What follows is an overview based on some of the notes I took. I wasn’t always writing as I sometimes just stopped to listen and I’ll probably follow up with more details later.
Data reduction was strong current running under the day. With the rise of newer libraries that enable more sophisticated capabilities in the browser, there is increased recognition of the need to reduce the amount of data passing over the wire. For geometry, TopoJSON is getting a lot of attention here. It delivers topologically correct data which reduces data by removing redundant geometry such coincident arcs between polygons and such. The output reminds me of an old ARC/INFO coverage and is providing stunning results in reducing the size of geometries when compared to GeoJSON or Esri JSON. Andrew Hill of Vizzuality mentioned that CartoDB currently has support for TopoJSON in staging. He also discussed strategies they use to reduce the size of feature and attribute data. This “data cubing” strategy turns out to similar to approaches that are commonplace in the financial industry. This serves as a reminder to look around and see what others are doing to solve similar problems.
One a related note, the reduction of application code being delivered to the browser is also a focus. This was probably best illustrated by Matt Priour’s talk on OpenLayers. The fact that there are now three builds (full, mobile, light) as well as a focus on being more modular in version 3. Peter Batty, in his demo of Ubisense, made a good case for intelligent balance between server-side and client-side processing to achieve good performance.
GeoJSON continues to get a lot of love. It continues to be, in my opinion, the motor oil for the engine that is web mapping. It will be interesting see how the separation of concerns between GeoJSON and TopoJSON evolves, but I think GeoJSON will continue to be an important syntax.
While Matt Priour and Aaron Ogle opened with great talks on OpenLayers and Leaflet, respectively, the clear star of the client-side show was D3 (although Aaron’s walkshed analysis in Leaflet was cool). There were lots of great visualizations (as well as nods to TopoJSON). While the D3’s support for projections is getting a lot of attention from geospatial developers, I am more interested in its holistic approach to data visualization. In D3, geo is one way to look at data, but not the central way. Because of this, D3 provides an integrated capability to have multiple data visualizations in an application without overloading with a ton of libraries.
I am already well past tl;dr on this post and I have to catch a plane. Many of the URL talks were great but I’ll follow up with some more details. All-in-all, I think the event was quite a success and I’m glad I came.