Meeting about to start

The ISO meeting during the weekend was, appropriately, held in an intercultural center, because the conference center is closed during the weekend. The meeting room did look a bit odd, but the discussion was sufficiently intense that nobody really noticed. And this committee is in any case used to some unusual meeting rooms.

TMDM-TMRM mapping

The purpose of this mapping is that in the TMQL standard TMQL queries are translated into mathematical operations on the TMRM. For this to be possible we need a mapping from TMDM into TMRM that TMQL can work with, and this is that mapping. I had originally written one, that Robert then improved and put into the official TMRM draft. He changed some things in the process, however, and we discusse these, keeping some changes, and dropping others.

There was also intense discussion of the relationship between inferencing and introspective queries. For example, everyone agrees that given the following CTM:

composer ako person .
puccini isa composer .

A query asking for all persons should find Puccini. That's easy. What's less obvious is what the answer to "how many associations does Puccini have?" There is just one in the CTM, but that one association implies another. Should the other be counted? If not, how can be we achieve this while still giving the correct answer to the first query? There was general agreement that this needed further thought and work.

Then we went to lunch at a local café, which was also a theatre, and came back to continue.

TMCL

Hard at work on TMCL

We worked through the Montréal comments on TMCL, discussed the relationship between scope and constraints, improved the general naming of PSIs and templates in TMCL, and did lots of work on the templating in CTM. TMCL itself changed very little, but we were able to move it quite a few steps further along.

It now really looks like TMCL might be finished after the Kyoto meeting in December. This, of course, means that people really should start looking at TMCL for real. With luck there may be some trial implementations soon that people can play with.

TMQL

We went through TMQL one more time, this time just discussing possible (and actual) holes in the specification. Hardly any changes were made to the language itself, but we found a couple of holes that were plugged, and added some missing syntax.

Like with TMCL it looks like TMQL will be finished after the Kyoto meeting in December. This means that if you have feedback on the language you better send it in now, because otherwise it will be too late and the language will be cast in stone.

The end

We finished late Sunday afternoon, and after that it was time for beer and relaxing, and then, the day after, the train to Berlin and the flight home.