Three (5?) cents:
The idea that you can make data self-describing solely by dint of its name is noble, but (it seems to me) undercut by the existence of the Attributes. Among other things.
A practice that doesn't scale will eventually be eliminated, whether by revolution or evolution. I prefer revolution myself, given the choice; but given that revolution is not possible for compatibility reasons, would encourage evolution now, to avoid worse outcomes later.
The base lists are already sufficiently inconsistent that exceptions (read: mapping) will be necessary for modeling; this is in the nature of following a "don't look forward" philosophy. (Not meant as a criticism, but to illustrate why inconsistencies will continue to happen.)
So my wish is to handle directions as an attribute going forward, and make them required for future cases if you feel that is necessary to make sure they are provided when necessary.
I also have the broader suggestion that CF start thinking about the concept of 'deprecated' with respect to CF terms, to cover practices that are no longer deemed desirable but which are still supported for backward compatibility reasons. In this manner CF won't break but users will be encouraged toward best practices.
John
At 4:42 PM +0100 6/28/07, Roy Lowry wrote:
>Hi Bryan,
>
>I would prefer any changes in Standard Names conventions/practices to be by revolution rather than evolution, even if that evolution is from bad to good practice. Semantic modelling and ontology creation are much easier if there are consistent patterns in the base lists. So, I agree with Jonathan even if my reasons are somewhat different.
>
>Cheers, Roy.
>
>>>> Bryan Lawrence <b.n.lawrence at rl.ac.uk> 6/28/2007 4:30 pm >>>
>
>> data would not be self- describing. Not knowing the sign convention is a
>> serious problem for analysis of data. The fact that the data would not be
>> CF-compliant would be no consolation for its being useless. It is better to
>> prevent problems happening.
>
>(Possible response: Make the sign mandatory. I think asking someone who didn't
>read the CF convention when they designed their model to rerun the model or
>rewrite the output is a bigger ask than adding one tiny attribute to each
>variable ... and one which which would make software deciding what to do with
>comparing data much easier to design).
>
--
----------
John Graybeal <mailto:graybeal at mbari.org> -- 831-775-1956
Monterey Bay Aquarium Research Institute
Marine Metadata Initiative: http://marinemetadata.org || Shore Side Data System: http://www.mbari.org/ssds
Received on Thu Jun 28 2007 - 10:17:36 BST
