We had a "Birds of a Feather" session at the end of the MeeGo Workgroup day, and I (Margie) was very pleased at the number of people who stayed to talk localization. I didn't catch everyone's name, but we kept it very informal, and the notes below are from memory, so I ask those that were there to add their notes, correct mine if I'm remembering things incorrectly, and add comments.
These notes are not in any particular order. We have a page here L10N/RoadmapProposal that covers many of the topics below.
We talked about languages and what constitutes a "priority" language from a community-contributed language, and how does a language get to be a priority language? What is the criteria to be a priority language? Margie mentioned that these languages are the ones that are asked for by customers, OEMs, Intel, or Nokia. These priority languages will be "officially supported." This led to a discussion of what that means. There needs to be a checklist or some sort of approval process so that those who would like to see their language become a "priority" know what to do to achieve that. Dave Neary mentioned that perhaps submitting a language at 100% for at least two releases shows a commitment on the part of the community for that language.
Working with upstream was a hot topic. All present agreed that the gnome l10n community in particular is a very tight-knit group, and working with them presents its own set of challenges. We discussed that there are three main branches (for lack of a better term): MeeGo components for which MeeGo *is* upstream; gnome upstream; and "extras" that are contributed by the community (sort of like the N900 extras on transifex.net). We talked about how to work with upstream--Dimitris had some ideas, but admitted that it is difficult. We didn't reach any firm conclusions about how to proceed.
How do we control the quality of translations? There are many established groups in the community that control who translates, who edits, who reviews, and who signs off on translations. We are nowhere near that at the moment for MeeGo except for the "priority languages." How do we prevent someone from overwriting existing, perhaps already-approved translations? Dimitris mentioned that the latest version of Transifex has language teams and leads, with a dashboard top track status. Miia talked about the quality (or lack of) of the Finnish translation she was looking at, and reiterated the need for some sort of review and approval process so bad translations don't get in. Margie asked that if anyone saw poor quality translations in any of the priority languages to send her an email with the specifics. Intel is using an internal translation team for MeeGo for this first release, and it has to be good.
MeeGo is now in string freeze, and all changes going forward must be approved by a Change Control board.However, Margie stated that in preliminary testing, she found a lot of English appearing, which means the code has not been internationalized, the strings have not been externalized, or the developers simply didn't integrate the latest PO files when they built their packages.
We are still working on the details for migrating translate.moblin.org to its new home on transifex.net. Dimitris and his team are working out the details.