[Openid-specs-ab] Lite Draft 9

Breno de Medeiros breno at google.com
Mon Aug 22 19:31:51 UTC 2011

On Mon, Aug 22, 2011 at 12:05, Allen Tom <allentomdude at gmail.com> wrote:
> I think it might be confusing to developers to have multiple access tokens.
> I don't think I've seen any other Connect/OAuth type implementations that
> return multiple access tokens. Are there any examples out there?

Yes. Facebook Connect uses signed_request as the id_token.

> Instead of using the id_token as the access_token for the check session
> endpoint, it might seem more natural to validate the id_token by submitting
> it as a regular parameter instead?
> Allen
> On Mon, Aug 22, 2011 at 11:22 AM, John Bradley <ve7jtb at ve7jtb.com> wrote:
>> In my discussion with Breno, his intention is that the Check Session
>> endpoint is a OAuth protected resource.  The id_token is sent as the access
>> token for accessing that resource, per the OAuth spec.
>> It is NOT a parameter for a API,  it is the access token.
>> Breno should correct if I have that wrong.
>> It makes the most sense to me to think of openID Connect returning two
>> access tokens for two protected resources, perhaps I am simple.
>> John
>> On 2011-08-19, at 10:06 PM, Nat Sakimura wrote:
>> question:
>> The messages spec states:
>> 3.4.1.  Check Session Request
>> To request the information about the authentication performed on the user,
>> the following parameters are sent to the Check Session endpoint:
>> id_tokenREQUIRED. The ID Token obtained from an OpenID Connect
>> authorization request.
>> and the Standard spec states:
>> 7.1.  Client Session Requests
>> Clients MUST make a HTTP POST request using transport-layer security with
>> the following application/x-www-form-urlencodedparameters in the request
>> body:
>> id_tokenREQUIRED. The ID Token obtained from an OpenID Connect
>> authorization request.
>> The Following is a non-normative example of an Check Session request:
>> I think these descriptions are correct from older drafts point of view.
>> Since it is just a parameter, it can be used in conjunction with
>> authentication.
>> If it is changed to be an access token, this description is wrong needs to
>> be updated.
>> Could you kindly clarify?
>> =nat via iPad
>> On 2011/08/20, at 4:12, John Bradley <ve7jtb at ve7jtb.com> wrote:
>> The id_token is the access token for the check session endpoint.
>> Only the id_token is sent to the check session endpoint.
>> As Oauth only has one access token we have to give the access token for
>> the session endpoints a separate name.  That is id_token.  I did rase the
>> possibility of calling it session but no one took me up on that.
>> The access token for the check session endpoint is a signed JWT that way a
>> client can inspect it directly and never use the check session endpoint.
>> John B.
>> On 2011-08-19, at 3:06 PM, Allen Tom wrote:
>> In section 3.3.1 - Are both the access_token and the id_token supposed to
>> be sent to the Check Session endpoint? The way that Section 3.3.1 in Draft 9
>> is currently written, it sounds like only the id_token is sent in the
>> request, and that the id_token is actually the access_token.
>> It would probably be helpful to have an example Check Session request in
>> the spec.
>> Allen
>> On Fri, Aug 19, 2011 at 12:02 PM, Allen Tom <allentomdude at gmail.com>
>> wrote:
>>> The explanation in Section 3 regarding when to use the Implicit vs Code
>>> flow is vague, because it's not clear as to what it means for a client to
>>> securely maintain state between itself and the auth server.
>>> It might be better to just say that the Code flow should be used if the
>>> redirect_uri doesn't use HTTPS, and if the client is able to securely store
>>> its client_secret.
>>> Allen
>> _______________________________________________
>> Openid-specs-ab mailing list
>> Openid-specs-ab at lists.openid.net
>> http://lists.openid.net/mailman/listinfo/openid-specs-ab


More information about the Openid-specs-ab mailing list