It works the same in multiplayer. If ennemy ask a tank to kill your mex, your game doesnt "receive" the information about that tank coordinates, firing pattern or anything like that. You just receive the information that other player selected the tank and then right clicked on your mex to attack it. Then your computer calculate what happens next. The "magic" is that the exact same thing always happens with the exact same starting context. Otherwise it desyncs and game is over since everyone is playing different versions of the same game. (think parallele universes where anything can happen ^^)
The only viable option is to allow external player connection to the P2P network of players in the game. > huge problems already, do they need to vote to take you back, how do you reach them, how do you maintain a list of active games, etc etc
Then everyone would have to be automatically paused. > not really cool for them
Then you have to replay all the commands since game start until you catch them back. > so if you have a crappy CPU, it can mean 40 minutes of waiting time
Then if it didn't desynced or anything like that, you are back in sync with people and game can resume...
AFAIK no multiplayer game that is based on exact simulation prediction with P2P netcode ever tried that ?
Even with source code access no one would develop that it is just too "useless", it would be better to spend efforts on improving network code in the first place.
If the replays were "seekable" somehow it would be more doable to think about that kind of feature since catching up would be feasible in a reasonable amount of time, but again better to implement better peer (re)connection code i think.
TLDR : even if FA was 100% open-source it would probably not happen
Statistics: Posted by Poch — 03 Feb 2015, 13:38
]]>