Appeal to ISO 639 RA in support of Elfdalian
lucp at skopos.be
Tue Apr 26 14:41:43 CEST 2016
On 25-04-16 18:51, Doug Ewell wrote:
>> Then if the RA is still no ready to assign a language code, then IETF
>> > should be ready to assign his own language subtag.
> Keep in mind that such a subtag would be 5 to 8 letters. You have said
> previously that this wouldn't work for you.
That is not my understanding.
Mats has even filed a registration form for a 5-letter non-ISO primary
language subtag "ovdal", that is pending since February 29.
And on 24-04-16, the day before you said the above, Mats wrote:
> I still don't understand why a 5-letter subtag would be a problem.
The draft appeal gives three reasons, but neither of them seem very
convincing to me. It says:
> (a) There is the possibility of conflict or redundancy if the RA later approves a code element.
> (b) Some processes are incompatible with 5- to 8-letter language subtags, which would not be beneficial to Elfdalian data.
> (c) While it is an option for BPC 47, it would decouple BCP 47 from ISO 639-3 in a way which neither we nor the ISO 639 RA may want.
As to (a), if/when such an approval ever happens, couldn't we simply
insert the new ISO code in the registry and deprecate "our" 5-letter
subtag in favour of it?
If yes, chances are that the RA will know that as well.
As to (b), I tend to agree that it might be inconvenient at first. But
in the end, either the broken process should be fixed, or users of
Elfdalian could look for alternatives.
They could be up and running long before we finish this byzantine debate
on how to tell the RA that he made a mistake without telling him that he
made a mistake.
Besides, we have Mats asking for a 5-letter subtag, so he seemingly
doesn't share our concern that it is "not beneficial to Elfdalian data".
If the RA is indeed monitoring this list, he will probably have noticed
that as well.
As to (c), that is the only real concern I can see, but unprecedented
situations do call for unprecedented solutions.
Besides, if (a) happens, we immediately deprecate, the decoupling would
be undone and (c) would evaporate.
So the RA may not see it as a problem if we drift apart for a "short"
What am I missing here?
More information about the Ietf-languages