> You sent me a link to the VisAD TextAdapter documention that showed examples
> on how to load a CSV file. It says that the only possible representations are
> 2-D arrays and 1,2, or 3-D points. In my .csv file, if I have 9 different
> input fields that can each be mapped to any visualization parameter, it seems
> as if I can only map at most 3 of these input fields at a time. Is this
> In that case, if I split up the 9 fields into 3 groups of 3, my data set would
> have to be broken down. However, in example #5 with the .bsv file, it looks
> like the range has 4 values. Is it possible to have a domain with 9 fields?
> Maybe I'm not understanding the meaning of these mappings because in example
> #7, I don't see why longitude and latitude are in the domain when they are a
> visualization parameters for the output in VisAD.
Field domains include independent variables, and Field ranges
include dependent variables. Thus in model output, temperature,
pressure, etc are dependent on latitude and longitude. However,
both dependent and independent variables may be "visualization
Also, I don't think there is a limit in TextAdapter on the
number of variables. VisAD can theoretically support any
dimension of Field domain, although only includes
implementations for up to 3-dimensional Sets (except for
LinearNDSet, which has arbitrary dimension). For model output,
you generally have 4 independent variables: 3 for space and
1 for time, usually nested as something like:
(time -> ((latitude, longitude, altitude) -> ...)). I am
not sure if TextAdapter can produce such nested domains.
Tom is the expert on TextAdapter, and perhaps he will answer
the rest of your question.