Github user JoshRosen commented on the pull request:
https://github.com/apache/spark/pull/12076#issuecomment-205690484
My understanding so far:
- [Marcelo
showed](https://github.com/apache/spark/pull/12076#issuecomment-204129446) that
we can use shaded Kryo in Hive.
- While the Protobuf shading might have originally been introduced for
Hadoop 1.x support, it's probably a good idea to keep the shading because
Protobuf is a conflict-prone dependency.
- We can't use the regular `hive-exec` artifacts because they package tons
of dependencies without relocation.
- We can't use the `core`-classified `hive-exec` 1.2.1 artifacts because
they don't shade the things that we need shaded. Even if we could use this
artifact, it is a pain to consume because its POM does not declare required
dependencies, so we have to manually add all of Hive's transitive deps as
direct deps. We've already sort of done this, but I believe that it's
unnecessary and am in the process of undoing a lot of those changes as part of
#1217.
Therefore, I think there's no avoiding having to publish another custom
Hive 1.2.1 build. I propose that we do so by editing the 1.2.1-spark POM to
restore Kryo shading. I'll work on this tomorrow and loop back to this
discussion once I've completed the required Hive dep. bumps.
---
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]
