Me and Cristi Silaghi had the honor of being invited to attend one of this year’s L10n hackathons organized by Mozilla Slovenia and the L10 drivers team. The event lasted for two days and focused on presenting coming changes to the L10n community, new workflows and the very central issue of translation quality raised by Jeff Beatty. Invited were members of the Balkan languages L10n communities within Mozila: Romanian, Hungarian, Serbian, Macedonian, Armenian and of course, Slovenian.
Our goals were lofty and I believe we did them justice given the available time.
Staś (L20n developer) did a full blow-by-blow on the Mozilla L10n blog, so I will use this opportunity to just give our perspective on the event.
I was pleasantly surprised by the amount of work done on improving the development-side of localization, which will make work easier for everyone involved. Tools and repository management is getting better and less complex for localizers. I also liked the honest discussions we had about workloads, deadlines and communities. Communities are messy to deal with (even more so volunteer ones), but Mozilla is committed to them and also receptive to feedback, good and bad.
On the technical side, I welcome the splitting of the monolithic string bases into “buckets” (like: devtools, UI, errors) which will make it easier to prioritize things which matter for each individual locale. For example, in the Romanian locale we will be removing devtools, which is something few if any actually need or use. Devtools actually makes most of the text of Firefox, and contains complex and highly technical strings. Removing devtools localization from Romanian will thus more than halve our workload, allowing us to focus more on the things that impact users directly. With “buckets” we will be able to show 100% completion on those bits of Firefox we find most relevant.
Also technically, I like the improvements made to Pontoon and the “secret” plan to move Firefox to L20n, a new internationalization format. L20n is more featureful and flexible than gettext. Although for us in Romanian we won’t need most of those features, they will definitely help Polish, Hungarian and the Slavic languages (and possibly many others).
I was particularly impressed by live-updated Firefox translations using L20n, Pontoon and Tinker. In the not-so-distant future we will have direct contextual information in our translation of programs and not just websites.
Delphine also detailed the plan to introduce MQM, a formal means to evaluate translations and a way to give people in the driver team a better overview of the quality of translations. It may sound fancy, but it just means when reviewers reject strings they will have a set of canned “reasons” for every action: typo, grammar error, space violation etc.
The discussion went a little into Terminology and the team asked me to put together a short presentation if possible. The next day I talked for 15 minute with some ideas on improving or adding terminology control to the tools we use.
On a lighter note, I am officially the 5th most active localizer for Bulgarian and 10th for Slovenian in Pontoon :) (the story of that, at a later date)
Later in the evenings we also had a chance to see the beautiful Slovene capital, with Nino and Gašper as guides.
The second day also included a trek up hill to Ljubjanskigrad castle, which tested us a little bit.
But up there, the view was worth it.
All in all, the event looked great, which is not a mean feat. I congratulate Gašper and Nino for the tremendous effort done in organizing this, as well as the rest of the Slovenian community for hosting us. I feel I’ve been a little slovenized and want to vising again :) Also, great thanks to Delphine, Flod, Matjaž and Staś from the drivers team who carved out the time to come.
Next post: what we didn’t get the chance to talk about.
(All photos and content in this post available under CC-BY-SA 4.0 or later unless otherwise specified)
