Re: an issue with prefix-less curies (also comment on the syntax document)

We can start a long philosophical discussion, and use lots of
bandwidths, and have heated debates on whether we would use

http://www.w3.org/1999/xhtml/relations#

or

http://www.w3.org/1999/xhtml/relations/

...:-) I would suggest to flip a coin...

Mark (and others), I would propose to try to get that done in the syntax
document as soon as possible, maybe before Amsterdam. You may have seen
that there is a surge of interest in RDFa (yay!) these days, and people
may begin to rely on the RDF generated by the various implementations.
Some of our other pending issues (instanceof, etc) are more edge cases
which would not invalidate graphs generated today (probably), but a
different namespace might hurt. Is it possible to settle this by email
or on the telco?

Ivan

Mark Birbeck wrote:
> Niklas,
> 
>> Mostly thinking out loud here; I hope I'm not confusing things.
>>
>> I wonder -- since this is in fact the introduction of a new set of
>> URIs for the unprefixed names in @rel -- would there be any merit in
>> distinguish it a bit more from the (namespace) URI of the xhtml
>> doctype? Like:
>>
>>     <http://www.w3.org/1999/xhtml/relations#>
> 
> I agree 110%. :)
> 
> We did a similar thing in XHTML M12N with the datatypes [1]. They used
> to be in the normal XHTML namespace, but we gave them the following
> mapping instead:
> 
>   <http://www.w3.org/1999/xhtml/datatypes/>
> 
> We did this because essentially, the definition of a datatype like
> 'Length'--defined as a non-negative integer--will almost certainly be
> unchanged in any future version of XHTML. So in M12N it is defined
> with the following URI:
> 
>   <http://www.w3.org/1999/xhtml/datatypes/Length>
> 
> Of course future versions of XHTML may not use different XHTML
> namespaces, in which case no problem would have arisen, but by
> isolating the datatypes like this we protect ourselves and any
> processors that make use of this information. We also allow anyone
> defining a new language based on XHTML M12N to re-use these datatypes.
> 
> 
> I think what you are saying, Niklas, is that we should consider doing
> exactly the same thing for the relationships between documents, and I
> would agree. There is a small tricky problem though, which is that
> unlike the definition of non-negative integers, the list of
> relationships may grow.
> 
> But this is actually only a problem for us over in XHTML-land, and not
> something we need to worry about here. In XHTML-land we will need to
> consider whether we should have different datatypes for xh11d:LinkType
> (which is what currently defines the content of @rel and @rev). At the
> moment the definition is just a list of tokens, but we might decide to
> define this more tightly in future versions, and could therefore end
> up with xh2d:LinkType, xh3d:LinkType, and so on.
> 
> But in RDFa-land all we need to know is what the appearance of one of
> these link types _generates_, and for that we only need to agree on a
> URI mapping. (Your suggestion seems as good as any. :))
> 
> This is easy to change in the syntax document, so it's not something
> we have to resolve now--it might even be something we want to raise at
> the Amsterdam face-to-face to see if others have any views.
> 
> Regards,
> 
> Mark
> 
> [1] http://www.w3.org/TR/xhtml-modularization/schema_module_defs.html#a_module_XHTML_Datatypes>
> 

-- 

Ivan Herman, W3C Semantic Web Activity Lead
Home: http://www.w3.org/People/Ivan/
PGP Key: http://www.ivan-herman.net/pgpkey.html
FOAF: http://www.ivan-herman.net/foaf.rdf