> Am 10.05.2019 um 22:27 schrieb George Fletcher <[email protected]>:
> 
> One thing to keep in mind with the "Push Request Object" model and the 
> concept of a more detailed scope structure, if the specified values are not 
> for a single transaction, then the AS will be required to keep the "Pushed 
> Request Object" for the "lifetime" of the access_token and possibly the 
> refresh_token depending on the use case. This could have some implementation 
> issues for the AS.

It’s not necessarily the request object but the client id and the scope. Can 
you please explain what implementation issue you expect? I don’t see much of a 
difference between storing/maintaining simple scope values and a JSON document 
for a grant.


> 
> Thanks,
> George
> 
>> On 5/10/19 9:52 AM, Dave Tonge wrote:
>> Thanks Torsten for this article - it is incredibly helpful.
>> 
>> I'm very much in favour of the "structured_scope" approach. 
>> 
>> While I understand George's point I think the line is very blurred between 
>> coarse-grained scopes and fine-grained transaction consent. In addition 
>> fine-grained authorisation metadata is needed for ongoing access APIs as 
>> well, e.g. how can a client ask for ongoing access to:
>>  - transactions in a users accounts with ids abc123 and abc124
>> 
>> From a UX perspective it is beneficial for the AS to ask the user for 
>> consent once. The AS therefore needs to have all the information about 
>> relating to the consent available when the user is redirected to the 
>> authorization endpoint. There should be a standard way for the Client to 
>> pass this data to the AS and I think structured scopes either sent as a 
>> query param or in a             request object are a neat way of doing this. 
>> 
>> Dave
>> 
> 

Attachment: smime.p7s
Description: S/MIME cryptographic signature

_______________________________________________
OAuth mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/oauth

Reply via email to