--On Tuesday, 06 September, 2005 15:14 -0400 Sam Hartman
<hartmans-ietf(_at_)mit(_dot_)edu> wrote:
John, what does it mean to put a registry document on the
standards track? In particular, how do you get multiple
implementations of a registry?
One is reminded of the story of Eeyore's birthday party.
Registering things --putting them into a registry so that they
can be retrieved and examined using whatever key was used to put
them there-- is always easy and, as you point out, untestable.
But it is also almost never the point: the point is whether the
right information is being placed in the registry to support the
relevant applications and whether those applications can use the
information in a way that promotes interoperability.
With regard to that driving issue, it is certainly possible to
have multiple implementations of matching rules. It is
certainly possible to examine, in practice, different uses of
the tagging system to determine whether its mechanisms are
sufficient and, if sufficient, whether they meet some "minimum
necessary" criteria or represent serious overkill and/or
redundancy.
With apologies to Spencer, let me turn your question around and
ask how something can be identified as a "Best Common Practice"
when
* when it has not been practiced at all,
* when there is no evidence that it is "best" (even if
one agrees that no better options are on the table or
even that none are likely to be searched out and found
unless, someone, this approach is shown to fail as RFC
3066 was shown to fail), and
* it certainly isn't "common" in the "widely
disseminated and used" sense of that term.
Things would be better if we had "PP" (proposed practice), "AGI"
(apparently good idea), or "WTOSITD" (well thought-out shot in
the dark), or "NRtBTWNW" (no reason to believe this will not
work) categories. But we don't. And the model and mechanisms
associated with Proposed Standard, when applied to the use cases
rather than the registry itself, much more nearly meet
application needs and community expectations than identifying a
document like that as a BCP.
john
_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/ietf