PETSc/SLEPc transition

Johannes Ring johannr at simula.no
Fri Jan 13 11:45:08 UTC 2012


On Sun, Jan 8, 2012 at 4:42 AM, Adam C Powell IV <hazelsct at debian.org> wrote:
> So I tried to build SLEPc, and got the following error:
>
> Checking environment...
> Checking PETSc installation...
> Checking ARPACK library...
> Checking LAPACK library...
> Traceback (most recent call last):
>  File "./configure", line 10, in <module>
>    execfile(os.path.join(os.path.dirname(__file__), 'config', 'configure.py'))
>  File "./config/configure.py", line 361, in <module>
>    import cmakeboot
>  File "/home/hazelsct/slepc-3.2-p3.dfsg/config/cmakeboot.py", line 7, in <module>
>    import script
> ImportError: No module named script
> make: *** [build-arch] Error 1
> dpkg-buildpackage: error: debian/rules build gave error exit status 2

I get the same error, but only when building for amd64. The problem
seems to be that SLEPc tries to build using CMake since
PETSC_BUILD_USING_CMAKE=1 is defined in
/usr/lib/petsc/<petscarch>/conf/petscvariables. On the other arches,
this variable is not defined and SLEPc builds fine (at least on i386).

> Any ideas?

Could the problem be that CMake was available when you built PETSc? I
guess that would be the only reason for PETSC_BUILD_USING_CMAKE=1 to
be set in petscvariables. I tried to rebuild PETSc for amd64 in
pbuilder and installing the resulting .deb packages fixed the problem
with SLEPc.

Johannes

PS: I have updated SLEPc to only build for the same arches as PETSc.



More information about the debian-science-maintainers mailing list