|
Over the last few months I've noticed minor changes to objects I created, that don't seem to show up in the history. Reviewing my edits confirms this. For example in this changeset I edited this crossing and this crossing, but their histories don't show the linked changeset. There were similar examples, where I tweaked the alignment or orthogonalization of an object, but I hadn't looked too closely as I wasn't sure if I was mixing up one intersection with another, until checking more closely today. I was spurred to do so by a very visible error, that has only existed a short time, but the objects' histories' show no recent edits. There's no way I missed an error like this for more than a day or two, but no changes for the last 4 moths show up. It's a common drag and drop error, someone has moved a node by a few hundred meters, almost a km, but the change does not show up in either way's history (this footway or this set of steps). Here is a screenshot for viewing after the error is fixed. Am I missing something? Has OSM changed the way it handles and reports objects' histories? It does not seem to be a temporary thing, (perhaps a result of overloaded servers) based on my experiences with less visible changes. Have I found a major bug? |
|
It might be seen as a bug, but I'm pretty sure moving nodes around never changed the way. As the way doesn't change, the changeset doesn't reference the way. 2
I've tried to find an example to the refute your statement that moving a node does not update the way's edit history, and have been unable to. I guess that edits that move nodes often add or remove one or more nodes, or change tags, in addition to the move, and I'd not realized until recently that ways that only have nodes moved don't get their edit history updated. Thank you for clarifying.
(10 Aug '21, 23:44)
keithonearth
2
If you move a node, the way history at osm.org will be unaffected. Some editors (at least Potlatch 2 and 3) do show node changes as part of the way history though - as a composite to changes to the way itself and to any nodes.
(10 Aug '21, 23:56)
SomeoneElse ♦
@keithonearth My pleasure, I've been puzzled by that before.
(11 Aug '21, 00:06)
H_mlet
"moving nodes around never changed the way" < I've never understood the logic behind this belief. If a node within a way moves, the vector of the way is bound to be amended. I believe Potlatch is the only editor to acknowledge this. Achavi also does.
(14 Aug '21, 14:54)
DaveF
1
It's not a belief, it's a fact. In the OSM database ways are not vectors, they are lists of nodes. Only the nodes bear the coordinates. Thus when you upload changes after moving nodes around, even if they are part of ways, the ways are not uploaded. Thus the changesets don't include ways. I think it makes sense in a database point of view. The osm.org website is mostly a front-end to the database, so some things might feel somewhat cranky. But I guess a pull request including ways implicitly modified in the changeset's page, probably apart from the others, would be accepted. There was some discussion about the API part of the problem.
(14 Aug '21, 16:03)
H_mlet
|
