Proposal for a subtag registration (fr-2004-ORTOGRAF or fr-ortograf ?)

David Starner prosfilaes at
Sat Dec 15 01:33:41 CET 2007

On Dec 14, 2007 12:48 PM, Mario Périard <mario_periard at> wrote:
>  For instance, if I want to translate Ubuntu (the Linux distribution) in
> order to allowed people to choose "Ortograf alternativ' during the install
> or update process of Ubuntu.
>  To do this I have to create a new translation team (directory):

No; you just need that to be an official translator. It's all open
source to do with as you will.

>  I don't know the technical or political reasons behind that, but a language
> code is often required to be a part of a translation sortware project.

Because computers need a way to label languages unambiguously and
consistently. That's why language codes were invented.

And the political reason is that there's always people coming up
wanting to translate major projects in Pig Latin, Klingon, Toki Pona,
their dialect of a standard language, etc. A language code is at least
one line of resistance against frivolous projects that no one will
really use and that will probably be abandoned.

Karen_Broome at wrote:
> Reviewing the Ubuntu site, they are not complying with ISO 639-1 or
> ISO 639-2 today and it appears they may already be assigning non-standard codes.

If you check out the links, all the apparently non-standard codes are
all language-country combinations.

> Or possibly they already use RFC 3066 or 4646 and would accept
> a private-use variant code.

That does not follow. First, I think right now they're probably using
the language-country subset. Secondly, private-use codes may be
forbidden as a matter of policy.

Honestly, I think the major projects like GNOME and Ubuntu are
overloaded on language translations; looking at the Ubuntu list sorted
by completeness, I count 56 languages that have not translated even
700 strings out 370,700. There are dozens of Wikipedias with a few
dozen tiny articles. They don't need more translations that few people
would use even if they were complete. So forbidding private-use codes
makes perfect sense.

More information about the Ietf-languages mailing list