Robert Barta rho@bigpond.net.au
Fri, 30 Apr 2004 12:47:05 +0200

Hi all,

Steve, as WG3 convenor, has asked me to lay out the next steps in putting TMQL on the road.

Quickly a recap, what happened in Ams'dam:

- we had a day of presenting and discussing existing proposals (Toma, TMPath, tolog, AsTMa?)
- while we originally had planned to break out into small groups to reduce the number of
languages to 3, we realized that the differences were rather on the surface and
decided to sketch a language in one go.
- We used tolog as starting point and added more flexibility and features where it had
weaknesses.
- This resulted in - what I would like to call - the 'clayman':

http://www.jtc1sc34.org/repository/0502.pdf

It is currently more a 80%/20% solution, but the basic features are:

- SELECT ... FROM .... WHERE for simple queries
- FLWR for the more general case
- querying of assoc like in tolog, querying of topics and components via path expressions
- more flexibility in formulating association predicates
- querying of multiple maps
- import of functions, user-defined functions for 'inferencing' and auxiliary means
- can return lists of tuples, XML structures & TM fragments
- support for EXISTS and FORALL (unlike pure tolog)
- namespace management for functions and subject identification

There are a few unresolved issues, though.

What happens next?

While larsbot is away, I will try to sort the drafty ideas and will squeeze them into
a 'woodman' version. I also try to reformulate the use cases in this language, not sure
how far I get with that.

I plan to

- use the tinman to formulate a 'surface syntax', mostly for convenience (and to
make the SQL lobby happy :-)
- define a rather orthogonal core language
- define a formal model onto which the core is mapped (doing this at the moment)
- come up with a non-XML notation for TMs (somewhat a mixture of LTM/AsTMa=)

This is supposed to be done until mid July. At that time we plan a
face2face meeting of the editors and interested parties, presumably in
Korea. This is organized by Steve P.

The result should be a 'tinman' which could be submitted as WD to SC34 in Oct this year.
The intention is to move this onwards to an 'ironman' until March 05 with a final ballot
for an 'steelman' mid next year.

I'll post anything worth here.

\rho