Train pathing penalty is incorrect if there are 2 or more trains in a segment.

[jgilmore42]

I think this is a minor issue. The workaround is simple, and figuring out what's going on isn't too difficult. But it's still a bug, so I figured I'd report it.
[0.17.79] (Build 47865, linux64, steam)

I've been attempting to recreate in survival mode the 10k-spm-cross-megabase, which is very heavily train-focused, and (like many "train waiting yard" scenarios) uses parked trains, without fuel and in manual mode, to force trains to path into the waiting area even if there aren't any "real destination" train stops that are ready to load/unload/whatever. This works, but if you try to place two trains in the false parking area instead of one, suddenly trains won't path to the correct destination anymore even when one is available, as the train stop with two parked trains is considered "closer". This can be resolved by either removing a train (why was I trying to double-check-make-extra-sure anyway?) or by placing a rail signal (of any type) between the two trains.

To reproduce:
1. Create a long straight rail. (doesn't have to be too long, mine were about 147 tiles)
2. On one end, place a train, and give it fuel.
3. create a spur rail going off to one side a short distance, so the train has to make a choice.
4. guard that intersection with signals.
5. place two stations, each with the same name, one on the spur and one at the far end of the long rail.
6. place two trains (NOT connected to eachother) on the spur. They can be lone cargo cars, or engines, doesn't matter.
7. observe as the signal protecting the spur correctly turns red.
8. tell the train placed in step 2 to go to the station.
9. Notice that the train paths to the spur station, even though there are two trains parked on it, in preference to the further station that's actually available.

The included screenshot includes the train waiting area, and the false stops, but doesn't include the further away true destinations.

Note that each false stop has two trains (engine and cargo on left, two cargos on the right) with a signal between them. If I remove that signal, all those waiting trains will just sit there even when there's stops available.

I'm fairly certain that this bug affects trains in automatic as well, but since such trains either never end up in the same segment, or aggressively destroy eachother if manually forced into the same segment, there's no real way for me to test that....

[DaleStan]

9 is almost certainly because, even with the various pathing penalties, the false station is closer than the true station. What problem are you trying to solve with these false stations?

[Loewchen]

Please post a minimalistic save file to produce the issue.

[disentius]

Not sure how OP uses them. I use them when i have several unload stations with the same name and configure them to disable when a train is present. Prevents no path error and keeps luring loaded ore trains into the stacker when all unload stations are occupied.
I make sure that trains change path the moment an unload station becomes free by guarding them with a triple signals connected by wire: the first one sets the back two to red, this gives the station a penalty of 2000 tiles.

[PyroFire]

Please post a minimalistic save file to produce the issue.

I have been able to replicate the issue.



And it does not matter which one you remove.



The type of signal used before the closer station (with the two trains in it) does not matter either.



Nor does it matter if a train is already parked in the closer station.
It just wants to get in there if there's 2 trains in the block.

[disentius]

I think you are right.
Wiki says:
When the path includes a manually controlled stopped train without a passenger -> Add a penalty of 7000.
And:
Before a train moves to a target (in this case, a Train stop), it calculates the best route based on the railway network at that time.

By these rules the train should never path to the north station.

Nice find:)

[boskid]

Thanks for the report. It is fixed for next release.