You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
In recent months and years, I have encountered several rare instances where closed changesets are not reflected as such in the changeset replication files.
Specifically, these changesets are listed with attributes such as created_at="..." and open="true" in the replication files, despite being closed in subsequent diff updates.
Unless I have overlooked something, none of the three changesets mentioned here will be listed as closed in a later replication file. As I understand it, however, this should not be allowed to happen because, at least on the basis of the replication diffs, the changeset is never marked as closed.
The text was updated successfully, but these errors were encountered:
Sounds like a bug in the changeset dumper rather than an operational issue.
Looking at the code everything except detection of new changesets (which is based on id) is essentially heuristic because there's no other way to do it, so it's just looking at everything closed or commented on in the last period based on time which means there is the potential for a small window between runs that isn't picked up.
none of the three changesets mentioned here will be listed as closed in a later replication file
Can you double-check this?
I've checked that the changesets listed above are not contained in the linked and following replication file. What I have not explicitly checked is whether the changesets mentioned were published as closed in a replication file much later in the day.
But my assumption would be: If the changesets were distributed as closed, they would also have been closed in my changeset DB.
In recent months and years, I have encountered several rare instances where closed changesets are not reflected as such in the changeset replication files.
Specifically, these changesets are listed with attributes such as created_at="..." and open="true" in the replication files, despite being closed in subsequent diff updates.
Here are three examples from yesterday:
OSM Changeset XML https://www.openstreetmap.org/api/0.6/changeset/155634311 and the replication file matching the changeset: https://planet.openstreetmap.org/replication/changesets/006/149/084.osm.gz
OSM Changeset XML https://www.openstreetmap.org/api/0.6/changeset/155637623 and the replication file matching the changeset: https://planet.openstreetmap.org/replication/changesets/006/149/202.osm.gz
OSM Changeset XML https://www.openstreetmap.org/api/0.6/changeset/155656638 and the replication file matching the changeset: https://planet.openstreetmap.org/replication/changesets/006/149/606.osm.gz
Unless I have overlooked something, none of the three changesets mentioned here will be listed as closed in a later replication file. As I understand it, however, this should not be allowed to happen because, at least on the basis of the replication diffs, the changeset is never marked as closed.
The text was updated successfully, but these errors were encountered: