Summary: de-DE-1996 is better than de-1996-DE

[Mark Davis]

The real issue is the practical implications of the ordering of the
subfields. A great body of software has the following fallback
behavior: If a resource is not found with the id 'X-Y-Z', then it will
try the id 'X-Y'. If it is still not found, it will try the id 'X'.
Similarly, if original id were 'X-Z-Y', then the fallback will be to
'X-Z', then to 'X'.

So, for the collection of designations {X, Y, Z}, if X-Y is a better
general fallback than X-Z, for most kinds of locale-based data lookup,
then X-Y-Z should be chosen. Otherwise, X-Z-Y should be chosen.

Mark
—————

Γνῶθι σαυτόν — Θαλῆς
[For transliteration, see http://oss.software.ibm.com/cgi-bin/icu/tr]

http://www.macchiato.com

----- Original Message -----
From: <Peter_Constable@sil.org>
To: <ietf-languages@iana.org>
Sent: Thursday, May 02, 2002 21:26
Subject: Re: Summary: de-DE-1996 is better than de-1996-DE


>
> On 04/30/2002 08:05:35 AM Michael Everson wrote:
>
> >Please give a clear and unambiguous "I prefer X" and indicate
whether
> >you can live with the other or not.
>
> Sorry, I've not been able to follow much email for several days.
>
> In view of other's comments, I will say that I have no preference
and can
> live with either.
>
> I will also come away from this discussion being persuaded that any
attempt
> to assign significance to position classes in IETF language tags
(apart
> from the source for 2- or 3-letter primary subtags and the source
for
> 2-letter secondary subtags) to be a lost cause. For me, this implies
all
> the more the importance of documenting precisely what the denotation
of any
> tag is, including consideration of the issues I discuss in terms of
types
> of language-related categories; and it also convinces me that RFC
3066 has
> a significant weakness in that, while it sanctions iso639-iso3166
> combinations without reguiring registration, it does not specify
anything
> in regard to such combinations in the absense of specific
registrations.
> Thus, if you encounter data tagged as (say) zh-CA, you'll know that
it has
> something to do with Chinese, and something to do with Canada, but
not a
> wit more, and therefore in relation to a number of processes, you
won't
> really have a clue what can be done with it; i.e., the tag doesn't
do a
> whole lot of good.
>
>
>
> >Doing my best,
>
> Appreciated.
>
>
> - Peter
>
>
> --------------------------------------------------------------------
-------
> Peter Constable
>
> Non-Roman Script Initiative, SIL International
> 7500 W. Camp Wisdom Rd., Dallas, TX 75236, USA
> Tel: +1 972 708 7485
> E-mail: <peter_constable@sil.org>
>
>
> _______________________________________________
> Ietf-languages mailing list
> Ietf-languages@alvestrand.no
> http://www.alvestrand.no/mailman/listinfo/ietf-languages
>