Ah well, I feared as much! I'll settle for waving my fists at IGS in impotent rage. :-) Thanks, Michael ----- Original Message ----- From: "Peter Strempel" <zotan@xxxxxx> To: <ggo-discussion@xxxxxxxxxxxxx> Sent: Tuesday, January 14, 2003 9:26 PM Subject: [ggo-discussion] Re: Argh! IGS lag -> loss on time. Any workaround? > > On Mon, Jan 13, 2003 at 05:15:02PM -0000, Michael Camacho wrote: > > > However... if the lagged player's byoyomi time expires and he still has > > moves to play, IGS seems happy to resign him without waiting to see if there > > is any lag. > > > > At least, that is how it seems to me. Is this what IGS is actually doing, > > and is there a way around it? I don't know if gGo has anything to do with > > this, but it seems unlikely. > > I fear that is how IGS is actually handling this and there is nothing the > client can do. > > When your time runs out, you don't lose instantly. The loss occurs when > either you make your move or your opponent does a 'refresh'. However, I > don't know if IGS would take the netlag compensation into account here. > > Example: > > 3 seconds left with 2 stones. > Your real thinking time is 2 seconds. So gGo sends out the move with the > 2 seconds tag. > Now assume you have netlag and the move takes 5 seconds to travel. > So when the move arrives at IGS, your server-side clock is at -4 seconds (3 > seconds left - 2 for real thinking time - 5 for netlag). > Now IGS *might* (or might not) take your netlag compensation into account > and notice your clock should in fact not be at -4 seconds but at +1 seconds. > Simply depends if IGS checks for (time_left <= 0) before or after the netlag > time correction. I don't know those details. Sounds difficult to find out by > testing. :) > > So far the theory. In practice, the client cannot do anything about that. > gGo uses the netlag compensation feature as the IGS protocol offers, but > what exactly the server finally is doing with the info is not the business > of the client. So even if IGS might handle this not correct, there is > nothing I can do about it. > > Peter > > >