Re: [Ltru] Re: Remove extlang from ABNF?

[Addison Phillips <addison at yahoo-inc.com>]

Felix Sasaki wrote:
> > In RFC 4646, we defined some tags as well-formed. In RFC 4646bis, we 
> > suddenly
> > say that some of these tags are not well-formed. We never would do 
> > this for
> > valid tags, so I'm really not sure it is a good thing to do for 
> > well-formed tags.
> >   
> +1

I agree.... except....

This is exactly what we did in RFC 4646. We made a vast array of 
"well-formed" but invalid tags illegal (by narrowing the ABNF). In 
4646bis, one could say that we were doing the same thing--making a (much 
smaller) array of tags (which were never valid) illegal. And we *have* 
changed the ABNF in a manner that narrows it in 4646bis already. We have 
narrowed the grandfathered production substantially.

I have no problem with XML Schema or others referencing the 4646 ANBF 
instead of the 4646bis ABNF for well-formness checking. It won't 
introduce anything particularly bad. And I am somewhat allergic to 
changing the ABNF because I have personally felt like we should resist 
tampering.

As an implementer, though, I really hate supporting extlang, now that it 
does nothing. So I'd propose:

1. Remove extlang from the ABNF in 4646bis. That would make the language 
production:

language      = (2*3ALPHA)             ; shortest ISO 639 code
              / 4ALPHA                 ; reserved for future use
              / 5*8ALPHA               ; registered language subtag


2. In the section on conformance, permit 4646bis well-formedness to 
reference either the current ABNF or an "obs-language" production that 
looks like:

obs-language  = (2*3ALPHA [ extlang ]) ; shortest ISO 639 code
              / 4ALPHA                 ; reserved for future use
              / 5*8ALPHA               ; registered language subtag

extlang       = *3("-" 3ALPHA)         ; removed in this version

3. Add a note saying that no tags were ever valid under obs-language, 
but that some processors may permit them. Also note that 3066 
well-formedness differed substantially from 4646/4646bis well-formedness 
and provide the 3066 production in that section for completeness.

Addison

--
Addison Phillips
Globalization Architect -- Yahoo! Inc.
Chair -- W3C Internationalization Core WG

Internationalization is an architecture.
It is not a feature.


_______________________________________________
Ltru mailing list
Ltru at ietf.org
https://www1.ietf.org/mailman/listinfo/ltru