Re: [closed] Re: N-Triples VS RDF/XML bNode identifiers

Dave Beckett wrote:

> On Tue, 28 Oct 2003 10:15:42 +0100, Arjohn Kampman <arjohn.kampman@aduna.biz> wrote:
> 
> 
>>Dear editors of the RDF Test Cases document,
>>
>>Last week, we stumbled across a problem in Sesame when RDF was read from
>>an RDF/XML document and then written as N-Triples. The problem was
>>related to the bNode identifiers, whose definition in RDF/XML and
>>N-Triples is slightly different:
>>
>>While parsing the RDF/XML, the parser generated bNode IDs that were
>>legal according to the RDF/XML specs and these were written as-is to the
>>N-Triples document. An example bNode identifier is "node09FC-1E4A-2". In
>>RDF/XML, the dashes (and underscores, etc.) are legal characters for
>>bNode identifiers. In N-Triples, however, only (ASCII-)letters and
>>number can be used. Thus the procedure sketched out above resulted in an
>>illegal N-Triples document.
>>
>>So, my question is: wouldn't it be convenient to make the two
>>definitions identical?
> 
> 
> It might be convienient but sadly, I don't think it's realistic.
> 
> An rdf:nodeID in RDF/XML defined at
>   http://www.w3.org/TR/rdf-syntax-grammar/#nodeIdAttr
> takes as a value a string compatible with an XML name:
>   http://www.w3.org/TR/rdf-syntax-grammar/#rdf-id
> which has a wide range of characters allowed and
> permits the full range of international characters.
> 
> It is also useful to re-use this XML definition since it enables checking
> using the standard XML NCName term (which can be then checked
> by XML schema languages which likely support that)
> 
> The subset that are also legal as N-Triples is thus rather narrow.
> 
> N-Triples also has had a (weak, but remains, "would be-nice")
> requirement to keep itself as a subset of N3.  N3 imposes several
> restrictions on the names that are allowed, and they vary between
> implementations.  The current set of allowed names define dby
>   http://www.w3.org/TR/rdf-testcases/#name
> is thus not likely to change.  
> 
> I think '-' in particular might have other uses in N3 for paths,
> arithmetic, a funky -_ encoding scheme or may be reserved or unwise.
> You'd have to ask the N3 developers, since I can't recall all the details.

It's acceptible. I have already modified our code to generate bNode IDs
that are compatible with the N-Triples so it's not a really big issue
for us.

Thanks for your comments,

Arjohn

-- 
arjohn.kampman@aduna.biz
Aduna B.V. (formerly known as aidministrator) - http://www.aduna.biz/
prinses julianaplein 14-b, 3817 cs amersfoort, the netherlands
tel. +31-(0)33-4659987  fax. +31-(0)33-4659987

Received on Thursday, 30 October 2003 06:06:05 UTC