Re: Language Tag Case Conflict (between RDF1.1 and BCP47)

Andy Seaborne <andy@apache.org> wrote:
>On 26/01/14 14:23, Richard Cyganiak wrote:
>> Thanks for the clarifications Ivan.
>>
>> So may I propose the following response to Vladimir? Maybe one of the
>> chairs may want to send it.
>>
>> [[ You make a good point. Unfortunately, RDF 1.1 Concepts and
>> Abstract Syntax is already in the Proposed Recommendation state of
>> the W3C process, and at this point it is no longer possible for the
>> editors to make changes (besides trivial typo/grammar issues).
>>
>> So this change may have to wait for a future update.
>>
>> If this is a showstopper for Ontotext, then technically speaking, as
>> a W3C member, you are entitled to object to the publication of the
>> standard, ie, to record a formal objection. The Ontotext AC can do
>> this by providing a set of changes that they request in order to get
>> the document published, or not to publish the Recommendations at all
>> for some reason. Just as on previous steps, the W3C Director has to
>> decide whether the objection is accepted or overwritten. ]]
>>
>> Best, Richard
>
>-1
>
>A/ technical
>
>In RDF 1.1, normalization is not required.  It was, sort of, in
>RDF-2004 
>(the note that an RDF implementation does not have to normalize then 
>talks about comparison).  I don't see that RDF is creating language
>tags 
>(BCP47 sec 4.5), only handling them.
>
>The comment however goes on to suggest requiring unique triples that is
>
>not required for other semantic equivalences in RDF.  The comment mixes
>
>syntax/storage and semantics.
>
>:x :prop 123 .
>:x :prop 0123 .
>
>B/ Process
>
>Vladimir has not said it is a problem for Ontotext - I read this
>comment 
>as a suggestion.  Ontotext did not raise a comment at last call. 
>Highlighting an objection at PR seems heavy.
>
>
>I suggest a response that explains it's rather late, and does not make
>a 
>judgement about the comment.
>
>[[
>RDF 1.1 Concepts and Abstract Syntax is already in the Proposed 
>Recommendation state of the W3C process, and at this point it is no 
>longer possible for the editors to make changes (besides trivial 
>typo/grammar issues).  Any design change will require going back to at 
>least the Last Call stage.
>
>The Working Group has considered the normalization of language tags 
>throughout its working phase and the design of semantic equivalence 
>(comparison equivalence) was the final outcome.  Normalization is one 
>way to achieve that, it is not the only way.
>
>Your comment will remain on the RDF comments list and any future
>working 
>group may decide to revisit this area.
>]]
>
> Andy
>
>(Yes - Jena stores lang tags as-is, based on feedback that some users 
>expect what gets parsed is also what's written out again.)

If be inclined to include that last parenthetical, expanded slightly, as additional explanation.  Something like:

For example, Jena stores lang tags without normalization, because users have requested this behavior, where they get back what they put in.   Jena remains compatible with the spec because when it does the tag comparison, it's case-insensitive.   If we were to change the spec to require normalization, systems like Jena would all have to be changed. The working group did not see sufficient problems with the current design to motivate such a code change.

    - Sandro

Received on Sunday, 26 January 2014 21:25:02 UTC