Re: Request for feedback about RDF-star

On 3/25/21 12:29 PM, Kjetil Kjernsmo wrote:
>>> "Any other names are not defined and SHOULD generate a warning when
>>> encountered, but should otherwise behave normally.", which seems to
>>> indicate that the expectation is that the names is a fixed set.
[ . . . ]
> To me, it is merely to say that systems should not break on unexpected
> vocabularies. When a system is designed to emit a warning under such
> conditions, it is a pretty strong statement to me (I tend to have "no
> warnings" as a test case in my test suite :-) ).

I respectfully disagree.  I think the purpose of such a warning is to 
notify the user of a possible typo -- not to discourage the intentional 
extension of the rdf: namespace in coordination with the RDF community. 
  After all, the whole philosophy of RDF is to be extensible, with the 
proviso that one should not squat on someone else's namespace.
https://www.w3.org/wiki/NamespaceSquatting
But this clearly is not squatting, since it is being done with the 
blessing of W3C.

BTW, there's a key piece of information missing from this discussion:
Exactly what names are being proposed, for addition to the rdf: 
namespace?  That is relevant because if they are names that are very 
unlikely to be wanted for some other purpose, then there is less risk in 
adding them to the rdf: namespace.  But if they are likely to be desired 
for some other (conflicting) purpose -- such as for a more general 
standard solution to the need for n-ary predicates, annotating multiple 
statements at once, etc. -- then we should be much more careful.  I, for 
one, am very much hoping that a more general mechanism *will* be 
developed for RDF, which would render RDF-star obsolete.

David Booth

> 
> I would certainly have appreciated it if the WG had foreseen the situation
> that people might add terms to the vocab, but I don't think they can be
> blamed. Given what we have, it seems that proceeding with caution is a good
> idea, and even though there might be general consensus that the suggested
> name belongs in the rdf: namespace, I think a cautious approach would be to
> bind it later :-) But it is just an opinion, not a research based advice.
> :-)
> 
> 
> Cheers,
> 
> Kjetil
> 
> 
>

Received on Thursday, 25 March 2021 22:40:58 UTC