BTW EHL, I haven't read all of -12 yet but I did read section 2 and, even
though I was among those that originally pushed for the client assertion
stuff, I think that where you've taken it is pretty good.

On Fri, Jan 21, 2011 at 9:45 AM, Brian Campbell
<[email protected]>wrote:

> I think we are generally in agreement on this Phil although there are
> legitimate cases where the client doesn't need to authenticate so having the
> spec require it is too restrictive.
>
> On Thu, Jan 20, 2011 at 3:50 PM, Phil Hunt <[email protected]> wrote:
>
>> I don't think this is a case of providing alternative methods.  It is just
>> a case of allowing ANY method that the Authentication server deems
>> sufficient. Kerberos, SAML, SSO token, basic auth, mutual SSL, etc.
>>
>> Thus you don't need to put specific authentication methods in the spec
>> other than to simply REQUIRE that the client be authenticated -- either
>> inbound with the POST as you suggest or OOB during a previous exchange.
>>
>> It may be a good idea to define one mandatory to implement methodology,
>> but to define more than one suggests limiting authentication approaches to
>> specific methods. I don't think this is the intent here.
>>
>> Phil
>> [email protected]
>>
>>
>>
>>
>> On 2011-01-20, at 2:28 PM, Marius Scurtescu wrote:
>>
>> > On Thu, Jan 20, 2011 at 2:14 PM, Phillip Hunt <[email protected]>
>> wrote:
>> >> The client does need to know how to authenticate. But given that it
>> already has to know a lot about the service, you would think acceptable
>> authentication types are well known to the client.
>> >
>> > Yes, true.
>> >
>> >
>> >> What is the problem with the client authenticating like any normal web
>> service client? (IE outside of oauth)
>> >>
>> >> Why involve oauth in any authentication for User or client?
>> >
>> > What is a normal web service client? Since the client has to POST a
>> > bunch of parameters as form encoded, it is simpler to also send
>> > authentication parameters there. Providing alternate methods of
>> > authentication is just asking for trouble. Of course, the spec could
>> > require that authentication is only sent through some other method,
>> > but I think it is way too late for such a change.
>> >
>> >
>> > Marius
>>
>>
>
_______________________________________________
OAuth mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/oauth

Reply via email to