On 19/11/2010, at 12:49 AM, Roque Gagliano wrote:

> Hi Geoff,
> 
> Thank you for your email, please see comments inline.
> 
>> 
>> I am of the opinion that the protocol already has sufficient knobs to 
>> support an environment of algorithm transition without further augmentation.
>> 
>> The existing protocol supports:
>> - multiple algorithms are implemented as multiple classes
> (Roque) agree.
>> - algorithms for each class are identified in the certificate provided by 
>> the server in the response to a LIST command
> (Roque) agree. I believe you meant that as we implement top-down you could 
> use the issuer's certificate, correct?

geoff: yes.

>> - the server performs a proof of possession text using the algorithm 
>> associated with the class that was nominated in the certificate request.
> (Roque) agree. And if we do not create any additional error code and if a 
> client sends a "issue" request using a non-supported algorithm for a given 
> class, the server will respond MUST an error code 1203: badly formed payload. 
> correct?

Geoff: Correct. In an algorithm transition context, the client MUST use the 
same algorithm as used in the server (issuer's) certificate within a resource 
class  (as provided by the Issuer in the LIST command response), and the server 
will response with a 1203 error code otherwise. You can write that up in the 
algorithm document. 



> 
> I will add text describing this mechanism in the next version of the agility 
> draft.

thank you

 Geoff

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

Reply via email to