A month ago, Drupal.org moved from CVS to Git for version control. There's a good chance you already knew that, as it's been widely discussed. Another significant change happened at the same time, and hasn't received as much attention: every Drupal.org user can now create "sandbox" projects on Drupal.org. Before this change, creating your first project on Drupal.org required someone to review the project, read and test your code, and make sure it's up to the standards of the Drupal community. Despite the tireless efforts of many volunteers, this review process often took months and turned into a bottleneck for new contributors, draining their enthusiasm before they even started.
Now, to create sandbox projects, new contributors can just check a box agreeing to Drupal's contribution policies, and immediately start contributing. There are, of course, major benefits to the community project review process, and that hasn't gone away for "full" projects. There are also a few barriers in place to prevent people from accidentally using sandbox code before it has gone through the review process. In theory, the newly available sandbox projects should be more inviting to new contributors and get them invested in the Drupal contributor community before their work is exposed to the larger Drupal user community.
I was curious how well sandboxes are accomplishing these goals in practice, so I did some research. All of the data I gathered here is publicly available on Drupal.org. I've simply aggregated it and applied it to answering some specific questions.
Are Sandboxes Being Used?
The new sandbox project functionality does no good, of course, if no one actually uses it. Far from that, Drupal.org has been dealing with the opposite problem: too many sandbox projects. What I wasn't sure about before looking at the numbers was whether this use was a flurry of activity at the beginning, or a sustained pattern. So far, creation of new sandboxes appears to be sustained, at an average of over 30 new projects per day, over four times the rate of new full projects created over the same time period.
There's also some early indication that sandboxes are being used for a wider variety of projects, with a larger proportion of themes and install profiles than we see among full projects. So we know sandboxes are being used, but how are they being used?
Are Sandbox Projects Being Actively Developed?
I looked at two factors here: number of Git commits and project development status, both compared to full projects created over the same four weeks. There's a slight trend toward less activity in sandbox projects, most pronounced in the large number of still-empty sandboxes, with no committed code at all.
The development status numbers show the same trend, with slightly less active development on sandbox projects.
One might have expected sandboxes to be more active than full projects, free from constraints like, say, working code. But community feedback is no doubt an important factor in driving Drupal.org project development. How much of that feedback are sandboxes getting, with their reduced visibility?
Are Sandbox Issue Queues Being Used?
One of the most promising benefits of sandbox projects, for me at least, was the exposure of the issue queue, the primary project management tool for Drupal, to all potential contributors. I wanted to see how issue queues were being used, so I looked at issue counts within both sandbox and full projects created over the past month.
While sandbox issue queues are being used much less than full project issue queues, the difference is actually more subtle than I was expecting. Despite the reduced visibility, the number of people finding, using, and providing feedback on sandbox projects is promising.
Who is using Sandboxes?
My last question: are sandboxes being used primarily by new contributors, who generally don't yet have access to create full projects, or by experienced contributors who simply don't feel these projects are ready for full project status? To answer this, I looked at the user IDs of project creators, and mapped those to the year they joined Drupal.org.
This seems to be the largest difference between sandbox projects and the full projects created over the same month: sandbox projects are being created by much newer contributors, with years less experience on Drupal.org. The number of sandbox projects created by people who created Drupal.org accounts this year is especially encouraging.
Are Sandboxes Successful?
My definition of "success" isn't necessarily shared by everyone, of course, and it's still very early to be drawing conclusions. But in general, the introduction of sandbox projects seems to be accomplishing what I've hoped it would: getting a lot of new contributors more actively involved in the Drupal community. There's still room for improvement, especially around keeping the new sandbox contributors involved in Drupal.org beyond initially creating a sandbox project. But these early indicators look very good.
Drupal 7 co-maintainer Angie "webchick" Byron recent wrote:
Based on my research, I'm hopeful many future Drupal contributors will be coming from among those starting with sandbox projects today.