Lots of false positives in an area

[emersonveenstra]

A couple weeks ago, lots of routing and structural errors started appearing in an area (Unconnected highways) and (open boundaries and rail crossings)

As far as I can tell, all of these are false positives, and i wanted to bring this up because it seems like a data issue on the backend.

[frodrigo]

As I can see the global number is issues is stable http://osmose.openstreetmap.fr/en/issues/graph.png?item=xxxx&country=usa_michigan_upper_peninsula (ignore the title of the graph)

A couple weeks ago, lots of routing and structural errors started appearing in an area

http://osmose.openstreetmap.fr/en/issues/graph.png?item=1210&country=usa_michigan_upper_peninsula

But something changed here.

It could be related to this
cefb7a9
#1760

@jocelynj I think we fixed Osmose on this but not the PBF extract

[jocelynj]

I've regenerated all extracts of Michigan, and relaunched this job here: https://buildbot.osmose.openstreetmap.fr/#/builders/975/builds/338

Should we also do something else to not use the diff-mode on the backend side?

[frodrigo]

I've regenerated all extracts of Michigan, and relaunched this job here: https://buildbot.osmose.openstreetmap.fr/#/builders/975/builds/338

Should we also do something else to not use the diff-mode on the backend side?

This analyser does not run in diff mode. So it is ok on that.

Nevertheless, the result are still the same.

@jocelynj We still have something wrong on the extract.
Here the .poly from download.fr

While on Osmose we still have and other wrong configuration :/
http://polygons.openstreetmap.fr/get_image.py?id=1582241,1582242,1801749,1801750,1898721,1907314,359765,582246,901500,901501&params=0

[jocelynj]

This time, it should be fine - I have updated the osmose-backend git clone on download.vm, updated the .poly file, and regenerated the extracts.

[frodrigo]

Markers and numbers look better now. Thank all.