holdenk commented on pull request #33784: URL: https://github.com/apache/spark/pull/33784#issuecomment-902141522
> Apache Solr is using Caffeine https://mvnrepository.com/artifact/org.apache.solr/solr-core/8.9.0 Sounds like a good example of potential conflict. I think it's good to distinguish between the client, which doesn't depend on solr-core or caffeine https://mvnrepository.com/artifact/org.apache.solr/solr-solrj/8.9.0 , and the server process (which seems like a less likely source of conflicts because I'd expect most people to be using the client). > > Anyway, we still cannot predict the Caffeine versions used by Spark applications directly since most of users won't publish their applications to maven central. Agreed, predicting conflicts from maven central is probably not the best way, but since you brought up maven statistics in your discussion I figured I'd point out that we use other libraries with more public usage than caffeine already. > > I'm definitely -1 for putting Caffeine on the Spark classpath directly. But +0 if other people think the extra work to shade the three libraries and the maintenance burden are the reasonable cost to pay. I'm not super sure how -1 on already committed but not released code works ( https://spark.apache.org/committers.html doesn't really call that out), but like I've said I'm only -0 on this revert. That said, there have been other reservations expressed by other committers you should check in with around consensus before merging. -- 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. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
