[CF-metadata] CF provisional standards

From: Jonathan Gregory <j.m.gregory>
Date: Wed, 15 Nov 2006 10:32:51 +0000

Dear Steve and Jamie

I agree with Jamie. I think we should add new stuff to the standard, just as
we always have. My example is the analysis/forecast time convention. We have
had as much discussion already about this on the email list as about many
aspects of CF, it is needed, it is in use, so it should appear in the document.
I will try to find some time to draft it properly.

We could mark new text as provisional or just "new" somehow (like Brian did),
until some time has elapsed. The rules for this should be debated and
then decided by the conventions committee, I suggest. My concern in my last
posting on this is that unless we follow fairly short timescales for agreeing
new features and testing them, it is not fair to users to insist on a
"provisional" status. If the CF process takes a year to decide something,
it is no use to turn round and say to someone, "Sorry, that was provisional,
your data is now deemed to be all wrong." It's only reasonable to do that if
we decide and test something on a timescale of a month or two, and they could
not wait for that short time before going ahead with producing data.

Also, I would say that experience shows we haven't made large blunders, in
the sense of agreeing something that was actually unworkable. Of course
further development may suggesting improvements and we might change our mind
later, but that is normal evolution, not because it was provisional - unless
we say that everything is provisional until the Ultimate Answer is known at
the end of time. :-)

Best wishes

Jonathan
Received on Wed Nov 15 2006 - 03:32:51 GMT