Debugging shape's incorrect first point

[RuiLoureiro]

I'm having a hard time understanding why some shapes have incorrectly snapped points. A lot of the errors seem to be in the first and last points of the trips.

Here is an example:
In the next plot, the blue line is the original GTFS shape points, and the red circle is the corresponding trip's first stop location.
The green line is the shape returned by pfaedle and the yellow circle is the corresponding trip's first stop location.

I was expecting the shape to route through the street Rua Quirino da Fonseca, like is shown here, in OSM:
https://www.openstreetmap.org/directions?engine=fossgis_osrm_car&route=38.73451%2C-9.13502%3B38.73729%2C-9.13377

I have used the -T flag to export the geojson graph and tried to change a few parameters on pfaedle.cfg but am having a hard time figuring out what's causing this behavior.

Here's a plot of the graph.json around that area:

Any help will be greatly appreciated

[marti1959it]

I have a similar problem.
I launch pfaedle with just the -D, -x, -i arguments, using a freshly updated italy-latest.osm; pfaedle.cfg is fully original, untouched.
Here's an example for a route.

This is the beginning of the route:

...and this is is the end of the route:

(The background map is OSM derived).

It looks like the first and the last points are "missed" from pfaedle.

Any help will be greatly appreciated.

Roberto

[patrickbr]

@RuiLoureiro That is strange, indeed. The Rua Quirino da Fonseca also has a bus relation, so it should definitely be chosen. Could you send me the GTFS file and the trip ID, to that I can reproduce this using -T?

@marti1959it Could you do the same?

Many thanks and sorry for the late answer.

[marti1959it]

Patrick,
thank you very much for the quick answer!
To keep things simple, here is a small GTFS including just one route (see my screenshots), just pick up any trip.
RL-TEST-gtfs-20201026-20201231-noArFer-route-170-new1.zip.
If you prefer I'll send you the full dataset (about 6MB).
Roberto

[RuiLoureiro]

I'm sorry for the late reply. I eventually figured out that the problem above was related to the GTFS feed and got it fixed.
However, I have a bunch of other errors that are very similar to the one described by @marti1959it.

Here's one example, with the gtfs' shape in blue, pfaedle's shape in green and the gtfs location of the serviced stops in red:

OSM for that area:

I was expecting for pfaedle to route through the 'Cais do Sodré' way, where the OSM and GTFS stops are located.

The trip_id is 25906598
Here's the gtfs:
gtfs_carris_02_2020.zip

Thank you for your attention