http://bugzilla.opensuse.org/show_bug.cgi?id=1055930
http://bugzilla.opensuse.org/show_bug.cgi?id=1055930#c6
--- Comment #6 from Edward Baudrez ---
Just for the record, I think this is the essence of the problem:
# which cdo
/usr/bin/cdo
# ldd /usr/bin/cdo
[...]
libnetcdf.so.11 => /usr/lib64/libnetcdf.so.11 (0x00007f9fee2b2000)
libhdf5.so.100 => /usr/lib64/libhdf5.so.100 (0x00007f9fed9bc000)
[...]
# ldd /usr/lib64/libnetcdf.so.11
[...]
libhdf5.so.10 => /usr/lib64/libhdf5.so.10 (0x00007f0a4c143000)
[...]
While the runtime linker can load libhdf5.so.10 and libhdf5.so.100 at the same
time, it cannot distinguish the symbols from either version, so it ends up
choosing whichever comes first, leading to the bizarre bindings that I've
shown. By using symbol versioning, this problem could be solved, but only at
the cost of rebuiling the HDF5 packages.
I'll immediately add that this is not a deficiency of 'cdo' proper! (I guess
that I should probably open a ticket on either the netCDF or HDF5 packages, but
'cdo' was the utility that exposed the problem.)
--
You are receiving this mail because:
You are on the CC list for the bug.