Contributors to Drupal.org CVS since 2000
One measure of the momentum of the fine Drupal project is the number of people who are creating contributed modules on drupal.org.
The Drupal contributed projects are stored in a system called CVS and data about that is stored in some database tables that keep track of each change by each person. At the request of some fine folks who are working on important things, I got interested in the idea of the trend related to people committing code to the drupal.org CVS server. Here is the data graphed by the number of committers per month. It is not the number of commits, which would show how active those people are, but the number of people which shows how big of a group of people is doing this work.
Also, this is only about the contributed module and theme area and not about Drupal core. Drupal core commits are done by a very small group of people after that small group reviews the code contributed by hundreds of contributers. So, this really shows activity of the non-core projects.
I've labeled 4 points on the graph.
1. 2006 through Drupal 5.0 slump
Point 1 shows a peak at June of 2006 followed by a slow down until the trough at August of 2006 and then some small increases until December of 2006. Then there is a huge increase in people in January and February of 2007 which is also when Drupal 5.0 was released.
2. 2007 Follows a similar contribution trend
Up next is a peak in February of 2007 followed by the valley of the trough in June 2007 lasting roughly until December and then the committers really expanded in January and February of 2008 mirroring the pattern from the previous year.
3. 2008 sees more growth echoing the shape of 2006 and 2007
At point 3 we see another peak in March 2008 just after the release of Drupal 6.0 in February. A trough comes in May of 2008 followed by the big increase in January, February and March of 2009.
4. 2009: a plateau of committers active on drupal.org
And finally at point number 4 we see a peak in March of 2009. The bottom of the trough doesn't come until December of 2009 and then in January, February and March there are monthly increaseses of 3%, 5% and 8% that pale in comparison to the growth of 2008 (8%, 20%, 15% for similar months).
Drupal's contrib audience is growing, momentum has slowed - why?
So, we can see a real increase in the growth of committers to the Drupal contributions code base around 2005. The trend has been rocketing upward in a pretty consistent annual pattern with some minor variations perhaps due to major releases of the Drupal core codebase.
Recently there has been a trend of people hosting code on other sites. Code is posted to places like Github, Launchpad, and other similar sites that offer project pages and issue tracking with a more modern version control system (and without the dreadfully nitpicky CVS review process).
It's also somewhat possible that we have hit the limit of new stuff. There's fewer places for innovation (though I think a chart of projects created per month may prove that wrong).
And then there's also the possibility that the long cycle between Drupal 6 and Drupal 7 has created a lull where people don't need to do work. I certainly feel like many of the contributed modules I maintain are "stable" and don't necessarily need major improvements at this point.
My perspective is that this data provides great motivation for the project to move from CVS to Git and to relax the CVS application reviews in favor of a per-project and perhaps per-release review process.
Previous similar statistics posts
You may also be interested in:
- December 2009: Contributors to Drupal 7.x - End of Code Freeze Edition
- August 14th 2009: Contributors to Drupal 7.x - Code Freeze Looming Update
- December 30th 2008: Drupal 7: Who is Providing Patches for the Next Release?
Edit: Source data now attached.
| Attachment | Size |
|---|---|
| 201006_committers_per_month.ods | 46.6 KB |
- Login to post comments
GVS projects
GVS is now part of Acquia.
Contact Acquia if you are interested in a Drupal Support or help with any products GVS offered such as the Conference Organizing Distribution (COD).





Comments
We'll definitely see a large
We'll definitely see a large increase once D7 is released (prompting contrib rewrites in order for them to use the D7 goodies like the Field API) and d.o is converted to Git.
A lot of "future" code (both d7, and contrib for d7) was, and still is being done on places like GitHub, Gitorious, private bzr branches, etc. I expect this fragmentation will disappear once CVS is dead.
Also, efforts have been consolidated a bit lately, which is a good thing. We no longer need to develop 150 different image handling solutions, for example.
Much bigger barrier to CVS accounts
I think this might also be explained in how much harder it is to be granted a CVS account -- I'm pretty sure that's the main factor.
right, which I said ;)
Yes, and I said that in the post ;)
Really, though, the two are interrelated.
When we get a more modern RCS we can do access control schemes so that people can get a sandbox easily and then use that sandbox to prove that they should be allowed to create projects.
projects added
I've graphed projects added on slide 28 at http://www.slideshare.net/gabor.hojtsy/come-for-the-software-stay-for-th.... It doubled up to 2008 and then slowed down in 2009. No data yet for 2010, since we are mid-year.
Limits & Trends
I find this really fascinating- thanks for sharing, Greg!
