Since this is a rather wide-ranging discussion of opinion, here are my thoughts
on netcdf3 vs netcdf4 and the role of Unidata:
(1) netcdf3 vs netcdf4:
- I appreciate and have used the on-the-fly compression in netcdf4. I don’t
have much experience with the other unique features of netcdf4.
- Nonetheless, when building applications on a new system, I always link
against a stable netcdf3 library, rather than netcdf4. Sometimes I have had
compatibility problems between cdo, nco, nccopy, and the netcdf files created
on another machine or application.
- I had to work around a problem earlier this year due to the underlying
reliance on the hdf library in netcdf4. Apparently some of the hdf attributes
were hidden to the netcdf interface, and there were problems with conversion of
netcdf3 to compressed netcdf4 format.
(2) Unidata support:
I have always found Unidata/UCAR staff knowledgeable and responsive to problems
I have encountered in netcdf releases. I am not entirely sure what language
bindings are officially supported at present; a previous post seemed to imply
that Python was supported. I think the C and Fortran libraries should be top
priority, but if there are resources to support any of the interpreted
languages, I vote for support for the Julia language. An interface to the C
library already exists, but it is immature and could use some dedicated effort.
-Ed
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Edward D. Zaron
Research Assistant Professor
Department of Civil and Environmental Engineering
Portland State University
Portland, OR 97207-0751
Phone: (503)-725-2435
FAX: (503)-725-5950
ezaron@xxxxxxx
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
