HeartSaVioR edited a comment on pull request #29069: URL: https://github.com/apache/spark/pull/29069#issuecomment-659144203
I guess we have several possible approaches here: 1. place the suite to the Hive-version specific directory (with new config on pom.xml to add the test source based on version) 2. create shim and place the shim to the Hive-version specific directory (same) 3. revert and try to place the suite on the list of running in separate JVM (less chance to encounter classloader issue) I guess the straightforward approach would be 1 - I guess it should work smoothly, though if we want to ensure the suite runs in both versions we'll end up with duplicating codes. If that doesn't matter much we can just do that, or even we can just enable the test on hive 2.3 only (as the suite is technically irrelevant to Hive 1.2 vs Hive 2.3). Less redundant but more complicated would be 2. I'm not 100% sure how much complexity is needed to make a shim for the suite, and not sure it worths to do instead of simply allowing redundant codes a bit. The simplest approach would be 3, but it's not guaranteed to fix the flakiness. I'd say it'd be better to leave as the last resort. WDYT? ---------------------------------------------------------------- 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: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org
