Re: example of options 3 & 4 simplifying code (ACTION-86)

I think the case insensitivity of language tags is a minor syntactic 
issue, whereas the relationship between the different XSDs is 
significantly more complicated (and requires significantly more code). 
Sandro is being disingenuous to pretend they are the same.

If I understand Sandro's desire correctly, then my suggestion is that 
this is an RDF application level issue, not a spec issue.
RDF is and should be speced as a WWW standard for knowledge, which 
necessarily involves I18N considerations as mandatory.

A simple RDF application (e.g. an editor that allows the user to publish 
their knowledge) could hide this mandatory I18N support in the following 
fashion:
- make an OS level call to find the users default I18N context
- allow the user to create the info to publish ignoring I18N info
- publish the data by combining the user's default I18N context with the 
info they have provided

I believe such an app is doable today, and meets Sandro's requirements 
without sacrificing the WW bit of WWW.
I also have a lot of time for curtailing this discussion of literals, 
since whatever we come with, will be, in most cases, mediated to the 
user by some other API or application layer, and essentially the same as 
what we have already. Thus, as the charter indicates, our time would be 
better spent in discussing new features that give new functionality such 
as named graphs

Jeremy

On 9/10/2011 3:32 PM, Sandro Hawke wrote:
> On Sat, 2011-09-10 at 16:40 +0200, Richard Cyganiak wrote:
>> Sandro,
>>
>> The built-in i18n support in RDF is a selling point that sets it apart from just about anything else on the data storage market. I'm not in favour of removing that advantage.
> I'm suggesting removing it from RDF and putting it in the same place as
> the understanding that "1"^^xs:int and "1"xs:integer are the same thing.
> I think both kinds of processing should be part of the standard datatype
> library that pretty much everyone uses, since RDF is pretty awkward to
> use without them.
>
>> Also:
>>
>> On 9 Sep 2011, at 17:19, Sandro Hawke wrote:
>>> I don't think language tag reasoning should be mandatory
>> I think you meant to say “language tag case folding”. It has nothing to do with “reasoning”, in any sense of that word.
> I called it that because recognizing that "1"^^xs:int and
> "1"^^xs:integer are the same thing is call "reasoning", and I think
> there is a strong analogy.
>
> Also, I'm not sure case folding is that useful; in any case where you
> want case folding for language tags, I suspect you should really be
> using language tag matching, which comes somewhat closer to deserving
> the term "reasoning".     But I'll stick to calling it "processing" in
> the future.
>
>    -- Sandro
>
>> Best,
>> Richard
>
>

Received on Monday, 12 September 2011 17:47:38 UTC