Friday started off with the “Open x 4″ panel, which had theoretical potential, but failed in the usual way panels do: lack of focus and excessive collegiality. For me the highlight was Peter Batty’s impersonation of Steve Coast (very convincing, in my opinion!).
Friday was also “PostGIS day” after a fashion, with 5 PostGIS-specific talks in a row. I started the series, with Tips for the PostGIS Power User (and here’s a link to the State of PostGIS talk I gave on Wednesday).
I was followed by Steven Singer with a talk on PostGIS replication. Really, a talk on PostgreSQL replication, since PostGIS support in PostgreSQL “just works”. The takeaway is that there are a couple means of replication available, the streaming replication available since PostgreSQL 9.0, which is suitable for replicating an entire instance, and the table-based replication available from packages like Slony and Skytools. For some use cases, like sharding, it makes more sense to use a table-based set-up. For other use cases, like high availability, using streaming replication of the whole instance is better.
Next up, Pierre Racine from Laval University gave an update on the new raster support in PostGIS. Raster will be released at PostGIS 2.0, and will bring a whole new toolbox for GIS analysis in the database. There was an incredible amount of excitement at Pierre’s talk, as the reality of the power of raster/vector analysis sunk in with folks in the audience. I predict that most of the new features added from 2.0 onwards are going to be in the raster side, as there are so many interesting areas for new features (contouring, surface fitting, density mapping, map algebras, etc, etc, etc).
After lunch, PostGIS day continued with Jim Mlodgenski presenting on the Stado (formerly GridSQL) parallel query system for PostgreSQL. Jim recently updated Stado to support PostGIS, which is a huge development. I was particularly piqued to hear this talk, since it is a direct response to my keynote at PgCon where I noted that an area of pain for PostGIS is the single-threaded query process in PostgreSQL. With Stado we can now leverage as many nodes as we want in processing spatial queries. For BI and spatial reporting, this is huge.
(My only regret about taking in Jim’s talk was missing GeoGlobalDomination: The Musical, which was going on at the same time.)
Finally, the most anticipated talk at FOSS4G, the review of new PostGIS 2.0 features from Regina Obe and Leo Hsu, my collaborators on the PostGIS steering committee. Though dense with SQL, the presentation got folks juices flowing: 3D support, new functions for slipping and snapping geometry, raster of course, topology models, the replacement of geometry_columns with a view, and more.
The conference wrapped up with thanks for the local committee and organizers, and very well deserved: the program, the venue, the parties, everything was excellent.
During the closing plenary, I was especially happy that my long-time Victoria colleague Martin Davis received the Sol Katz Award for his work on the JTS geometry library and work in supporting computational algorithm development throughout the geo open source community.
I’ll wrap Saturday into this final post: the code sprint was very well attended, and I was impressed to see folks in MapServer and OpenLayers with laptops cracked open and code being hacked. The PostGIS team huddled and coordinated our goals for the 2.0 release, setting a feature freeze date (end of November) and a first beta date (early January). But I personally only had the energy for one trivial bug fix, it’s been a long (and enjoyable) week.
Thanks everyone for an amazing FOSS4G!