[whatwg] RDFa Features
ben at adida.net
Wed Aug 27 08:56:59 PDT 2008
Kristof Zelechovski wrote:
> This amounts to saying that URLs take precedence over CURIEs and CURIEs can
> be enclosed in brackets in case of any ambiguity. This sounds ridiculous
> given the weight you put on avoiding ambiguities and name clashes. Since
> the author does not control the URL scheme registration process, he can
> never be sure that a particular prefix is "safe", therefore using unsafe
> CURIEs is just asking for trouble. However, Manu's examples DO NOT use safe
> CURIEs, nor do any examples I have seen on this discussion. Good heavens!~
All examples written up by Manu are completely unambiguous, because we
only parse @rel, @rev, and @property as CURIES, never as URIs (and if
your value doesn't encode a valid, prefixed CURIE, we ignore it.)
(Reading the CURIE spec on its own is not really the right target,
especially since that is still being worked on, whereas the RDFa spec
stands on its own and clearly explains how CURIEs are used *in RDFa*.)
The only time we need disambiguation is for @about and @resource, which,
since they specify objects, we wanted to have accept URIs as a default,
just like @href. The disambiguation step of Safe CURIEs only applies for
those, and only when you want a CURIE in there, which is rare.
So, the processing rules are actually trivial and quite robust.
And all of Manu's examples function without any disambiguation.
More information about the whatwg