David Wells

We should make the rest of the changes to the Q vectors in IBFEMethod once this is done: see #938.

When we/I finally get around to writing this test we should validate it against the 0.6 output just to triple-check the new force spreading code.

Well, I still haven't added this test. I want to get the release out on schedule (and adding this test doesn't really impact the content of the release) so I...

Do you remember which test it was?

Oh, I found it - `explicit_ex4_2d.source_test.input`. Let me still leave this open - I want to add a test that uses this and the scratch hierarchy before closing it.

I tried this this morning and found at least one issue with our interface to hypre. I found another interesting problem in the stokes solve: we get floating point exceptions...

I ran the test suite with `feenableexcept(FE_DIVBYZERO|FE_INVALID)` - in this case it looks like we get infs in some other way because the programs abort when we do `inf +...

I did not - that's a good idea.

The standalone navier-stokes tests pass so I suspect its not an initialization problem but there is only one way to find out!

valgrind doesn't complain so it is not an initialization problem.