Re: [Ltru] my technical position on extlang

John Cowan 2008-05-18 03.57:
> Martin Duerst scripsit:
>
> > In my understanding, one of the main points against extlangs was
> > that the "remove from the right" fallback hierarchy may not be
> > right. 
>
> The problem is, AFAIU, that too much information is thrown away:
> zh-yue-Hant does not fall back to zh-Hant, but first to zh-yue and
> then to zh.
>   

If we say the tag must EITHER be formed as 'zh-yue-Hant' OR as 
'yue-(region)-Hant', then we can have different fallback rules for 
either. For 'zh-yue-Hant', then "take away the second subtag" could be 
the rule.

I.e. the use of 'macroLang-encompLang' could defined as the preferred 
construction when a fallback to the macrolanguage is recommended (by the 
author and/or the spec) as a good choice and/or when the 
"macroLang-encompLang" construct is considered more important/relevant 
than "encompLang-Region".

Thus this way we would permit the tagger to state whether fallback to 
the macrolanguage tag SHOULD happen or not.

In case the tagger uses 'encompLang-(whaterver)' then the same rules as 
currently proposed counts: Then implementations MAY use the 
macrolanguage information for fallback decisions.

> > I'm opposed to allowing both extlang and independent tags
> > (i.e. both zh-yue and yue), but if that's what we as a WG
> > decide, I can live with it.
>
> I don't think anyone has ever suggested that: it's either/or.
>   

See Phillips, Addison 2008-05-13 01.11, under the subject "Re: [Ltru] 
extlang or not extlang", and Karen's reply. Above, and in my previous 
post, I have hopefully given the "better spelled out usage rules" that 
Karen asked for.
-- 
leif halvard silli
_______________________________________________
Ltru mailing list
Ltru@ietf.org
https://www.ietf.org/mailman/listinfo/ltru