[CF-metadata] CF provisional standards

From: Jonathan Gregory <j.m.gregory>
Date: Fri, 24 Nov 2006 18:12:57 +0000

Dear All

> our challenge is *interoperability*. That and only that is
> the reason that we are working as a community instead of working
> individually.
I agree with that.

> The question of timeliness -- how long does it take for a desperately
> needed new idea to become a part of the standard -- is perhaps the key point.
I agree with that too. As I've said, I'm happy with the idea of provisional
status so long as it doesn't last long. I would say that two months is
reasonable. It gives us a "cooling-off" period after the debate as well as
the opportunity for testing.

Jamie raised the question of what constitutes an implementation. It seems to
me that a metadata standard is not the same as software in this respect. You
want to be sure that the metadata standard is not wrong or inadequate, which
would mean that it is unclear what to write in the file in some circumstance,
or that what you have written is ambiguous. Although it would be a good test
to interpret files by program, we can get a long way just by looking at them
as humans. I'd say it's pretty easy to tell, when you do that, whether the
standard is OK, whereas testing software against a spec is usually harder. In
this respect it's interesting what Egil and John say about the importance of
the metadata for *humans* to read. Perhaps one implementation, therefore, is
just that a human looks at the metadata and decides whether it makes sense. For
a second implementation, the CF checker could be extended to read a candidate
file containing an instance of the new convention.

I would also point out that the evidence suggests that our debates are pretty
good at reaching satisfactory decisions. Although we're aware of other choices
we could have made, I can't see evidence from this email list of complaints
that CF metadata cannot be successfully interpreted.

Cheers

Jonathan
Received on Fri Nov 24 2006 - 11:12:57 GMT