Re: RDF Namespaces and Attributes (again)

Jason Diamond wrote:

> The truly unfortunate thing is that _we_ really can't do anything. The only
> entity capable of putting an end to this issue is the W3C.

But my point is that even the W3C can't really do anything but follow up
with recommended best practices and deprecations.  The spec is out of the
bag. Follow-ups help, but don't let us off the hook of dealing with the
large number of contradictory interpretations of the spec.  If we want
bullet-proof RDF processors, we're going to have to deal with multiple
versions of reality.  So, even if a practice is deprecated, we need to
seriously consider supporting it anyway, particularly if a generator of
"gray area" RDF becomes common.

> Your approach actually breaks several examples from the M&S (thus, your
> argument to always qualify rdf "keywords".
> 
> Techincally, though, your solution is sound and I would be happy to
> implement it if that was the approach that the W3C decided to take.

I wouldn't call anything that breaks on spec examples technically sound! :-)
However, remember that the thrust of my approach is leniency. Our processor
will recognize cases like <rdf:type resource="..."> and <rdf:li
resource="...">, even though the approach would prefer that rdf:resource
were qualified.

Where our processor does get confused is cases like this:

  <rdf:Description type="Foo" value="1" prop="2" xmlns="schema:" .../>

Is that schema:type or rdf:type?  Is that schema:value or rdf:value?  Our
processor thinks that they are schema:type and schema:value.

So, another thing that the FAQ might usefully deprecate is the use of the
default namespace when unqualified RDF keywords are in scope.  The Values
Containing Markup example in section 7.5 exemplifies this practice.

> Oops. You just found a bug in my parser. It's amazing how you can spend so
> much time scrutinizing the tiniest details and miss something that obvious.

Tell me about it!  For the longest time, I was treating rdf:li as equivalent
to rdf:Description, in terms of the presence or absence of abbreviated
properties (attributes).  That inconsistency still bugs me -- I can see no
syntactic reason why abbreviated properties aren't allowed for rdf:li.

Perry

Received on Monday, 2 October 2000 14:06:56 UTC