Re: Path finding

[Eric McDonald <mcdonald at phy dot cmich dot edu>]

Hi Brian,

On Tue, 27 Apr 2004, Brian Dunn wrote:

> A bit of a problem since the path-finding stuff went in a month 
>or two ago:

His Astar implementation has been in the source tree since the 
closing months of last year. We started earnestly debating about 
withdrawing it about a month or two ago.

>where armies may cross.  This makes the new pathfinding code seem 
>to take much much longer per turn to think things through.  
>Before that path finding change, the turns went quite a bit 
>faster.

This is symptomatic of all the games when using the new 
pathfinding code, though some feel the pain worse than others. 
However, in theory, it should do a better job at actually finding 
a path.

> Another thing is that the standard game doesn't make or use 
>transports.  Try setting up a standard game between a few 
>computer players and watch what they do for a few turns.

I believe this has something to do with Hans fiddling with 
roll_dice or some such call. It was part of a hack to get rid of 
some of the sync errors that the pathfinding created.

When I get around to working on the pathfinding and movement code, 
I will probably roll back the new pathfinding, if Hans doesn't do 
it first. At that time, I will undo any other little hacks that 
hobbled Xconq in an attempt to get rid of sync errors.

> STILL no crashes on my Linux box since your fixes to the TCL 
>map redrawing code!

Good.

  Regards,
    Eric