You are viewing a plain text version of this content. The canonical link for it is here.
Posted to reviews@spark.apache.org by GitBox <gi...@apache.org> on 2020/03/20 02:59:10 UTC
[GitHub] [spark] HeartSaVioR edited a comment on issue #27860:
[SPARK-31101][SQL][BUILD] Upgrade Janino to 3.1.2
HeartSaVioR edited a comment on issue #27860: [SPARK-31101][SQL][BUILD] Upgrade Janino to 3.1.2
URL: https://github.com/apache/spark/pull/27860#issuecomment-601504689
So that's the matter of maintenance; I wouldn't expect we will get Janino 3.0.17 easily, even though we fix the issue and provide the patch. The 3.0.x version line wasn't even maintained and even after branching it's defined as "deprecated".
Regarding confidence of stability, yeah, Spark is used to work with Janino 3.0.x so Janino 3.0.16 gives more confidence, though what I originally fixed in Janino is actually long-standing bug. There're still some known bugs in Janino 3.0.x (SPARK-25987, and https://github.com/janino-compiler/janino/issues/90#issuecomment-509020465 Josh Rosen reported to Janino) which will never be fixed in Janino 3.0.x I think, because these bugs don't exist in Janino 3.1.x.
The question remains: have we ever constructed the way to justify the confidence of Janino stability? Any tests fail to catch the bugs I mentioned, so I don't think we have stability tests which verify edge-cases. In normal cases, a bunch of Spark UTs generate codes and Janino will be tested by these codes whenever we integrate Janino. If we evaluate the confidence from just test coverages, nothing looks to be changed.
I'd agree about general fear of adopting minor version which actually had major changes; that's why I asked about opinions in other PR. In the long run I think Janino 3.1.x is the way to go, but if adopting Janino 3.1.x becomes the burden for Spark 3.0.0, I'm OK to go with Janino 3.0.16 in Spark 3.0 - any further Janino bugs may require Spark 3.1 to resolve.
----------------------------------------------------------------
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:
users@infra.apache.org
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org