On Oct 31, 2006, at 5:46 AM, Ed Hartnett wrote:
Simon Wood <simon.wood@xxxxxxxxxx> writes:Hi NetCDF-4 folks, We're been thinking about using compound types in some of our newdatasets and have been puzzling over how to add attributes to compoundvariables. We expected to be able to specify attributes on a per *field* basis but I get the definite impression this is not actually going to be possible... attributes can be global, per group, per variable but it seems thats it. Is this so?Has anybody else been thinking about metadata for compound types? Howis metadata expected to be represented for individual fields? Surely something as basic the ability to specify different units for each field has been considered? Or have we got this whole compound types idea wrong? Any thoughts most gratefully received. regards Simon Wood
We've also considered adding attributes on compound datatype fields a few times, but haven't had a serious interest from a funding source to get it done right. We'd be happy to talk with people about the design issues involved...
Quincey Koziol koziol@xxxxxxxxxxxx
We have considered adding specific support for units attributes for compound types. However, this would expand the interface quite a bit. We have also considered that one could create an attribute of compound type, in which each element was a unit for an element of the data compound type. Ed -- Ed Hartnett -- ed@xxxxxxxxxxxxxxxx====================================================================== ========To unsubscribe netcdf-hdf, visit: http://www.unidata.ucar.edu/mailing-list-delete-form.html====================================================================== ========
Description: S/MIME cryptographic signature