Consider this a dry run for eventually rolling out variants that use
SHA-3-?? instead of SHA-256.  The code for aliasing should give you
confidence that you can ship new challenges without disrupting
usability.

In a way, it's a shame that the challenges didn't change more often
during development.  Testing concurrent support for old and new
versions is pretty valuable.



On 14 March 2017 at 15:28, Roland Shoemaker <[email protected]> wrote:
> I'd argue that removing the challenge version numbers adds unnecessary
> complexity to the specification and any existing implementations going
> forward.
>
> Existing servers and clients will need to have some kind of mapping from
> the draft names to the final un-versioned names and any protocol
> revisions going forward will need to know about both the versioned and
> un-versioned names so that they don't name a new version of a challenge
> using one of the draft version numbers, leading to existing
> implementations thinking an older version is actually being used.
>
> I agree dropping the version numbers is somewhat more aesthetically
> pleasing but I can't really see any actual technical reasons to do so,
> while keeping them seems like it'd save a lot of headaches down the road.
>
> On 03/13/2017 11:56 AM, Richard Barnes wrote:
>> I would prefer we stick with dropping the version number, just because
>> it's cleaner and the future is bigger than the past.
>>
>>
>> On Mar 13, 2017 2:27 PM, "Jacob Hoffman-Andrews" <[email protected]
>> <mailto:[email protected]>> wrote:
>>
>>     Roland posted a PR tweaking the challenge names for the final RFC:
>>     https://github.com/ietf-wg-acme/acme/pull/272
>>     <https://github.com/ietf-wg-acme/acme/pull/272>.
>>
>>     This raised the question: What do we want the challenge names to be in
>>     the final RFC? I think we've been assuming that "http-01" would become
>>     "http" once the RFC is published. However, this does create a slightly
>>     deployment headache, in that draft-compatible implementations have to
>>     use one name, and RFC-compatible implementations have to use another, so
>>     it's hard to test working code with the final names before the RFC is
>>     really finalized.
>>
>>     I don't have a strong opinion on this one way or another. What do folks
>>     on this list think?
>>
>>     _______________________________________________
>>     Acme mailing list
>>     [email protected] <mailto:[email protected]>
>>     https://www.ietf.org/mailman/listinfo/acme
>>     <https://www.ietf.org/mailman/listinfo/acme>
>>
>>
>>
>> _______________________________________________
>> Acme mailing list
>> [email protected]
>> https://www.ietf.org/mailman/listinfo/acme
>>
>
> --
> Roland Bracewell Shoemaker
> Software Engineer
> Linux Foundation / Internet Security Research Group
>
> _______________________________________________
> 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