R24409: Soldier moves to the same square as civilian

[Kaz]

While testing another bug, I found this one. As I started he Dam map, I moved a soldier in the first turn to square (136,137,0) as shown under debug_map 3. As the soldier entes the building the square shows to have been occupied by a civilian, yet the soldier keeps moving to that square, until at the end of the motion, both the soldier and the civilian are standing in the same square, as shown in the first screenshot. Bad as this is, if you leave the soldier at that spot and end the turn, a little while later (presumably as the civilian's tries to move or something else) the game crashes claiming a failed assertion in file src\game\g_ai.c, line 516. It is my idea that the check for "square occupied" shows unoccupied for squares the player does not see, and is not rechecked while the soldier is moving.

Cheers!

[attachment deleted by admin]

[odie]

While testing another bug, I found this one. As I started he Dam map, I moved a soldier in the first turn to square (136,137,0) as shown under debug_map 3. As the soldier entes the building the square shows to have been occupied by a civilian, yet the soldier keeps moving to that square, until at the end of the motion, both the soldier and the civilian are standing in the same square, as shown in the first screenshot. Bad as this is, if you leave the soldier at that spot and end the turn, a little while later (presumably as the civilian's tries to move or something else) the game crashes claiming a failed assertion in file src\game\g_ai.c, line 516. It is my idea that the check for "square occupied" shows unoccupied for squares the player does not see, and is not rechecked while the soldier is moving.

Cheers!

Hi Kaz,

Whilst looking at this entry, i cant help but wonder about:
1) The dev revision of the svn u r using? (Revision nos, or a built by one of the uploaders?)
2) Your current operating system? (Xp, ubuntu, mac, version and Service pack?)
3) Was this recreatable?

Maybe it would help to provide a backtrace too? Or a savegame to this?

[Kaz]

Hi Kaz,

Whilst looking at this entry, i cant help but wonder about:
1) The dev revision of the svn u r using? (Revision nos, or a built by one of the uploaders?)
2) Your current operating system? (Xp, ubuntu, mac, version and Service pack?)
3) Was this recreatable?

Maybe it would help to provide a backtrace too? Or a savegame to this?

Hi Odie,
1) As in the topic's subject, SVN revision 24409.
2) Windows Vista SP1.
3) Like clockwork. I just selected the soldier closest to the building, disabled reaction fire (to have all TU available) and set him to move to the specified square.

The game was started in skirmish mode. Since you can't save in the middle of a battle, I can't provide a savegame.

Btw, a quick inspection of the code reveals that the failed assert that crashes the game seems to be checking that a distance not be zero, which given that it's the distance between the soldier and the civilian, seems to be that.

Cheers!

[Duke]

Kaz,
did you get that crash with or without Mattn's patch from the 'out of sync'-thread ? The patch seems to be about visibility...

[Kaz]

@Duke,
Without the patch. I reverted it once it did nothing for the other bug.

Cheers!

[Mattn]

i would also be interested in the information whether it's in your round or in the aliens/civilians round (your actors are moving, or ai controlled actors)

there's a cvar cl_logevent (or cl_logevents) which creates an event log file in your home dir - it would be interesting which was the last event when this bug occurs.

[Kaz]

It's during my turn that I move my soldier to where the civilian is. Then, once I end my turn, it will move a few aliens and then crash. Anyway, I think the ctash is totally normal, given that the soldier should not have moved to the square occupied by the civilian in the first place.

Cheers!

[odie]

It's during my turn that I move my soldier to where the civilian is. Then, once I end my turn, it will move a few aliens and then crash. Anyway, I think the ctash is totally normal, given that the soldier should not have moved to the square occupied by the civilian in the first place.

Cheers!

Hmmm, aniwae, i recompiled a new version today, with a clean rebuilt of all the maps (took me abt 4 hrs plus i estimate w the upload.....).

U might want to keep a lookout for that if u r not self-compiling the CB..... Its only @ 22% and est another 1.5 hrs...... Cheers!

[Kaz]

I update from svn and compile everything myself. Last update yesterday and the issue is still there.

Cheers!

[Duke]

Kaz,
would you please try Mattn's patch and see if this problem persists ?