Re: RDF-ISSUE-93 (non-langString): Give a name to "literals that are not language-tagged strings"

On 20 Aug 2012, at 15:23, RDF Working Group Issue Tracker wrote:
> Do we want/need a name for literals that are not language-tagged strings?
> Language-tagged strings are sometimes treated differently that non-languaged-tagged strings, so that there will probably be cases where it is necessary to refer to "literals that are not language-tagged strings".

Well, there might sometimes be cases where literals that are not of a numeric datatype need to be treated differently from anything else. That doesn't mean we should define a term for it.

Why do you say "probably"? What are those cases?

> The phrase is terrible and could be given a shorter name. If the RDF WG does not define a name for this, another WG may do it (cf. the notion of "simple literal" in SPARQL).

If another WG needs it, then they should define it. If it sounds useful, a future RDF WG can roll it back into RDF Concepts. We shouldn't add it unless we have a good bit of evidence that it's needed.

> A proposal: "typed literals".


> This will avoid countless confusions of people who are learning RDF with older tutorials and publications. Plus, they *are* typed literals in the sense that they *do* have a formal datatype (as opposed to "No datatype is formally defined for [rdf:langString]").

The goal of the literal redesign was to have less exceptions and to streamline stuff. Because of that, all literals in RDF 1.1 now have a datatype IRI. Explaining why certain literals are not "typed literals" despite having a "datatype IRI" seems counter to that goal.

I tend to call them "normal literals". "Normal literals, as opposed to language-tagged ones". Found that to be sufficient so far.

Thinking about it, "untagged literals" would be reasonably accurate and short. I would still rather avoid that term if possible.


Received on Wednesday, 22 August 2012 21:25:16 UTC