Ed. note: This blog post is reprinted with permission.
The delegates are heading home, or wandering the outback, and our
OpenGeo conference team of seven has dispersed as well. The conference
was a great success this year, bringing the message of open source
geospatial to well over 400 delegates (the final number might take a
little while to compute; apparently there was also a large contingent
of walk-up registrations).
The OpenGeo team presented Web mapping concepts (SLD, performance,
production tuning, tiling) and our core projects (OpenLayers,
GeoWebCache, GeoExt, Geoserver, PostGIS). We got a feel for the future
from the presentations of others.
The second day keynotes gave some good hints.
First, Senator Kate Lundy
spoke on Australian government open data and access policies. We are
seeing this trend throughout the world - governments are recognizing
the power of data access to improve services and engagement, and the
democratic imperative of transparency in government. In my home
province of British Columbia, the city of Nanaimo has been providing practical examples of data openness, while the City of Vancouver has made open data explicit policy, passed by council.
Second, Raul Vera from Google gave a roundup of the emerging spatial
technology consensus - ubiquitous geolocation, Web-based information
infrastructures. At OpenGeo, we like the way the open source model fits
technology for user interfaces, so this is a vision of the future we
share. Vera also talked about the future non-map-based location
services. That is something we are working on for our next revisions -
server-side scripting to provide direct access to all the spatial power
of Geoserver and PostGIS.
Out on the exhibition floor, it was great to see Autodesk still pushing
their open source agenda strongly. Jackie Ng, a local developer,
presented his FDO Toolkit
application at the conference, a cool desktop take to the FDO library
that probably wouldn't exist if Autodesk hadn't open sourced that
technology two years ago. Autodesk was a Gold sponsor this year along
with ourselves and Ingres. I also had a chance to talk with
representatives of both MapInfo and ESRI on the floor - the world of
open source is definitely getting larger (Intergraph, where are you?)
and more inclusive. I hope next year in Barcelona we will see
representatives of European companies like Geoconcept and Cadcorp on the floor.
Dale Lutz of Safe Software
highlighted the new inclusiveness of open source in his lightning talk
about the "End of Religion", in this case the religion of open source.
It does seem that we are entering a new era of pragmatism (or at least
acknowledging that we were always at our core pragmatic), where what
matters is the best tool for the job, as measured by some combination
of features and value and organizational capability.
Reports from the tutorial sessions (1.5 hour technical deep dives)
indicated overwhelming demand for some topics. The "Making Maps Fast"
and "Making Maps Pretty" sessions our team participated in had people
on the floors and out the doors. The demand for information about
practical operational topics remains very high - an important thing for
next years organizers to think about when soliciting talks and
My personal favorite technical talk was given by our own Tim Schaub.
"Openlayers: Vector Mayhem" was a concise tour of the vector subsystem
and clarified a number of issues that I personally found a little
confusing when trying to figure out the system guided only by the API
documentation. When his slides go online on the conference site, be
sure to check them out. For the database lovers, particularly folks who
have to work with Oracle, Simon Greener's "PostGIS and Oracle Spatial" slides will be a treasure trove of techniques for making Oracle Spatial easier to work with.
Finally, the technical highlight was the anticipated Mapserver/Geoserver performance benchmark, presented
(OpenDocument Presentation) at the closing plenary. The end result
showed little difference on the smaller point/polygon test layers, but
some differences when the testing moved to a larger 5M line roads
layers, and larger differences still for raster layers. In the end,
Mapserver was faster in most cases, but usually by small amounts, the
exception being raster layers, where it had a significant advantage.
The database backends also provided very similar performance numbers.
PostGIS was the fastest, but never by very large margins. I'm already
planning for next year's process, thinking of new ways to exercise the
servers and bring their differences to the fore.
Also at the closing plenary, this year's Sol Katz Award
was presented to Daniel Morissette. Daniel pre-dates me in open source
geospatial involvement; when I showed up, he was already a leader in
the Mapserver community, and had contributed some very important format
support options (MapInfo TAB, and Arc/INFO coverage) to the GDAL/OGR
library. Over the years, he's continued to lead Mapserver (he's the 5.6
release manager, once again), has been a leader in two important open
source companies (first DM Solutions and more recently MapGears), and
has become involved in OSGeo outreach activities, in particular in his
home province of Quebec. They are having regular meetings with 40
people in attendance now, which is very impressive. Many
congratulations to Daniel!
So, the 2009 takeaways: yet more cross pollination between open source
and proprietary; the inexorable growth of the Web as primary platform;
and the insatiable demand for information on Web publishing best
We look forward to participating next year, at FOSS4G 2010 in Barcelona , and hope to see you there!