[tmql-wg] Re: [tmcl-wg] TMPath - experiments summary

Robert Barta rho@bigpond.net.au
Fri, 1 Aug 2003 07:04:11 +1000


On Thu, Jul 31, 2003 at 11:15:48AM -0400, Dmitry wrote:

> Robert, thank you for your time and your comments. I tried to learn
> and reuse as much as I can from AsTMa? :-)

:-) And I ... borrowed from XQuery what I could so ...

> My original goal was to investigate functional, "XPath"-style
> language which can be embedded into other constructs such as TMCL.
> So I tried to stay on "pure" functional ground.

... also AsTMa? should be purely functional.

> My main interest was TMCL. For TMCL I tried to find equivalent of
> FOL functors. I think that "path" expressions can play this role
> very well.  For TMCL my plan is to have "pure" assertion-based
> language vs pattern-based language.

This makes a lot of sense to me.

I believe that all AsTMa* patterns can be compiled into path
expressions like TMPath or AsTMaPath. This means that the core formal
semantics of AsTMa? can be based on this. I am not too happy with the
current semantics definition in AsTMa! It is too convoluted.  If you
can base your TMpath more firmly on a formal model, then we are in
business.

> Currently I try to extend TMPath with pluggable serializers. Two samples for
> XML and Wiki:
> http://homepage.mac.com/dmitryv/TopicMaps/TMPath/TMPathSerializerSamples.txt

Yes, I was thinking exactly along these lines. As I said TMPath can be
made into a query language (more or less) easily.

The wiki idea is pretty cool. I wonder, though, whether the knowledge
that this is wiki-code is relevant for the query language. It is flat
text and inside the {}'s there is TMPath code.

\rho