Inconsistent use of ISO language codes and country codes for localizations

[jrm]

Hello,

Most of the localisations in trunk/roundcubemail/program/localization/ uses only the ISO language code which I guess is okay, but some localisations like Swedish, Czech and Armenian uses the ISO country code.

For example the abbreviation used for the Swedish translation is 'se', but in the ISO language codes se=Northern Sami where Swedish should be 'sv'. On the other hand 'SE' is the country code for Sweden.

The same is true for the Czech translation 'cz'. 'CZ' is the country code for the Czech Republic, but 'cs' is the language code for the Czech language.

And the Armenian is language code is 'hy' and country code 'AM'. The language code 'am' which is used in RoundCube is Amharic.

Some languages use a combination of country code and language code, for example en_GB for British English. I find this the most natural and elegant approach, but at least the localisations should use the ISO language code instead of country code as the strings are translated into another language - not another country.

I have found the ISO codes on the gettext web page:

ISO 639 language codes - ​http://www.gnu.org/software/gettext/manual/html_chapter/gettext_15.html#SEC221
ISO 3166 country codes - ​http://www.gnu.org/software/gettext/manual/html_chapter/gettext_16.html#SEC222

It should be trivial for the maintainers of the individual translations to fix/update this.

Kind regards,
Jesper

[seansan]

Review in 0.1.1

[till]

Postponing since it's not urgent. But we'll get to it!

[alec]

I think, we should rely on HTTP protocol specification of Accept-Language header defined in RFC2616

[thomasb]

Language codes are now based on ISO 639-1 and ISO 3166-1