[MMTK] A rather annoying newbie question on the netCDF interface

Peter mmtk at maubp.freeserve.co.uk
Wed Jul 14 02:35:07 CEST 2004

>> If anyone can help me with the following problem then it would be 
>> much  appreciated.  I've been putting myself through the wringer 
>> trying to  get the netCDF interface working in Scientific.  I've 
>> installed the  following on windows...
>> 1. Python 2.3.3 using Python-2.3.3.exe from python.org
>> 2. Scientific python using ScientificPython-2.4.6.win32-py2.3.exe
>> 3. netCDF for win32 - (\\python23\include\netcdf.h and netcdf.inc;  
>> \\python23\libs\netcdf.lib; also netcdf.dll in PATH)
> That looks fine, from memory, perhaps some Windows user can confirm.  
> Which version of netCDF do you use? It must be the same that was used  
> when ScientificPython was compiled (I guess 3.5, which is the current  
> one).

It was 3.5.0, see below.

> Otherwise, no idea, sorry. Windows is a mystery to me.

Have you looked at the Windows Installation instructions on the Wiki?


This should also explain other bits and pieces like Mark Hammond's 
Windows Extensions and setting up visualisations:

I've been meaning to make the NetCDF installation bit more explicit.

Anyway, the ScientificPython-2.4.6.win32-py2.3.exe installer was built 
using the prebuilt binary of netcdf-3.5.0 available here:


If you are using the Scientific Python and MMTK installers, then I think 
all you need to do for netCDF is download that file, and extract just 
netcdf.dll into:


for Windows XP, or on Windows 2000 or NT:


Did that help?

If not, then some additional information like what version of Windows 
you are using, and if you have a C compiler, wouldn't hurt.



P.S.  In fact, anywhere on the path should work... I seem to have put it 
next too Scientific_netcdf.pyd on my Windows 2000 machine:


I was trying to find out what would go wrong if the DLL was in the wrong 
place, see this also email:


More information about the mmtk mailing list