[Bug]: numpy.int64 types are not serialized correctly

[jrmccluskey]

What happened?

Relevant repro of the problem:

Relevant error:

WARNING:apache_beam.coders.coder_impl:Using fallback deterministic coder for type '<class 'numpy.int64'>' in '[7]: Create/MaybeReshuffle/Reshuffle/ReshufflePerKey/GroupByKey'.
ERROR:apache_beam.runners.common:Unable to deterministically encode '0' of type '<class 'numpy.int64'>', please provide a type hint for the input of '[7]: Create/MaybeReshuffle/Reshuffle/ReshufflePerKey/GroupByKey' [while running '[7]: Create/Map(decode)']

The problem appears to be that the coder does not know how to handle the numpy int64 type and the fallback coder (PickleCoder, I believe) cannot encode the type deterministically so it clobbers the content inside the class to its base value of 0.

Issue Priority

Priority: 2 (default / most bugs should be filed as P2)

Issue Components

• Component: Python SDK
• Component: Java SDK
• Component: Go SDK
• Component: Typescript SDK
• Component: IO connector
• Component: Beam YAML
• Component: Beam examples
• Component: Beam playground
• Component: Beam katas
• Component: Website
• Component: Infrastructure
• Component: Spark Runner
• Component: Flink Runner
• Component: Samza Runner
• Component: Twister2 Runner
• Component: Hazelcast Jet Runner
• Component: Google Cloud Dataflow Runner

[shoyer]

Here's an another case which is clearly broken:

>>> [(np.int64(1), 'first'), (np.int64(2), 'second')] | beam.GroupByKey()
[(0, ['first', 'second'])]