[0.17.14] Train signals with a gap of two tiles between them behave differently if they're offset by one tile.

[knightelite]

I was testing ways to detect which branch of an intersection a train came from, and I noticed this inconsistency in signal behavior depending on where a signal is placed.

Scenario 1, signals are one tile apart:


The leftmost signal in both cases is yellow; behavior is consistent.

Scenario 2, signals are three tiles apart:


The leftmost signal in both cases is red, behavior is consistent.

Scenario 3, signals are two tiles apart:


In the top case where both rail signals turn red, the rightmost rail signal is aligned with the "outgoing" part of the rail segment. In the bottom case (where the left signal is yellow, and right is red) the rightmost signal is aligned with the "incoming" part of the rail segment.

What I expect: Behavior to be the same with equally spaced signals.

[TruePikachu]

Actually, the behaviour is consistent when you consider the location of the signal with respect to the train itself. It is also considered NaB that the train stops in a different location "relative" to the red signal in both instances; the train is stopping itself from entering the tile of track that exists just behind the red signal (recall that rail tiles are 2x2 large; the little red circle lies on the boundary just to the left of the tile the train isn't permitted to enter).

[knightelite]

I agree that this one may be intended behavior, and I also figured it was the difference on whether the train was stopping before or in the segment of rail.


Looking more closely at my screenshots here, you're right that one can actually see that the train braking point is past the rightmost signal on the top (so the far edge of the rail segment), while it is on the left of the rightmost signal on the bottom. This means even though the signals are offset by one tile, the train on the top actually stops in line with the train on the bottom which has its signal set one tile further forward.