[2.0.10] Signal splits block on wrong side of crossing
Posted: Fri Oct 25, 2024 3:45 pm
by Bocian
Sometimes a signal will split a rail block in an incorrect position, offset by a few tiles. It can lead issues like this, where the split happens on the opposite side of the crossing to where the signal was placed.
Re: [2.0.10] Signal splits block on wrong side of crossing
Re: [2.0.10] Signal splits block on wrong side of crossing
Posted: Tue Nov 12, 2024 9:09 pm
by boskid
I can tell you entire history of why this signal spot exists to conclude why this is a minor/wont fix.
During 2.0 development we tried if its possible to build a 2-gap T-junction and it was possible, however it was only possible to correctly put signals if the rail system was left hand drive. This made one of devs angry and alternative signal spots were added to allow for correct T-junction build:
11-12-2024, 22-04-37.png (675.55 KiB) Viewed 339 times
Here you may notice this exact case where signal is built slightly far from the ralted rail end:
11-12-2024, 22-05-33.png (251.95 KiB) Viewed 339 times
This is why this specific rail has alternative signal spots that are binding to the same rail end:
11-12-2024, 22-06-13.png (370.93 KiB) Viewed 339 times
Problem is that this signal spot is far enough from one of its ends that it is possible to build rails between a signal and rail piece end.
I am not willing to add bits of logic to make this signal decide to bind to the other end of a rail as in this case both ends would still remain valid and i would have to add some extra collision condition to prevent it from selecting split on the "wrong" side.
It is slightly annoying given that it breaks one of the expectations but i think this outweights the benefits of having this extra signal spot available and as such i am going to move this to Minor issues.
Re: [2.0.10] Signal splits block on wrong side of crossing
Posted: Fri Nov 22, 2024 6:23 am
by Abex
This leads to very confusing situations, like the following blueprint where the short curved rail causes the signal to become invalid despite being visually quite far away.
2024-11-21_23-09-28.png (185.14 KiB) Viewed 242 times
This is especially hard to debug when you place down a bunch of T junctions that overlap each other and they suddenly all become invalid.
2024-11-21_23-11-50.png (1003.28 KiB) Viewed 242 times
Ideally it would split the segments where the signal actually lives, but failing that, it would be nice if there was a bit more feedback on how the rail is failing to split. The "sensor" across the track disappears on the problematic signals, but they all show the same in the signal planner and have the same "Cannot divide segments" message. It would be nice if the problematic ones more explicitly said that their split point was covered by the same track hitbox on both sides, and/or the ones that were able to split indicated that in the signal planner.
