Now, when compare errors raised during `in`, dict.keys(), dict.values() and set all propagate the errors. But dict.items() will swallow the errors(only key compare):

>>> class BadEq:
...     def __hash__(self):
...             return 7
...     def __eq__(self, other):
...             raise RuntimeError
... 
>>> k1, k2, v1, v2 = BadEq(), BadEq(), BadEq(), BadEq()
>>> d = {k1: v1}
>>> k2 in d.keys()
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "<stdin>", line 5, in __eq__
RuntimeError
>>> v2 in d.values()
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "<stdin>", line 5, in __eq__
RuntimeError
>>> k2 in {k1}
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "<stdin>", line 5, in __eq__
RuntimeError
>>> (k2, v2) in d.items()
False
>>> (k2, v1) in d.items()
False

dictitems_contains.patch tries to fix this.