Oh, absolutely not arguing that it’s bad to do the very detailed mapping. But I might need to choose between mapping one intersection to that detail, or ten in a more simplified way, and then I’m not sure that doing the one is overall more valuable. So I’d leave some room for multiple tiers of mapping like the Pedestrian Working Group uses.

“fourteen” should be 12, right?

I wonder how verifiable the button_operated values are. V-plans aren’t usually (as far as I’m aware, in Flanders) public by default, though perhaps they’re FOI-eligible? And that’s assuming they’re implemented correctly, and none of the sensors are broken (I’m aware of both happening here).

I wasn’t aware of traffic_island=yes being used at all given we have crossing:island; I removed one case in Brussels.

Splitting pedestrian and cyclist crossing seems reasonable, though I don’t always want to spend the time to do it - especially in the case where the sidewalk and cycleway parallel to the road are not mapped separately.

Another fun example that I see quite often, e.g. at Antwerpsesteenweg x Oude Bareelstraat:

Here the pedestrian crossings over the cycleway don’t have traffic lights, but those over the car lanes do. I guess the fully detailed mapping would add footway=sidewalks parallel to the cycleway, and tag the crossings as (NW to SE) 1. bit of footway=sidewalk 2. footway=crossing;crossing=uncontrolled 3. footway=traffic_island 4. footway=crossing;crossing=traffic_signal (just one, because no traffic island in this case, but often two with another footway=traffic_island between) 5. footway=traffic_island 6. footway=crossing;crossing=uncontrolled 7. bit of footway=sidewalk

which is… a bit much :)

Hey juminet, fantastic work! I’d love to use your data to help complete addresses in Gent. I think a plain list of missing addresses would be simplest for me, but whatever format you can share without much effort would be much appreciated.