failure running pnetcdf tests

[michael bane]

On Tue, 2007-01-09 at 21:31 +0000, michael bane wrote:
> -------- Forwarded Message --------
> > From: michael bane <michael.bane at manchester.ac.uk>
> > To: parallel-netcdf at mcs.anl.gov
> > Subject: failure running pnetcdf tests
> > Date: Sat, 06 Jan 2007 18:11:27 +0000
> > 
> > Folks
> > 
> > I've just downloaded the pnetCDF tarball and attempted to install on my
> > box with mpich2 and Intel compilers. When running
> >         ~/src/parallel-netcdf-1.0.1/test/nf_test$ ./nf_test
> > I get numerous errors immediately, starting with
> >         *** Testing nfmpi_strerror ...
> >         *** Testing nfmpi_open ...
> >          0: MPI_File_open error = File does not exist, error stack:
> >         ADIOI_UFS_OPEN(53): File tooth-fairy.nc does not exist
> >          nfmpi_open of nonexistent file should    have returned system
> >         error
> >          0: MPI_File_open error = File does not exist, error stack:
> >         ADIOI_UFS_OPEN(53): File test.nc does not exist
> >          nfmpi_open: Can not open/create file
> >          nfmpi_redef of read-only file should fail
> >          0: MPI_File_open error = File does not exist, error stack:
> >         ADIOI_UFS_OPEN(53): File test.nc does not exist
> >          nfmpi_open: Can not open/create file
> >          nfmpi_close: Not a netCDF id
> >         
> >            ###            8  FAILURES TESTING nfmpi_open! ###
> > 
> > I am at a lost as to what this errors mean or if I've misunderstood
> > instructions so any tips welcome!
> > 
> > Thanks, Michael
> 
> 
> Thanks to Robert who advised I should be running
>  nf_test -c && nf_test -v
> However, for both FORTRAN and C I get many more errors than
> "expected"... anybody know where I should be looking to see what I've
> done wrong? Thanks, Michael
> 
> michael at manchester-campaigns:~/src/parallel-netcdf-1.0.1/test/nf_test$
> mpirun -np 1 ./nf_test -c
> michael at manchester-campaigns:~/src/parallel-netcdf-1.0.1/test/nf_test$
> mpirun -np 1 ./nf_test -v > nf_test.out 2>&1
> 
REMOVED for brevity: {numerous errors, ditto for nc_test}
> 

Okay - found the requirement to use '-mp' for Intel compilers
(README.ifc_linux) and now get the expected number of errors!

Sorry about the diversion folks!

Michael