I have run several tests.
Operating system: Win XP Pro; 2 GB memory (I just tested it for errors)
BfW had been run in the debug mode with the -d switch.
Sometimes I wasn't able to reproduce the crash even with an identical scenario, but I always found ways how to reproduce it again.
After tests I reported in my last posts, the original scenario and the container 'unit_spawns' still crashed the GI (Gamestate Inspector).
Then I created a tiny scenario "minimum WML code" (it is in the attachment, in a directory 'Scenarios'), that just created the container 'unit_spawns' - and now the GI didn't crash. So a given variable content can cause the crash or not.
Then I removed all scenarios and all add-ons, except the one that crashed before ("full scenario" in the attachment), and it didn't crash now. So a given scenario can cause the crash or not.
Then I returned to the original state of BfW (I moved the addons back) and now the previously 'bad' scenario stopped to crash at all. So a given state of BfW can cause the crash or not.
Most tests were done with the resolution 1600x1144x32. Resolutions 1044x749x32 and
FullScreen (800*600?) usually didn't crash GI.
It appears that a higher chance to a crash cause
- a higher resolution
- a larger container that is inspected
- more WML installed
- more programs that run in the system besides Wesnoth
- rather "capitol mode" than the simpler "all-villages mode" (maybe because "capitol mode" produces a bit bigger container 'unit_spawns' that crashes the GI)
Maybe the crash is related to some incorrect memory management? Like this bug, which is reported by more players: Very strange problem with many add ons.Please try to help