“There are three kinds of lies: lies, damned lies, and statistics.”

Mark Twain, originally by Benjamin Disraeli

Seeing some nifty statistics from the Fedora Ambassadors group, I decided to jot down some statistics for the Localization group too. Once in a while, it helps taking a step back, and be honest, judgemental and stern with ourselves, before finally concluding what an awesome job we did.

The dear reader is encouraged, with Mark Twain’s words in mind, to go on and read the following ultra-deceptive lies about FLP. The Sirens disguised as charts ruthlessly suggest and persuasively imply that the FLP, just like the Ambassadors group, is indeed growing and improving. It has been reported that the text has even created impossible feelings like, for example, that the grass is not always greener on the other side of the fence. Or that the circle has only one side.

On to the data now.

(In some of the following graphs you’ll notice the absence of some data between last February and today. Feel free to blame my today’s lazyness and stiffness from yesterday’s rock climbing.)

Community

In the past 2 years we’ve changed a lot of things on how the L10n project is structured, how the language teams are organized, and how the actual work is being done.

On March 2007, the group ‘cvsl10n’ was created on the Fedora Account System. We used this group to identify translators and give them access to all projects hosted on the Fedora CVS server. Since then the Fedora Account System group has evolved and became the ID of a “Fedora Translator”. The following chart shows how the number of the Fedora Translators grew the past years.

Approximately 20 new translators join the group each month on a steady rate.

On communication, fedora-trans-list is the main channel of the FLP (together with #fedora-l10n of course). The following graph shows the volume of emails sent to the list for the past four years.

This graph reflects the “bandwidth” increase in the FLP’s main communication channel. Compared to 2005, the list’s content for 2006 grew by 200%, and each consecutive year saw a further increase of 70%. More issues get discussed and resolved, more people participate.

Uptake

With the term “project coverage” and “language coverage” we refer to the number of the resources (software, documentation, websites) we are translating, and the number of languages we are translating these resources to. High project coverage means more translated interfaces, and high language coverage means more Fedora users happy.

One of the best translatable resources Fedora builds are the release notes. I like looking at their coverage when I talk about the health of the FLP, since they are a tough resource to translate, both because of its size and its importance for every Fedora release. Here’s a graph showing in how many languages we shipped it since day 1.

Fedora is upstream for a number of projects. The project coverage measures how many projects we can translate. Increasing this metric was one of the reasons we migrated from elvis to the open Fedora Infrastructure: to allow more projects to be translated. One and a half year later, our infrastructure has improved, the community has grown, and more and more projects request translations. The following chart shows the number of projects that were available for translation on elvis and today.

Finally, the following chart shows the number of commits which happened through Transifex in the past months. In total, 5400 commits took place in the past 15 months, averaging 360 commits per month.

The string freeze periods where translators push their contributions can be seen as the ups in the chart.

L10n engineering

The main FLP tools live in cvs.fedoraproject.org (for now [hg soon], muahaha), ‘L10N’ repo. Files needed by the FLP to work properly are maintained there, such as the translation interface and the owners.list file that syncs with Bugzilla to have components created.

By February 2009, around 3000 commits have landed on cvs/L10n measuring around 125 commits per month, or 4 per day. It’s worth noting that this repository receives a lot of love by our community members (thumbs-up to Piotr and Diego). These guys have been tirelessly maintaining the tools and keeping them in shape to serve our translation community as best as possible. The following chart helps illustrate the involvement of the volunteer community (non Red Hat employees) in the Fedora Localization toolchain.

Extremely happy to see so much community involvement in the admin side of FLP.

$me happy, listening to “Side” by Travis.