If the Flush call is in a part of fleur that is not exercised when hdf5 is not used some linkers will recognize this and not report an error. You should be able to use the nm command as Scot suggested to find the location from which the flush command is being called.
On Wed, Feb 19, 2014 at 10:41 AM, Samir ABDELOUAHED < [email protected]> wrote: > Dear Scott, > I understand what do you mean. But at the same time I compiled Fleur > without using HDF successfully. > So the problem comes from the HDF libraries and/or how to link Fleur to it. > I need to use HDF to make Fleur faster and to have small size output files. > Thanks, > Samir > > On 19/02/14 16:21, Scot Breitenfeld wrote: > >> Do you mean you have a >> >> CALL FLUSH (lunit) >> >> in your Fortran program? HDF5 does not use FLUSH and does not have a >> function called flush. You might want to 'nm' the other libraries you have >> and see what provides flush__ >> >> Scot >> ________________________________________ >> From: Hdf-forum <[email protected]> on behalf of >> Samir ABDELOUAHED <[email protected]> >> Sent: Wednesday, February 19, 2014 8:48 AM >> To: [email protected] >> Subject: [Hdf-forum] linking problem >> >> Dear Sir/Madam, >> >> I am trying to use HDF5/parallel for a Fortran Code (Fleur code). >> the code uses MKL (which contains scalapack openmpi/blacs). I built HDF5 >> using the mpif90 (the ifort intel compiler wrapper). the compilation of >> the code looks ok. But right at the end when it tries to link to >> libraries I have the error message: >> ssubredist2.o:ssubredist2.F:(.text+0x1606): more undefined references to >> `flush__' follow >> make: *** [fleur.x] Error 1 >> >> I would be grateful if you could help me solving the problem. >> Many thanks. >> >> -- >> Dr. Samir ABDELOUAHED >> Postdoctoral Research Fellow >> Department of Physics >> University of York >> Heslington, York, YO10 5DD, UK >> Tel: 0044 (0) 1904 322208 >> Office: N003 >> >> >> _______________________________________________ >> Hdf-forum is for HDF software users discussion. >> [email protected] >> http://mail.lists.hdfgroup.org/mailman/listinfo/hdf- >> forum_lists.hdfgroup.org >> >> _______________________________________________ >> Hdf-forum is for HDF software users discussion. >> [email protected] >> http://mail.lists.hdfgroup.org/mailman/listinfo/hdf- >> forum_lists.hdfgroup.org >> > > -- > Dr. Samir ABDELOUAHED > Postdoctoral Research Fellow > Department of Physics > University of York > Heslington, York, YO10 5DD, UK > Tel: 0044 (0) 1904 322208 > Office: N003 > > > _______________________________________________ > Hdf-forum is for HDF software users discussion. > [email protected] > http://mail.lists.hdfgroup.org/mailman/listinfo/hdf- > forum_lists.hdfgroup.org > -- Jim Edwards CESM Software Engineering Group National Center for Atmospheric Research Boulder, CO 303-497-1842
_______________________________________________ Hdf-forum is for HDF software users discussion. [email protected] http://mail.lists.hdfgroup.org/mailman/listinfo/hdf-forum_lists.hdfgroup.org
