It has seemed to me for a while now that the number of nodes presented on a street page is generally higher than the nodes visible on the map. Example: Bijlmerpleinpad
4 nodes are visible, but the page says 5. Of course it could be that there are 2 overlapping nodes, but I find the same discrepancy in almost all streets I look at closely.
Now, it is not a big issue, because I think as far as completion is concerned the count does not seem to have role, so it is purely cosmetic.
Interesting observation. I suspect @JamesChevalier will appreciate it, as often small errors may mean bigger issues.
That street does have 5 nodes in the database. It looks like these two are duplicates:
The node count is extremely relevant, because a street is marked as complete after finishing at least 90% of its nodes. If the count is off, it could result in the inability to complete it.
I can look into why there are duplicate nodes…
I can’t really confirm it, but my suspicion is that nodes are duplicated that join 2 OSM ways with the same name, but not anything else. In my example, there are 2 ways in OSM, with 2 and 3 nodes respectively. But 1 node is the same in both nodes.
When looking at other streets I haven’t found a high node count for streets that consist of 1 way in OSM.
Yeah, that’s what’s happening.
If you visit the Node in OSM ( Node: 3104127129 | OpenStreetMap ) the left column displays its associated Ways:
This may be worthwhile adding somewhere in wiki. It’s probably likely this is happening all over the place. Personally I’ve merged quite a few ways together over the years so that it is easier to edit a longer portion of a street at once vs lots of separate ways for same street. I didn’t think that it could potentially be helping CS with node count.
However, the above example, is likely very common where ways shouldn’t be merged; even though they are connected. It looks like one of the ways is part of a tunnel so they need to be separate ways in order to have different tags. Alas this means this ‘miscount’ of nodes in CS can’t be fixed on the OSM side.
@JamesChevalier Would first pulling all nodes in from OSM for a street and then filtering to only distinct OSM node IDs be a good way to get a street’s node count?
From what you found, it doesn’t seem that bad. The node count is technically correct, it is just that nodes are duplicated. Those will be completed at the same time, and because they are where “ways” touch each other I expect they will not be in hard to reach places. So the overall effect might be that it is easier to complete streets, e.g. a street with 9 “real” nodes will now have 10, so if you don’t run 1 of the nodes at the end it counts.
Well, hypothetically you could find this situation: there are 15 nodes in a street, but a duplication like this makes it say 16 in CS. If you have caught all nodes except this duplicate one, you really have got 14 out of 15, which is 93%, so street should be marked as Complete. But the duplicate makes the ratio 14 out of 16 instead, which is 87%, so street is Incomplete