(Siebrand, I am unsure if this will arrive at mediawiki-i18n, feel free to forward it you consider it interesting to them).

OK, I've written a few lines of Python [1] which actually helped me answer my questions. Sorry to bother.

And the answers are yes, yes, no, but close, and i hope so.

There are a small number of wikis which use a different language code than their site code is, namely:

crh -> crh-latn
als -> gsw
be-x-old -> be-tarask
roa-rup -> rup
simple -> en

But, at the same time, the given *site* codes exist as *language* codes as well, i.e. the languages/messages files exist for them, but they just fallback to the given language code (i.e. MessagesAls.php just names gsw as a fallback).

I would not be surprised if each of these five examples would have an anecdote to explain why they are the way they are :)

Thanks,
Denny

P.S.: There is one thing I do not understand though. According to https://simple.wikipedia.org/w/api.php?action=query&meta=siteinfo the language of simple.wp is "en", but MessagesSimple.php seems to be taken into account (instead of "edit" it has "change" in the UI, one of only two changes in MessagesSimple to MessagesEn). So it seems that the language is "simple" -- why does it say "en" in the siteinfo?


[1] available here: <http://pastebin.com/JpApSmNX>

2012/5/9 Siebrand Mazeland <s.mazeland@xs4all.nl>
Forwarded. Please cc Denny,  as he is not on this list.

Begin doorgestuurd bericht:

Van: Denny Vrandečić <denny.vrandecic@wikimedia.de>
Datum: 9 mei 2012 01:36:01 GMT+02:00
Aan: MediaWiki Tech list <wikitech-l@lists.wikimedia.org>
Onderwerp: [Wikitech-l] Language codes vs site codes
Antwoord aan: Wikimedia developers <wikitech-l@lists.wikimedia.org>

Hi,

Wikimedia projects like Wikipedia exist for a large number of languages
(>280, wow). Let us call them the sites. They are identified by the site
code used in the subdomain and in language and interwiki links, e.g. "en"
for en.wikipedia.org, i.e. the English Wikipedia.

MediaWiki has a number of interface languages. These are selected through
the preferences. They also have short language codes that identify them and
which are used, e.g. in the localization of the code. "en" is used for
English.

My four questions:
* are the language codes a proper superset of the site codes?
* if the code is the same in both cases, does it always refer to the same
language?
* does the Wikimedia project identified by a specific site code also always
use the same language code as its default interface language?
* are the answers to the previous three questions accidental or by design,
i.e. can we expect that this will stay like this in the future?

I hope that the answers are yes,yes,yes,yes :)

Sorry for the nitpicking questions, I hope someone knows the answers.

Cheers,
Denny

--
Project director Wikidata
Wikimedia Deutschland e.V. | Obentrautstr. 2 | 10963 Berlin
Tel. +49-30-219 158 26-0 | http://wikimedia.de



--
Project director Wikidata
Wikimedia Deutschland e.V. | Obentrautstr. 2 | 10963 Berlin
Tel. +49-30-219 158 26-0 | http://wikimedia.de

Wikimedia Deutschland - Gesellschaft zur Förderung Freien Wissens e.V. Eingetragen im Vereinsregister des Amtsgerichts Berlin-Charlottenburg unter der Nummer 23855 B. Als gemeinnützig anerkannt durch das Finanzamt für Körperschaften I Berlin, Steuernummer 27/681/51985.