The 2017 International FOSS4G Conference (Free and Open Source Software for Geospatial) took place in Boston, MA with a record attendance of around 1500 attendees. That likely makes FOSS4G the largest GIS tech conference not produced by Esri. The conference sessions featured both developer and user sessions, grouped into three day long tracks - PostGIS, R, and QGIS. At first, UGRC’s connection to these topics might not be readily apparent. To date, UGRC has only set up a successful proof of concept with PostGIS. Although UGRC stores a lot of data that could be used for exciting statistical analysis, we are rarely tasked with projects that could take advantage of R. And, QGIS is something we’ve installed to ensure that our services support its’ users. But we had no problem finding plenty of interesting and relevant topics, rooted in an environment that for which we are gaining appreciation.

Across the 3 themed days, important topic threads touched on open data and portals, containerized GIS, jupyter notebooks, integrating and/or moving to open source GIS solutions, vector tiles, serverless architectures, big data storage and query, and offline mapping capabilities.

At a big picture level, the open source geospatial community of developers and users has so many parallels with the broader GIS community. The ethos of the group – learn, share, partner, contribute – is really what powers the most impactful work in our field, whether using OS or proprietary technology bases. And the emphasis on building communal code and infrastructure that is available for all sectors to leverage, is so similar to what’s so important to enterprise-level GIS and open data initiatives.

On the big stage, an excellent keynote by Paul Ramsey on the attention, gift, and cash economies of open source software was particularly insightful slides and video. Holly St.Clair, chief data officer for Massachusetts shared her innovative initiative to put geospatial and other data to work to improve service delivery to citizens. And, in an unusual moment, the founding father of GNU, OS celebrity Richard Stallman preached his thought-provoking, if not controversial warnings on the inherent losses of freedom that results from closed-source software implementations.

We heard the term GeoPackage used during the sessions. If you missed this news, in 2014 the Open Geospatial Consortium (OGC) approved an open standard using SQLite 3 for storing multiple vector and raster spatial datasets in a single file similar to an Esri personal geodatabase. The data is stored in a single *.gpkg file! SQLite removes a lot of the limitations and missing features of it’s predecessor the shapefile and has mobile device support. I was even more surprised to find out that GeoPackages have been supported in Esri desktop software since version 10.2.1. Will GeoPackage adoption eclipse shapefiles? Should gis.utah.gov replace its’ shapefiles with GeoPackages? The open source and private sector GIS systems are on board, it will hinge on engineering software.

The vendor area had an excellent variety of companies showcasing very cool software and services. GIS Cloud demonstrated a well thought-out web first GIS desktop and mobile data editing platform that puts the editing of data in an environment that leverages the power and efficiency of vector tiles.

QGIS is maturing gracefully. If you have tried QGIS in the past and were not satisfied, it might be a good time to give the software another chance. For new and occasional users the price of entry is hard to argue with and the user community continues to build supporting resources.

Mapzen was showcasing their search service. It is built on top of open data and the open source pelias project. Using nodejs and elastic search for geocoding is definitely something to keep an eye on. Their routing engine is also worth noting with among other things, its low stress bike routing.

Esri was also in attendance talking about open data, geohubs, cedar, and esri leaflet. Andrew Turner gave a compelling talk showcasing how geographic information can be dissemenated without interacting with a map. There is a sample where you can input an address to find out useful information derived from GIS - like trash pickup and recycling days. He took it a step further with Amazon Alexa integration using a very useful Echo simulator. The days of solely getting GIS data information via a map click and pop up window are over.

Descartes Labs showed off some amazing raster analytics. My favorite being their GeoVisual Search where a user can click a box on a raster image and see other areas that are visually similar to your selection. I used it to find nearby tennis courts.

Other Noteworthy Items

  • JSGeo is looking for new organizers to continue hosting a great conference! If you are interested - get in touch with Brian Timoney, Steve Citron-Pousty, or Chris Helm.
  • t-rex looks like a great tool for creating vector tiles from PostGIS data. It is also a web server for serving tiles and allows for visual styling with Maputnik. It’s written in Rust so it is a single executable and fast!
  • North Carolina’s GIS data ships natively with the OSGeo Live compilation of geospatial applications - desktop to server - that all FOSS4G attendees received again this year on a bootable USB drive. Why shouldn’t be Utah data in the near future?!.

Lastly, Vladimir Agafonkin, aka, mourner - the creator of leaflet, presented on some awesome work he has been doing and I will highlight the ones that stood out to me.

  • concaveman is a very fast 2D concave hull algorithm in JavaScript.
  • tile-reduce is a geoprocessing library that allows you to run spatial analysis over vector tile grids in parallel “lightning fast”
  • supercluster is a crazy fast geospatial point clustering library for browsers and Nodejs.
  • cheap-ruler is a fast approximate geodesic measuring module
  • linematch is a super-fast algorithm for comparing road networks.