Jan Ciger wrote:
in that case the design evolved from a single player game, where the
weapon is a particle system and the individual particles cause
damage. when we needed to quickly make that a dogfight thing that
works over LAN, was simple and optimally efficient to keep the
client-side damage calculations, and currently trust is no issue
there. dunno if that is interesting also w.r.t p2p ideas.
This will work until the first player connects with a hacked client
reporting vastly increased damage and ruins the game play for everybody.
This is precisely the thing you *do not* want to be on the client or the
client has to be locked down tightly (cryptographically signed, closed
source, etc.). Even then you will face cheating - see games such as
Counterstrike where various aimbots (most of them are client hacks)
almost wiped the scene until some countermeasures appeared.
For your example about client side computation of damage, wouldn't it
be possible to compute the damage on several clients and have clients
to verify if the result they get from other clients is good ?
Player A is playing with a hacked client, it compute a damage of 1000
HP instead of 10 HP as it should be, a good way to cheat.
Player B and C are playing on good client. Those clients compute that
the damage should be 10 HP, when they get a result of 1000 HP from
client A they both know that something is wrong. Then client B and C
report client A as a cheater to the server, the server ban client A.
A mutual anti cheat system so to speak :)
Souvarine.
|
_______________________________________________
Soya-user mailing list
Soya-user@gna.org
https://mail.gna.org/listinfo/soya-user
