eap Mailing List: Re: reviewers for new methods

[Jari Arkko (jari.arkko]

Thomas Otto wrote:

> Who is now an expert? I guess of course the authors
> of the respective EAP method, which leads to the question:
> May the authors volunteer self or must the reviewer be some
> unbiased person? 
>  
I think the point of the review is that its done by
some person other than the authors of the proposal!
But if you have a proposal, here's how you can help:
review some of the other proposals, then it is easier
for us to find a reviewer for your proposal.

We already have one volunteer reviewer from yesterday's
call. But ideally I'd like to get one reviewer for each
method.

> Assume the authors should abstain from review, then 
> *who* is an expert? Which skills, grades are needed?
>  
The reviewer should thoroughly understand RFC 3748.
There are no formal requirements. Most members of
this WG should be able to do it. Of course, the rules of
the IETF indicate that ADs have the final word in
the appoitment of expert reviewers.

> That is, if the EAP method meet all criterias in the 
> catalogue, then all obstacles are overcome and an
> IANA EAP Type Number can be assigned. (Right? Please
> correct me..)
>  
Yes. All obstacles from the point of view of
EAP number assignment are then overcome.
Note that there may also be general IETF
requirements on actual Individual RFC
submissions, as usual.

> Then, the method can be used in real world deployments
> (because supplicant and AAA both recognize the method by 
> this Type Number) and authors and users are perfectly
> happy.
>  
The chairs of the EAP WG would be even happier
if people brought their methods to RFC status, so
that a permanent specification of methods would
be available to all. Note that the RFC 3748 rules
for EAP method type allocation actually call for
this (or some other permanent reference) since
Specification Required is indicated. From RFC 2434:

   Specification Required - Values and their meaning must be
          documented in an RFC or other permanent and readily available
          reference, in sufficient detail so that interoperability
          between independent implementations is possible.

--Jari