Context

For Tails to be correctly localized, we need to have a list of valid languages.

How to fix a FTBFS

If you are reading this, it's likely you just met a FTBFS. Ouch!

If your build logs include a line that starts with The following supported browser locales lack descriptions in, then this section is for you, so you can read on. Otherwise, you hit a different kind of bug, that we still haven't documented how to fix.

First thing: jump to the "How to add new locales" section, do it all, build again. If Tails still FTBFS, go on with this.

In the logs, look for The following supported browser locales lack descriptions in line. The last two characters in that line will tell you which is the missing language.

Now, we must determine what territories are relevant for that language: see How to manually associate a territory to a language.

Add it to po/po-to-mozilla.toml, in the extra_languages list (which itself is in the [extra] section).

Finally, run

./bin/locale-descriptions generate > config/chroot_local-includes/usr/share/tails/browser-localization/descriptions

How to add new locales

Our goal here is to add more locales to the [map] section of po/po-to-mozilla.toml.

Running this will provide you with suggestions:

./bin/locale-descriptions suggest

The first part of the output gives you suggestions. For each language, a list of possible locales is shown. Pick the ones that you consider more fid, using the methods and principles from How to manually associate a territory to a language.

The second part of the output is introduced by ## You can add the following block as-is, but please verify it first!.

As it says, you need to verify. Hopefully, copy-pasting will be enough.

Finally, run

./bin/locale-descriptions generate > config/chroot_local-includes/usr/share/tails/browser-localization/descriptions

How to manually associate a territory to a language

Given a language, our goal is to decide which territories it needs to be associated with.

If the language does not include any underscore or hyphen

You look at wikipedia's list of iso-639-1 codes, find the corresponding line. Then, you visit the wikipedia page about that language, and try to understand where it is most commonly spoken.

If there is a single prevalent territory

This is, for example, the case of ki (Kikuyu) or it (Italian).

Look for the 2-letter code of the territory in wikipedia's list of iso-3166-1 codes

Result: ${mozLocale}:${territory} (ie: it:IT, or ki:KE)

If it's spoken in multiple territories

This is, for example, the case of es (Spanish). How do we go here? We do pick all territories which are "important enough". This definition is very vague, not purely technical (political in fact). Those principles can help:

• we at least include the single territory which includes the most native speakers (ie: Mexico for Spanish, Egypt for Arabic)

• we at least include the territory which is more commonly associated with the language, if there is one (ie: Spain for Spanish)

If the language includes an underscore or an hyphen

For example, pt_BR. In this case, the territory is the part after the hyphen: BR.