roystgnr

elements_begin would be the most appropriate range to use on a distributed mesh. I'm afraid Elem::is_semilocal and Mesh::semilocal_elements_begin are atavistically named; IIRC they predate GhostingFunctor. They basically ask the question...

> maybe there is some iterator invalidating we are somehow missing. I think I've figured this out: The semilocal_ iterators are finding point neighbors of elements, which they do by...

Pinging @friedmud - got any time to check this out?

It's not nearly as bad as the "will be reverted the next time someone runs bootstrap" changeset, but this still seems to be asking for regressions. Making these sorts of...

Are you solving a linear system that you build with `constrain_element_matrix_and_vector` (or one of the other homogeneous-only APIs) rather than `heterogeneously_constrain_element_matrix_and_vector`?

Except that would leave you without the BC contribution to the problem at all ... I assume that interior solution isn't purely coming from a convective term plus forcing function...

Is `enforce_constraints_exactly` being called somewhere with the `homogeneous` parameter set to true?

> For example, /systems_of_equations/ex7 This would seem to be an example of what I was talking about: ``` ~/l/g/normal (master)> grep constrain ../examples/systems*/*ex7/*.C dof_map.constrain_element_matrix (Ke, dof_indices); dof_map.constrain_element_vector (Re, dof_indices); ~/l/g/normal...

https://www.gnu.org/software/autoconf-archive/ax_gcc_archflag.html doesn't show any updates since the version you pulled. I'm surprised it doesn't seem to have a simple `march=native` fallback, though; gcc can try to autodetect available instruction sets...

> For 3D mortar the dual shape functions depend on element deformation so need to be recomputed on each element. Technically shouldn't this dependence occur for nearly any non-affine element?...