Spring RTS Engine

Hey peoples,

I just discovered theres a bug in 0.72b1 which you can easily work around (I introduced it myself , I'm sorry).

If you're hosting and your PC is much faster then others then it can happen that you finished loading before the others even started loading. This will cause them to drop out, they won't be able to connect.

If you notice this happens, just try again but erase the pathing info for the map you're going to play (only host does this). This will make loading slower but things will work, and since you had a fast computer anyway it doesn't matter too much.

Tobi

Good you found it.

I was puzzled by this bug happening, tried to rehost million times, and also hole punching lots of times..

what is funny, is it happened to me.. well.. i have a 1.7 athlonxp...

so must be the lag of my internet, like the packets not synched or smthng.

This is not easily worked around: ive done as perscribed, but it doesnt seem to help much, and ive hosted for people with much better PCs than me.

ETA on a fix?

I was wondering what was causing this. I caused about a million specs to get kicked wrongfully.

It is very well possible some people are experiencing other problems.

This bug only occurs when you finish loading before the other players start loading. So I think it's pretty much impossible to trigger this bug when using the described workaround. You'd need to have a PC which is so fast that it can load and path within 1 round trip time to the slowest player in your game (assuming the processing time needed by Lobby to start spring.exe is negligible compared to network lag).

I think you can see whether people started loading in the .info list in the right-bottom. If people show up there, everything is fine, if they aren't in there they didn't start loading. So if they are there but drop out later, it's another unrelated bug. Also if people just can't connect because of port forwarding/firewalls and stuff they won't show up...

As for a fix, next release I'd say, I can't give ETA for that...

you've fixed this in SVN already?

yes

wd (again )