Epetra crashes with sor

[Patrick Farrell]

Consider the following code:

from dolfin import *
import random
import sys

mesh = UnitInterval(40)
V = FunctionSpace(mesh, "CG", 2)

parameters["linear_algebra_backend"] = "Epetra"
assert ('sor', 'Successive over-relaxation') in krylov_solver_preconditioners()

def main(ic):

  out = Function(V)
  u = TrialFunction(V)
  v = TestFunction(V)
  A = assemble(inner(u, v)*dx)
  b = assemble(inner(ic, v)*dx)
  solve(A, out.vector(), b, "cg", "sor")

  return out

if __name__ == "__main__":
  ic = Function(V)
  vec = ic.vector()
  for i in range(len(vec)):
    vec[i] = random.random()

  out = main(ic)

It crashes with:
Solving linear system of size 81 x 81 (Epetra Krylov solver).
AZ_check_options: ERROR: sym GS preconditioning can only be used with MSR matrices.
       data_org[AZ_matrix_type] = 3
...
*** -------------------------------------------------------------------------
*** Error: Unable to solve linear system using Epetra Krylov solver.
*** Reason: Solution failed to converge (error code 1).
*** Where: This error was encountered inside EpetraKrylovSolver.cpp.
*** -------------------------------------------------------------------------

It works if you either
a) remove the "sor" string
b) change the backend to PETSc.