[Openid-specs-ab] MTI section in Messages Draft 15

Nat Sakimura sakimura at gmail.com
Thu Jan 31 23:28:12 UTC 2013


Needs for fine grained claims request gets in other jurisdictions than the
US, I can imagine.
With the Data collection limitation and Data minimization requirements
combined with PIA requirement after 2014, that gets pretty important in EU.

As to the perceived complexity of the building of the consent screen, I
really do not understand.
It has been there for a long time since the age of SREG and AX.
Our implementation has been doing it since June 2007 and our implementation
on the interop does it also. It has not been a big deal. (On the Shib.
land, uApprove's been developed to support fine grained authorization.)

Distributed and Aggregated claims support is a bit complex, I would admit,
though.

Nat


2013/2/1 Brian Campbell <bcampbell at pingidentity.com>

> It's almost like we don't work for the same company sometimes John... ;)
>
> As I said on the call today, we probably will support it eventually just
> because our shtick as a vendor is supporting standards. But that doesn't
> mean I'm a fan of the complexity and ROI (as I perceive it anyway).
>
> Tim's fish and bicycle analogy resonated with me too and I'd definitely be
> supportive of some kind of a distinction between support for the fine
> grained authorization stuff and some of the more straightforward (for lack
> of a better word) features of the Request Object.
>
>
> On Wed, Jan 30, 2013 at 4:32 PM, John Bradley <ve7jtb at ve7jtb.com> wrote:
>
>> Yes my main reason for wanting request object MTI is to hold it over you.
>>
>> I think it is different to say that it won't be available in the firs
>> release vs it will never be implemented.
>>
>> I expect that Ping will implement it.  (Product people don't talk to me
>> so take nothing I say as gospel)
>>
>> John
>>
>> On 2013-01-30, at 8:18 PM, Brian Campbell <bcampbell at pingidentity.com>
>> wrote:
>>
>> I'm chiming in late on this one but I did want to offer my 2 cents and
>> say that I'm pretty sympathetic to the viewpoint that Tim/Google put forth.
>> In fact, there's a good chance that our initial OP product offering will
>> not ship with support for the Request Object.
>>
>>
>> On Wed, Jan 30, 2013 at 3:21 PM, John Bradley <ve7jtb at ve7jtb.com> wrote:
>>
>>> That is probably not unreasonable if the OP doesn't want to look in the
>>> request object, top treat it as a request for all the default claims.
>>>
>>> The RP should still know via discovery if the OP supports it.
>>>
>>> The other thing that is in there is the max_age that the client wouldn't
>>> know if it was being ignored unless it also asked for auth_time.
>>>
>>> It is probably OK to have that happen as any RP wanting those enhanced
>>> features iOS going to need to know if it can trust the OP and have some
>>> info about them.
>>>
>>> So I would be OK  with request_object/request_url triggering a default
>>> in the OP if it doesn't support it.  What the default is will probably need
>>> more discussion.
>>> We would need to add discovery information and a warning about trusting
>>> max_age if you haven't asked for auth_time and don't know if the OP
>>> supports the request object.
>>>
>>> John
>>> On 2013-01-30, at 7:05 PM, Mike Jones <Michael.Jones at microsoft.com>
>>> wrote:
>>>
>>> At the very minimum, I believe that OPs must be required to not reject
>>> requests that include “request” or “request_file” parameters.  A more
>>> reasonable minimum might be to also require that if “request” or
>>> “request_file” parameters are present and the OP doesn’t choose to process
>>> the contents of an OpenID Request Object, that their presence be treated as
>>> a request for an OP-specified default set of claims.****
>>>
>>> That way, if, for instance, an RP uses “scope=openid” and a request
>>> object containing {“userinfo”:{“claims”:{“given_name”:null}}} that the RP
>>> will get some claims – even if they’re not the ones that it is expecting.
>>> ****
>>>
>>> In short, the OP should still treat the presence of “request” or
>>> “request_file” as requests for claims, even if it ignores the contents of
>>> the requests.****
>>>
>>>                                                                 -- Mike*
>>> ***
>>>
>>> *From:* John Bradley [mailto:ve7jtb at ve7jtb.com]
