kennknowles commented on a change in pull request #13124:
URL: https://github.com/apache/beam/pull/13124#discussion_r508878813
##########
File path:
.test-infra/jenkins/job_PostCommit_Java_ValidatesRunner_Dataflow.groovy
##########
@@ -23,7 +23,7 @@ import PostcommitJobBuilder
// This job runs the suite of ValidatesRunner tests against the Dataflow
// runner.
PostcommitJobBuilder.postCommitJob('beam_PostCommit_Java_ValidatesRunner_Dataflow',
- 'Run Dataflow ValidatesRunner', 'Google Cloud Dataflow Runner
ValidatesRunner Tests', this) {
+ 'Run Dataflow ValidatesRunner', 'Google Cloud Dataflow Runner
ValidatesRunner Tests (streaming/batch auto)', this) {
Review comment:
When `--streaming=true` is not specified, this happens:
https://github.com/apache/beam/blob/996c8d2ed77375061c90417cf4018fde604e364a/runners/google-cloud-dataflow-java/src/main/java/org/apache/beam/runners/dataflow/DataflowRunner.java#L869-L871
So the unmodified `ValidatesRunner` test suite uses batch mode when possible
and streaming when necessary. We expect essentially all users to use the
unmodified form and let it "just work". But very few of our `ValidatesRunner`
tests have unbounded PCollections so streaming is largely untested in this mode.
It may be overkill to change this title. I got worried it would be confused
for a "batch only" mode, which it is not.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
