Re: Breaking matchmaking deadlocks

On Thu, Jul 8, 2010 at 10:41 AM, James Robertson <[email protected]> wrote:
> All you're looking for is
> a way to reject one connection and keep the other, with a fallback for the
> (extremely) rare instances that both times are identical.  The timestamps
> could be years apart and the mechanism would still give the required result.

If you just want a meaningless number with minimal chances of a
collision, timestamps sound like one of the worst possible choices -
people put a lot of effort into making clocks accurate and
synchronised, but you'll get more collisions as they get more
accurate. And once you get one collision, it's much more likely that
you'll get another collision when you try again a little later. An
n-bit random ID (seeded by more than just the timestamp) would have a
much lower chance of collision than an n-bit timestamp.

-- 
Philip Taylor
[email protected]
_______________________________________________
Sweng-Gamedev mailing list
[email protected]
http://lists.midnightryder.com/listinfo.cgi/sweng-gamedev-midnightryder.com