Re: RDFa WG telecon minutes for 2010-05-13

On 5/15/2010 5:19 AM, Ivan Herman wrote:
> On May 15, 2010, at 24:32 , Ben Adida wrote:
>
>    
>> On 5/13/10 9:07 AM, Manu Sporny wrote:
>>      
>>
>> "When both lang and xml:lang are specified on the same element, they MUST have the same value"
>>
>> I don't think we can *will* the markup to be the way we want it to be. If both @lang and @xml:lang have the same value, then that line in the spec is superfluous. If they have different values, then the spec is unhelpful in resolving the conflict. I would propose that, instead of this resolution, we specify what happens when both are defined to be different values. Does the parser blow up? Do we give priority to one of them? Do we ignore both?
>>
>>
>>      
> The argument at the meeting (I think it came from Shane) was that, in general, we do not specify what an RDFa processor does in case of an invalid markup. In specific languages, like in XHTML5, this is, in fact, and invalid markup which validators should handle. I am just conveying the arguments. Reading through this again, I would be perfectly happy to say that @lang has a priority for the (X)HTML(5) cases.
>
> Note that there is now a separate issue on whether an RDFa processor should give error messages or not. If we decide that there are (limited) cases when an RDFa processor should give those, this may be a perfect case example for it...
>    

I think you are both considering this text in isolation.  The complete 
clause is (to me) crystal clear as to what takes precedence.  The MUST 
requirement on documents is a nod toward how HTML5 does processing - 
where @xml:lang is effectively ignored.  So, in an HTML5 processor, it 
is *possible* that a document could have both xml:lang and lang 
specified and, were they to have different values, the HTML5 processor 
would treat the language as the @lang value, while the RDFa processor 
would treat it as the @xml:lang value.  Frankly, this doesn't bother me 
at all.  All we care about is what an RDFa processor looks at, and we 
want consistent triples to be emitted regardless of the host language 
when faced with @xml:lang.

This is what the spec says:

    This specification also adds the |lang| attribute to the I18N
    attribute collection as defined in [XHTML-MODULARIZATION11-2e
    <http://localhost:8080/rdfa/sources/xhtml-rdfa/Overview-src.html#bib-XHTML-MODULARIZATION11-2e>].
    The |lang| attribute is defined in [HTML401
    <http://localhost:8080/rdfa/sources/xhtml-rdfa/Overview-src.html#bib-HTML401>].
    When this attribute and the |xml:lang| attribute are specified on
    the same element, the |xml:lang| attribute takes precedence. When
    both |lang| and |xml:lang| are specified on the same element, they
    /must/ have the same value.



>    
>>
>> "For prefixes defined via xmlns: and @prefix, the prefix text should be converted to lowercase by the RDFa Processor."
>>
>> So what happens in the following case:
>>
>> <div xmlns:foo="http://a.com" xmlns:Foo="http://b.com">
>> ...
>> </div>
>>
>>
>> Clearly one of these @xmlns will override the other, but which one? Is it alphabetical? I don't think we've ever defined an order for handling these @xmlns attributes, have we?
>>      
> And I do not think we can. By the time any processor gets to the attributes themselves in the DOM tree, the order, afaik, is unpredictable. The only thing we can do is to add that to the list of possible errors...
>    

I agree.  Actually, in the html4 and html5 case I imagine that we would 
only ever see one of those... and I don't know which one it would be.


>    
>>
>> It probably gets more complicated with @prefix:
>>
>> <div prefix="foo: http://a.com Foo: http://b.com">
>> ...
>> </div>
>>      
> That is actually the case when we can define an order...
>
> We did discuss at some point what is the order within a @profile for the various files, ie, left-to-right or right-to-left; this is one of our open issues:
>
> http://www.w3.org/2010/02/rdfa/track/issues/23
>
> I guess for the sake of consistency we should probably follow the same order whatever we decide for ISSUE-23...
>    

Yes.  I think the conclusion is that it needs to be the same as for 
microformats and @profile...

-- 
Shane P. McCarron                          Phone: +1 763 786-8160 x120
Managing Director                            Fax: +1 763 786-8180
ApTest Minnesota                            Inet: shane@aptest.com