I'm generally skeptical of trying to enumerate what clients should **not**
do.  But this seems fine.  I approved the PR.

On Tue, Apr 17, 2018 at 12:53 PM, Daniel McCarney <[email protected]>
wrote:

> There has been some confusion[0][1] around the relationship between the
> "identifiers" and "authorizations" arrays in an order object. Because one
> particular ACME implementation[2] returns an authorization for each
> identifier in an order some developers made assumptions about the order of
> the "authorizations" and "identifiers" fields in an order matching.
>
> There is no language in-spec that dictates the order of elements in
> fields. Since server policy allows for a design in which an order object
> has less authorizations than identifiers defining a sort order isn't
> especially helpful anyway. There is no guaranteed 1:1 relation between
> identifiers and authorizations.
>
> I opened a PR[3] to indicate client developers SHOULD NOT assume a sort
> order for the "identifiers", "authorizations" or "challenges" fields in
> server responses. This PR also includes an explicit mention that there is
> no guaranteed 1:1 relationship between an order's identifiers and its
> authorizations.
>
> - Daniel / cpu
>
> [0] - https://community.letsencrypt.org/t/dns-based-
> validation-fails-on-renew/59027
> [1] - https://github.com/ietf-wg-acme/acme/issues/419
> [2] - https://github.com/letsencrypt/boulder
> [3] - https://github.com/ietf-wg-acme/acme/pull/421
>
> _______________________________________________
> 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