suggestions

Hi Brian,

I'm sorry I didn't get back to you sooner!

Mmmm, this is very interesting.  Seems like it is best to constrain the
ball one more cell inside the boundary; that would have less of an effect 
on the players.  I also like your idea regarding TIME_OUT, I'll apply
those to the next release.

Regrettably, I'm going to have to postpone the International Contest.
My Ph.D. advisor is on me to crank out some more work, and he knows about
ASCII soccer!!!!!   

I still welcome your reports and will slowly apply many of your changes.

How about if you provide an amiga release for the ASCII soccer homepage?
You'll be world famous!  Does amiga have tar?  Let me know.  If you
can pack it up, I'll stick it on the page.


--Tucker


> I wrote to you yesterday about strange behaviour in soccer, near the top and
> bottom of the pitch.
> 
> I have found the problem, and I think it is a bug. However, it is a very long
> standing bug and fixing it might change the performance of some teams.
> 	(In a very small way. Maybe)
> 
> The situation is this: if the ball comes to rest against the top or bottom edge
> (i.e. y is min or max legal value) and a player approaches it along the adjacent
> row, but makes no attempt to move it or kick it (i.e. the player is trying to
> get directly behind the ball before moving or kicking it) then that player may
> be in contact with the ball for several cycles. Your current code which spots
> stalemates is oversensitive to this behaviour. It does not distinguish between
> failed attempts to move the ball and manouvering around it. The presence of
> _any_ player close to the ball, when the ball remains stationary, increments
> 'count'. This happens for _every_ player on _every_ team. So, from one point of
> view, the value for TIME_LIMIT at 20 is simply too small.  
> 
> 	This can also happen if a player manouvers around a ball which is in the
> 	middle of the pitch. The effect is that the ball jumps just before it is
> 	kicked - sometimes completely out of range of the kicker!
> 
> One fix is to change TIME_LIMIT to 30 or 40. However, this causes 'real'
> stalemates to linger longer than is sensible, and may cause unnecessary
> clustering of players at the ball.
> 
> I have 'fixed' the behaviour in my version by incrementing count in a different
> place - well, 2 places, to be exact. In this version 'count' is used to count
> only ATTEMPTS TO MOVE THE BALL, and I have set TIME_LIMIT at 8:
> 
> 	I increment count whenever any player legitimately attempts to kick the
> 	ball.
> 
> 	I increment count whenever any player move would take it through a
> 	location occupied by the ball (regardless of what is on the other side
> 	of the ball)
> 
> This gives a smoothly flowing game, but has the effect that the ball follows the
> sidewalls rather more than it used to. In fact, once the ball hits the side it
> can only get into the middle again by getting trapped between two players.
> 
> This is a shame, because the occasional unexpected movement of the ball from the
> side into the middle was much more realistic and interesting!!!!
> On the other hand, the 'random' calls to nudge() while a player was manouvering
> around a ball in the middle of the pitch has completely scuppered one of my
> strategies.
> 
> How would you like to play this?
> 
> 	I would almost like to suggest allowing players (but not the ball) to
> 	stray into the sidelines for the purposes of kicking the ball out into
> 	the middle. This would also allow more realistic defending if we applied
> 	the same logic at the goal-line. However, this would be a substantial
> 	change, I think.
> 
> 
> brian
> 
> 


-- 
-------------------------------------------------------------------------------
Tucker Balch, PhD Student, CoC Office 123A                        home 350-0520
tucker@cc.gatech.edu, http://www.cc.gatech.edu/grads/b/Tucker.Balch fx 894-9846
Mobile Robot Laboratory, Georgia Institute of Technology         (404) 894-5613
-------------------------------------------------------------------------------