Re: 3 letter language codes when ambiguous

"Eric Kow" <bc4dyjp02-O/[email protected]> Sun, 21 Dec 2003 19:36:25 +0100
Newsgroups gmane.science.linguistics.wikipedia.international
Message-ID <[email protected]>
On dimanche, déc 21, 2003, at 12:56 Europe/Paris, Brion Vibber 
brion-at-pobox.com |Wikipedia| wrote:

> Please see RFC 3066 - Tags for the Identification of Languages, 
> section 2.3 Choice of language tag:
> http://www.faqs.org/rfcs/rfc3066.html

Ok fine.  Sorry for making you rehash.

1) Following standards is great, but this is one of those times that we 
need to recognise that the standards are broken and that we should 
allow ourselves the immodesty of doing things our way.  Besides, we'll 
still be using ISO codes, just not the ones 3066 tells us to.

2) Language tags are meant for document meta-data, but nobody ever said 
anything about domain names!  I agree that a hypothetical XML export 
should be tagged "en" for English instead of "eng" (so that linguistic 
tools can process them, fine) but why on Earth would we have to apply 
the RFC to domain names?

3) Brion argues on meta (http://meta.wikipedia.org/wiki/Language_codes) 
that even 3 letter codes are not sufficient, so we'd be effectively 
replacing one broken thing with another.  Good point, but the issue at 
hand is not linguistic coverage, but the simple matter of *avoiding 
confusion*

In short, there is an RFC but we should not be slavish because 1) it's 
broken 2) it does not "jurisdiction" over domain names, and 3) it does 
not make sense from a purely pragmatic standpoint.

I quote Tobias Condradi from meta:
   "Wikipedia is young and there is no need to repeat the mistake of 
using ISO 639-1"

-- 
Eric Kow                     http://www.loria.fr/~kow
PGP Key ID: 08AC04F9         Merci de corriger mon français.