[bf1942] Linux server status report: 2003-03-31

Fredriksson, Andreas andreas.fredriksson at dice.se
Mon Mar 31 11:47:32 EST 2003


John,
 
this sounds like packets were being dropped or delayed along the way
causing the client side prediction to advance and then "snap back"
just as you say.

The same effect could easily occur with nearly anything predicted
on the client side; i.e. nearly all objects.

As I mentioned we'll be testing both the current server and the old
1.31 server to try to verify these problems but I'm fairly sure they
are related to the networking differences between win32 and linux in
the old versions.

Regards,
Andreas

-----Original Message-----
From: Stewart, John
To: 'bf1942 at icculus.org'
Cc: 'andreas.fredriksson at dice.se'
Sent: 3/31/2003 6:35 PM
Subject: RE: [bf1942] Linux server status report: 2003-03-31


> no, we haven't been able to reproduce it yet on the old server, but
> we will surely try. Focus so far has been on integration and 
> development
> rather than more extensive testing, that phase will follow soon.

FYI, the most reproducible symptom of the "bungee effect" has been for
me as an engineer. Especially on Coral Sea, for example.

You'll start repairing a ship and as the engineer's meter goes down the
ship's goes up. This seems to continue for a bit, but apparently the
feedback you get is only your client-side prediction - after a while you
see the ship's meter "jump", usually down from where the client thought
it was at, as the client-side has predicted too much repairing. It
happens every time for me when repairing, whereas other effects of the
bungee bug appear to be more intermittent (like the zoom in/zoom out
issue).
 
> We've had a lot of success with other things regarding the 
> linux port lately
> so I'm very optimistic :-)

I am too; been saying to my admins and users things should be getting
better Real Soon Now.

Good luck and thank you very much!

johnS



More information about the Bf1942 mailing list