Removing OrderedDict is not an urgent task, however it makes the code more readable at a few instances. So I would love to remove it, but really want to avoid regressions. Codecov shows that there are a couple of code paths which I needed to change, which actually do not have any test coverage.

I would see three ways forward:

1. wait till Convert all tests to pytest #468 is completed
2. add test for those missing paths in this PR (could be cumbersome)
3. Force merge it as test coverage actually does not change. We would have to be fairly confident, that the source changes are not breaking anything.

@n2ygk Any thoughts?

Removing OrderedDict is not an urgent task, however it makes the code more readable at a few instances. So I would love to remove it, but really want to avoid regressions. Codecov shows that there are a couple of code paths which I needed to change, which actually do not have any test coverage.

I would see three ways forward:

1. wait till Convert all tests to pytest #468 is completed
2. add test for those missing paths in this PR (could be cumbersome)
3. Force merge it as test coverage actually does not change. We would have to be fairly confident, that the source changes are not breaking anything.

@n2ygk Any thoughts?

@sliverc I think it will be pretty safe to replace OrderedDict and certainly improves readability, especially when it is invoked with a list of tuples vs. the much more readable dict syntax. A force merge in this case seems fine.

Sounds good. I made the PR ready then for merging. I went through all changes again, but please have a close look whether everything looks good to you.

Per our conversation doing a force merge to get around failed codecov target.