[CF-metadata] CF "datasets"

[Jonathan Gregory]

Dear Steve

>    As a general matter, I think we ought to get to the heart of this
>    "dataset" concept.  CF "datasets", that are not simple files, have
>    been in routine use for a long time.  It is routine that the data
>    object we think of as a "CF thingy" is actually an aggregation of many
>    files.

I agree. I think it would be good to say something explicitly about it in the
standard. It is an obvious and convenient idea that a collection of files can
be treated as one file. However, certain conventions are needed to know how
to aggregate the variables and dimensions, and probably software in use may
make different assumptions.

Best wishes

Jonathan