On 05/08/2002 04:19:08 AM Keld Jørn Simonsen wrote:


>I think that in some cases it is a sensible thing to add a 2-letter code
>for a language that already has a 3-letter code. This could be of quite
>some versability eg in the UNIX world. A recent addition was Walloon that
>got the code "wa".
>
>I really do not see any problems for the internet with such an addition.
>All other codes would still work.
>
>> This is to  support the internet application of language tagging,
>> where we have a legacy of two-letter codes for Danish and English,
>> but also now a set of three-letter codes which are currently being
>> used. So far, codes are all unique. Changing this rule would
>> introduce ambiguity for e.g. Hawai'ian.
>
>Non sequitur. The iso 639-1 codes are not legacy, but are very much in
>use. The 3-letter codes are only emerging, and not really used for
>mainstream production in many applications/usage areas.
>I do not see how this would change anything for Hawai'ian on the
>internet.


The point that Michael is making is that 3-letter codes such as *haw* are
legacy codes, and that if a 2-letter code for them were suddently created,
then Internet protocols based on RFC 3066 would require the 2-letter code
to be used, causing existing documents to become incorrectly tagged wrt.
the requirements of the RFC.

There are three ways to address the issue:

- We can ensure that those 2-letter codes do not ever get introduced.
Havard has pointed out the possibility that some day Internet technologies
may have changed to the point that that requirement is no longer a concern.
This is true, though I can also understand Michael's concern: when no such
changes in Internet technologies are anticipated, it would be a concern to
know that the "freeze" is not permafrost.

- IETF could revise their technologies so that synonyms are not a concern,
as Havard suggested would in general be a good thing to be doing. (I state
this as a possible solution, not as a recommendation.)

- The safest way for IETF to ensure that 2-letter codes do not become
available for use on the internet for cases in which 3-letter codes already
exist would be to revise RFC 3066 and enumerate the specific 3-letter codes
for which 2-letter codes do not exist and state that these are always to be
used, thus representing an exception to the norm that 2-letter codes are
always to be given precidence. If this were done, then Havard's reluctance
to predict the future would be of any cause for concern to IETF.



BTW, I certainly hope that the unix world will one day learn that 2-letter
codes are insufficient to cover user needs.




- 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: <[log in to unmask]>