OAuth.  OpenID uses
> RP and OP.  I assume by SP you mean RP here...
>
> The OpenID 2.0 spec section 15<http://openid.net/specs/openid-authentication-2_0.html#security_considerations>calls out several opt-in measures that an RP or OP can take to increase
> security for the authentication process.  But no, as has been stated an RP
> is not obliged per (my reading of) the spec to require that if discovery is
> done using HTTPS that authentication must also be done using HTTPS.
> --
> Andrew Arnott
> "I [may] not agree with what you have to say, but I'll defend to the death
> your right to say it." - Voltaire
>
>  On Thu, Jan 1, 2009 at 4:07 PM, Peter Williams <pwilliams at rapattoni.com>
> wrote:
>
> Is the SP right or wrong to redirect to an http OP endpoint, given an https
> openid, in your understanding of the spec (and what it means to be a
> conforming implementation)?
>
>
>
>
>
> Though operating in openid1 legacy mode, neither SP not OP objected –
> probably because the spec does call out for code to raise an exception –
> presumably because it isn't one, formally.
>
>
>
> *From:* general-bounces at openid.net [mailto:general-bounces at openid.net] *On
> Behalf Of *Eddy Nigg (StartCom Ltd.)
> *Sent:* Thursday, January 01, 2009 3:00 PM
> *Cc:* OpenID List
>
>
> *Subject:* Re: [OpenID] Bug in OpenID RP implementations
>
>
>
>
>
> On 01/02/2009 12:49 AM, Eric Norman:
>
> On Jan 1, 2009, at 4:28 PM, Peter Williams wrote:
>
>
>
>
>
> The openid 2 spec says in section 15 (a non-normative must, note):
>
>
>
> "In order to get protection from SSL, SSL must be used for all parts
>
> of the interaction, including interaction with the end user through
>
> the User-Agent."
>
>
>
>
>
> When I include "https:" in my OpenID, I'm saying that I
>
> want protection by SSL, right?
>
>
>
>
> Your OpenID is https:// then, it's not ncesseraly the same as http and the
> other way around too. It has been many times already mentioned.
>
>
>
> So if something elsewhere decides not to use SSL for
>
> whatever reason, that would be incorrect behavior, right?
>
>
>
>
> Correct. However an OP may return the claimed OpenID as https (there are
> for example some OPs which don't do plain http, only https via redirect.
>
>
>
> And let's not forget that the error message I quoted is
>
> clearly inappropriate.
>
>
>
>
>
>
>
>
>
> Regards
>
>
>
> Signer:
>
> Eddy Nigg, StartCom Ltd. <http://www.startcom.org>
>
> Jabber:
>
> startcom at startcom.org
>
> Blog:
>
> Join the Revolution! <http://blog.startcom.org>
>
> Phone:
>
> +1.213.341.0390
>
>
>
>
>
>
> _______________________________________________
> general mailing list
> general at openid.net
> http://openid.net/mailman/listinfo/general
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openid.net/pipermail/openid-general/attachments/20090101/baae734e/attachment-0002.htm>