GNU Savannah Statistics

• Commits data includes nearly all git/hg/cvs repositories hosted on GNU Savannah servers, but few repositories were not included due to technical issues.
• Commits data does not include Bazaar (bzr) projects. If you are a Bazaar-expert and interested in helping, please write to savannah-hackers-public@gnu.org.
• Data for 2014 includes January-2014 to September-2014. You can either extrapolate, or ignore the year 2014 for now.
• For questions, suggestions and comments please write to savannah-hackers-public@gnu.org.
• See also more GNU savannah project statistics

• User counts based on commit logs is roughly estimated. No attempt was made to reconcile multiple IDs as one user.
• GNU Savannah website was created in the year 2000. CVS Commits to GNU projects (hosted on GNU Savannah) go back to 1985.
• When counting by unique user names, foobar, foobar@gnu.org and foobar@gmail.com are counted as three distinct users.
• When counting by unique real names, "Foo J. Bar" and "F. Bar" are counted as two distinct users.
• Therefore, the counts should be taken as a likely maximum number of unique users, with the actual number a bit lower due to duplicated users.
• The Max. active users is the sum of the users who submitted bugs/patches/tasks and the highest of the two user counts based on commits (implicitly assuming there was no overlap between the two groups). The actual number is likely a bit lower.

• There is a large portion of inactive users on GNU Savannah website

• The charts counts items which were submitted though the GNU Savannah web interface.
• Bugs submitted though other methods (e.g. 'GNU Coreutils' which uses the mailing list and debbugs) are not counted.

• Despite increasing number of hosted projects and users (see below), the number of items (bugs/patches/tasks/support tickets) has not increased, perhaps even declined.

• Data is based on commit logs: If a project had any commits in a given year, it is counted in this chart. Project registered on GNU Savannah but without code commits are not counted.
• web sites refers to GNU Savannah's CVS/Web mechanism - commits to this repository update the project's website on www.gnu.org or www.nongnu.org .
• Some projects do not host their websites on GNU Savannah, and are not counted here.
• Projects can be active (or at least not decomissioned) but not have any commits in a given year. For example, GNU gperf had no code commits in 2013. As such it was not counted towards the total in the year 2013.
• "Impossible" git/hg commits before 2005 - project which switched to git/hg from previous CVS repositories will have commits with dates prior to 2005 (when git was released).

• Sharp decline in CVS after 2005 (when git/hg/bzr appeared).
• Noticeable decline in activity since ~2009 (when GitHub/BitBucket/Gitorious gained momentum)

• These are software projects which officially part of the GNU project.
• As of September 2014, The GNU website lists 378 active projects, and 78 decommissioned projects.

• The GNU Savannah website hosts many projects which are not officially part of the GNU project. These are dubbed non-gnu.
• To host a project on GNU Savannah servers, the project must use Free-Sofware license (see hosting requirements). Unlike other hosting services (e.g. GitHub/BitBucket), projects must be approved by GNU Savannah administrators (users can not create their own projects without approval).

• Many non-gnu projects are "abandoned" after a while.
• Some non-gnu projects have migrated to other hosting services

• The correct way to read this chart is: "For a project registered on GNU Savannah in year X, which VCS does it use today".
  Example: GNU Coreutils has been registered in 2002 (using CVS at the time). It has since switch to using git. In this chart, Coreutils will contribute "+1" to the value of git since 2002.
• VCS Usage data is based on the GNU Savannah website database information, regardless of whether actual code was commited or not.
  Example 1: GNU a2ps's website indicates both CVS and GIT usage. In this chart, it will contribute "+1" to both cvs and git since since 2002.
  Example 2: FIM's website indicates SVN usage. The project does not have any code commited in the SVN repository (but have files available for download elsewhere). In this chart it will contribute "+1" to svn since 2007.
• This chart should not be used to evaluate the number of active/inactive projects. Project activity status is a difficult question, so no graph for it is presented yet.