Examples of where two nodes on top of each other is likely correct:
when the nodes have level or layer tags
situations in indoor mapping
passenger lifts which are part of the pedestrian street network
Also certain data loaded into OSM was ‘noded’ so where any features crossed a node was created in each. The main example is TIGER data in the US, but early imports of shape files often resulted in the same feature. Joining such nodes had undesireable properties (such as allowing routing from a bridge on to the road underneath).
Thanks both. Is there a way of exploding the nodes like you can PowerPoint layers (like this), to inspect each one in turn? Maybe in JOSM?
I’m still on iD and the only way I know is to ‘disconnect’ a node, drag it sideways to inspect it and the one underneath, and then reassemble if necessary. The result is immaterial changes mixed in with the material ones, which have to be uploaded to the database nonetheless.
Also iD-related: if a way seems needlessly complicated, I tend to drag one node on top of another, and one of them magically disappears - I assumed it was quietly deleted once merged with its sister, but I’m wondering now if actually I’m piling nodes on top of each other?
Not with that visual effect, but you can use the middle mouse button to select from stacked features in JOSM.
If you want to avoid such immaterial changes in iD, I believe you can undo the action(s) with Ctrl+Z instead of dragging it back by hand after you have inspected the situation.
(In JOSM, this is particularly useful because undo doesn’t change your selection. So if the middle mouse button doesn’t work as well because the nodes have no tags, you can move the nodes apart, select the one you want to delete/edit, use undo to get them back to their original location, and you will still have that node selected. Not sure if that applies to iD as well.)
From what I understand, iD automatically snaps and merges nodes you drag close to each other like that, so it should do what you expected.