There is two ways to write the comparison when we search an item equal to the needle:

    item == needle

and

    needle == item

In most cases they get the same result. But if __eq__ of the item or the needle is not correctly implemented (returns False instead of NotImplemented for unsupported types) the order can matter. We want the behavior be consistent in all cases.

The majority of the code uses needle on the right side of the == operator. But there are few outliners:

* tuple.__contains__ and list.__contains__. They code is different from the code used in methods count(), index(), remove() where needle is on the right.

* The general implementation of the "in" operator for iterators (see PySequence_Contains).

* dictitems.__contains__.

* The copy of list_contains in _ssl.c.

* The C implementation of _asyncio.Future.remove_done_callback. It differs from the Python implementation.

In all other code (many tens of occurrences) needle is on the right.

I think it is worth to fix the minor inconsistency in these few places.

See also the discussion on Python-Dev: https://mail.python.org/archives/list/python-dev@python.org/thread/VSV4K4AOKM4CBQMOELPFV5VMYALPH464/.

New changeset 18b711c5a7f90d88fb74748f18fa8ef49d8486c7 by Serhiy Storchaka in branch 'master':
bpo-37648: Fixed minor inconsistency in some __contains__. (GH-14904)
https://github.com/python/cpython/commit/18b711c5a7f90d88fb74748f18fa8ef49d8486c7

Just to add list.__contains__ and others might cause subtle changes as noted. One example is around discussion at https://github.com/python/cpython/pull/14700#discussion_r324054193 where my code to fix mock.ANY was depending on this behavior to work on master and didn't work with 3.8 since it's not backported. mock.ANY depends on the order of operands.