Waterway=riverbank only left in Norway

Seems like we have been called out for being the main country left to use the deprecated waterway=riverbank tag

Maybe something we should comply with? Apparently there was an issue of dual tagging water and sand, but that can be solved by having two objects on top of each other

1 Like

It has already been discussed in the community: Rivers update in Norway / users: Norway / OpenStreetMap Forum

Could you provide a link to the formal proposal for deprecating waterway=riverbank please?

See Tag:waterway=riverbank - OpenStreetMap Wiki (also added to first post)

Maybe something we should comply with? Apparently there was an issue of dual tagging water and sand, but that can be solved by having two objects on top of each other

I’m upgrading to the new standard whenever I randomly come across instances that don’t also have a natural=* tag. Simply because it annoys me to see the warnings in iD.

Eventually I’d also like to tackle the ones that arguably were the main cause of the lack of consensus in Norway, but I feel like the alternative mapping style isn’t as clear cut as you make it seem. Sure, you can map it using two superimposed areas, and I’ve done so myself when going for the same effect in my own mapping. (See e.g. this horrible micromapping masterpiece: Relation: ‪Øvrebotnen‬ (‪6173044‬) | OpenStreetMap) But, I haven’t actually found this mapping technique to have gotten any formal recognition anywhere, and it almost feels like it just works by accident in a lot of rendering engines. I’d like to see this technique, or a variation of it gain a formal approval before I would risk updating any of these cases.

@eisa01 If you want to, we could revisit the conversation that was had in the community, that @NKA mentions.

The conversation revolved mostly around what to about river areas that also have a natural!=water tag. My personal takeaway from that conversation was the following:

  • Gazer75 has a couple of good point regarding disadvantages of the proposed new schemas.
    • The old schema is correctly rendered by carto (for now). Some of the proposed alternatives would not be rendered correctly.
    • The old schema allows for mapping these areas as a single area, or a single multipolygon. The alternatives that are correctly rendered today would require two areas. This requires more effort from the mapper interested in mapping riverbed.
  • Despite point no.1 I feel like we should be able to find some working solution that allows us to use the modern natural=water water=river schema. That would IMO imply one of two options:
    1. Going through the formal proposal process with a single-area schema, like the riverbed=* suggestion mentioned in the conversation. Get said proposal to be rendered correctly, at least in carto. Then finish modernizing the remaining river areas in Norway.
    2. As a community, accept the slight increase in pain of a dual-area schema, where river and riverbed are mapped separately, such that we can modernize immediately while keeping rendering the same.

Right now, the dual-area approach, is my preferred option. This is because I actually like being able to treat the river and riverbed as distinct objects, which they are. It also allows for incrementally increasing the detail of the riverbed mapping, while not having to touch the river itself.

This is not a clear cut case, though. I’m tempted to reach out to a few data consumers, like carto, and/or mapbox to hear whether they have preferences in terms of what they’d find easier to render. And even if the community accepted the dual area approach, it would probably be best to get it through the proposal process, and document it properly on the wiki, to actually give it some authority.

I’d like to take the temperature on the Norwegian community in general, though. Does anyone besides Gazer75, @eisa01 and me have strong feelings on this? Please share your thoughts.