[sc34wg3] TMQL - Comments against draft dtd 2007-07-13

[Lars Marius Garshol]

* Lars Heuer
>
> No, my suggestion was to loose the need for the 'anchor'.
>
>     tm:name ako tm:characteristic
>
>     tm:occurrence ako tm:characteristic
>
> So, you'll get with "tm:characteristic" names and occurrences.
>
> And you can use the postfix filter for filtering by type (see my
> examples in the original mail).
>
> The 'anchor' seems to be useful for the roles / players navigation but
> all other productions can live easily without it (and some ignore it
> at all as already stated above).
>
> And iff we drop the (ignored) anchor, we get only one production which
> need it. And your argument, that you'll need 10 more productions would
> be disproved. ;)
>
> Seriously: I do not see an advantage if the 'anchor' is kept if you
> get the same results with postfix filters.

This makes a lot of sense. I've made the issue tmql-anchor-needed for  
this. Will return.

> [tmql-tm-content-wrapper]
>
> Hmm... Does this answer my question? ;)

No, Robert didn't answer it, but it's too early to answer it. We need  
to look at CTM first. It may be we can remove the problem there. In  
fact, we may already have removed it in Leipzig. I need to look at  
the meeting notes to remind myself.

> Sure, but I wonder how a user can define the 'default' topic map if
> the FROM clause is missing. TMQL refers often to the 'current context
> map' and I thought this topic map is declared within the environment
> somehow.

In these cases the current context map is determined by the  
environment outside the query. This is analogous to how XPath  
expressions in XSLT don't specify what document they are querying.

--Lars M.