ISO meeting in Seoul, day 3
Posted in Technology on 2006-06-01 06:03
The meeting is now beginning to settle into a kind of daily rhythm that seems to run: two separate WG3 meetings, enormous lunch, plenary WG3 meeting, vast dinner, drinks, sleep, then start over. This already gives you the big picture of the day.
The time before lunch Robert, Graham, and I had an editors' meeting where Robert took us deeper into his TMQL-to-T+-mapping, which has actually become very nice and simple. He doesn't yet have any written document that describes the last (and best) iteration of this, but this will be his TMRA'06 paper, so it should emerge before too long. Overall, I am growing more and more pleased with the formal underpinning that Robert has provided for TMQL, and we are now (after three attempts) beginning to see the correct shape of the TMDM-to-TMRM mapping as well.
The rest of the Working Group spent the pre-lunch session discussing PSIs. Steve Pepper did a presentation about PSIs at a workshop in Edinburgh, and the result is that now both OASIS and the W3C are seriously interested in the concept. To do this, he changed the acronym from PSI to PRI (Published Resource Identifier), however. I wasn't present at the pre-lunch session, but from what I could tell WG3 and James Bryce Clark of OASIS agreed to work on some specification for this. Exactly what will happen was (as far as I could tell) not decided there.
Sam Oh and James Bryce Clark
Suffice it to say that lunch was huge, delicious, and Korean.
After lunch the meeting was walked through the latest version of the TMRM, which has now become very simple. My subjective summary of it is that it contains the T+ model, four simple operations on that model, and fluff. In addition, there are annexes containing more operations on the model, and a mapping from TMDM to TMRM written by Patrick. (The T+ model is often also referred to as the TMRM, which is slightly inaccurate, as the TMRM also contains other things, but the inaccuracy is convenient, so...)
The T+ model is actually simple enough that it can be explained in a paragraph. A T+ model is a set of proxies. Each proxy is a set of properties (where property is what would elsewhere be called a property value assignment) in the form of a (key, value)-pair. The key is always a proxy, and the value is always either an atomic value (string, number, ...) or a proxy. And that's it! There isn't anything more than this!
The four operations are also very simple:
- Find all the keys in a proxy. (The symbol for this is down-arrow.)
- Find all the keys of which this proxy is a value somewhere. (Up arrow.)
- Find all values for a key in a proxy. (Right arrow.)
- Find all proxies which contain (k, v). (Left arrow.)
This, together with some operations for dealing with sequences and sequences of tuple values, is what TMQL is eventually mapped down to, via some intermediate layers.
In between other things, we also had a little get-together where we practiced the chant that Korean soccer supporters use when rooting for the national team. See the picture below if this was confusing.
The last day of the Seoul meeting also had the closing plenary in the morning, with lots of excitement around who would vote what regarding the OpenXML submission from ECMA and Microsoft
Read | 2006-06-02 04:21
The ISO meeting during the weekend was, appropriately, held in an intercultural center, because the conference center is closed during the weekend
Read | 2007-10-16 00:24