This archive contains answers to questions sent to Unidata support through mid-2025. Note that the archive is no longer being updated. We provide the archive for reference; many of the answers presented here remain technically correct, even if somewhat outdated. For the most up-to-date information on the use of NSF Unidata software and data services, please consult the Software Documentation first.
> Date: Fri, 10 May 2002 10:01:26 +0200 > From: Ole Holm Nielsen <address@hidden> > To: address@hidden > Subject: Re: 20020508: netcdf 3.5.1 Beta 3: Intel Fortran patch needed Ole, > I'm trying to port our production code to the Intel Fortran 95 > compiler, and for this I need the NetCDF libraries. Intel Fortran > won't link with the libnetcdf.a library that I get from standard > Linux RPMs, so I'm trying out the NetCDF 3.5.0 source. > > I have succeeded in building NetCDF 3.5.0 with the Intel compilers > ifc and icc (disabling the C++ part of NetCDF). But when I run > "make test", a number of errors are reported by the test code, > and finally some test bombs out. I wonder if you know of > a way to make NetCDF work 100% correctly with the Intel compilers ?? > FYI, I'm using the new version 6.0 of Intel compilers. No, sorry, we don't have the new Intel compilers installed here yet and don't even have a Pentium 4 platform here to test on. I'm told we will eventually get one, but for now questions like this might best be answered by asking the netcdfgroup mailing list or some of the other people who have been working with netCDF and the Intel compilers. For example, I found this reply in our support archives that sounds like it may refer to the same problem: http://www.unidata.ucar.edu/cgi-bin/mfs/70/4566 It may be this is a symptom of a bug in the Intel compilers, since it doesn't seem to be a problem with other compilers and platforms. > By the way, do you have any understanding of why the Intel Fortran > doesn't recognize the libnetcdf.a library from a Linux RPM ? > There seems to be a problem about library formats or something ? > This is a really bothersome feature of the Intel compiler, > and we have no such problems with the Portland Group compiler. I suspect the compiler used to create the Fortran interface for the libnetcdf.a you are using (maybe g77) has different conventions for calling C from Fortran than the Intel compiler. There is no standard for how Fortran programs should call C functions, so there are wide variations in how it's done, but the Portland Group may have decided to do it the same way as g77/gcc Most of these variations are encapsulated in the cfortran.h header we distribute with netCDF, but the Intel compilers you're using haven't been integrated into cfortran.h yet. --Russ _____________________________________________________________________ Russ Rew UCAR Unidata Program address@hidden http://www.unidata.ucar.edu