Some suggestions.


How about the following features in the next version of netCDF ?

        1. String data type
                I heard this is in the works.
        2. Hidden variables:
                These can be used to make up an attribute list, but would 
                not "exist" or atleast would not be displayed by ncdump.
        3. Adopt TeX/LaTeX styles as standard for labels (ie B_{x,z})
                So that plotting programs are able to do sub/superscript 
                labelling of plots by picking the info. off a netCDF.

I can't do much about 1. and 2., but I have adopted 3. as a local standard.
Your views would be greatly appreciated.

        best regards
                Ata <(|)>.