>>> *Sent:* Wednesday, January 30, 2013 1:41 PM
>>> *To:* Mike Jones
>>> *Cc:* Tim Bray; <openid-specs-ab at lists.openid.net>
>>> *Subject:* Re: [Openid-specs-ab] MTI section in Messages Draft 15****
>>> ** **
>>> I understand that from a simplicity point of view only scopes is easier,
>>> ****
>>> ** **
>>> Though Google currently stringing together a long list of URI in scopes
>>> to represent individual claims is not any more elegent.****
>>> ** **
>>> This separates the fine grained claims with parameters e.g. age over x
>>>  in the request object from the core grain scopes that can't have
>>> structure. ****
>>> ** **
>>> The other issues revolve around needing signed requests for higher LoA
>>> (I know most people don't care), the authentication context (though that
>>> can be set in registration).****
>>> ** **
>>> I have always suspected that the request object as MTI would het push
>>> back.  ****
>>> ** **
>>> If we were to drop it from MTI then discovery would need to say if the
>>> IdP supports it.   I don't think that is the end of the world for IdP who
>>> don't care about the addition;l functionality.****
>>> ** **
>>> However I would want to avoid having people invent other ways to encode
>>> claim requests that would not be interoperable.****
>>> ** **
>>> John B.****
>>> ** **
>>> ** **
>>> ** **
>>> ** **
>>> On 2013-01-30, at 2:07 PM, Mike Jones <Michael.Jones at microsoft.com>
>>> wrote:****
>>>
>>>
>>> ****
>>> Interesting.  The point of the Request Object is to give RPs control
>>> over the information they’re asking for and receiving.  For instance, if
>>> all my RP wants is your first name and the Request Object isn’t supported,
>>> it would have to use “openid profile” to get your first name, which also
>>> comes with middle name, last name, full name, nickname, preferred_username,
>>> profile URL, picture URL, website URL, gender, birthdate, time zone,
>>> locale, and time last updated.  That seems like overkill and doesn’t
>>> minimize disclosure of information to the RP.****
>>>  ****
>>> But I understand the simplicity/minimality argument for your position.**
>>> **
>>>  ****
>>> Let’s make this a discussion topic on tomorrow’s call.****
>>>  ****
>>>                                                             Thanks,****
>>>                                                             -- Mike****
>>>  ****
>>> *From:* openid-specs-ab-bounces at lists.openid.net [mailto:openid-
>>> specs-ab-bounces at lists.openid.net] *On Behalf Of *Tim Bray
>>> *Sent:* Wednesday, January 30, 2013 8:40 AM
>>> *To:* <openid-specs-ab at lists.openid.net>
>>> *Subject:* [Openid-specs-ab] MTI section in Messages Draft 15****
>>>  ****
>>>
>>> I refer to the material in
>>> http://openid.net/specs/openid-connect-messages-1_0.html#ImplementationConsiderations
>>>
>>> We’ve been discussing this at some length and probably would not ship a
>>> OP conforming to this draft, because our plans do not include support for
>>> OpenID Request Objects.  It seems perfectly possible to implement an
>>> Internet-scale federated-login system with good interoperability, security,
>>> user-experience, and developer-experience properties, entirely without the
>>> use of Request Objects.
>>>
>>> Given this, why are they considered essential for the MTI section?
>>> Absent Request Objects, our chances of shipping a conforming OP are pretty
>>> good.****
>>>   -Tim****
>>>
>>>  ****
>>> _______________________________________________
>>> Openid-specs-ab mailing list
>>> Openid-specs-ab at lists.openid.net
>>> http://lists.openid.net/mailman/listinfo/openid-specs-ab****
>>>
>>>
>>>
>>> _______________________________________________
>>> Openid-specs-ab mailing list
>>> Openid-specs-ab at lists.openid.net
>>> http://lists.openid.net/mailman/listinfo/openid-specs-ab
>>>
>>>
>>
>>
>
> _______________________________________________
> Openid-specs-ab mailing list
> Openid-specs-ab at lists.openid.net
> http://lists.openid.net/mailman/listinfo/openid-specs-ab
>
>


-- 
Nat Sakimura (=nat)
Chairman, OpenID Foundation
http://nat.sakimura.org/
@_nat_en
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openid.net/pipermail/openid-specs-ab/attachments/20130201/7180dd36/attachment-0001.html>


More information about the Openid-specs-ab mailing list