Open rbooth200 opened 7 years ago
Then you need to update also the expected error of the test in nose...
It's basically the same. I suppose I could add a test for both dust algorithms though...
If you look at the errors it seems that there is a very small error (10^-9) in the hydro when running with MPI, which is causing the tests to fail. Any ideas?
Is this pull request still needed or have the changes been merged elsewhere (since there have been other dust branches merged)?
I never Merged these changes because this pull request highlights a likely small, but un fixed, bug with mpi
On Sat, 3 Jun 2017, 12:25 dhubber, notifications@github.com wrote:
Is this pull request still needed or have the changes been merged elsewhere (since there have been other dust branches merged)?
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/gandalfcode/gandalf/pull/123#issuecomment-305966343, or mute the thread https://github.com/notifications/unsubscribe-auth/AOqItDcFC8GBb9D5fEC6VTgPsmEFgUaIks5sATSugaJpZM4LRAjE .
I've changed the dustybox test to be in 1D. While I accept that we do need some 3D tests, this one is very slow on travis. It's essentially a 0D problem, so 1D should be fine.