Re: Expert review for AAA Diamater assignments

[Jari Arkko <jari.arkko@piuha.net>]

Bernard Aboba wrote:
> > Anyway, expert help for 3GPP's use of Diameter is very
> > useful. I spend surprisingly lot of time in talking to
> > either our implementation folks or the 3GPP participants
> > about their use of Diameter. One of the recent topics
> > has been the application identifier usage of some of
> > their newer Diameter-based interfaces.
>
>
> I'd agree.  I'd also note that there has been some controversy relating to
> usage of the 'M'andatory bit.  For example, my understanding is that some
> implementations return terminal "Missing AVP" errors in response to
> missing AVPs classified as optional in the Diameter specifications.
> This is a troubling development, since could affect interoperability at
> the most basic level.

Yes. But its possible to err in two directions; part of
my recent experience has been that you can also follow
the rules too narrowly, and that its easier to fall back
to a new application identifier than think about whether
you can live with an existing standard application. This
is equally bad to the Missing AVP fatal error, because two
boxes will be unable to interoperate -- only the manner
in which they fail is different, early application id mismatch
vs. late error message.

--Jari