dolfin team mailing list archive
-
dolfin team
-
Mailing list archive
-
Message #02748
Re: _PETSC_COMM_WORLD and _MPI_COMM_WORLD
I tried linking against PETSc without any add-on solvers (no Hypre,
Spooles, etc) and received the same error, so it doesn't seem related to
the external PETSc packages.
Garth
On Fri, 2006-06-16 at 13:36 +0200, arvenk@xxxxxxxxx wrote:
> It seems like libtool is difficult to integrate with non-libtoolized
> dependencies (e.g. PETSc) :| I'll try and inquire with someone who knows
> libtool better than I do. Is there any way to determine which of the PETSc
> libraries are static btw?
>
> Arve
>
> > I rolled back to what we had before. Arve, tell me if you find another
> > solution.
> >
> > /Anders
> >
> >
> > On Fri, Jun 16, 2006 at 10:06:42AM +0200, Garth N. Wells wrote:
> >> On Fri, 2006-06-16 at 09:26 +0200, Anders Logg wrote:
> >> > On Thu, Jun 15, 2006 at 05:51:20PM +0200, Beni Stamm wrote:
> >> > > In src/lib/Makefile.am on the below mentioned line there is already
> >> > >
> >> > > libdolfin_la_LIBADD = @DOLFIN_LTLIBOBJS@
> >> >
> >> > src/lib/Makefile.am should look as follows:
> >> >
> >> > INCLUDES = @KERNEL_CFLAGS@
> >> > lib_LTLIBRARIES = libdolfin.la
> >> > libdolfin_la_SOURCES =
> >> > libdolfin_la_LIBADD = @DOLFIN_LTLIBOBJS@ @DOLFIN_EXT_LIBS@
> >> > libdolfin_la_DEPENDENCIES = @DOLFIN_LTLIBOBJS@
> >> >
> >> > The change I made yesterday was to add @DOLFIN_EXT_LIBS@ to fix
> >> > some compilation problems related to linking of PETSc libraries.
> >> >
> >> > If this does not work for everyone, we'll just roll back the fix and
> >> > find another solution. Can anyone else confirm this problem?
> >> >
> >> > Garth, did you do a configure + make clean + make install and it still
> >> > doesn't work?
> >>
> >> I cloned a new version and it doesn't work.Deep in the output when
> >> compiling DOLFIN I do get the warning message
> >>
> >>
> >> *** Warning: Linking the shared library libdolfin.la against the
> >> *** static
> >> library
> >> /usr/local/lib/petsc-2.3.1-p14/externalpackages/spooles-2.2/linux-gnu-c-real-debug/MPI/src/spoolesMPI.a
> >> is not portable!
> >>
> >> *** Warning: Linking the shared library libdolfin.la against the
> >> *** static
> >> library
> >> /usr/local/lib/petsc-2.3.1-p14/externalpackages/spooles-2.2/linux-gnu-c-real-debug/spooles.a
> >> is not portable!
> >>
> >>
> >> I didn't configure Spooles by hand. I just let PETSc download and
> >> configure it. This could be the problem.
> >>
> >> Garth
> >>
> >>
> >>
> >> >
> >> > > (without @DOLFIN_EXT_LIBS@)
> >> > > So, I just added the @DOLFIN_EXT_LIBS@ to check the other
> >> possibility. but
> >> > > while
> >> > > running ./scripts/preconfigure the following error occurs
> >> >
> >> > This is a known bug in the script preconfigure. You need to run
> >> >
> >> > scripts/preconfigure
> >> >
> >> > without the ./ or otherwise it will break... It's a stupid test to see
> >> > from where the script is run. I didn't have time to find a better
> >> > solution when I wrote it:
> >> >
> >> > CHECK=`echo $0 | cut -d'/' -f1`
> >> > if [ "$CHECK" != "scripts" ]; then
> >> > echo "This script must be run from the top level directory."
> >> > exit 1
> >> > fi
> >> >
> >> > If anyone has a better solution, just name it.
> >> >
> >> > /Anders
> >> >
> >> >
> >> > > iacspc79:/Users/stamm/Documents/dolfin-0.6.1 stamm#
> >> ./scripts/preconfigure
> >> > > This script must be run from the top level directory.
> >> > >
> >> > > Although I running this command from the dolfin-0.6.1-level.
> >> > > And I tested with only (without scripts/preconfigure):
> >> > > configure
> >> > > make
> >> > > make install
> >> > > make demo
> >> > >
> >> > > and it doesn't solve the problem.
> >> > >
> >> > > ben
> >> > >
> >> > >
> >> > > On Jun 15, 2006, at 5:30 PM, Anders Logg wrote:
> >> > >
> >> > >
> >> > > Strange... works for me. I checked it before committing.
> >> > >
> >> > > Does it work if you change in src/lib/Makefile.am on the line
> >> > >
> >> > > libdolfin_la_LIBADD = @DOLFIN_LTLIBOBJS@ @DOLFIN_EXT_LIBS@
> >> > >
> >> > > Remove @DOLFIN_EXT_LIBS@ and then run scripts/preconfigure and
> >> > > configure.
