Savegame & pak folder to demonstrate this bug: https://dl.dropboxusercontent.com/u/61716/stationsignals.rar
Observe the furthest right platform at Stevenage, and press 'B' to show track reservations.
Notice how the following train reserves a path into the platform before the other has left -- seemingly reserving a path which it should not be able to. The train should not be able to get the all-clear from the signal it is waiting at.
Although this is relatively harmless here (though of course it should not be allowed), it causes havoc when a train is trying to reverse in a through platform and another comes in behind it where it should not. This will lead to deadlock.
This does not occur in 10.27.
Ah, this is that bug I was mentioning the other day, I see! Good. Now I won't have to try to get a reliable recreation myself. I've had it cause some major jams.
Thank you for the report. This is not quite what it seems at first glance: the signals at the end of the platform are misleading. This is because, at present in Simutrans, where a train stops at a station, a signal is implied in the station tile at which the front of the train stops, quite apart from any actual signals that might be present. So, when the first train in the right hand platform clears the section, the section that it is clearing is not the section to the signal that we can see at the end of the platform, but the section to the implicit signal one tile before the end of the platform.
On top of this, there was a bug where trains would clear their reservations when the first carriage passed over the relevant parts of track, rather than when the last carriage passed over them. I have now fixed this bug: thank you for the report.
It would be good one day to change the implicit signal behaviour, which can be confusing, but it would require quite a complicated re-write of the signalling code, and there are more pressing considerations.
Hi James, thanks for the clarification. I had assumed that the implicit signal was at the end of the platform -- but of course until recently, "the station tile at which the front of the train stops" always coincided with the end of the platform.
Thanks also for the bug fix!