Ways with same tags as mp relation

[joto]

Sometimes (inner or outer) ways have the same tags as the multipolygon relation. This could be the result of an incomplete cleanup of the old-style multipolygon or have other reasons. In northern Italy for instance there are a lot of cases of forest polygons inside forest polygons, possibly from some import.

I have been tracking inner ways with the same tags, but not outer ones.

This was brought up here:
gravitystorm/openstreetmap-carto#2533 (comment)
which mentions this example:
http://www.openstreetmap.org/way/100827142

This needs to be addressed. One problem here is that a lof of these are water polygons (as in the example above) and there are different taggings (water, waterway, riverbank, ...) involved here. If we are doing challenges for these, it would be better to make sure we also correct inconsistent tagging in the same go.

[olt]

I did noticed that as well while I wanted to remove more old-style multipolygon code from Imposm.

There are a lot of relations where the outer way contains the same type tag and often some more specific tags like source, wood, leaf_type, etc.:

• http://www.openstreetmap.org/relation/49631
• http://www.openstreetmap.org/relation/56699
• http://www.openstreetmap.org/relation/67378

I'm counting more then 900 relations in Lower Saxony, Germany.
An importer needs to keep track of which outer ways were already imported as part of a relation, as long as these errors exist in the dataset.

osm2pgsql calls this members_superseded and use of this feature will be dropped when gravitystorm/openstreetmap-carto@cb23299 is rolled out with the update to the openstreetmap-carto styling 4.0 (@pnorman, please correct me if I'm wrong).

[pnorman]

osm2pgsql calls this members_superseded and use of this feature will be dropped when gravitystorm/openstreetmap-carto@cb23299 is rolled out with the update to the openstreetmap-carto styling 4.0 (@pnorman, please correct me if I'm wrong).

osm2pgsql and openstreetmap-carto are different projects. the osm2pgsql default C tag transforms are not yet changing, osm-carto changes are independent of that.

[joto]

I have started to create challenges for the about 50,000 cases where one or more ways have the same tags as the multipolygon relation.

[imagico]

For the more complicated cases (i.e. MPs with several members) i think an organized effort in fixing would go more smoothly if the standard map (and ideally also other widely used maps) as well as common software (including libosmium - see osmcode/libosmium#155 😉) have fully arrived in the post-old-style-MP world so you get correct feedback for fixes in terms of data interpretation. As long as member way tags are still discarded in anticipation of old style tagging this is much more difficult.

[joto]

@imagico OSMF servers are in the process of being changed to carto 4 which uses the new interpretation. As I write this 3 of the 4 OSMF tile servers have been switched over, I assume the last one will follow soon. This should have the biggest impact.

Libosmium has support for this old-style-mp-less world in the master branch, but docs are sparse and the code has to settle some more before a release. If somebody wants to experiment, please do. Other software developers are aware of these issues (see #23).

[olt]

@pnorman What I meant was, that carto 4 will not use this members_superseded feature.
It's good to hear that the servers are already updating.