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

Re: netCDF for W2K

>To: address@hidden
>From: "James K Lewis" <address@hidden>
>Subject: Re: 20040311: netCDF for W2K
>Organization: Scientific Solutions, Inc.
>Keywords: 200403110123.i2B1NgrV000584

James,

> I'm looking into trying to generate the Win32 binaries, but apparently there
> has been much in the way of problems in getting a netcdf.lib file that works
> on a Windows machine.  Some questions:
> 
> 1) Where's this src/f90/ directory you mentioned several times?  It wasn't
> in the netcdf-f90win32.tgz file you had me download, I didn't found it in
> the netcdf-3.5.0.win32bin.zip file, and it's not pointed to on your web page
> that discusses "Experimental netCDF Fortran 90 Interface".

It's in the source distribution of netCDF, from which all the binaries
for various platforms are generated.  The source distribution is
available either as a compressed tar file:

  ftp://ftp.unidata.ucar.edu/pub/netcdf/netcdf.tar.Z

or as a ZIP file:

  ftp://ftp.unidata.ucar.edu/pub/netcdf/netcdf.zip


> 2) What was used to generate the f90 binaries you have to download?  If you
> go to
> http://www.per.marine.csiro.au/staff/Dirk.Slawinski/cwr_web/work/netcdfinfo.
> html, you'll find how someone finally generated a netCDF library for the
> Lahey Fortran using Microsoft Visual C++.  Did you use Microsoft Visual C++?
> I'm using Lahey, but I can't get it to recognize the *.lib file in either
> the f90 distribution or the win32 distribution.

Those binaries were contributed by a netCDF user. who gives the
details of how he generated the library here:

  http://www.unidata.ucar.edu/cgi-bin/msgout?/glimpse/netcdf/5355

> If you have Microsoft Visual C++, you could follow the above example to
> generate a binary that should work with my Lahey Fortran.  But Lahey says
> that are only a few binary libraries that they have developed capabilities
> to digest during linking.  Apparently, having an f90 binary *.mod or *.lib
> file for netCDF doesn't guaranty that one can then do f90 netCDF.

I don't understand that, I would think you could link with any library
generated by their compiler, otherwise what good is generating the
library?  We don't have access to a Lahey compiler under Windows for
testing, unfortunately.  If you manage to succeed in building a
Fortran 90 interface you can use for Windows, please keep notes and
let us know how you did it.  Thanks!

--Russ