James, thanks for testing. I know these crashed are hard to debug, since, well, they are random. Anyway, no problem, we'll see how 9.0 holds up :)
As to the teleporting trains: AEO is right. The consist will move forward a few tiles (depending on consist length, I'd say about half it's own length). This can be particularly annoying when the departing train p****es a red signal this way, or when it starts blocking the path of an incoming train.
By the way: this issue is also present in Standard (occurs frequently in my old game Standard 102.2/pakBritain).
Mod note: split from this (http://forum.simutrans.com/index.php?topic=5976.new;topicseen#new) thread in the Experimental board, as this problem is reproducible in Standard.
Might I suggest that the teleporting trains issue be posted on the Standard board?
Please, first test whether this appears in a standard-nightly :)
I read this message, as the bug appears in the stable release.
@Brambo: If this bug happens in standard nightlies, could you upload this savegame please?
Teleporting still occurs in r3851. See save:
http://www.megaupload.com/?d=JQPWCC4T
One convoy that I've seen teleporting a lot is (65) BR Cl**** 165 "Network Turbo" DMU (nice to be able to ctrl-c this, btw ;-)), running the Bridgewater - Kinross IC service. Note that this train is not always teleporting, but it appears to be more likely at some stations (Cardiff, Reading). I had suspected this might be a pakBritain issue, and to be related to this: http://forum.simutrans.com/index.php?topic=3376.0, but reading it again, I'm not so sure.
thank you, the savegame helped. Should be fixed in rev 3852.
Thanks Dwachs! It seems to work in the new nightly. Out of curiosity: what was the problem here?
James, will this also be incorporated in experimental?
This will be incorporated into the next version when I merge the latest changes from Standard.
@Brambo: There is a routine in the code that checks whether a train is corrupted. Then it moves the train some tiles forwards and starts the train. That procedure ensures that the train is fully on track again. This routine had an error that was triggered in your situation, which makes the train moving forward an entire train length. The problem was that for all length calculations the length of the last vehicle is not taken into account, but this routine was different in this aspect.