[CF-metadata] CF and the CDM

From: Jonathan Gregory <j.m.gregory>
Date: Thu, 14 Sep 2006 08:02:05 +0100

Dear Simon

I don't know about the CDM myself. However, I expect CF will use netCDF-4
features when they are the best way to do something. If I remember correctly,
we have been advised not to rush in to committing ourselves to use them
while it is still experimental, but in at least one case netCDF-4 would seem
to me to help with a particular known problem: we could use ragged dimensions
to get round the problem of wasteful storage of sets of timeseries which have
different lengths but which we want to put in a single variable. In general
CF is extended when someone comes up with a new specific need, rather than
our trying in advance to propose solutions for which we don't have an extant
problem. It will be interesting to see whether your application turns up such
structurally new needs. We will have to consider the implications for
backwards compatibility carefully.

Best wishes

Jonathan
Received on Thu Sep 14 2006 - 01:02:05 BST