Over the last few months I’ve noticed a pattern emerging on Google Code’s Project Hosting. Almost every single major project (excluding, although sometimes including, Google’s own projects) on the site seems to have migrated over to GitHub. I decided that I would do a little investigation to calculate the proportion of projects that have moved over.
Google Code allows you to search for products by tags so you can view projects written in a specific language. On their homepage they have a list of tags that I assume are most popular/most committed to/most downloads. I then went through this list and recorded the total number of projects for each tag. Obviously there will be some overlap. I then carried out each search again with GitHub appended onto the end, hopefully identifying projects that mentioned that they had moved over. Obviously a more accurate way would be to get a full listing of all projects on Google Code and then to try and find the same project on GitHub, but my method seemed accurate enough – all of the results I clicked on definitely showed themselves as having migrated.
I then recorded all of this in a spreadsheet:
Obviously there are a few things to note from this. Firstly, almost 30% of all projects mentioned GitHub in their description. Secondly, the total repos thrown up by a search often produced very, very similar numbers so I cannot be certain that these numbers are accurate (on the other hand it could be that every Android project is also tagged with Database, but that is clearly not true).
A very important thing to note is the order. I assume that the tags are listed in some kind of popularity/activeness order on the Google Code homepage. It certainly isn’t total number of repos. Therefore we may assume that more popular projects written in Java and Python have also been more active historically, however these projects are now moving to GitHub.
Very little can be concluded from this except that there is definitely some migration of projects away from Google Code and onto GitHub (and I get almost 4 million search results for ‘google code to github’) and even without statistics the open source community is definitely converging on GitHub.