parallel-netcdf-0.9.3 release

[Rob Ross]

Rene,

Just a quick note -- everyone here is on break until after the first of 
the year, so we're not going to be much help for a few days.

This is an interesting problem; we'll have to see what we can do to help 
with it.  Do you know if the serial netCDF guys have a solution to this 
problem?

Thanks,

Rob

On Tue, 23 Dec 2003, Rene Redler wrote:

> hello,
> 
>     I succeeded in using pnetcdf jointly with netcdf in some
> ocean model code where we use pnetcdf to read our own input
> files for data assimilation while all other model specific file I/O
> is still done via the usual netcdf. No naming conflicts, no problems
> from this side. The code runs and produces reasonable results.
> 
> During my exercises I tried to build a pnetcdf library using Fortran
> compiler options that set all integer and reals to 64 bits (-ew on NEC SX).
> Although the configure recognized the appropriate lengths from Fortran to C
> (e.g. size of Integer is long) the execution of the Fortran test cases
> failed, probably because this information is not used for building
> the c routines. As far as I can see all integer variables in the c routines
> are still declared as int.
> 
> I then found a way to build the ocean model without -ew (by only setting
> the reals to 64 bit) which enabled me to use the standard pnetcdf lib.
> When compiling the ocean code and including pnetcdf.inc I got an compiler
> error which was related to the last lines in pnetcdf.inc.  These lines
> tell me to define the C macros NF_*_IS_C_*. Why and where is this needed
> for?
> 
> Rene
> 
>