Description
SparkContext.hadoopConfiguration should not be used directly in SQL module.
Instead, one should always use sessionState.newHadoopConfiguration(), which blends in configs set per session.
The idea is to add the linter rule to ban it.
- Restrict the linter rule to the components that use SQL. use the parameter `scalastyleSources`?
- Exclude genuinely valid uses, like e.g. in SessionState (ok, that can be done per case with scalastyle:off in the code.
Attachments
Attachments
Issue Links
- is duplicated by
-
SPARK-24919 Scala linter rule for sparkContext.hadoopConfiguration
- Resolved