Hi Daniel,

I thought that there shouldn't be extra complexity arising from my PR,
because a "stub account object" contains a "contact" field anyways.
Hence, a server implementation should check if it is contained in the
payload. I think I was implicitly reading a MUST for the "contact" field
which is not there (yet).


On 22/03/17 17:35, Daniel McCarney wrote:
> Thanks for following up on your initial message with PR's Sophie.
> 
> For PR 281 I left a small comment but otherwise LGTM.
> 
> For PR 280 - I think the proposed changes are great with the exception of
> one question I have RE: the mechanism to differentiate the account URI
> recovery POSTs from account creation POSTs.
> Do you think it would be clearer to separate the account URI recovery to a
> new endpoint?
> It seems like the empty JWS payload approach would work, but we're adding
> more complexity to the account creation endpoint to allow it to double as
> the account URI recovery endpoint. Maybe it's time to separate it properly.
> What does the list think?
> 
> On Mon, Mar 20, 2017 at 4:50 AM, Sophie Herold <[email protected]>
> wrote:
> 
>> PR for 1.-3.
>> <https://github.com/ietf-wg-acme/acme/pull/280>
>>
>> PR for 4.
>> <https://github.com/ietf-wg-acme/acme/pull/281>
>>
>> Issue related to 4.
>> <https://github.com/ietf-wg-acme/acme/issues/278>
>>
>> On 20/02/17 16:17, Sophie Herold wrote:
>>> 1.
>>>   I cannot find where the response to account updates is specified.
>>>
>>>   Boulder replies with "202 Accepted" and an account object. Since
>>>   updates with trivial payload should be used for account information,
>>>   it seems to be expected that an account object is returned. The
>>>   specified response status for "Account deactivation" is "200 OK".
>>>   It would be consistent to specify the same response status for
>>>   account updates I guess?
>>>
>>> 2.
>>>   The client should be able to choose between "Account URI recovery"
>>>   and "Account Creation", if needed. A possible way would be to clarify
>>>   that account creations with trivial payload MUST be rejected. However,
>>>   if an account exists, the "Content-Location" header and "200 OK" is
>>>   returned, independent of the payloads content. Therefore a trivial
>>>   payload could be used to check if an account with the used key exists,
>>>   without creating it.
>>>
>>> 3.
>>>   As "6.3.4.  Account deactivation", other queries should get their own
>>>   sub-sub sections. Something like
>>>    - "6.3.x Account URI recovery" ("new-account", maybe trivial payload)
>>>    - "6.3.x Account update"
>>>    - "6.3.x Account information" (update with trivial payload)
>>>
>>> 4.
>>>   Is it really an account URI or rather an account URL? Some fields
>>>   changed from "uri" to "url". I guess, all occurrences of URIs should
>>>   be URLs. The only exception that I see is the "type" field in the
>>>   "Problem Details" responses.
>>>
>>
>> _______________________________________________
>> Acme mailing list
>> [email protected]
>> https://www.ietf.org/mailman/listinfo/acme
>>
> 
> 
> 
> _______________________________________________
> Acme mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/acme
> 

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

Reply via email to