Re: Proposal for ISSUE-12 language-tagged literals

Hi Antoine,

On 17 Jul 2011, at 09:47, Antoine Zimmermann wrote:
> Le 14/07/2011 15:08, Richard Cyganiak a écrit :
>> On 14 Jul 2011, at 08:36, Antoine Zimmermann wrote:
>>> In your proposal, rdf:Text is equivalent to rdf:PlainLiteral so it
>>> seems redundent.
>> 
>> No. rdf:Text is a class, rdf:PlainLiteral is a datatype.
> 
> But a rdfs:Datatype is also a rdfs:Class, and rdf:Text would be
> owl:equivalentClass of rdf:PlainLiteral (according to the OWL semantics).

Not according to RDF semantics.

>> Using rdf:PlainLiteral with rdfs:range creates expectations that its
>> impractical lexical form is used.
> 
> I'd like to see examples of usage to be convinced, but at least
> the spec of rdf:PlainLiteral clearly says that it is forbidden to use its lexical form, so normally it should never happen.

Here's another thing that normally should never happen: a spec that defines a lexical form, and then forbids its use. How weird is that?

>> "Plain literals without language tag" are now simply xsd:strings.
>> "Plain literals with language tag" are not actually plain. So the
>> name "plain literal" doesn't make sense any more.
>> 
>> For these reasons, rdf:PlainLiteral is inadequate for use in
>> rdfs:range statements.
> 
> Weird conclusion. Using rdf:PlainLiteral in a range statement exactly has the same effect as using rdf:Text.

No it doesn't have the same effect. One has the effect of typing xsd:string literals as rdf:Text. The other has the effect of typing xsd:string literals as rdf:PlainLiteral. The difference is that one of these makes sense, the other doesn't.

> rdf:PlainLiteral is a recommendation, you can't just ignore it and remake a new almost equivalent proposal just because the name is inadequate.

I'm not ignoring it. I'm proposing to change it. This group is chartered to reduce the confusion around string literals (time permitting), and the rdf:PlainLiteral recommendation is a part of that confusion.

> Moreover, i repeat that rdf:PlainLiteral was intended to be named rdf:text, but this name was rejected.

Had they known that the plain untagged literal form would no longer exist in the RDF 1.1 abstract syntax, I'm pretty sure they would have rejected the name rdf:PlainLiteral.

The rdf:PlainLiteral is an accident of history. I guess it was the best that these WGs could do given their schedules. But it is clear that it could not have been published in this form one year later. It ought to be fixed. If there's too much opposition to that, then I believe indeed that it is best ignored. It wouldn't be the first W3C Recommendation that suffers this fate.

Best,
Richard

Received on Wednesday, 20 July 2011 19:58:41 UTC