This is the mail archive of the xconq7@sources.redhat.com mailing list for the Xconq project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Standard game problems in windows (20031214) version


On Tue, Dec 16, 2003 at 06:18:55PM +0100, Bruno Boettcher wrote:
> On Tue, Dec 16, 2003 at 11:42:51AM -0500, Eric McDonald wrote:
> Hello!
> 
> > This behavior was present in Xconq prior to the new path 
> yep!
> i address it with standing orders :D
> 
> > code. I agree that it is annoying. If you move any unit past its 
> true!
> 
> > SupplyLow threshold, you will see this behavior. Again, I am going 
> > to wait for Peter to make his move first before trying to do 
> > anything about it.
> that would be really nice!

I better reply in some detail about how I intend to address this.
The path-finding node state will include not only the geometric position
but also the amount of material in each category. As the unit passes a
refueling point, the supply will be repleneshed. Two nodes will have
equivalent state if their material supply and geometric position are
equivalent, also their transportation state. I am currently modifying
path.c and task.c, but certainly cant yet compile the code, its still
in a design phase really.

I still havent quite worked out how to ensure the unit gets to point B
with enough fuel to fly back to a resupply point. Possibly a pseudo
target of getting to a resupply point after reaching the destination
would work here, so that a destination is re-fuelable, for the AI anyway.
The statenode, when it reaches its x,y target, gets a flag set, so
that the final target is a refueling point, and the unit has enough fuel to return
across whatever terrain is there.

> 
> > There is at least one complication here:
> > Suppose that a unit's path becomes blocked or its destination 
> > is captured while it is en route. If we blindly follow the path, 
> > ignoring the SupplyLow indicators, then we risk losing or 
> > paralyzing the unit due to overconsumption of supplies. I suppose 
> > that one could recheck the path for blockage after each move, but 
> > this might be costly in terms of CPU (it would still be cheaper 
> > than recomputing the path after each move, I think).
> what about recomputing the path to the nearest resupply source when we
> hit the low supply case (and eventually compare it to the destination)?

The attention post 7.5 is to move the path-finding above the networking,
so the cpu problem isnt so bad. It should also be possible to inspect
the cached path to ensure that all nodes are outside zoc and also that
all refueling points are available, and if waiting for a ferry if the
ferry is not sunk. Also a path recalculation every N
hexes or perhaps every move probably is reasonable, if a ferry isnt
scheduled to pick the unit up.

Unfortunately the AI seems to generate distant targets in some games.

Peter


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]