[sc34wg3] TMQL: Atoms

[Robert Barta]

On Sun, Jun 22, 2008 at 06:18:26PM +0200, Lars Heuer wrote:
> Hi all,
> 
> Comments against TMQL draft dtd. 2007-07-13.
> 
> Acc. to the *grammar*, the following queries are valid:
> 
>      24 << types
> 
>      "Hello World" >> supertypes
> 
>      42.0 -> member
> 
> The texts gives the impression that at least the latter is undefined,
> but it would be nice if the grammar can reflect that.
> 
> As far as I can see, the only use case were an atom should be allowed
> in front of a 'step' is the 'atomification' (resp. de-atomification)
> axes. For all other axes an atom seems to be useless.
> 
> IMO a more explicit grammar would be helpful, even if it increases the
> productions.

I think we agree that you cannot convey any semantics with a grammar,
so why start there?

The TMDM model has _MANY_ cases where navigation from a point A will
_NOT_ result in anything, we would have to introduce _A LOT_ of
special productions. For example:

  topic item and "give me the roles"

  assoc and give me the names

This is _EXACTLY_ what the axes syntax in TMQL tries to avoid with
introducing a mini language covering the axes ... richness of TMDM.

\rho