BCP 47 reference (was Re: Last Call: draft-nottingham-http-link-header from digest)

Hello Mark,

> Regarding hreflang - looking through the history, it's been discussed  
> in a fairly positive light a few times, but never made it in. I think  
> it does make some sense, since it's both in Atom and HTML. 

I think hreflang would be useful to add. You might want to consider calling it just 'lang', since that locution is more familiar.

I think you might want to make it more than a single language tag. The purpose of http-link-header is to provide metadata about a link in addition to the URI. This is more like providing the author's intended target audience rather than the document processing language. That is, it's like the Content-Language header and/or <meta> element, rather than like the <html> lang attribute. Cf. http://www.w3.org/International/questions/qa-http-and-lang#answer 

> I'm a bit  
> concerned about what the appropriate reference is for the value space;  
> ATM I'm thinking BCP47 directly, rather than to a specific RFC, to  
> allow it to evolve*.

Although there is a new RFC-to-be (4646bis) now in the RFC-editor's pipeline, I rather think that future changes to BCP 47 will tend to be limited to the production of extensions defined by 4646/4646bis, rather than changes to the grammar of language tags. There are some people who think that a revision might happen to use some reserved subtags for ISO 639-6 if/when that standard reaches maturity, so the possibility of revision does remain.

> ...
> * Often, a reference to an RFC is preferable, so that software can be  
> reliably written to a specific set of identifiers. My initial feeling  
> is that here that's not appropriate to do that, because language tags  
> are labels, not something that you're going to hardcode into  
> infrastructure software. Feedback appreciated, especially from the  
> i18n community.

Language tags are, as you note, labels that should not be hardcoded into infrastructure. However, BCP 47 defines the grammar for language tags themselves. Implementers need a reference to determine if the list of language tags they generate or receive is well-formed or not. That might be a good reason for a specific reference. Otherwise, I tend to recommend that people reference BCP 47 (it avoids that old chestnut "RFC WXYZ or its successor").

Best Regards,

Addison

Addison Phillips
Globalization Architect -- Lab126
Chair -- W3C Internationalization WG

Internationalization is not a feature.
It is an architecture.

Received on Friday, 24 July 2009 15:35:30 UTC