next meeting about DOMAINs, and last night's meeting report

Hello!

Our next meeting on February 20th will be about DOMAINs, led by the
inimitable David Wheeler. David has not been able to attend our last
few meetings, but we are looking forward to his return. I will post
a more detailed description in the next week or so.

Key points from below:

* Bill - can we have a copy of your slides?
* Volunteers: Anyone want to help record the next meeting?
* Volunteers: Anyone want to present an EXPLAIN operator? (5 minutes
max!)

--
Interested in the Database Reading Group that meets on Fridays at
10am? Here's the URL:

http://datalab.cs.pdx.edu

Click on Reading Group!

I've been told that they like newcomers. The next paper is
"Community-Driven Ontology Matching" and it is available on the site.
--

And here's my meeting report:

Thank you everyone who attended last night. I truly enjoyed Bill's
presentation and hope that he will provide a copy of his slides to
the group. I did not manage to make a recording of the
presentation. Sorry! Are there any volunteers to set this up and
run it next time? I can provide my laptop if we can't use FreeGeek's
stuff. I have proven that I can't be trusted with recording
responsibility.

I began the meeting by showing lots of pictures of my chickens in the
snow, and with a really lame EXPLAIN operator: UNIQUE. Anyone want
to volunteer to talk about a better operator next time?

Also, I brought up an annoying problem I've had building DBD::mysql
on OS X 10.4. It was more of a Perl issue - but still! What are the
MySQL people thinking to not have this working!? DBD::Pg works great.

Bill's presentation started with a detailed overview of CMOP and its
place in oceanography research more generally. He showed examples of
the equipment used to gather data and talked about some of the issues
the data gatherers encounter (uh - echo location and the Navy?).
Many of the pretty graphs were visualizations of temperature and
salinity levels, which are important indicators of salmon habitat.

Then, he dove into the schemas he's developed to handle the different
data sources and showed how he came up with a set of generic sensor
data tables. He expected more heckles from the crowd on a slide
about not using surrogate keys, but we all just rolled over -
mostly. Perhaps we can arrange for Bill to have a debate sometime
with someone who has stronger feelings -- he seemed to be itching for
a fight.

I liked in particular a slide that had his tables, and then a phrase
describing the purpose and relations of the table over the top. This
was a nice way of explaining the thinking behind the schema without
using a bunch of jargon. Useful!

Bill also talked about the issues with having generic data tables.
To oversimplify an example: certain devices move around and have
arbitrary xyz coordinates, while others don't move around or have a
fixed x, y and/or z position. Bill uses triggers on insert to help
manage these cases. Also, as the schema changes, using views to
support legacy applications is very useful. This brought up the need
for rules in conjunction with the views for writing data back. There
was some talk of a package that automates some of the rule creation -
I'd be interested if someone wanted to post more information about
this to the list! Finally, Bill showed some of the magic needed for
partitioning tables and the reasoning behind a particular
partitioning scheme based on data collection stations.

Of course, Bill had lots of pretty slides of the products he
generates from the data. I asked about using the salinity 3D graph
thingie for making realistic water in video games. Sadly, Bill
thought that the data was too coarse for detailed modeling. This
also evoked shouts of "What, do you want to play Grand Theft
Salmon?" and "Extreme Bass Fishing" from the crowd.

Bill cut his presentation short - there was more about PostGIS that
we didn't see at all and he skipped through a few slides toward the
end. Maybe he will do a shorter talk on PostGIS sometime in the future!

-selena