[sc34wg3] TMQL: Item reference vs. IRI

[Robert Barta]

On Sun, Jun 22, 2008 at 05:12:14PM +0200, Lars Heuer wrote:
> Comments against TMQL draft dtd. 2007-07-13.
> 
> Maybe this is too micrological, but anyway ... ;)

Life on this planet is all microscopic.

> Acc. to section 4.3.:
> 
>      """
>      [...]
>      an absolute IRI or a QName is interpreted as subject identifier
>      [...]
>      """
> 
> And it seems that "music:ec" in
> 
>     %prefix music http://psi.example.org/music/
> 
>     member-of(member: music:ec, group: $g)
> 
> is interpreted as a reference to the topic with the subject identifier
> http://psi.example.org/music/ec, there is *no need* to append a "~" or
> "<< indicators" to "music:ec" since the interpretation as topic is
> implicit.

That's exactly how it is, yes.

> Acc. to section 4.4: "indicators"
>      """
>      If the value is a topic item, in forward direction
>      this step retrieves all subject indicators of this item.
>      If the value is an IRI, in backward direction this step produces
>      the topic which has this IRI as subject indicator.
>      """
> 
> Hmm... isn't that a bit unlogical? If every IRI is already interpreted
> as reference to a topic with that subject identifier, the 'value' can
> never be an IRI, since it is already a topic, so the 'backward
> direction' seems to be useless.

It would be illogical, yes. But "IRI" above refers to IRI literals ala
production [10]

   iri -> " QIRI "

So, this ...

>    http://www.example.org/ << locators

... will not render anything as you correctly point out, but

  "http://www.example.org/" << locators

does. The prose is definitely suboptimal (will be fixed). But you should
have been alerted by the fact that

   "Robert Barta" \ name >> locators

does obviously return IRI literals. It _MUST_ be symmetric.

> If TMQL introduces a notation of subject locators, i.e. the '=' in
> front of the IRI TMQL could drop the backward directions for
> "indicators" and "locators". Would we loose something?

Probably. The one which comes to mind is _consistency_: If the whole
language is based on postfixes having prefixes in the mix is not a
good idea.

\rho