[sc34wg3] TMQL: XML content vs. < > delimited IRIs

[Robert Barta]

On Thu, Jul 10, 2008 at 04:21:52PM +0200, Lars Heuer wrote:
> Hi Robert,
> 
> [...]
> >> ... but <> are needed for rootless IRIs like "mailto", "mid", "urn"
> >> etc., right?
> 
> > For TMQL I have now:
> 
> >   "http://whatever/"    # interpreted as IRI literal
> 
> How does the parser know that "http://whatever.com/" is not meant as
> string?

Because the string follows a certain syntax. We have the same game
with decimal vs. integers. Integers are special decimals, detected
by a more specific grammar. Same with IRI-string vs. string.

> >   "http://whatever/" ~  # IRI literal used as subject identifier
> 
> >   http://whatever/      # stand-alone IRI used as subject identifier
> 
> The "" around IRIs is necessary, if someone wants to refer to a topic
> by its subject locator / subject identifier and wants to use the axes?

_IF_ you want to explicitly use axes, yes.

> The:
>    http://whatever/ ~
> and
>    http://whatever/ =
> 
> would be an error then?

Exactly.

> <> for rootless IRIs and http://... for subject identifiers is aligned
> with CTM. But "http://www.example.org/" is not supported by CTM, it
> would be interpreted as string, unless the datatype is made explicit:
> 
>       "http://www.example.org/"^^xsd:anyURI

Ok, here is a difference. One of many :-)

\rho