fix bug a bug in the MCMGMapper contains method for blocksize > 1

Can you add a test that shows that this is the correct fix?

I'll have to think of a good test - I tend to do more tests now with dune-python that is how I found it....

changed milestone to %DUNE 2.7.0

added bug label

A simple test would be to check whether the contains method returns the same index as the subIndex method if blocksize>1.

Is this ready to be merged?

I looked at the fix and it looks okay. But for me the fact that nobody noticed the problem earlier is a clear sign that the unit test is not good enough. Therefore I would like to get an appropriate test together with the fix.

Sounds good. Should not be to hard to come up with something. I'll take a look.

added 1 commit

c91254e1 - [bugfix][MCMGMapper] added test for blocksize > 1.

Compare with previous version

I added a test that hopefully triggers the blockSize > 1. I couldn't test because I did not have UGGrid installed.

Thanks @robert.kloefkorn but I don't think UGGrid was the issue - I'll have a look over the weekend.

@robert, the problem is independent of the grid manager. What happens is that nowadays MCMGMapper can attach more than one dof to an entity. That is what needs testing---as Andreas found the contains method does not implement that correctly.

The test itself is implemented such that if fails if UGGrid is not found and in the test only UGGrid is used. That should be changed anyway to use YaspGrid in addition. I can add that.