This is a good idea, especially in light of some of the frustrations around Avro. From a project perspective, what are your thoughts related to plugins that could be released as experimental? Do you think we should setup a separate project from the Drill project? Should a goal be to move plugins from this project into Drill project after they meet certain criteria? I would want to see some level Apache like oversight to ensure an experimental plugin project asks for all contributions to be Apache licensed, have source code fully available, and be somewhat transparent. I am curious what your thoughts are on this, and if this is something that could be done as part of the Drill project, something separate, but adhering to the Apache way, or just "wild west".
I think if we did this, we should ensure it's easy enough for non-developers to easily compile (or use "released" versions as part of the project) so they can use them without being a developer. That can be challenge for some of us. :) Also, should this be just plugins, or should we create a community around UDFs as well? I think there is a huge opportunity for both. On Sat, Apr 2, 2016 at 5:49 AM, Magnus Pierre <[email protected]> wrote: > Experimental plugins: > I strongly believe in the open architecture offered by Drill, and I would > welcome a plugin dictionary separated from core Apache Drill where > "experimental" and / or even commercial plugins could be shared. (similar > to how Spark is dealing with packages) I believe experimental plugins to be > central to drive adoption and new features, but I am not sure if they need > to be controlled the same way as the core or by the same developers. As a > hobby plugin developer it is a tad bit frustrating to not be able to share > work and by that find interested participants to bring a new features into > the Apache Drill community. Does anyone else have a view on this? > > Regards, > Magnus >
