Giving up on mapstraction
Today we’re going live with the major test (50% of users I believe) of a new design of Nestoria in four of our eight markets. I’m proud of this not just because I think the new design is a nice improvement, but also because it’s a major user facing project that I had no input into (I’ve been able to largely turn my attention to new projects, for which I’m very grateful). It’s been a great effort by all involved. It’s a total visual refresh, including new branding and logo, while staying true to our ethos of making property search as simple as possible.
But the market situation mapstraction was built to address has changed. Fundamentally, it seems developers now face two choices - Google or OpenStreetMap (one could argue Yandex is a viable choice in certain markets, and to be honest I have no clue about China and Japan). Frankly, none of the other mapping providers have critical mass, at least in the digital circles I frequent. Some, like Yahoo, have disappeared completely ** lone purple tear runs down my cheek **. Almost all of the innovative stuff I see is being done using OpenStreetMap. And while I still come across the occasional use of OpenLayers for displaying OSM tiles, by a wide majority devs have switched to using the excellent leaflet library, not least because of the emphasis on lightness and the excellent documentation.
With leaflet using tiles from any of the OSM tile services, or indeed any tile service, is trivial (at Nestoria we use MapQuest’s OSM tiles). There are more and more OSM tile providers (see MapBox or Thunderforest), I can’t really foresee a scenario in which we’ll move away from it to another commercial service.
And so, rather than build mapstraction into our site - at the cost of additional bandwidth for the user to download and additional maintenance burden for our team (having to stay on top of what’s new in mapstraction), we’ve moved to the assumption that we will always use leaflet or Google.
All good wishes to those who continue to use mapstraction, and heartfelt thanks to all the contributors - on Nestoria alone your code has been used by millions of people, be proud. But as the online mapping landscape changes the cost of an abstraction layer no longer justifies the benefit.