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 Note that the archive is no longer being updated. We provide the archive for reference; many of the answers presented here remain technically correct, even if somewhat outdated.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.
Hi Stefan, > > > 4. The actual build process starts: > > >> mkdir build && cd build > > >> CC=`which mpicc` ../configure --enable-parallel-tests > > --prefix=/opt/netcdf > > >> make -j -l2 > > >> make -j -l2 check Although our "make -j" used to work with our Makefiles, that broke recently, for some reason I could never determine. So although it takes longer, I'd suggest not using parallel make. I'm not sure that's a cause for some of the problems you're seeing, but I've seen similar symptoms with "make -j", so just use "make". > > > Oh no: > > >> cat: t_dap.c: No such file or directory > > > Well, this seems to be about DAP. Actually, curl is not installed so why > > > should dap tests run? Lets simply try again with: > > >> CC=`which mpicc` ../configure --enable-parallel-tests --disable-dap > > --prefix=/opt/netcdf > > >> make -j -l2 > > >> make -j -l2 check > > > Whoo, all tests passed, even all the parallel tests run through. > > >> make install > > That is very good news, although I am puzzled by the lack of t_dap.c. > > Following your steps so far (with the exception of running autoreconf > > instead of the libtoolize command) and I did not encounter a missing > > t_dap.c. > > > That DAP thing seems to be not a big deal. Just wanted to let you know. As > mentioned above, on OpenSUSE it works and with the --disable-dap it works > on Ubuntu, too. This might be due to some missing dependencies. > > > > > 5. So far so good. NetCDF-fortran is needed as well > > >> svn checkout https://github.com/Unidata/netcdf-fortran/trunk > > >> libtoolize && aclocal && autoheader && autoconf && automake > > --add-missing > > >> mkdir build && cd build > > >> CC=`which mpicc` FC=`which mpif90` CPPFLAGS=-I/opt/netcdf/include > > LDFLAGS=-L/opt/netcdf/lib ../configure --enable-parallel-tests > > --prefix=/opt/netcdf > > >> make -j -l2 > > > Oh no, just a ton of errors: > > >> ... > > >> make[2]: Entering directory `/opt/netcdf/netcdf-fortran/build/fortran' > > >> DHAVE_CONFIG_H -I. -I../../fortran -I.. -I../libsrc > > -I/opt/netcdf/include -g -O2 -c -o module_netcdf_nc_data.o > > ../../fortran/module_netcdf_nc_data.F90 > > >> /bin/bash: DHAVE_CONFIG_H: command not found > > >> make[2]: [module_netcdf_nc_data.o] Error 127 (ignored) > > >> g -O2 -c -o module_netcdf_nc_interfaces.o > > ../../fortran/module_netcdf_nc_interfaces.f90 > > >> /bin/bash: g: command not found > > >> ... > > > Well, all those lines should begin with: > > >> /usr/bin/mpif90 -DHAVE_CONFIG_H ... > > > Something went wrong with the FC variable. If I am using > > >> FC=mpif90 > > > instead it works. The issue with that is, that I can not specify an > > > non-system compiler. > > Ok, I am not a fortran developer, but Russ is out of the office today; > > let me provide what help I can. > > > All right, I am on hold. Lets see what Russ says. > Anyways, here is a little more information. Maybe you want to forward this > to Russ. I'll try to see if I can reproduce your other symptoms. Sending us the Makefile doesn't help, and your config.log just indicates that the configure script worked OK, but it might help to see the output from "make", at least up to the point where there are lots of errors. --Russ Russ Rew UCAR Unidata Program address@hidden http://www.unidata.ucar.edu Ticket Details =================== Ticket ID: ISB-918823 Department: Support netCDF Priority: Normal Status: Closed