Re: [OAUTH-WG] Review of draft-ietf-oauth-dyn-reg-metadata-00

[Justin Richer <[e-mail]>]

On 04/24/2014 05:06 AM, Hannes Tschofenig wrote:
> Hi Justin,
>
> thanks again for the quick response.
>
> A few notes below.
>
>
> On 04/23/2014 10:25 PM, Justin Richer wrote:
>> Thank you for the review, responses inline (and this draft still needs
>> to be factored back into the main one):
>>
>> On 04/23/2014 03:14 PM, Hannes Tschofenig wrote:
>>> Hi all,
>>>
>>> I read through the document as part of my shepherding task; it is nicely
>>> written and easy to understand.
>>>
>>> I only have a few minor suggestions:
>>>
>>> * client_uri: URL of the homepage of the client.
>>>
>>> Would it be better to say that this is the URI provides further
>>> information about the client software (provided by the client software
>>> developer)?
>> I personally think "homepage" is a clear designator of that, but would
>> not be adverse to extending the definition somewhat.
>>
>>> * logo_uri: The value of this field MUST point to a valid image file.
>>>
>>> Would it make sense to provide a type field here as well, such as in
>>> HTML (e.g., type="image/png")?
>> Unless we're going to allow the client to register and manage multiple
>> images (please, no), then we shouldn't go out of our way to store
>> mimetypes. This is a URL to be fetched and/or stored -- its content
>> doesn't much matter.
>>
>>> * contacts: Would these email addresses be in the format of
>>> mailto:[e-mail] or would you just use [e-mail]?
>>> I am asking because with the URI scheme one could potentially provide
>>> other contact information here as well, such as XMPP URIs or so.
>> I think you could do either, but what I've seen deployed is the non-URI
>> version of [e-mail]. I think it would make sense for it to be
>> listed as not just email addresses but also other contact URIs --
>> however, I don't think it's particularly useful to devolve into a full
>> contact record. It should be simple and actionable by a person, and
>> email fits that bill (as well as XMPP might, arguably).
> Maybe you want to say that this is a list of email addresses without the
> mailto URI scheme prefix. I am just seeing others putting all sorts of
> different stuff in there and expect it to work.

 From what I've seen, it's meant to be a human-facing string. So we 
should either call it a plain text string that MAY be an email address 
or other contact URI, or a formatted string with a particular format 
such as email without the mailto: scheme. I would lean toward the former 
and wash our hands of it. If people want a structured contact thing, 
people can standardize around an extension on a new metadata field.

>
>
>>> * policy_uri: Would it be better to call this a privacy notice rather
>>> than policy document?
>>> Here is a short description what a privacy notice is:
>>> https://www.privacyassociation.org/resource_center/privacy_glossary/privacy_notice
>> I would think that policy document is a superset of privacy notice, right?
> I am not sure what a policy document is.
>
> When you register your application with Facebook then you have to
> provide a link to your privacy notice. That made sense to me given the
> nature of the data sharing that is going on.
>
>>> * jwks_uri: The text provides little information about how this element
>>> is used. I believe that this is an alternative way of using the PoP
>>> architecture, where the client registers keys with the authorization
>>> server that can then be tied to access tokens. Right? I could add some
>>> text in the PoP overview document to explain this and maybe you could
>>> include a reference to the PoP document (as an informative reference,
>>> for example).
>> The text states that this is there for the use of higher level
>> protocols. Several of which (including OIDC and BB+, not to mention more
>> advanced token types and client authentication methods) have a need to
>> tie a key to a client. This, and the proposed jwks value, provide a hook
>> for that kind of stuff.
> It is difficult to describe the semantic if it is not further explained
> how it is supposed to be used. Is this field used in OpenID Connect?

Yes, it's used in OpenID Connect and other protocols that add 
key-related stuff on top of it.

>
> Ciao
> hannes
>
>
>>   -- Justin
>>
>>> Ciao
>>> Hannes
>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> OAuth mailing list
>>> [e-mail]
>>> https://www.ietf.org/mailman/listinfo/oauth