Re: RDFa and Microformats

Martin,

Wow, this thread has certainly flourished.

Okay, first things first: let's keep the personal attacks and innuendos
out, please. We've managed to keep a very cordial mood here on this
list, and I'd like to keep it that way. We welcome all comments and
disagreements, even strong disagreements, but let's not start implying
that such-and-such content was "ripped" from somewhere else (when the
author trail is obvious), or making comparisons to the Borg (which
strikes me as very close to Godwin's Law.)

Regarding individual participation: I agree that it would be nice for
individuals to participate more easily. I will communicate this feedback
to the W3C team.

I want to highlight a note that Mark sent:

> If we have this:
> 
>   <span
>     class="p1" content="v1"
>     property="p2">v2</span>
> 
> How does the RDFa parser know that @content was added for use in a
> Microformat, rather than for use in RDFa?
> 
> You can argue that you wouldn't do this, and that you'd advise people
> to add extra elements to make things clear, but how do you absolutely
> ensure against it?
> 
> This kind of pollution is exactly the problem that Microformats has
> been unable to solve, and exactly the kind of problem that we put a
> lot of energy into solving in RDFa.

I could not have said it better.

RDFa has introduced certain attributes specifically for the embedding of
RDF in XHTML. Those attributes trigger the generation of RDF triples, as
per the RDFa rules. They should not be used for other purposes. A
publisher who uses RDFa attributes should expect the corresponding
triples, as specified by RDFa, to be implied. Always.

The reason why that is is simple: we want well-formed RDFa-in-XHTML
snippets to mean the same thing, no matter where they appear. We want
RDFa parsers to be consistent and always vocabulary-independent. These
were core design goals when we started, and they haven't changed.

The word "overloading" is incorrect, and the word "squatting" is too
nice, in my opinion. I would consider these alternative proposal more of
an *abuse* of RDFa attributes. By using them in ways that are
inconsistent with RDFa, you would be diluting our efforts at a single
metadata syntax for HTML, effectively weakening the meaning of other
folks who want to use RDFa consistently.

That's bad practice, and we would certainly speak out strongly against
it, much like many accessibility folks spoke out strongly against the
abuse of @title.

So, to reinforce what Mark said extremely well: we will continue to find
ways to simplify the expression of RDFa, but always with a consistent
parsing model and a consistent meaning for all attributes.

Feedback on this is welcome, of course, but seeing as how this has been
a consistent principle of RDFa for the last 3-4 years, through Last
Call, CR, and now PR, I doubt that we would give serious consideration
to proposals that so substantially weaken RDFa.

-Ben

Received on Tuesday, 16 September 2008 01:44:03 UTC