Coverage increased (+0.7%) to 46.03% when pulling c99781f on petschki/issue-71 into d5af7b1 on master.

@malthe maybe you can take a look at this?

I'm not sure I understand the solution here. It seems like all the code just deleted.

Either a clear analysis of the change should be presented, or a test case that shows the issue and how it's fixed with this change. It's been a while since I touched this code so it's hard for me to understand the correctness of it.

@malthe yes you're right and I'm sorry for this code cut. The problem is, I cannot clearly reproduce the strange behavior of miss-ordered taxonomies. but it happens ... and with this simpler code it doesn't happen again. maybe it's because of python3 ... need to make some more investigations.

@petschki should we merge this or is there anything more to do?

There is a confusing ordering problem in a project of @agitator and me and this PR tries to fix it but AFAIK there are still some edge cases ... so not ready to merge for now