[CF-metadata] Parameter Ontology Project

From: Jonathan Gregory <j.m.gregory>
Date: Fri, 22 Sep 2006 08:17:31 +0100

Dear Bryan

I don't disagree with anything you say. My question was aimed at clarifying
what kind of versioning is needed, and I think your comments are the kind of
considerations which are relevant. No, I have no problem with versioning
per se. It seems like what one may need to know is what was the standard name
table at the time the name was chosen. How can we arrange that? One way would
be to give the table itself a version number and require that to be recorded
in the netCDF file as an attribute, like Conventions, or as part of it.
However I don't think that would work, because data-writers probably would not
do it consistently. I expect that writing the Conventions attribute is hard-
wired in most programs, and it would be too much work to keep up with the
standard name table version number. Another way might be simply to use the
date at which the file was created and have an archive which allowed you to
reconstruct the standard name table of any date. That does not guarantee that
the data-writer will actually have used the most up-to-date standard names,
of course.

These are not comments which come from any careful consideration, I have to
say. What we need really depends on what it is needed for.

Best wishes

Jonathan
Received on Fri Sep 22 2006 - 01:17:31 BST