-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Am Freitag, 16. Mai 2003 23:16 schrieb Mark Rivers:
> Reimar wrote:
> > Yes it's all correct but the difference to the variables is
> > thats the reader of the data has no idea how long the vector is. If the
> > vector is defined as a netCDF variable we have to look first at the
> > belonging dimension name and we are able to read it's length.
> > This means the data type, length and dimension are described
> > with variables in the file. By a vector as attribute there is no rule
> > given.
>
> That's not true. There are functions available to inquire about the data
> type and length of any attribute in the file.
>
> Here is the interface you need:
> int nc_inq_att (int ncid, int varid, const char *name,
> nc_type *xtypep, size_t *lenp);
>
> So you can inquire about the type and size of the attribute using this
> function.
>
Oh, I missed this. Then it's well described.
But I am not lucky with that. Because someone is able to use these attributes
for all parameters only. Many people don't know about the reasons why to
store more as the values into datafiles.
If you set up a three dimensional dataset with three coordination variables
then it is for all netCDF readers clear what's to read. But I got files
where the time dependent coordination variable (the third dimension) is set
as global attribute.
I know it depends on the user who has decided to define it this way and it's
not the fault of the format itselfs. But it's terrible.
Reimar
> Mark Rivers
- --
Forschungszentrum Juelich
email: R.Bauer@xxxxxxxxxxxxx
http://www.fz-juelich.de/icg/icg-i/
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.2-rc1-SuSE (GNU/Linux)
iD8DBQE+xgYW5aOc3Q9hk/kRAokXAJ9DICSgAcx72ryWWpdthL2Fp3HI1wCgvOaG
DFV3nZ12lNP4ofVjJRhkXCk
=IXIt
-----END PGP SIGNATURE-----
