[
https://issues.apache.org/jira/browse/BEAM-583?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15435766#comment-15435766
]
Luke Cwik commented on BEAM-583:
--------------------------------
It would make sense for me for testing reasons that they should really use
TestPipeline. Having Test[Spark|Flink]Runners just adds needless noise to
the set of runners that pipeline authors can choose from. We also expect a
certain interaction pattern between TestPipeline and TestYYYRunner which a
pipeline author is unlikely to do.
> Auto Register TestDataflowRunner
> ---------------------------------
>
> Key: BEAM-583
> URL: https://issues.apache.org/jira/browse/BEAM-583
> Project: Beam
> Issue Type: Improvement
> Reporter: Mark Liu
> Assignee: Mark Liu
>
> Register TestDataflowRunner automatically.
> Simplify option's arguments when using TestDataflowRunner run end-to-end test
> against dataflow service. Instead of using
> {code}--runner=org.apache.beam.runners.dataflow.testing.TestDataflowRunner{code},
> we can also use {code}--runner=TestDataflowRunner{code}.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
