After the preview of the analysis of Allura, we present another analysis on a software forge. And again, one derived from the software running SourceForge, although in this case derived from the old software that was running it around 2000: FusionForge. This software project is an offspring from GForge, which was itself an offspring of the last free version of the primitive SourceForge software.
FusionForge was born in February 2009, when some of the developers of GForge decided to fork its GPL branch into a new project. GForge in turn was born in 2002, when the last free software version of the original SourceForge project (release 2.6) was merged with one of its forks, the Debian-sf project. Fortunately enough, the git repository maintained by the FusionForge team has all this history (which by the way, is described in more detail in the Wikipedia article about GForge), back to August 2001.
We at Bitergia have analyzed this git repository, along with the project mailing list archive, which starts on January 2009, with the birth of FusionForge. You can read some more details about this work-in-progress report in the rest of this post, or go straight to the preview we have prepared for you.
The charts about commits and committers tell a lot about the history of GForge GPL and FusionForge. It can be observed how the project started with a really small team (2-3 committers) for about one year, but soon was joined by others, until it stayed around 6-8 active committers per month in the period 2002-2006. Activity during these years varies, but is usually above 50 commits per month.
In 2007, the pattern of activity clearly changes. The team is reduced back to 2-4 committers, with some months having almost no activity. Even when the number of active committers grows to 4-6 during 2008, the number of commits is still low.
Only in 2009, when FusionForge is launched, the activity recovers the highest levels of 2003, and more. Developers seem to work in bursts, since large peaks of more than 500 commits/month are followed by less active periods of around 200 (still clearly higher than most of the pre-2009 months) . The team of active committers per month comes back to 6-8 or more. After the 5.1 release, in mid 2011, activity stabilizes in lower numbers, of around 150 commits/month (except for January 2011, when very little activity is observed).
Recently, the project seems to be maintained by 5-6 committers, and activity seems more stable (even still having the usual peaks and valleys).
During all these years the project moved from svn to git, so not all commits are comparable (people tend to work differently with both systems), but except for this, we are not aware of project policies or other constraints that could explain the differences in commits and committers except for the sheer activity (of course, knowing that not all commits are equal).
The analysis of mailing lists shows also a great peak of activity (both in traffic and in people sending messages) the first months of the project, supposedly while they were organizing all the new stuff.
After a sharp decrease which rendered the lists as almost unused for some months, activity again bloomed during the first half of 2010, wich peaks of traffic of 147 messages and 22 active senders per month in February 2010. Since mid 2010, activity has been lower, but also more stable, until a new increase around Summer 2012.
Comparing the community participating in mailing lists, with around 15 persons per month during the life of FusionForge, with the team committing, a ratio of about 2:1 appears, which is maybe a bit low when compared to other communities. Maybe this is due to the kind of software the project is producing: A software forge is not something that a random user installs and configures, so it is reasonable that only a few people are interested enough in the project to participate in mailing lists, even just to make questions, when compared with an application targeted directly to any desktop user, for example.
All in all, the project seems to feature a small but engaged community, which has been able of forking (twice!) when they were not happy with the situation of the software they work with, and of finding time and resources to maintain it alive during more than a decade since its first producer (SourceForge) stopped to distribute it as free software.
Now that SourceForge is again pushing for a free software system (Allura), built with a more modern software architecture, and willing to be covered by the Apache umbrella, it is interesting to see what’s going to happen with FusionForge, and how both are going to compete for users (if they compete at all, since both could be targeting different populations).