On 13/04/12 10:15, David Hassell wrote: > That sounds reasonable to me. From a backwards compatibility view > point, it sounds like any existing software built on the common netCDF > libraries will already correctly interpret these attributes on > coordinates ... is that a fair assumption?
>From what John wrote, I think anything written using the Java netCDF libraries will unpack coordinates and interpret it correctly. However, everything else (e.g. using the C libraries) will have to have coded in their own unpacking support. I imagine a sizeable number (probably the majority) haven't done so, even for data variables. If one is purely seeking to reduce data size, and if the data are suitable, it's certainly better to use netCDF4 internal compression as this is fully and transparently supported in all libraries and requires no specific application level code. Perhaps this could be included in the unpacking section as a recommendation/point to consider? I'm aware that the original poster was looking for other conveniences though. Cheers, Mike. _______________________________________________ CF-metadata mailing list [email protected] http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata
