I've run into a problem with visad.data.FileFlatField. I've adapted
a netCDF "function" to a FileFlatField by using an inner-class
visad.data.FileAdapter. When the time comes for the FileFlatField to
instantiate the FlatField, it calls the Accessor to get the MathType of
the FlatField. This MathType is a Function whose domain is an abstract
space: the Sets associated with the RealType components in the domain's
RealTupleType are of type FloatSet. The FileFlatField attempts to
construct a FlatField using these FloatSets. The FlatField constructor
will not allow FloatSets (or DoubleSets) in the domain of a FlatField
and throws an exception.
I think I either have to modify the MathType of the domain of the
function I'm adapting, so that its components don't contain FloatSet's
(which means that the domain space of the function won't be very
abstract) or I have to modify FileFlatField so that it uses the domain
sampling Set of the function I'm adapting (which would require Bill's
Is this correct?