game out of sync

[Alan Kenwright]

Stan Shebs wrote:

> Alan Kenwright wrote:
> >
> > I'm fairly new to this, so apologies if this is naive....
> >
> > I've been trying to play Xconq-7.3.3 over a LAN between two machines
> > both running RedHat6.2 (one is a 50MHz 486 and the other is a Sparc-2,
> > so they're both a bit slow).  After a relatively small number of turns
> > (like 10) playing the Introductory game, I start getting error messages
> > about "game is out of sync".  This does not happen if I just play on one
> > machine against an AI side.
> >
> > Questions:
> >
> > Is this a known problem with network play - does anybody else see it?
>
> Yes, unfortunately it happens pretty consistently.
>
> > Is there a way to re-sync a game once it has got out of sync?  (Should
> > there be?)
>
> No way to resync, but there should be.  The "host" program has the
> official accurate state, so it ought to be possible for other programs
> to reload from it.
>
> > Should I try to save the game on both machines at the point where the
> > error occurs - is it possible to determine the problem from that?
>
> I've collected some examples already, thanks.  The common cause seems
> to be that one program is a step ahead of the other, as if the checksum
> was sent too early - in one game a unit has completed its action, but in
> the other it's scheduled to do the action, but hasn't done it yet.  But
> there's a real bug somewhere, since if the checksum is disabled, the
> programs eventually end up out of sync in more serious ways.
>
> If someone has ideas for how to debug this sort of thing, I'd
> love to hear them!
>
> > Looks like a great game.....
>
> If only for that d*mn networking... :-)
>
> Stan

Well, it's good to know that it's not just me!  One additional thought - I
tried setting up a game across the network where I asked for an AI side on one
machine to play an AI side on the other machine and that really fouls it up.
It doesn't even get the game set up properly before it gets out of sync.
Could this be a clue?  I think I read somewhere in the mailing list archive
that you had recently produced a version which had taken out the "auto-attack
behaviour" - has anyone tried playing that version over a network?  I'm
wondering whether some action the computer takes (either as AI or in
auto-attack mode) causes the problem.  I know that's a long way from a fix,
but if it helps localise the bug....


--

Alan.