Re: rdf-ns-prefix-confusion

>>>Uche Ogbuji said:
> I think I've pretty much responded to everything in this message,
> so it looks as if we're narrowing in on "agree to disagree".

Looks like that.


Uche said:
> Dave said:
> > and you propose (1)
> >   <rdf:Description about="http://example.org/somewhere/" a:b="x" c:d="y" />
> > or even with xmlns= the RDF namespace
> >   <Description about="http://example.org/somewhere/" a:b="x" c:d="y" />
> > 
> > which looks odd; why would 'about' be bare, a:b and c:d require prefixes.
> 
> I don't see a problem here at all.  Clearly the person viewing the document 
> understands default namespaces, otherwise they have far bigget conceptual 
> hurdles in front of them.

The problems and confusion with default namespaces are well known and
complex.  I would doubt most XML developers, never mind users, really
do understand them properly.
  
  [[What is clear is that no mapping is defined for unprefixed
    attribute name -- this is the meaning of the statement,
    "...default namespaces do not apply directly to attributes..." --
    and therefore that there is no way to construct universal names
    from unprefixed attribute names.]]

  [[Note that elsewhere the Namespaces FAQ states that unprefixed
    attribute names are not in any XML namespace.]]

   --  Myth #4: Unprefixed attribute names are in XML namespaces
       in Namespaces Myths, Ronald Bourret, XML.com
          http://www.xml.com/pub/a/2000/03/08/namespaces/


But I'm sure you know all this.

I've seen the above myth implemented in some RDF systems.

So for formats that were designed from the start to generate
universal names (URIs), requiring XML Namespaces support, expecting
to be used inside/containing other XML formats, i.e. RDF, it is best
avoid this area.

This namespace confusion stuff has come up on this list several
times.  The solution of always using prefixed attributes is clear,
simple and quite precise.


Uche said:
> Dave said:
> > <rdf:Description about="http://example.org/somewhere/" a:b="x" c:d="y" />
>
> > That just seems horrible.
> > 
> > I think this is something to do with RDF/XML expecting to see lots
> > elements and attributes with many namespaces defined, so picking the
> > rdf ones out of them should be easy.  What about embedding?  Which
> > one gets the attributes?  The innermost XML standard?  Just a mess.
> 
> I don't follow you here.  Do you have an example?

I'm mostly talking about reducing confusion.

If RDF/XML was being used with another XML format and both used
non-namespace attributes extensively, it would be confusing at a
glance to see which one was meant, especially if the local names were
similar such as xsl:id, rdf:ID.

If they were the same local name, it might cause applications to
choose the wrong method of using the attribute if the two formats
(RDF and the other one) used these unprefixed attributes in different
ways - since noting the Myth above, this isn't defined in XMLNS.

Dave

Received on Tuesday, 22 January 2002 10:49:56 UTC