Moving towards graduation of Apache Stanbol – Our Story

Apache Stanbol is one of the great open-source projects that was founded during the lifetime of IKS project. Apache Stanbol has currently incubating status at the Apache Software Foundation (ASF). An incubating project, also called a podling, is a project that is not yet accepted by the ASF.

The ASF incubator has two primary goals:

So one of the major challenges of a podling is to build a community in order to be accepted. The ASF has a strong focus that a podling has a community that is diverse and strong enough to drive the project once it is accepted. With the IKS project and our large community of CMS vendors and developers the Apache Stanbol project developed a very strong community over time. With this community it should be possible to graduate Apache Stanbol in the summer of 2012.

Another aspect of graduation is to ensure that the people inside the Apache Stanbol podling understand how the ASF works. One important part is doing releases which conform to the ASF policies, especially in terms of licensing. We have spent a lot of time in Apache Stanbol reviewing and fixing all the dependencies that we have to third party components. We had to ensure that each used component and library has a license that is compatible with the Apache Software License version 2.0. In cases, where this was not the case, we had to find solutions and workarounds.

Apache Stanbol is written in Java and built using the popular build management tool Apache Maven. A consequence from this is, that all dependencies that are used by Apache Stanbol have to be available from a central repository server. If dependencies are not available from this server, the dependencies have to be packaged separately to the sources of Apache Stanbol in cases of releases. These “-deps” package have to be provided for download to ensure that people are able to build the software from its source. We ensured that all dependencies are available from the central repository and contacted projects to upload their components there to make them available. Another process that took some time.

A lean and easy to maintain process to create source releases was another challenge. Apache Stanbol is quite complex with a lot of bundles organized in sub-modules. The build management based on Apache Maven is a nice tool but we had to figure out how to configure and tweak it in some details to make it really easy to cut release candidates. This is closely related to the infrastructure that is offered by the ASF to stage and publish releases. The ASF offers a great infrastructure but people need to learn how to use it efficiently during incubation.

Apache Stanbol started as an incubating project in November 2010. One and a half years later we have managed to create two releases. The first release was Apache Stanbol 0.9.0-incubating which contained the complete Apache Stanbol software stack. The second release was an attempt to ensure that the project is also able to release single components of the software stack independently. With the release of the Apache Stanbol Entityhub 0.10.0-incubating the project has shown that its release management process works efficiently. The project members did understand how a release vote is managed and which formal steps are required to publish a release. To let the community verify and vote for a release candidate is a very fundamental principle at the ASF to ensure that releases conform to the policies. The first release took us seven release candidates to get it right but the second release required only three candidates. A good indicator that the Apache Stanbol podling process has worked.

Overall, Apache Stanbol seems to be ready for graduation and there are first attempts on the mailing list to start the process. From the perspective of the IKS project this would be a great achievement that underlines the impact of the IKS project within the open-source community of CMS vendors. With Apache Stanbol as a successful project that will live on at the ASF, the CMS community now has a great software suite with the potential to influence the field of semantic CMS for the next years. The IKS project will end in December 2012 but the Apache Stanbol project is hopefully just going to start its long life as a top level project at the ASF.

Author: Wernher

Wernher Behrendt is senior researcher at Salzburg Research and the coordinator of the IKS project

Comments are closed.