Route snapper refuses to use a specific edge when district drawing

[TFCx]

In default Montpellier Geojson, when defining a district, I can't use the whole Avenue de la Justice de Castelnau as a limit. The cause may be due to a topologically identical trajectory ? (edge-case of the route planner ?)

Both edge E783 and E1009 links the same node (N957, N958).
Route planner wants to use E1009 and never E783.

[TFCx]

For each double-edge (2 edges connecting the same 2 nodes), should there be an inserted mid-node that can be used for district editing ?

[dabreegster]

Ooh, good catch. As you identified, it's because edges are identified by node pairs, but this isn't sufficient in this case. It's a bug in the other library; I'll take a look.

should there be an inserted mid-node that can be used for district editing ?

Could be useful. I had some experiments to also snap to the outline of big waterway polygons, but it doesn't work well, since the lines along the polygon are so long. dabreegster/route_snapper#18 would help

[dabreegster]

Please try again. Still the default snapping can't handle two edges linking the same nodes:

But now if you make a few extra waypoints and click one to turn it blue and drag it anywhere, you can work around it:

Resulting in:

(I also made style changes recently to try and match the old tool; let me know what you think of them!)

[TFCx]

Yes it works :)

IIRC, in LTN v1, there was a way to display the entrances/exit of the drawn district. Is there something similar on LTN v2? The thing that doesn't seem to me intuitive is: should I generally include the road circling my district or not? And, in the case of a double road circling my district, what should i do ?

[dabreegster]

#68 for the arrows showing entrances/exits.

I don't have great advice yet on handling the roads circling your district. There's a new checkbox to "Include perimeter roads" -- it lets you try including these or not in the analysis. I'm not sure the shortcuts calculation makes sense with it.