language codes

[Frederik Fouvry <fouvry CoLi Uni-SB DE>]

Hi,

A question about your usage of language codes.  In KDE, we are
about to decide on the proper names of language codes and
locales.  There are two points of view currently: stick to RFC
3066 or stick to ISO 639-2.  The first one more or less means
that the existing situation is continued, at least as far as
language codes are concerned.  (I'm rather ignorant where locales
are concerned.)  The second one requires conversion of all
language codes/tags to the 3-character code of ISO 639-2.

I noticed that your indications on the web pages are as ambiguous
as ours: you refer however to ISO 639-2, which technically lists
3-character language codes, but you use 2-character codes.  From
the list of codes, e.g. "wa" is not an ISO code, "x-wa" is; you
use "no", which is a correct code, but ambiguous (as far as I can
see - for us no and nb mean the same); you happen to use zh_CN,
which is probably based on the locale name (RFC 3066 prescribes
"zh-CN").  (Sorry for the nitpicking ;-)

What are your plans with this?  One of the reason that I'm asking
is that if we have divergent language codes, we need to
explicitly consider the way in which the other(s) use(s) the
codes, and implement the mappings.  Not really nice, and prone to
being forgotten.

Personally, I'd like to use RFC 3066, but not everyone is
agreeing with this, because it is some mix of the two ISO 639
lists (i.e. use ISO 639-1 where available, otherwise ISO 639-2).
The disadvantage with pure ISO 639-2 is that most software uses
the 2-character codes, HTML requires RFC 1766 (successor is RFC
3066), so everything needs to be mapped to the usage described in
the RFC anyway.

For our documentation, changing over would not be too
complicated: the language codes need to be changed in the DTD
(they are stored in parameter entities, and as such they are
invisible for our translators, as are any changes to the codes),
and a mapping function in the style sheets would need to be
written (to find the style sheet for the appropriate language).
For the locales, this is probably more complicated, but I don't
really have a clue about these.

Reactions?

PS: I'm not on the list, so please send them directly to me, or
to the KDE DocBook list.  And if you know of any other projects
that will have to deal with the same problem, feel free to
forward the message.

--
Frederik Fouvry		-	fouvry@coli.uni-sb.de
KDE DocBook Team	-	kde-docbook@master.kde.org