[CF-metadata] CF Conventions and netCDF-4 enhanced model

From: Jonathan Gregory <j.m.gregory>
Date: Thu, 11 Sep 2014 17:34:24 +0100

Dear Jim

> I think the point of CF 2.x would be to openly embrace new netCDF
> features, and if strong backward compatibility would make it
> awkward, then backward compatibility would lose. CF 1.x could
> continue to evolve along side it. CF 2.x would be a refactoring that
> took new features and lessons learned into account.

I don't think that would be a good idea, myself. It would increase (by maybe
a factor of two) the amount of work required to maintain CF, and it would cause
a lot of debates about which changes should be made in both, etc. I think if
we step to CF-2.x then the final CF-1.x should be the end of that line. I am
not completely opposed to backward incompatibility, as I said before. I just
think that there has to be a very strong case for it. A complete refactoring,
starting all over again and repeating the discussions of the last 15 years,
would be a tremendous amount of work and I don't see a good reason for it.
I think we have to keep in mind that a convention, such as CF, is at some
level arbitrary. We make the best decisions we can at the time, and I would
rather spend time extending CF to new purposes than revisiting old decisions,
even though in many cases a different decision could have been made, unless
the old decision has truly proved itself to be a nuisance. But backward-
incompatible changes that materially improve the usefulness of the standard
should be considered, I agree.

Best wishes

Jonathan
Received on Thu Sep 11 2014 - 10:34:24 BST