OSGeo @ Where 2.0

OSGeo (your Open Source Geospatial Foundation) will be on the exhibition floor at Where 2.0 this year. I volunteered to be the "event manager" for OSGeo at Where 2.0 this year, but all the heavy lifting was done by Alex Mandel preparing for the American Geographical Society conference earlier this year. So thanks to Alex's hard work, we'll have some posters and brochures and handouts for the exhibit hall.

Most importantly, we'll have ourselves. Not booth bunnies, per se, but we have beautiful minds. Arnulf Christl, the OSGeo President, is coming in from Germany to attend, so there will be lots of good conversation to be had. Come ask us any question about open source, and we'll be sure to give you an answer!

I'm also giving a workshop on spatial databases and web mapping. In fact, I'm preparing the material right now. Oh, how I hate preparing material – the being done is so much better than the doing.
 

FOSS4G 2009

The best conference of the year just got better! They've chosen me as a keynote speaker! (My turnoffs include soggy popcorn, rainy days and false modesty.) See you in Sydney, mates! Throw another shrimp on the barbie! That's not a keynote address, this is a keynote address.

Postscript: I just realized I learned everything I know about Australia before 1985.
 

The Fifth Circle of Marketing

What have I been doing for the past month? Not much PostGIS, as the growing list of 1.4 milestone issues attests. Mostly residing in the fifth circle of marketing – web content.
 

OpenStreetMap moves to PostgreSQL

Recently, the OpenStreetMap project put out a very successful call for donations to upgrade their physical database infrastructure, from a dual-core Athlon with 8Gb of RAM and lots (~1Tb) of disk, to a quad-core Xeon with 32Gb of RAM and heaps (4Tb) of (15K RPM) disk.

The speedy success of the hardware appeal (target reached in less than three days) was pretty impressive, but what really perked my (PostgreSQL fanboi) ears up was the news that the new hardware was going to run PostgreSQL, instead of the MySQL database OSM has used from the start. As of April 19, OSM is running their new API live on PostgreSQL.

So, why has OSM abandoned the worlds most popular open source database? I asked the OSM folks, and this is what Tom Hughes of OSM told me:

Personally I've been very frustrated with MySQL from when I first got involved with running things. Some of the problem was of our own making in that we had a mix of MyISAM and InnoDB tables (originally everything was in MyISAM) and some tables were using MyISAM features that meant they couldn't be easily moved to InnoDB.

On top of that it seemed that virtually any non-trivial query would completely defeat MySQL's optimiser.

The comment about a mix of tables really hits home, since so many MySQL features are split across table types. Want transactions? InnoDB! Want full text search? MyISAM! Want spatial? MyISAM! Want spatial or full-text and transactions? Tough. The devil is in the details. When asked: does MySQL support spatial, transactions, full-text? the MySQL answer is "yes", "yes", "yes", but the reality in production is not nearly so clear-cut.

Note that OSM is not using PostGIS for the main database at this time (their current data model of nodes and ways wouldn't get much leverage from it) but it is used for other processes like OSM tile generation. And a growing number of people on the PostGIS users list seem to be using osm2pgsql to extract data from the OSM production server for rendering / analysis in PostGIS.

So, welcome OSM, to the PostgreSQL community!
 

Diaboogical

Someone, please call the FTC! I may post my blog on Google, store my mail and calendar and notes on Google, teach people how to make maps using Google, but even I have my limits.

The Google Profile leverages Google search monopoly (you listening, FTC?!?!) into a you-must-join social network. I like LinkedIn as a professional social network, but this is going to blow all that out of the water. Want to be found? To control your professional image? Yes, having a decent personal site will still be important, but now the starting point for a personal search (Google Search) is also the end point (Google Profile).

Can you afford to not have Google Profile? I don't think I can, and I think that stinks.

Update: Apparently I wasn't clear about my point: because Google controls search, their Profile is a must-have for anyone who expects to be looked up on the internet (independent contractors, mobile professionals, me). It is not optional. It's a Google world. Comply.
 

Earth Sandwich

Happy Earth Day! If you missed the earth sandwich three years ago, here's a chance to re-live the dream!

