Re: PROPOSAL: Errata text regarding defining a prefix of '_'

Well, it depends on the implementation environment. I rely on rdflib,
which has its own bnode management. So I generate bnodes into the graph
using its mechanism. The only thing I have to do is to have a 'mapping'
between the RDFa author's bnode id-s and the internal instances. The
result is, of course, that the generated RDF serialization will _not_
use the same bnode id-s than the RDFa source, but whatever the rdflib
environment produces as unique names. But that is all right and no RDFa
authors should expect to see the same bnode URI-s in the output.

If you do the whole bnode management than, of course, the situation
becomes your responsibility. But it is all right if, internally, you do
a mapping from the user's name to anything inside and output that one.
The author's bnode id-s do not matter, in this sense.

Ivan

Shane McCarron wrote:
> Oh, and a slightly related question for implementors.  In RDFa it is
> possible to create a *specific* bnode at any time.  They sort of
> "auto-vivify" when you reference them (e.g., about="_:myNewBnode").  At
> least, I think that is how it works.
> 
> There are also *implicit* bnodes that get created all the time when
> certain attributes are present.  The names of these are not important. 
> What is important is that they become the implicit subject or object of
> triples in their child nodes. These bnodes are by definition CANNOT be
> explicitly referenced, since you have no idea what the implementation
> will name them.
> 
> My question is this:  How do you ensure that your implicit,
> automatically created bnode names never collide with a documents
> explicit, auto-vivified bnode names?  And should we have some sort of a
> test case to ensure this happens?
> 
> Shane McCarron wrote:
>> Just curious - how does it issue a warning?  As part of the graph?
>>
>> Ivan Herman wrote:
>>> Looks fine to me.
>>>
>>> (I was unsure on the call and I checked: my implementation already does
>>> that and issues a warning in case of reusing '_')
>>>
>>> Thanks
>>>
>>> ivan
>>>
>>> Shane McCarron wrote:
>>>  
>>>> At the 12 November meeting, the group agreed that the RDFa Syntax
>>>> Specification reserves the prefix of '_', but that it could be clearer
>>>> about this.  We agreed to put a clarification in the Errata.  I propose
>>>> the following text:
>>>>
>>>> In sections 5.4.5 and 7 the specification indicates that the prefix '_'
>>>> is reserved and is used to create / reference blank nodes (bnodes).
>>>> Because this prefix is reserved, authors SHOULD NOT declare a mapping
>>>> for the prefix '_' and conforming processors MUST NOT incorporate
>>>> such a
>>>> prefix mapping into the 'list of URI mappings' as defined in section
>>>> 5.5.
>>>>
>>>>
>>>>     
>>>
>>>   
>>
> 

-- 

Ivan Herman, W3C Semantic Web Activity Lead
Home: http://www.w3.org/People/Ivan/
mobile: +31-641044153
PGP Key: http://www.ivan-herman.net/pgpkey.html
FOAF: http://www.ivan-herman.net/foaf.rdf