Re: [Libmesh-devel] compilation error

[John P. <pet...@cf...>]

pco...@ue... writes:
 > > On Wed, 13 Jun 2007, pco...@ue... wrote:
 > >
 > >> Compiling C++ (in optimized mode) src/numerics/petsc=5Flinear=5Fs=
olver.C...
 > >> Compiling C++ (in optimized mode) src/numerics/petsc=5Fmatrix.C..=
=2E
 > >> Compiling C++ (in optimized mode)
 > >> src/numerics/petsc=5Fnonlinear=5Fsolver.C...
 > >> src/numerics/petsc=5Fnonlinear=5Fsolver.C: In member function =91=
void
 > >> PetscNonlinearSolver<T>::init() [w ith T =3D double]=92:
 > >> src/numerics/petsc=5Fnonlinear=5Fsolver.C:252:   instantiated fro=
m here
 > >> src/numerics/petsc=5Fnonlinear=5Fsolver.C:179: error: =91SNESSetM=
onitor=92 was
 > >> not
 > >> declared in this scope
 > >> make: *** [src/numerics/petsc=5Fnonlinear=5Fsolver.i686-pc-linux-=
gnu.opt.o]
 > >> Error 1
 > >>
 > >> Will it related with petsc configuration or libmesh=3F
 > >
 > > It's related to the combination of the two.  The PETSc developers =
seem
 > > to make new incompatible changes to their APIs with each new minor=

 > > release, and a couple changes in PETSc 2.3.3 aren't supported by
 > > libMesh yet.  You'll need to revert to PETSc 2.3.2 for now.
 > > ---
 > > Roy
 >=20
 > Ok, I compiled with petsc-2.3.2-p10 and everyyhing work fine. Howeve=
r I
 > remarked that in example 13 where once it was called the PetscLinear=
Solver
 > now it calls the LaspackLinearSolver. What's the reason for this cha=
nge=3F

LibMesh probably didn't detect the new petsc correctly during ./configu=
re.

-John