Github user mallman commented on the issue:
https://github.com/apache/spark/pull/16122
I'm not sure configuration-level rollback will guarantee an absence of
interactions with other tests. For one thing, I think we need to create and
clean up an independent metastore directory for Derby. Otherwise we may end up
with
https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/69669/testReport/junit/org.apache.spark.sql.hive/HiveExternalCatalogSuite/_It_is_not_a_test_/.
I guess I'd like a stronger assurance that we're doing this the right way.
The only call to `Hive.get` in the Hive codebase I can find is in a private
method in `HiveClientImpl`. Spark seems to be going to great lengths to
discourage calling that method directly.
I've also thought this might be better suited to testing in the
thriftserver codebase, where there seems to be a stronger sense of
client/server separation in testing. I haven't gone too far in investigating
that direction.
---
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 [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
