selective receives

Many programming languages these days support actors, and that's a good thing. Erlang has been designed around this model and Scala borrows heavily from Erlang's approach. The main idea is simple: every process/thread/super-lightweight thread etc. has a mailbox and may send/receive messages. But pattern matching on a queue gets slower and slower as the message queue length increases, especially if you are not consuming all messages.

Here are a few links to blogs discussing the issue:

inversion of control

what a year! I have been too busy/lazy/engrossed to write. Too many ups and downs. I have been trying too hard. So has my software done! I'm preparing an improved release of pg51g focused on in-place, cross-database diffs. This take is closer to Fabien's original approach. I have been using it in production for a few weeks and I'm a lot happier performance-wise. There's a nice little web service, too. In Erlang, of course!

pg51g @ PGDay Europe 2009

PGDay Europe 2009 was my first open source database event ever. And it was such an interesting event, with people from around the world and the PostgreSQL ecosystem, in particular. There was a French track, too, I actually went to a French lecture, which I understood! Then again, I had already read Fabien's technical report in English beforehand. I also tried to present my pg51g efforts in a short talk.

Having never done a lightning talk before, I was naturally quite stressed about squeezing as much information as possible into 5 minutes. Then, I became absolutely terrified when I realized the lightning talks were bundled together with the closing session, which meant I had to present in front of 150+ people in festive mood. The situation got worse when one of the lightning talk speakers did not show up and Greg did yet another stand-in. This was followed by David's talk on lightning talks, whose lolcats complemented the meerkats of Gavin's talk :-)

In the end, I managed to cover three out of five slides, so mission accomplished. And the audience was very kind. Here are the slides to my talk: pg51g-in-five.odp

You may find the slides from all talks here: http://wiki.postgresql.org/wiki/PGDay.EU,_Paris_2009

pg51g into the wild

Software is always "beta" but there's no point holding this baby back anymore. No guarantees, obviously, but it is probably fine for web applications and small-to-medium sized business databases. For more information, please see the relevant page: http://pgdba.net/pg51g

data migration

Yet another data migration, but this time it's a personal project. Over the next weeks, I shall try to migrate all content from http://freeyourtech.wordpress.com here. This will be a personal tech-oriented blog on my favourite topics, such as PostgreSQL, data-driven programming, Erlang and open source.

Michael Nacos