[BEAM-10217] Suboptimal type inference on Python 3.6

• Type: Bug
• Status: Resolved
• Priority: P2
• Resolution: Fixed
• Fix Version/s: Not applicable
• Component/s: sdk-py-core
• Assignee: Udi Meiri
• Reporter: Valentyn Tymofieiev
• Created: 09/Jun/20 01:10
• Updated: 24/Jul/20 20:01
• Resolved: 23/Jun/20 01:42

Description

Following trivial_inference_test fails for me with AssertionError: :: List[Tuple[Person, int]] != List[Any] on Python 3.6.8; Passed on 3.5.2, 3.5.6 and 3.7.3.

  def testMethodReturnsTuple(self):
    class Person(object):
      def __init__(self, name):
        self.name = name

    def parse_record(name_and_age):
      name, age = name_and_age.split(',')
      return Person(name), int(age)

    self.assertReturnType(typehints.List[typehints.Tuple[Person, int]],
                          lambda record: [parse_record(record)], [str])

See: https://github.com/apache/beam/pull/11954.

Issue Links

• links to: GitHub Pull Request #11954
• links to: GitHub Pull Request #11966