rabbitoid: The solution is very simple - the site shouldn't set the timeout of games according to vacation days like now but according to their original time settings (much simpler too).

So, if you have the next month set as vacation days and your opponent moves on a 3 days per move game, the deadline would still be set as in 3 days for now (instead of in 1 month as it is now). However, in 3 days from now, when the game was about to timeout, the system would notice that it was a day marked as vacation day and add another 24 hours (this is already working like this now, I think). And it would keep doing that until either a non-vacation day was reached (timeout) or you moved (business as usual).
Naturally, you shouldn't be able to mark the current day as a vacation day (which you can't already).

So, the site already has everything it needs to fight this problem. This would be one of those rare events where to solve it, all it takes is actually de-implementing something

pauloaguia: That's not how I would do it. What I would do is: if one times out on a game, and the player has set that date for a vacation day, or has autovacation set (and still has vacation days set), I'd add 24 hours to the clock on every game the player has running, regardless whether it's his/her turn. Then subtract 1 from his number of vacation days for the year. You only need to do this at most once every 24 hours. After a player moves, check whether his clock is above the max. time that the game setting allows (this may happen after a player was on vacation while it wasn't his move) - and set the clock to the max if over.

The key point is that 24 hours only get added to the clock when the number of vacation days in decreased.

AbigailII: I was just commenting on current site's implementation (as I think it is. I'm not really sure, for all I know it can be as you described already). I aggree that setting it for all the games is probably more efficient, my focus was on not having to implement anything different (but, like I said, for all I know it can already work as you described)

Why subtract 1 vacation day? I suppose you're thinking about autovacation, but pawns have no such feature, they need to set their vacation days manually (and so can paying members). I can't see why a vacation day should be subtracted if one is already set. Activating autovac and actually using vacation days are two different features right now (and will probably remain so)

EDIT: On second though, I don't aggree. The site should only add 24 hours to the games that would expire on the same day, not to all games. Otherwise, we'll get back to the same problem (a 30 days per move game has no need to geting extra 24 hours daily from almost 1 month in advance). Also, if it's not the player's turn, the time settings shouldn't be changed - otherwise I could have extra time added to my games because my opponent was on vacation, even if I'm out of vacation days already).

pauloaguia: Also, if it's not the player's turn, the time settings shouldn't be changed

Of course they should! Suppose you have 2 games running. In one it's my move, and you have 5 hours left (assume Fisher clock). The other times out at noon. Now, you want 24 hours on both your clocks, otherwise, I make a move at 1 PM, and you'll time out at 6 PM. Since you've already spend your vacation day this day, you would lose the game. That's why all clocks should get 24 hours in my scheme, not just the ones where it's your move.

AbigailII: Fischer clock settings means no vacation days on this site. So I'll assume it's not a Fischer clock game
The fact that you already spend your vacation day is not a problem - it's set on your account as a vacation day. Any other game that would otherwise timeout during that day will also get extended time when the moment arrives without wasting another vacation day, this is already how it works today.

nodnarbo: After a site's downtime no time checking functions are running on the site, if I understand correctly. That's why games go into negative time - there's no function running that will make them timeout in the background. However, if you allow games to go into negative times in normal conditions (while all these processes are executing) changes could indeed be profound.


Again, my main point was that NO CHANGES in the timeout or vacation days mechanisms need to be done. The only change would be the calculation of the deadline for timeout. Everything else would continue to work as currently. But, of course, only Fencer can say for sure if my reasoning is correct or not - he's the master of all time around here

pauloaguia: What if, instead of adding time to games the games just continue into negative time, but they aren't checked for time out if a vacation day is activated. This is also already implemented on the site with the case of site downtime, and wouldn't think it would take a huge coding change.