[Trilinos-Users] Problem compiling Trilinos against recent PETSc

[Heroux, Mike]

Jens,

Given that PETSc 3.3 was released more than 2 years ago, this is a section
of ML code that is not exercised often.  The right fix for this would be
to support both the pre 3.3 version of these interfaces and the new, and
then compile the proper one depending on the version information in the
petscversion.h file.

For now, I suggest you just make the change Jed suggested.  Please let me
know if the change works for you.

Thanks.

Mike

On 8/27/14, 10:30 AM, "Jed Brown" <jed at 59A2.org> wrote:

>Jens Timmerman <jens.timmerman at ugent.be> writes:
>
>> Dear lists subscribers,
>>
>> I ran into trouble compiling trilinos against a recent PETSc version (>=
>> 3.3)
>> since they changed:
>>
>>   * VecCreateSeqWithArray() and VecCreateMPIWithArray() now take a
>>     blocksize argument before the local length
>
>Use bs=1 if you aren't using block size in your existing code.  If you
>know a block size (like elasticity), then it is preferable to set it so
>that the information can be used by solvers such as GAMG and Hypre (via
>HYPRE_BoomerAMGSetNumFunctions).