[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[netCDF #TLB-677315]: FW: Net CDF-4 Issues

Coy,

In ansswer to your last question:

> Maybe there is a better way to go about my end goal then having a lot of
> unlimited dimensions. The data I writ to file is acquired using
> different modes (sets of parameters). When I create the file for the
> first time, I do not know what modes I will be trying to write to the
> file in the future. In NetCDF-3, I am forced to guess and often make
> unnecessary room in the file for future modes. With NetCDF-4, I thought
> I would just make the number of modes unlimited so I could add future
> modes without issue. This also requires I make other dimensions
> unlimited as some of the parameters may change the amount of data
> acquired. Is there a better approach to this problem then using
> unlimited dimensions?

I believe the better way is to use variable-length user-defined types,
which are intended for representing such structures as ragged arrays.

An example of the use of variable-length types in C is in the source
distribution, ncdump/tst_vlen_data.c.  In addition to the documentation
in the C Users Guide on the functions used, there's also some material
on use of variable-length data types in the white paper "Developing
Conventions for netCDF-4 (Draft)" at

  http://www.unidata.ucar.edu/netcdf/papers/nc4_conventions.html

An example CDL file showing how ncdump represents variable-length types
is in the source release in ncdump/ref_tst_vlen_data.cdl.

We need to provide more guidance on use of multiple unlimited
dimensions versus variable-length types, and plan to do so in
the future, but for now, I hope this helps.

--Russ


Russ Rew                                         UCAR Unidata Program
address@hidden                     http://www.unidata.ucar.edu



Ticket Details
===================
Ticket ID: TLB-677315
Department: Support netCDF
Priority: Normal
Status: Closed