Re: Test 0123 now incompatible with RDFa 1.1

We recognized this issue VERY EARLY in the RDFa deployment cycle, and we 
put a warning in the errata that the direction was going to be 
case-insensitivity  - mostly because lots of user agents seem to lose 
the case of attribute declarations when in HTML mode.  I think we are 
okay breaking backward compatibility on this because, to be honest, it 
should never have been this way in the first place.

However, if the group feels otherwise I think Gregg's suggested model is 
excellent.

Gregg Kellogg wrote:
> Section 7.5 item 4 indicates that prefixes, whether from @prefix or 
> @xmlns, MUST be converted to lower case. However, XML namespace 
> prefixes are case-sensitive. Test 0123 checks to ensure that this case 
> sensitivity is retained in namespace mappings. I presume that this 
> test is obsoleted by RDFa 1.1, and it's okay to break these 
> dependencies (which were discouraged anyway).
>
> This creates a backwards-comability issue with RDFa 1.0, where 
> prefixes are case sensitive.
>
> I'd suggest the following:
>
>     * Prefixes established through either @prefix or @profile MUST be
>       mapped to lower case
>     * Prefixes established through @xmlns MUST remain case sensitive,
>       unless specifically indicated otherwise by the Host language.
>       Implementers SHOULD NOT use case-sensitive prefixes.
>     * CUIRE and QName processing MUST first attempt to find a
>       case-sensitive prefix mapping and fall-back to using lower-case
>       if a mixed-case mapping is not found, also as indicated by the
>       Host language.
>
>
> Or, we just decide to be incompatible on this issue and universally 
> require that all prefixes be case-insensitive.
>
> Gregg
>

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

Received on Monday, 3 May 2010 21:50:36 UTC