Hi Carl,
thanks for your review comments.
Reviewer: Carl Wallace Review result: Has Nits
Comments/questions - Section 2.2 states "When different common data fields are used in combination, the permissions the client requests are the cartesian product of the values." What would that mean for the example in Figure 7 where geolocation is asserted?
The example in figure 7 asks for read and write access to metadata and images managed by a photo api offered at two different endpoints if the geo locations in the respective pictures are at one of the coordinates given in the geolocations array.
Are data fields other than the common types not processed this way? Yes, they are.
- The MUST in the first sentence of the third paragraph in Section 3.1 should probably be a SHOULD.
From an interoperability standpoint, it is important the client can expect the AS to consider both sets of requirements. The paragraph before recommends against using both scope and authorization_details.
for the same API. However, if the same client asks for access to multiple APIs in the same authorization request and different APIs use one or the other method, that is nevertheless be supported by the spec. That’s also important to allow pre-existing APIs and new APIs to co-exist.
For example, OpenID Connect uses scopes but newer APIs for Open Banking might use RAR. Asking for identity claims and the permission to execute a payment in the same transaction might make a lot of sense. The rest of the paragraph leaves open how to handle the combination (presumably including ignoring one or the other).
The normative requirement is to support both in the same authorization request. - It is unclear to me how the "MUST consider both sets of requirements in combination with each other" language in Section 3.2 squares with the "resource parameter does not have any impact on the way the AS processes the authorization_details" language in Section 3.3 for a request that includes scope, resource and authorization_details w/locations. If scope and resource were used before, it seems odd to only consider scope during a migration period.
good point. What we want to get across is: Whatever was valid before, is not rendered invalid by RAR. scope and resource can be combined as defined in RFC 8707 - that's the first sentence. What 3.2 tries to convey is that while the resource has an impact on the interpretation of the scope, it does not have any impact on the authorization details - that’s the second sentence. - In section 7, the first paragraph has a MUST regarding sending "the authorization details as granted by the resource owner". The third paragraph leaves room for discretion. Should the MUST be a SHOULD?
That makes sense. Changed the wording to MUST.
Nits - Expand AS and RS on first use
done
- The security considerations should echo the requirement that the AS ensure that there is no collision between different authorization details types that it supports.
added sentence to the second paragraph in the security considerations section - In section 6, the bulleted list of privileges is incomplete. The example in Section 3 also allowed for checking the status of and canceling a payment.
- Section 11.1 should probably include a bullet regarding client use of the authorization_details_types metadata.
Added “(if needed) Entitle clients to use certain authorization details types” - In the second paragraph of Section 12, "all strings" should probably be "all strings in an authorization_details parameter".
done
best regards, |
--
last-call mailing list
last-call@xxxxxxxx
https://www.ietf.org/mailman/listinfo/last-call