> On Feb 19, 2017, at 12:27 PM, Josh Soref <[email protected]> wrote:
> 
>> A client should attempt to fulfill at most one of these challenges,
> 
> fulfill is an odd word. And "attempt" is an odd word in concert. I'm
> pretty sure you're trying to say to a client "once you've fulfilled a
> challenge, you do not need to fulfill any additional challenges", not
> "you should only try one challenge, and if you fail, you should not
> try to complete any of the others".
> 
> The "at most one" text is odd... I suppose a client could attempt to
> fulfill zero challenges, but that seems pointless.
> 
>> and a server should consider any one of the challenges sufficient to make 
>> the authorization valid.
> 
> I think something like:
> 
> A server SHOULD treat the challenges portion satisfied when a client
> fulfills one challenge.
> 
> That should be sufficient to tell client implementations that they
> need to complete one, and that they don't need to complete more than
> one. Without telling them that if they try one and fail, they
> shouldn't try a different one.
> 
> FWIW, as a user, I run into this portion of the spec often. Typically
> my client tries https or http. But a friendly client would be willing
> to try both, stopping if the first one it tries completes, but
> continuing to the second if the first fails.

+1.  This is a significant improvement over the current text.

Russ
_______________________________________________
Acme mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/acme

Reply via email to