On 19.10.2010 20:55, Mike Grant wrote:
> Out of curiosity, why do you want to store time as strings?
I believe the motivation was mentioned as human readability?
> It's easy to create those strings from numerical values, and
> numerical values are easier to handle in code (and in netcdf-3,
> as Seth said).
I think this is a very good point. One could argue that the ISO8601
strings are just one way to _present_ such values. Since binary netCDF
files aren't directly human-readable anyway, the values don't need to be
human-readable as they are stored there.
Another reason NOT to store timestamps this way, would be if there is
also a 'time' variable holding the same udunits compatible values. That
would create a redundancy as well as a potential ambiguity for the
reader [software] when/if the times don't match. That's another mistake
to check for during reading, and another potential "point of failure".
One may argue, though, that an option to output time values formatted as
ISO8601 would be very useful in tools like 'ncdump'. In that case it
should be optional as I believe the basic idea is to "dump" the internal
values as directly as possible from how they are actually stored
internally. (And of course, if 'ncdump' does it, then 'ncgen' must be
equally able to parse them, so that the two tools are fully
"symmetrical" to each other.)
--
Regards,
-+-Ben-+-
Received on Wed Oct 20 2010 - 02:38:15 BST
