Re: language tags

Hi Andy,


On May 10, 2013, at 08:46, Andy Seaborne <andy.seaborne@epimorphics.com> wrote:

> We have a number of approaches to language tags.
> 
> * Option 1
> Language tags are lowercase strings in the abstract data model.
> 
> Pro:
>  It is the 2004 spec.
> Con:
>  It's not what most system do.
> 
> * Option 2:
> Language tags are compared case insensitively.
> 
> Con:
>  Unclear as to when "compare" happens
> 
> :s :p "xyz"@en .
> :s :p "xyz"@EN .
> 
> 2A: When the set of triples that is the graph is created
> => one triple, but what is written out is then not clear
> 2B: On access:
> => two triples


This choice is purely system dependent.  Who cares?

> 
> * Option 3:
> Language tags are values - they can be represented in different ways (c.f. datatypes).
> 
>  Do the work in the definition of rdf:langString.
>  Don't mandate rdf:langString understanding.
> 
> Pro:
>  It seems to be what systems actually do.
>  Some collapse case; most do not.


So I think I'm fine with any of the three, with a slight preference for options 2 or 3.

Regards,
Dave
--
http://about.me/david_wood


> 
> I'll send a message with the document consequences for option 3 separately.
> 
> 	Andy
>