Lessons Learned from CityOpenSource
This document discusses lessons learned from using the Django framework in the CityOS open data platform. It summarizes that CityOS aims to enable crowd mapping of transitory information and communities around open data practices. The platform originated from a project in Naples to map street art that was later restarted using Django, AWS, and modern technologies with a goal of a meta-level approach. Django was chosen for its geographic features like GeoDjango but caching, optimizations, and real-time updates presented challenges. Additional issues involved customizing map markers, choosing base maps, and geocoding street addresses from various APIs. The document stresses that rushing geographic systems is bad, geographic data problems are often in the data itself, and keeping GIS simple
Lessons Learned from CityOpenSource
- 1. Django in the CityOS Platform Lessons learned, «Do»s and «Don’t»s Marco Montanari – CTO CityOpenSource
- 2. Hey who’s that guy?? • Marco Montanari • Django developer, GIS developer • Open Data «Evangelist»
- 3. What is CityOpenSource? • A platform to create platforms • Enable crowd mapping of transitory information • Enable the creation of practice communities Such meta!
- 4. It starts from Mappi-Na (http://mappi-na.it/) • Project born in 2012 in Naples • Aims to collect data on the street art in cities. • Was a heavily tweaked «off the shelf» Wordpress Theme...
- 6. It starts from Mappi-Na (http://mappi-na.it/) • Project born in 2013 in Naples during the Open Data Day • Aims to collect data on the street art in cities. • Was a heavily tweaked «off the shelf» Wordpress Theme... • ... that was hacked last summer and we decided to keep the remaining files (around 75% was safe) and restart with a new platform • Using django, AWS, «modern tech» • Looking for a meta-level approach => CityOpenSource
- 7. Django... why Django? • Django.contrib.gis • Python • Fast development needed • Reliable infrastructures • Easy social integration • Pip install whatever < AMAZING • Yet...
- 8. Caching, optimizing, stuff Everything can be accomplished with a given amount of code => time BUT Accomplishing it means more points of failure.
- 9. Users expect real time updates! • Cool, but it breaks caching archives • Cool, but it breaks the single page with the «nearby» feature • Enter vector-ish «tiles» • And/Or very soon GeoREDIS
- 10. What’s in a marker? • Custom markers? • Generic markers? • MAKI markers? BTW: thanks Mapbox for adding forcefully the mapbox id to the request breaking ALL apps that used the markers... I still love you anyway :P
- 11. Where are we displaying the markers!? • OpenStreetMap • OpenStreetMap France • OpenStreetMap Germany • Stamen Toner • OpenCycleMap • OpenTransportMap • Stamen Watercolor • Stamen Toner • Stamen Toner Lite • Mapquest • Mapquest Aerial • Esri Aerial Open • Mapbox... Oh, Mapbox... :D Bing, Google, Here... But we don’t really <3 them...
- 14. Where on earth is that piece of street art? • Non-standard APIs • Interesting wrapper for python • Different data quality • Google, nice but tricky • Bing, nice but limited • Nominatim (OSM), nice API but VERY limited Via Mascarella 34, Bologna, Italy Hear that with the tune «Where on earth is Carmen Sandiego»
- 16. Lesson 0 Rushing Geographic Information Systems is BAD
- 17. Lesson 1 The problem is never in the Geographic data
- 18. Lesson 2 The problem is almost always in the Geographic data
- 19. Lesson 1.5 There’s Geodata and Geodata
- 20. Lesson 3 It’s difficult to KISS GIS