I agree with Gian. Calcite uses JMH. We don’t distribute the binaries it produces — because we only need them for running benchmarks.
It’s not an identical situation to MySQL because while there are other implementations of JDBC — e.g. Postgres — there is only one implementation of JMH. But JMH is only used during testing, not production. So, let’s not do anything rash, like starting to remove code. I think we will have clarity in a few days as various folks in Apache react to what Owen has done. We can revisit this as part of the first Druid release. Julian > On Apr 23, 2018, at 2:14 PM, Gian Merlino <[email protected]> wrote: > > I'm not sure why the ORC folks decided this was necessary. I found these > two links, > > - https://issues.apache.org/jira/browse/ORC-298 > - https://lists.apache.org/thread.html/c27267a6a3df9b76c5e036fc181c86 > 5150639f4cf15ffc5cca27b690@%3Cdev.orc.apache.org%3E > > It looks like they ended up moving the code to a different, non-Apache > repository. > > The relevant Apache policy, I think, is: http://www.apache.org/ > legal/resolved.html#optional. From what I've seen it comes up most often > with MySQL drivers (which we also use). See also https://issues.apache.org/ > jira/browse/LEGAL-200. > > It seems to me like JMH and MySQL are in the same boat. Both are optional > dependencies -- there is no reason that a "normal" user needs to run > druid-benchmarks. If this understanding is correct, then we should just > make sure we aren't distributing JMH (or MySQL drivers) as part of binary > releases. But I would be interested in understanding the thought process of > the ORC folks in more detail. > > On Mon, Apr 23, 2018 at 1:44 PM, Roman Leventov <[email protected]> > wrote: > >> See this: >> http://mail.openjdk.java.net/pipermail/jmh-dev/2018-April/002743.html JMH >> might be not welcome in an Apache-licensed project. >> --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
