Oliver Roick


Take-aways from FOSSGIS 2019 conference

FOSSGIS is the local, German-speaking chapter of OSGeo. Every year, they hold an annual conference, like a German-speaking FOSS4G. I used to go quite regularly, and I even helped to host one but haven’t been able to attend in the last few years. FOSSGIS 2019 was held in Dresden, which was reasonably close for me to go again.

Here are some random notes from the conference.

There’s a new standard that defines how “view service” (a web service that provides access to geo-data visualisations, also known as WMS) should be implemented within the German spatial data infrastructure (GDI-DE) to fulfill INSPIRE regulations. I do understand the need to define a baseline of metadata, map projections or languages that need to be supported by a standardised web service that is part of an interoperable infrastructure. I was surprised, though, that the details services to deliver web maps are still debated. The ins and outs of rendering and delivering visualisations of spatial data as images were a solved problem, I thought, as Tile Map Services do this very well. Apparently, there’s still a great need for custom maps to be delivered over the Web.

The talk on GeoStyler was probably my favourite one. GeoStyler is a set of modular JavaScript libraries that provides user interface components (in React) to create and modify map styles, data parsers (for GeoJSON and WFS) as well as parsers to read and write style definitions (SLD, Mapbox, or OpenLayers).

When I wrote my thesis 12 years ago, I tried to build something similar: A web app to edit SLD styles. I remember how hard it was even to get the most basic interface right to edit colours, strokes, and fills. This project is on an excellent way to solve the problem in a flexible and re-usable way.

It was the first time I heard about QField, a QGIS port for Android devices, which supports field-based data collection and management. It’s great to see a growing number of open-source solutions to mobile data management as a free and open alternative to various products from the big players.

It’s been ages since I last worked with OpenLayers and I only loosely follow news around the project these days. It looks like the main feature set hasn’t changed all that much in recent years, which only shows that the project’s scope is very well defined, which is a good thing. Instead, they have made considerable progress under the hood improving performance and developer experience, which included several rewrites of the library. I like the adoption of ES6 modules, which now allows developers to pick the functionality they need in their projects instead of importing the complete library.

OpenDroneMap has made impressive progress since I last checked the project. It has grown into an impressive set of tools to process aerial images:

  • A command line toolkit to process images
  • A web-based app that can run on local machines or servers that provides a graphical user interface to ODM, which is also available as a bootable ISO image.
  • A Node library that exposes a REST API to access ODM and a minimal web interface.
  • A command line tool to process images on the Web via the Node REST API
  • Also, a Python SDK.

It looks like I need to buy a drone after all to have a reason to play with OpenDroneMap tools.

Volker Mische gave an introduction to IPFS and demo’ed how to use IPFS to serve Sentinal data. I can see applications for land- and property-rights data: It keeps track of every version of a file, making it possible to trace the full history of a data set. Data can easily be mirrored across different nodes in the network, making it harder to permanently delete valuable information, or into a local network, so you access and edit data without the need to connect to the Internet. I know next to nothing about distributed file systems but IPFS is definitely something, I want to explore further.