The Original Challenge
The Find-the-Opposite Tool
The Winning Entry Announced
The Winning Entry

Mmmmmmm. Earth sandwich.
 

Map of the World as We Know It

Somehow, I missed this video in the geoblog meme-stream, so for all the rest of you who also missed it:



And since I'm posting YouTube links, you must watch this unlikely Scottish talent show contestant blow the audience and judges away (figuratively).
 

Where? There. When? Soon.

For a number of years now, I have watched with longing as Where 2.0 slipped past me on the calendar. I love the idea of a geo-conference where most of the attendees are not in the government/land management/asset management realm. Unfortunately, trivial matters like the births of children and so on have gotten in the way of my attending, thus far. But not this year! This time out, I'm going to Where, and I'm going to be teaching a short course on spatial databases. Spatial SQL is a powerful tool, so powerful that you don't require much more than a spatial database and a web mapping API to do some nifty stuff. OSGeo will have a booth this year, so if you're at Where and you want to talk databases, track me down there.

Reminder: If you're reading this before Wednesday, you still have time to register before the early bird rates go away.
 

Recession Signs

On my last trip through SeaTac, I thought my thin copy of WiReD was a physical harbinger of the recession. This time, the state of my flight from Victoria was a pretty clear sign.



Oh boy.
 

A small thing...

Leaving the house this evening, my daughter presented me with a daffodil from the garden, and I put it into my lapel. Little did I know that this would cause me to be sent through the "special line" at customs: the daffodil is an agricultural product.

On the up-side the officers were uniformly good humored and enjoyable and basically waved me through (who can be cross with a man with a daffodil from his daughter?) so it was actually a positive in an otherwise normal drab transit of SeaTac.
 

ESRI Formats - Back to the Future?

Remember when "coverage" was the holy grail? "Boy, if I could just read/write coverage, I could interoperate with the GIS folks no problem."

And then things seemed to get better. Most data was in "shape files" and there was a published specification which more-or-less matched the data generated by the ESRI tools. And lots of folks were using ArcView, so you could share "avl" files (layer style) and "apr" files (project file) and even write third-party tools to consume them directly (they were just a funny text format after all).

And now they are back-sliding. Layers and projects are "lyr" and "mxd" binary formats generated by ArcGIS and not consumable by third party tools. Data is starting to be moved into "file-based geodatabase" (FGDB) and the only tools that can peer into those are Arc*. (The old geodatabase had a lot of drawbacks, but at least you could open it in a third-party tool.) There are no specifications for the data formats, for the layer formats, for the project formats.

I thought we in the computer profession had this discussion and came to a consensus: vendors that tried to use proprietary file formats as a lock-in were to be avoided if possible and castigated if necessary. Even old Microsoft had to play the "open format" game, at least in mixed company.

It's not just data, though the rise of the closed FGDB format over the last couple years is bad enough. The "metadata", the style files and the project files, are a key piece of "operational data" that can be incorporated into workflows automatically – if you can read and write them. With format lock-in in place, the only way to automate your work flows is through the vendor-approved channel.

Really, I thought it was decided – they don't get to do this to us anymore.
 

Geometry Invalidity Examined

Why do we get so uptight about rings that touch themselves? After all, masturbation is perfectly normal for a young ring. Healthy even.
 

Spring is officially here

You know Spring has really returned to Victoria when the cherry blossoms come out. I can't begin to say how happy I am not to live in a place where turkey vultures are the harbingers of Spring.

"Procuring" Open Source

Open source has some real problems in the market, and most of them seem to reside between the ears of the people making corporate decisions on software. Really good article on procurement perceptions here from a Plone support company (what, Plone needs support? :)

The bit that Simon said that made the bulb go on in my head? He said that most companies set out on a procurement process to procure a software license. That is what software is to them: licenses. That is what they know, and that is how they think. If you've ever had the fun of dealing with the contracts/procurement department of a large company you will recognise this thinking. And so with that goal in mind, of course they end up succeeding in their quest to procure a license for software
—Matt Hamilton