Hello,
I'm trying to move the sandbox servers, but I have problems with british savegames. The swedish pak game works fine. But I get segfaults, both on server and client. Commit 8ad124ca88c729fd406f4bd5d66ce7bb727e9824.
I struggle to get a consistent result. Some games run fine locally and crash only when connected online, some do not start at all.
http://list.extended.simutrans.org/~vladki/debug-saves/client5-network.sve
http://list.extended.simutrans.org/~vladki/debug-saves/server13353-network.sve
Thank you for letting me know. I had somebody else report something similar on Github that occurred only when the ~/simutrans directory was deleted. Can you check whether this is the cause?
I have fresh ~/simutrans on the server, extracted from http://bridgewater-brunel.me.uk/downloads/nightly/packages/Simutrans-Extended-Complete.zip
This is odd. May I ask -does this occur on all or only some paksets; does it occur on all or only some saved games? Does it crash (1) only in network mode; (2) only when a server; or (3) even in single player mode?
Pak sweden seems to be ok in both single player and network.
Pak britain is crashing even in single player. I took the demo game, did a few fixes, saved, did more stuff and crashed. Start again, load, OK. Move over the map, load again the same game and I get crash... Start load, crash. Start load, OK... WTF???
Does this crash with a saved game other than the demo game? That game has some corrupted buildings (their type changed from a 1 tile building to a 2 tile building at some point after the game was created, so they now overlap other buildings). I have known this to cause memory corruption in the past.
Yes it crashes also with my test game "Trnava" the one I used for reporting bugs about signalling, and also with one of the old big maps "bridgewter-brunel copy".
The fixes I spoke about were deleting the department stores, that went from 1x1 to 1x3 tiles.
Sorry for double post...
I tried to connect to James' bridgewater brunel game, and got the same error as in $subj, just with different number.
I don't know if it killed the server or not. It is running now, but I think there is an automatic restart...
I have had some trouble reproducing this. However, I found a memory management error (I am not sure whether it is the same error), which I have corrected on the master branch. Would you be able to re-test this? I should be most grateful.
Seems fixed
Excellent, thank you for checking.