[Freeciv-Dev] Re: Network bandwidth, and other things.

[Christian Knoke <chrisk@xxxxxxxx>]

On Wed, Nov 06, 2002 at 06:30:46PM +0100, Thomas Strub wrote:
> On Tue, Nov 05, 2002 at 03:54:49PM -0500, Jason Dorje Short wrote:
> > 
> > I still have no answer for how you would go about showing that freeciv 
> > was "faster" or "more responsive" under one of these patches.  But as I 
> > said, an anecdote would be a good start.  Unfortunately one needs the 
> > right network to get such an anecdote (or real numbers, for that 
> > matter), so not anyone can do it (unless they are clever and simulate a 
> > slow network).
> 
> Perhaps someone should try my pingpatch ... but changes like that are
> only visible with an offset < 1 min ...

This remembers me of another idea I had. I've thought about TCP/IP, and I
may be completely wrong, so tell me. But listen:

What happens with TCP/IP, when a single packet got lost? I think, in the
first place, _nothing_ happens. Neither of both endpoints of the connection
gets aware of this (But I may be wrong here!). Only after another packet is
sent in the same direction, or a specific amount of time passes, the
receiver gets notified of this and resends the packet.

So: What happens when I try to move a unit (cursor key) and the packet gets
lost? I wait a long time, getting nervous, losing needed time to fight the
battle. That has actually happened to me. Finally, the TCP/IP layer gets
aware that a packet is missing and resends it. Civserver answers, the unit 
moves on the screen, the battle's lost. Shit happens.

What can we do? Just send a ping every second (if nothing else is sent
within a second). TCP/IP notices the packet loss when the ping arrives, and
retransmit of the lost packet is initiated. This should keep the delay below
2 seconds. - - And now the flames, please :)

Christian

-- 
Christian Knoke     * * *      http://www.enter.de/~c.knoke/
* * * * * * * * *  Ceterum censeo Microsoft esse dividendum.