I attended (and spoke at) the Association for Geographic Information (AGI) Technical Special Interest Group Open Source Event yesterday- down at the British Antarctic Survey headquarters in Cambridge (I want to work there, they have skiddoos parked in their carpark). The event was designed to kick off the newly invigorated Tech SIG, after a hiatus of several years. I can understand why there was a hiatus- the other SIGs have a more defined focus, such as the Environment, or Crime and Disorder and so on. The Technical SIG covers all areas, from a methodological point of view at least, and yet doesn’t want to tread on the other group’s feet.

Anyhow, the Open Source event was fairly well attended- with 30-odd people there, most of whom were new to the group, which bodes well. Alongside my own talk (on “Going Open-available on the talks page), there were talks from Paul Cooper (a roadmap of Open Source components for GIS web services and clients, and the South Georgia GIS) and Andrew Fleming from the BAS (access and delivery of Polar Satellite Imagery), Andrew Mackay from IPL (Using Open Source software in operational systems) and Gillan Arnold and Egbe Equavoen from the Ordnance Survey (OS Web Map tools).

A general observation- almost everyone talked about using some combination of geoserver/postgresql/openlayers. Other packages were mentioned in passing, but these were the big 3, the packages du jour. That’s not to say that there aren’t perfectly viable alternatives, but these ones definitely seem to have the momentum at the moment. It was nice to spread the word about mapfish though, which most people hadn’t heard of!

Andrew Mackay talked about using open source to deliver crime mapping to Kent Police. The thing about Andrew’s talk that stood out was the discussion about software development cycles. At first glance this seems like overkill for the majority of small-scale mapping applications. But then IPL are involved in some very large-scale application development, and probably understand more than most the need to get it exactly right.  I did begin to think that a little more discipline in my software development would be a good idea, even if the full cycle as per the textbook is a little too much!

Andrew Fleming’s talk about Polar satellite imagery was interesting. He raised a lot of issues around the delivery of the imagery, as near to real-time as possible, with less than brilliant communication lines (not much broadband in Antarctica).  For their needs, georss was a better bet than wms, due to speed and the size of the data set. They are still looking for a solution for deploying 3D and radar data, but have yet to come up with an optimal solution. Answers on a penguin to…

The demo of the Ordnance Survey’s web mapping tools was interesting, but ultimately frustrating because whatever you do with it you’re still hamstrung by the data license. The basic tool itself is a integrated postgresql/openlayers setup, with some nice extras for cutting their mapping tiles into handy chunks, and built-in geocoding (if you have the license…). I was hoping to hear that they are ready to start rolling out wms access to their data with an innovative new pricing structure, but no, so I’ll keep waiting for that.

The end discussion revolved around people’s concerns over the “risk” of open source either in general or in their organisation, and the need for traditional support agreements. In simple terms there was a basic division between those organisations with in-house IT staff who were quite flexible and willing to try out, or see the advantage of, open source tools, versus those who had out-sourced their IT to external companies with 3 or 5 year contracts to use particular packages. It’s a chicken and egg scenario put far more succinctly by other people than I ever can- but basically without the critical mass of open source adopters, how can large “traditional support companies” be supported?

All in all, an enjoyable and informative day.