[OSM-talk] Identical Duplicate buildings

[Wayne Emerson, Jr.]

To analyze this situation I decided to check my local area and examine 
the problematic changesets in detail. So far I have found a few 
different scenarios:

1.) Duplicate ways uploaded in one changeset which took a long time to 
process.
Changeset 92289777 created 65 duplicate parking spaces (created 
2020-10-11 00:43:02; closed 2020-10-11 00:46:45) by the iD editor. We 
can see it took 3 minutes and 43 seconds for this changeset to upload 
which is unusually long)
The 2 sets of duplicate parking spaces had a gap in their numbering 
sequence:
Set 1 was numbered ID# w857666299 through w857666363
Set 2 was numbered ID# w857666698 through w857666762
These were the only new ways uploaded in this changeset.

So I decided to check the way ID's that fell between the above sequences
w857666364 from changeset 92289780 (created 2020-10-11 00:43:16; closed 
2020-10-11 00:43:17)
w857666697 from changeset 92289805 (created 2020-10-11 00:46:12; closed 
2020-10-11 00:46:13)
This helps us to see the timing of the 2 upload attempts of the duplicates:
Set 1 was uploaded between 2020-10-11 00:43:02 and 00:43:16
Set 2 was uploaded between 2020-10-11 00:46:13 and 00:46:45

I have sent a message to the mapper of the changeset in question to see 
if they can recall what happened during their upload. But no response yet.


2.) Another set of duplicates I found were uploaded in 2 different 
changesets, 36 seconds apart by the iD editor.
https://overpass-api.de/achavi/?changeset=76618180
2019-11-05 Created: 00:36:10 Closed: 00:36:11
4 ways deleted; 10 ways modified; 7 new ways

https://overpass-api.de/achavi/?changeset=76618192
2019-11-05 Created: 00:36:47 Closed: 00:37:06
7 ways modified; 7 new ways(duplicates)
Only 7 out 10 modified ways were re-uploaded. Each of the modified ways 
had the same tags & geometry as the previous changeset but still had 
their version number increased by 1.
The 3 ways from the first changeset that were not re-uploaded:
https://pewu.github.io/osm-history/#/way/569219081
https://pewu.github.io/osm-history/#/way/568979084
https://pewu.github.io/osm-history/#/way/214527867


3.) Potlach 2 created these duplicate ways
Changeset: 75951029
original: https://pewu.github.io/osm-history/#/way/737004106
duplicate: https://pewu.github.io/osm-history/#/way/737376779

Changeset: 76010386
original: https://pewu.github.io/osm-history/#/way/737004108
duplicate: https://pewu.github.io/osm-history/#/way/737360688
In these 2 changesets the mapper moved the pre-existing swimming pools 
but they were uploaded as duplicates.

Does this provide any more clues about what is going on? One possible 
conundrum for using a bot to remove duplicates is that some duplicates I 
found ( but not referenced above) were later added to relations, so a 
bot would need to pick the duplicate not in a relation to delete.

-Wayne


-- 
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus