.resync
Moderator: Moderators
.resync
I just finished playing a great game. Desynced...
Could there be some kind of .resync command that resends all the relevant data from the host's master copy to all players involved? If it took 5 minutes, or even 10 minutes, people would still wait. It's worth it on a good game.
Could there be some kind of .resync command that resends all the relevant data from the host's master copy to all players involved? If it took 5 minutes, or even 10 minutes, people would still wait. It's worth it on a good game.
- SwiftSpear
- Classic Community Lead
- Posts: 7287
- Joined: 12 Aug 2005, 09:29
This has been in research since before SJ left the lead position in spring. The unfortunate reality is that the scale of spring is such that transmitting the game state in one block would take up to a minute of download time from the host to each client(if not significantly more) and require a massive amount of proprietary code that doesn't exist right now. Realistically a more efficient system is needed, and that system has yet to be figured out. There's a few options that may work that have been discussed, but nothing concrete yet.
Suffice to say that this issue is well known about, and it's not being ignored for any other reason that a fix at this point in time isn't feasible based on the way the engine handles multi play.
Suffice to say that this issue is well known about, and it's not being ignored for any other reason that a fix at this point in time isn't feasible based on the way the engine handles multi play.
tbh it's out of the research phase already, it just is a lot of work to implement properly, and not quite interesting work either...
And the bandwidth usage of resync probably wouldn't be *that* bad either, we'd probably use some (multidimensional) checksumming matrix/vector thing (as the syncdebugger and some code in branches/tvo1 already uses IIRC) to determine which units/which areas of heightmap to resync, so not the entire gamestate including multiple heightmaps has to be send over...
And the bandwidth usage of resync probably wouldn't be *that* bad either, we'd probably use some (multidimensional) checksumming matrix/vector thing (as the syncdebugger and some code in branches/tvo1 already uses IIRC) to determine which units/which areas of heightmap to resync, so not the entire gamestate including multiple heightmaps has to be send over...
- Drone_Fragger
- Posts: 1341
- Joined: 04 Dec 2005, 15:49
read the threads! This discussion has been held over and over again, and loads of ppl put valuable time in it. Asking them, to explain it every time again and again, certainly won't help fixing it.TradeMark wrote:I want explanation what causes sync errors.
Does anyone even know it actually? Is it cause of lost data packets or what?
So, maybe it's best to make a sticky for sync problems, containing some information, and how to file reports?