Open robertwb opened 1 year ago
Does this mean currently Python PubsubIO can only run on Dataflow runner (in addition to Python direct runner as there is a local overwrite) ?
Correct, currently they only run on Dataflow (that swaps out PubSub) and the olde Python direct runner (that also swaps them out).
What needs to happen?
Reading from and writing to PubSub should be proper (if primitive) transforms, rather than being represented by "special" parameters of Read/Write operations necessitating workarounds like https://github.com/apache/beam/blob/release-2.48.0/sdks/python/apache_beam/io/iobase.py#L946.
Issue Priority
Priority: 3 (nice-to-have improvement)
Issue Components