I think it's a good idea. I have two related comments:
First: one of the exercises we went (actually, I think it's still
going on) through in McIDAS was to try to pick names for parameters
that could be translated via tables to the names used in specific
databases. For example, "T" was chosen to be the system-wide name for
temperature. The user and software could ask for "T" and the server,
if set up properly, would be able to fetch the temperature from the
selected database regardless of what it's name might have been in
there. Part of this exercise was to also define "default" physical
units for each of the parameters. This is a maintence issue (each
database 'schema' has to have a translation table), but it is also
consistent in keeping the intimate knowledge of the file "format"
at the server side.
Second: One of the most interesting (and useful) charts that I received
during my undergrad 'career' was a representation of quantity names
using a 3-D matrix for powers (and inverse powers) of Length, Mass, and
Time. In each box, the "unit dimensional" quantity was shown (like:
Not every box had a name, but the ability to relate to a quantity
without introducing specific units systems was an extraordinary aid to
dimensional analysis. I would think this sort of thing would be
critical when users start combining parameters with units attached.
Steve, has anyone looked into this type of thing for use with udunits?
Tom Whittaker tomw@xxxxxxxxxxxxx
Space Science and Engineering Center University of Wisconsin-Madison
Phone/VoiceMail: 608/262-2759 Fax: 608/263-6738