I am curious why you say that Drupal contributers have peaked, since this graph would appear to indicate (to me) that we're on the edge of another big uptick in contributors (mind you I'm just looking at the graph, not the raw #s, so there could be something I don't see). You mention that the last two large bumps happened after a major version was released, and I'm curious why you'd discount what appears to be a thread when projecting forward.
When I look at this graph, and see the beginnings of the uptick before a release (which already has moved above the previous peak), what it says to me is that we may be on the edge of a large jump in contributers. Either way I don't see any indication that the slump for D6 has changed the overall trajectory. It's not surprising we're in a bit of a lull- D7 has been cooking for ~2 years now.
Interesting Information
I love seeing people break down statistics like this. I hate to sound like a parrot, but I agree entirely with your assessment. I look forward to seeing D7 get released and see the overall activity as a result, and I can't wait to see changes in the version control system.
If nothing else, I do take heart that our overall trajectory as a community has been pointed in the right direction even if it comes in fits and starts.
Thanks Greg for the research!
Cool, 100% disagreement
Less is better. Quality is desired. Kill the quantity. Users love us.
Compare the crappy WP plugin list to Drupal's. Compare current and future trends to 10 year old, completely different marketing sectors. Compare whatnot, but please don't try to see a negative trend in a "slowed momentum".
The reality is entirely reversed. Countless of contributed projects provide evidence. The world of Drupal contributions is constantly changing and improving. Thousands of new users and developers are changing Drupal's community nature. From a scratch-your-own-itch-developer-corner into a happy and friendly let's-join-forces-and-provide-real-value culture.
Less momentum here effectively means more momentum there. Elsewhere, before the shit hits the fan.
What's worse, we just need more humans in this anonymous pile of product.
Remix conclusions.
great insight!
This is some great insight - of course there are more ways to measure "success" than the one I'm using.
However I'm not sure I totally buy into the idea that "less is more" in this case. Certainly we need some contributors and more is better up to a point...it's only at the extreme that it starts to become a problem. Even then, many of the "problems" with having "too many contributors" are more about systems, social norms, and other problems that can potentially be removed allowing greater involvement by more people.
I think to really confirm this idea you have to provide some evidence (even anecdotal) behind the viewpoint that less is better in this particular case even though it has some merit for me.
Also, note that one of the complaints about WP's many plugins is that they are scattered all over the internet and not on a central site. I hope that with this post I can help encourage the infrastructure and social norms to keep all of our code as centralized as possible.
In-depth discussion
=> http://drupal.org/node/703116
Worth reading from the beginning.
I don't have hard numbers to share. Actually I don't care. I care about taking the time and energy to work within the facets we are talking about. Developing and reviewing for core, maintaining contributed projects, co-maintaining contributed projects, contributing to projects, and lastly, and perhaps even most importantly, also reviewing new contributions via CVS applications is key to understanding the current state.
Less is more meant the version control activity. -- Joining forces in building better APIs and products, a new understanding and importance of unit tests and user testing, better code review tools, and an overall increase of developers and users naturally lead to a shift in processes and change management.
Sure, we have and see more contributors. Not so sure about the quality of the tools we provide them to perform their job ("contribute") best. A new look/design won't improve the situation.
sun
PS: Total agreement on the importance of centralized hosting of extensions. Crucial to Drupal's success.
Github as central repo
Seems like the easiest solution to bridge this gap would be to allow releases to be tagged to a github link instead of just to a CVS tag. Since I haven't put any of my modules on d.o (mostly because I don't want to deal with CVS btw) I don't know how much is involved there. Only thing I see is that the number of commits would have to be scraped in which could cause more headaches. I'll have to talk it over at the Portland drupal meetup and see what other people think about trying to make this patch.
drupal.org moving to git
In case my subtle links weren't clear enough, drupal.org is moving to git :)
So, in the near future you'll be all set to contribute on drupal.org itself!
It's also somewhat possible
Everything has been invented, then? ;) I'm curious to see the follow-up chart of projects created per month...
We don't have and didn't
We don't have and didn't create everything yet, but given the currently available projects/modules, people start to understand that there is much more value in joining forces on existing projects, abstracting functionality to make it re-usable, and that duplication of efforts hurts everyone.
With fields in core, Views, Flag, Panels, (O)Groups, Feeds, Services, Migrate, Wysiwyg, and other game changers, we'll see a huge rise of features, instead of modules, very soon. We are slowly but certainly reaching the limit of atomic functionality, but of course not the unlimited possibilities of business logic, combinations, and use-cases.
Our challenge is to direct contributing forces into the right paths and provide a conceptually prepared environment for the add-on product functionalities.