Hi Tim,
I agree with your assessment of the low priority for cryptographically
secure hashes if there is a run-time penalty. We would probably start
with a glibc-supported hash.
Anyhow, I really like the idea of incorporating something like this into
ncks. Of course it goes without saying that the user will have to make a
new MD5sum every time the file is changed, so most users will probably
just do this after the last operation on the NetCDF file prior to it being
archived.
A global attribute should be used for the file hash.
If it seems useful, we could allow per-variable hash too, stored as
variable attributes, so that processing could continue after hashing.
Charlie
--
Charlie Zender, surname@xxxxxxx, (949) 824-2987, Department of Earth
System Science, University of California, Irvine CA 92697-3100
****** Visiting NCAR 12/18/04--1/30/05 **** Same e-mail ************
******* Phone: (303) 497-1380 **** FAX: (303) 497-1324 *************
