Added a simplified version of ts/ex11, for a 5-point Laplacian with convergence test with a synthetic problem. Serial. Uses differencing to produce matrix without a solve, fixed a false positive error in SNES to support this.
I would like to work on ex33 but this pull request is polluted with garbage. Non of these conflicts are real. THis is in 'next' so I don't know to touch it really. I think you can take 'their conflicts' for all of these.
The bitbucket merge dialog is bogus: there are no conflicts in merging to 'master'. Is the branch "done" in the sense that it should be in 'master', or is it a work in progress (in which case it should not merge yet)?
It will probably get completly superseded when Matt and I get a chance to DMDAPlexify it. It doesn't sound like Matt is going to get to it until next month. But don't I want it in master so that I can branch from master to work on it?