There are several tests where the expected output
relies on order of strings generated from the arbitrary
ordering of dictionaries or sets. The new version of
tox causes that ordering to be different between tests
runs so tests must be resilient and test either the
content of the dicts or sets or their sorted output.
Note: These fixes are only of those cases where a failure
was generated. If there are issues in skipped tests, this
patch has not found nor fixed them. The vast majority of
skipped tests are related to availability of functionality
so there's little that can be done about that with the
current testing setup. The best we can do is be on the
lookout for failures that could be related PYTHONHASHSEED
down the road and fix them as they happen.
Closes-Bug: 1348818
Change-Id: I6b27ca2597c51b0656f441f325f9ffd0e31a606d
3a009031be