Hello all (particularly our mentors!), I'm working on the release packaging for 0.13-incubating and have a question around best practices for distributing a library with a license not compatible with Apache.
This library is a dependency of an extension to use MySQL as Druid's metadata storage. Historically, the way we handled this is by separating the extension from the core distribution and providing a link to download the extension and instructions on how to install it. For reference, see Druid's download page: http://druid.io/downloads As I understand, release artifacts are distributed through Apache's network of mirror servers, and I suspect it would not be appropriate to distribute this GPL-licensed library in this way. My question is if it would be appropriate to continue to do something similar to what we're doing - that is, to package our compiled MySQL extension + the Java connector library and to distribute it from a non-Apache network (such as through S3 + CloudFront). Otherwise, would we have to separate our compiled code from the GPL library, distribute the non-functioning MySQL extension alongside the rest of the release on Apache's servers, and then direct users to fetch, say, mysql-connector-java-5.1.38.jar from elsewhere and drop it in the appropriate location? If the latter, would it be appropriate to include instructions and links to that library from the Druid website / docs? Thank you very much! David
