Detecting some very wrong "Impossible angle in a highway" problems

[naoliv]

Take a look at the flagged features from https://osmcha.mapbox.com/changesets/50705989/

In one of them we have this:

Which is this way https://www.openstreetmap.org/way/511591015

There is really nothing wrong with it, while osmcha says that it has an "impossible angle" This same problem can be seen with other ways and in other changesets.

One problem that it causes is when people blindly believes in what osmcha says and marks the changeset as "bad", without opening and verifying the object in JOSM.

[vgeorge]

It seems the same problem is happening here:

https://osmcha.mapbox.com/changesets/66565198/

Can someone provide a link to the source code that evaluates this type of error? I couldn't find it by searching "Impossible angle in a highway" on OSMCha related repos.

[willemarcel]

@vgeorge The impossible angle is on that way: https://www.openstreetmap.org/way/665347283

It's common that people map road links that way and it generates some alerts that are, on some way, a false positive, however I yet think that is a useful alert, as it can detect some accidental dragged nodes.

The code is on: https://github.com/mapbox/osm-compare/blob/master/lib/impossible_angle.js

[vgeorge]

To avoid false positives the test could verify if the angle vertex node is part of another way. Is that possible to do in osm-compare?

[vgeorge]

My issue is not a problem with osmcha-frontend and doesn't seem to be related to the one pointed by @naoliv and is being discussed here:

https://github.com/mapbox/osm-compare/issues/213