search

osmlab / fixing-polygons-in-osm

Fixing (multi)polygons in OpenStreetMap
25 stars 4 forks source link

Richier stats about MP would help #12

Closed d1g closed 7 years ago

d1g commented 8 years ago

More light on this topic will help mappers and data consumers to write custom software/complex custom queries.

We have taginfo to inspect primitive stats about tags, but we don't have stats or even yearly public reports about what do we have in "relations".

mboeringa commented 8 years ago

How many valid MP inner rings are used in (other MP) as outer rings (inner-outer-inner implemented using multiple MP); according to our current page this is still valid (clearly problematic for some data consumers)

While maybe an interesting statistic, I think this particular condition is not an issue, and certainly doesn't require "fixing" of the multipolygons by duplicating data and ways. That would be even counter productive.

Ways are ways, they should be treated irrespective of to how many relations they are assigned, and in what role.

E.g. we also wouldn't want routes to duplicate ways if the route follows a way that is already used for another route? That would quickly lead to a completely unmanageable edit situation, with potentially dozens of independent ways stacked on top of each other.

If a data consumer application can't handle this, then the application needs fixing, not the OSM data.

joto commented 7 years ago

A lot of statistics are available on the http://area.jochentopf.com site. More are created or can be created based on the tools at https://github.com/osmcode/osm-area-tools .

Regarding the touching inner rings: This is allowed in OSM and there is really no problem to correctly identify those cases and create proper multipolgons from them data all data consumers can use.

joto commented 7 years ago

I agree that we do want more stats, but the area.jochentopf.com site is only a temporary site and I am not planning to add more. I have been working on these tools which can create some more detailed stats. Not sure yet how and where I'll show the stats generated from these programs, but I'll want to get them somewhere. Closing here.