Github user mariobriggs commented on the issue:
https://github.com/apache/spark/pull/14214
What i tried to do as a 'side fix' was like this,
eliminate [1] since it was a lazy val.
Move [2] out of the code path of the main thread i.e. let ListenerBus
thread pay the penalty of producing the physical plan for logging ( i was
coming from a performance test scenario, so it allowed me to proceed :-) ) . So
the change was that SparkListenerSQLExecutionStart only take QueryExecution as
a input parameter and not physicalPlanDescription & SparkPlanInfo . However
this cannot be the solution since SparkListenerSQLExecutionStart is a public
API already.
[3] remains.
As you might have already noticed ConsoleSink also suffers from the same
problem of [2] and these are inside Dataset.withTypedCallback/withCallback, but
it is only for Debug purposes
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