> >> > >
> >> > > Does it work then?
> >> > >
> >> > > /Anders
> >> > >
> >> > > On Thu, Jun 15, 2006 at 05:02:46PM +0200, Garth N. Wells wrote:
> >> > >
> >> > > Looks like something got messed up with the changes (made
> >> today) to how
> >> > > libraries are linked. I can't link anything, and get the
> >> error
> >> > >
> >> > > . . . . /libdolfin.so: undefined reference to `MAIN__'
> >> > > collect2: ld returned 1 exit status
> >> > > make: *** [dolfin-poisson] Error 1
> >> > >
> >> > >
> >> > > Garth
> >> > >
> >> > > On Thu, 2006-06-15 at 15:39 +0200, Beni Stamm wrote:
> >> > >
> >> > > Hello,
> >> > > I've just successfully compiled dolfin on my mac osx
> >> 10.5
> >> > > (intel-based). everything worked fine.
> >> > > I also did successfully "make demo" but while executing
> >> for example
> >> > > "dolfin-poisson" the
> >> > > following error occurs:
> >> > >
> >> > >
> >> > > iacspc79:/Users/stamm/Documents/dolfin-0.6.1/src/demo/pde/poisson
> >> > > root# ./dolfin-poisson
> >> > > dyld: Symbol not found: _PETSC_COMM_WORLD
> >> > > Referenced from: /usr/local/lib/libdolfin.0.dylib
> >> > > Expected in: flat namespace
> >> > >
> >> > >
> >> > > Trace/BPT trap
> >> > >
> >> > >
> >> > > Is there a conflict between _PETSC_COMM_WORLD and
> >> > > _MPI_COMM_WORLD?
> >> > > Do I have
> >> > > to change something in my petsc/mpi installation? Does
> >> anyone knows
> >> > > how to handle this
> >> > > problem?
> >> > >
> >> > >
> >> > > thanks for your answer
> >> > >
> >> > >
> >> > > benjamin stamm
> >> > >
> >> > >
> >> > > p.s. I had to downgrade the gcc compiler from 4.2 to
> >> 4.0.1 as the
> >> > > 4.2
> >> > > seems to be more strict than the older versions
> >> > > (I had some errors during compilation).
> >> > >
> >> > >
> >> > >
> >> > >
> >> > > _________________________________________________________
> >> > >
> >> > >
> >> > > Benjamin Stamm
> >> > > Modeling and Scientific Computing
> >> > > CMCS-IACS-EPFL
> >> > > Station 8
> >> > > CH-1015 Lausanne
> >> > > Switzerland
> >> > >
> >> > >
> >> > > Tel: +41 21 693 55 09,
> >> > > http://iacs.epfl.ch/~stamm/
> >> > > _________________________________________________________
> >> > >
> >> > >
> >> > >
> >> > >
> >> > > _______________________________________________
> >> > > DOLFIN-dev mailing list
> >> > > DOLFIN-dev@xxxxxxxxxx
> >> > > http://www.fenics.org/cgi-bin/mailman/listinfo/dolfin-dev
> >> > >
> >> > >
> >> > >
> >> > > _______________________________________________
> >> > > DOLFIN-dev mailing list
> >> > > DOLFIN-dev@xxxxxxxxxx
> >> > > http://www.fenics.org/cgi-bin/mailman/listinfo/dolfin-dev
> >> > >
> >> > >
> >> > > _______________________________________________
> >> > > DOLFIN-dev mailing list
> >> > > DOLFIN-dev@xxxxxxxxxx
> >> > > http://www.fenics.org/cgi-bin/mailman/listinfo/dolfin-dev
> >> > >
> >> > >
> >> > > _________________________________________________________
> >> > >
> >> > > Benjamin Stamm
> >> > > Modeling and Scientific Computing
> >> > > CMCS-IACS-EPFL
> >> > > Station 8
> >> > > CH-1015 Lausanne
> >> > > Switzerland
> >> > >
> >> > > Tel: +41 21 693 55 09,
> >> > > http://iacs.epfl.ch/~stamm/
> >> > > _________________________________________________________
> >> > >
> >> > >
> >> >
> >> > > _______________________________________________
> >> > > DOLFIN-dev mailing list
> >> > > DOLFIN-dev@xxxxxxxxxx
> >> > > http://www.fenics.org/cgi-bin/mailman/listinfo/dolfin-dev
> >> >
> >> >
> >> > _______________________________________________
> >> > DOLFIN-dev mailing list
> >> > DOLFIN-dev@xxxxxxxxxx
> >> > http://www.fenics.org/cgi-bin/mailman/listinfo/dolfin-dev
> >>
> >>
> >> _______________________________________________
> >> DOLFIN-dev mailing list
> >> DOLFIN-dev@xxxxxxxxxx
> >> http://www.fenics.org/cgi-bin/mailman/listinfo/dolfin-dev
> >
>
>
>
> _______________________________________________
> DOLFIN-dev mailing list
> DOLFIN-dev@xxxxxxxxxx
> http://www.fenics.org/cgi-bin/mailman/listinfo/dolfin-dev
References