2011/7/20 Fabian Christ <[email protected]>: > Hi, > > I chew the problem of OpenNLP models and EntityHub data over and came > to the conclusion that it may be a good solution to split releases of > the Stanbol Framework from releases of enhancement engines and data > models. > > The Stanbol Framework is usable software even it comes without any > Entityhub data models or enhancement engines. I could imagine that > there is a group of users who would like to have an empty Stanbol > environment that they can fill up with their own data and specific > engines. So releasing Stanbol framework independent of any data models > and engines would make sense. > > Additionally, it would solve our current problems and we could start > with our release cycle. Once we have an engine or data model that is > ready for use and can be released under Apache, we can release just > that engine. We should focus on making the installation of new engines > in Stanbol as easy as possible in the future. Rupert already did a lot > of work on this for the data models. > > This would require no change at all in our current trunk development. > I would only add a minimal Stanbol framework launcher that comes with > no data models and no engines. It's just to start up the framework. > This framework launcher would be part of the release. All other > launchers that already include engines and data models will not be > released. > > WDYT?
I think this is an ok solution for the short term, hoping that we will be able to package default models under a clear license for the next release. But I don't expect many users that really want an "empty" Stanbol. We will need to make it clear on the release notes that to be able to use the engines you need to build the data bundles yourself or download them from a non-ASF hosted site. However, why not embed the default, small dbpedia_43k index? It does not have any licensing issue. -- Olivier http://twitter.com/ogrisel - http://github.com/ogrisel
