mapbox / minjur

Osmium-based converter of OSM data to GeoJSON
ISC License
59 stars 14 forks source link

Drop points with only source and/or created_by tags? #12

Closed rclark closed 8 years ago

rclark commented 8 years ago

There are a lot of these points, but I don't believe that they really represent anything thematically interesting. Would it be acceptable to drop these from the output geojson?

example: http://www.openstreetmap.org/node/23137494

Are there other tags that nodes might have that should not whitelist them for inclusion in the output geojson?

joto commented 8 years ago

These are QA tiles, so maybe we want to keep strange tag combinations, so we can write checks for them or so. It is certainly doable to filter more, though.

rclark commented 8 years ago

cc @geohacker @mikelmaron curious what you think about these kinds of nodes from the QA perspective.

mikelmaron commented 8 years ago

What volume are we talking about here? And in what system --- dynamosm? OSMQATiles?

A node with only these tags --- well that node shouldn't have any tags on it actually. The only QA check we'd want to do is find them, and remove the unnecessary tags. Cleaning these up isn't a high priority, so not a big deal if they're not represented.

geohacker commented 8 years ago

@rclark I'd keep them in QA tiles, if it's not making us hit a performance wall. There's soon going to be a tag validator which will help us flag and clean these on a regular basis.

rclark commented 8 years ago

Sounds good let's keep 'em.