Re: 20020228: Netcdf-C library use (Trouble adding more dimensions to variables)

>To: address@hidden
>From: Naveenta Anand <address@hidden>
>Subject: Re: 20020228: Netcdf-C library use (Trouble adding more dimensions to 
>variables)
>Keywords: 200202282208.g1SM8j809246

Naveenta,

> Oh, I am trying to add dimensions by basically
> 1) Increasing the RANK of the variables
> 2) Then add another array subscript to nc_def_var
> 3) Increase start position to # of new dimensions at the time of nc_put_var

But if you were doing this to an existing variable, the nc_def_var()
call would return an error.  So either you are not checking the error
return from the nc_def_var() call, or you have found a bug where it
should return an error status but doesn't.  If nc_def_var() is
returning NC_NOERR in this case, please send us your code so we can
track down the problem.  If you are not checking the return status
from netCDF calls, please change your code to check them in the
future, otherwise you can expect problems will continue.

> And what I get from your email is that this approach is totally wrong. I 
> should either use 1) ncgen approach or create a new netcdf data set and 
> redefine the variable in the new data set.
> 
> It's making sense to me now. Please tell me if my interpretation as above 
> is right.

It sounds right.  I'm still not sure whether you only need to change
one file with a small amount of data, or you need to do this for many
files or lots of data, which would determine which approach is best.

--Russ