This is the third part of the report from the recent RMLL/LSM 2010 “Development” technical track (see links to previous parts at the bottom of this post)
This part deals with the forges afternoon that took place on the 7th, co-authored with my colleague Madhumita Dhar (who’s also working on the COCLICO project).
The first forge to be presented by Sylvain Beucler was Savane, which is the software that runs the FSF’s forge Savannah (EDIT: I have uploaded a preview video of the talk). He spoke about the previous and future road-map of the Savane project. Breaking away from its legacy PHP code (sourcing from the old libre Sourceforge codebase in PHP), Savane is going in for a complete code rewrite. This is quite a bold decision and awaits a lot of work for everyone involved. However it also allows the choice for newer tools (Django), languages (Python), practices, and should provide a more modern, easy-to-wield forge at the end of the day. Main novelty should be in a new model of forge that would be decentralized (following recent trends ala git, and guidelines about freedom in the Cloud by Moglen and others), but I’ll blog on that topic later.
The second presentation was that of fusionforge, by Roland Mas (EDIT: I have uploaded a preview video of the talk). Like Savane, Fusionforge’s ancestry also started as a fork of the old Sourceforge. The path followed by them in recent times, interestingly however is quite different from Savane’s plans. Instead of building a new fusionforge, they are trying to make the current one better, providing more features and upgrades, through code clean-up and smooth refactoring. As Fusionforge has a good user-base (often as old heavily patched GForge 4.x deployments), the effort is also more to merge in the plugins/patches developed by its users and bring about more contributions (hence the fusion term). A considerable amount has aleady been done, with more on the way. It was also mentioned that as part of efforts conducted in the COCLICO project, there’s now some converging code (plugins, for instance) between Codendi and FusionForge to try and reduce implementation efforts (Codendi is another of the cousins of Savane and FusionForge forked-off from the Sourceforge base).
Beyond efforts on FusionForge, together with other COCLICO project participants, Roland and me initiated the PlanetForge initiative, that seeks for more discussion, interoperability and other experience exchange among forge implementers and users. It was discussed during the Q&A session, and we tried to clarify that the goal is not to make other forges merge to FusionForge (even if different forks of the same PHP codebase could benefit from merging together). The goal of PlanetForge is to create an interest community, so that each FLOSS forge project can continue to live its own life, but that, if possible, all can better share information, APIs specs and other interoperability means. In short, be different, while still interoperable to each-other.
The next presentation was about Launchpad, by Jonathan Lange (EDIT: I have uploaded a preview video of the talk). It was a general presentation, explaining how Lanchpad came to be developed, the current priorities, motivations of the people who are behind launchpad, its different features, and so on. There was some discussion afterwards on whether the goals of openness seeked by LaunchPad authors can be compatible with the technical difficulties in setting up LaunchPad instances for one’s own use.
The last presentation of the afternoon was on How to choose a forge for your needs by Joseph Roumier. A good talk outlining the why and how of the methodology used by his team at CETIC, to go about selecting a forge, for their specific needs. Any guesses for the winning forge ? You’ll have to watch the video, or read the slides.
Previous parts :
- (1/4) Report on programming languages and techniques in Development session of RMLL/LSM 2010 (1/4)
- (2/4)Report on ALM, QA and industrialization methods and tools presented in “development” track at RMLL 2010 (2/4)