Setting cell orientation fails for 1d mesh in 2d, but the problems run deeper

[Martin Sandve Alnæs]

I tried to make a fix here:

https://bitbucket.org/fenics-project/dolfin/branch/martinal/topic-fix-1d-in-2d-init-cell-orientation

However this makes other tests fail, and I don't know how deep this goes.

The problem is that around the changes in the above commit but also other places in the dolfin code, there's a strange assumption of a geometric dimension 3 several places. See also MeshEntity::midpoint(), I don't know how deep this goes.

[Martin Sandve Alnæs]

The weird preference for 3D is even in Mesh::point(), and the Point class itself.

I'm going for a duct-tape fix for this particular issue and not touching the deeper thing here.

[Martin Sandve Alnæs]

Fixed the immediate problem.

[Anders Logg (Chalmers)]

Removing milestone: 1.7 (automated comment)