Mobile/Projects/Non-multilang build localizations
Make it possible to ship an increasing number of Fennec localizations without having a significant negative impact on shipping APK size.
Bugzilla query to come.
- Figure out who outside of the Fennec team should be involved in the initial brainstorming & planning of this feature (joduinn, axel, ??)
- Book a workshop/meeting to start brainstorming potential solutions (during Fennec workweek in Sept)
- Compile ideas for potential technical approaches & possibilities (use the "Brainstorming" section below for now)
- Start defining the user stories & impact -- what's the ideal user experience? How close can we come to it given technical restraints?
This is currently for brainstorming & potential approaches -- will be replaced with a "Technical strategy" section later when we have chosen one.
Use language packs. This is hard for the Android side of things, as long as we continue to use the Android l10n system. We can't really use a gecko system for l10n, but we can use anything else, including, but not limited to, l20n.
L20n could be natively implemented in java, and thus start up without having to wait for gecko to be up.
Perf impact of this is completely unknown, of course.
At Black Hat, Chris attended a presentation that discussed the number of android users that actively download apps outside of the Google Play store. The presenter estimates that nearly 50-60% download the majority of their apps from other marketplaces and estimates that this is due to the way that Amazon has driven their own marketplace for Android. Perhaps this trend indicates that hosting our own shippable builds is a viable option.
- Additional resources
- Minimal increase to shipping APK size
- No user-noticeable performance impact