[Patch awaiting end of stabilization phase] Split Crossing Ways test into Crossing Ways and Crossing Waterways

[avar]

Waterways should cross other ways (e.g. highways) without having connecting nodes. They should be left out of the crossing ways test.

[anonymous]

Use layer! Waterway also can be on bridge or in tunnel.

[stoecker]

When a waterway crosses a street on same layer something is wrong. Usually a bridge is missing. So this test is perfectly valid and data should be fixed.

[avar]

Replying to stoecker:

When a waterway crosses a street on same layer something is wrong. Usually a bridge is missing. So this test is perfectly valid and data should be fixed.

Actually a layer=* is missing. It'll still flag it if there's a bridge and no layer=* tag on either objects.

In any case I'd like to reopen this as an enhancement request and ask that when ways are crossing and one of them is a waterway=* that it be reported under another category (Crossing Waterways).

Usually Crossing Ways that aren't waterways are errors in the data like a self-overlapping highway made from GPX data, or two residential streets that cross but don't share a node. Those errors are important and can screw up routing, but crossing waterways without a layer=* are something that's handled by all the renderers and no other user of the data really cares about.

[LamoniDave]

Perhaps just a JOSM beginner question, but in the case where a road fords a waterway, would both be at the same "layer"? After all, the road traveler's (car, biker, walker, etc.) tires/ shoes/ feet "hit the road" at the same level as the water from the waterway. Is this a case where "a waterway crosses a street on the same layer" where something is not wrong? Should this produce an error? Or is there another way to represent a ford that should address this problem?

[avar]

Replying to LamoniDave:

Perhaps just a JOSM beginner question, but in the case where a road fords a waterway, would both be at the same "layer"? After all, the road traveler's (car, biker, walker, etc.) tires/ shoes/ feet "hit the road" at the same level as the water from the waterway. Is this a case where "a waterway crosses a street on the same layer" where something is not wrong? Should this produce an error? Or is there another way to represent a ford that should address this problem?

If the waterway and road meet with a ​highway=ford node then crossing ways won't complain. But if there's a highway=ford way over the waterway then the test will be triggered.

I'm attaching a test file which demonstrates this.

[avar]

ways crossing on a ford

[simon04]

In [4678/josm]:

fix #3736 - Split Crossing Ways test into Crossing Ways and Crossing Waterways