[BEAM-9574] NamedTuple instances generated from schemas cannot be pickled

• Type: Bug
• Status: Resolved
• Priority: P2
• Resolution: Fixed
• Fix Version/s: 2.21.0
• Component/s: sdk-py-core
• Assignee: Brian Hulette
• Reporter: Brian Hulette
• Created: 23/Mar/20 16:53
• Updated: 16/May/20 14:18
• Resolved: 27/Mar/20 00:37

Description

Attempting to pickle an instance of a generated NamedTuple class results in the following:

_pickle.PicklingError: Can't pickle <class 'apache_beam.typehints.schemas.BeamSchema_a7de91e0_ae11_4c52_a041_0b58ada35ac1'>: attribute lookup BeamSchema_a7de91e0_ae11_4c52_a041_0b58ada35ac1 on apache_beam.typehints.schemas failed

In general, we shouldn't be pickling these instances, but occasionally it may be necessary, and we should just do it rather than failing hard.

Issue Links

• is related to: BEAM-9576 FastPrimitivesCoder should be able to encode types from schema'd PCollections without encoding the schema
• links to: GitHub Pull Request #11196