[freeciv-ai] Re: (PR#6807) [Bug] fstk suggests to attack non-enemy

["Gregory Berkolaiko" <Gregory.Berkolaiko@xxxxxxxxxxxx>]

<URL: http://rt.freeciv.org/Ticket/Display.html?id=6807 >

On Sun, 9 Nov 2003, Per I. Mathisen wrote:

> <URL: http://rt.freeciv.org/Ticket/Display.html?id=6807 >
> 
> On Sun, 9 Nov 2003, Gregory Berkolaiko wrote:
> > The following is a part of the problem.
> > From Paulz profile I can see that each execution of ai_military_attack
> > calls fstk around 5 times on average.This is because fstk suggests a
> > victim which is not our enemy now, we are next to it, but we cannot
> > attack it, so we call fstk again.
> >
> > There is a reason why fstk suggests a non-enemy, it's because this
> > non-enemy is our "war target" or "enemy-to-be". Because fstk is also
> > used to evaluate building wants, it views the "enemy-to-be" as an enemy
> > so we can build units against them before we actually declare war on them.
> >
> > When it comes to finding stk for a real unit, this behaviour is not so
> > desirable though.
> 
> Since fstk is going the way of the dodo anyway, I think maybe we should
> just revert the AI diplomatic clerverness I put in there to make it build
> up against 'war targets'.
> 
> Diff:
> -    if (!HOSTILE_PLAYER(pplayer, ai, aplayer)) {
> +    if (!pplayers_at_war(pplayer, aplayer)) {
> 
> Or we could fix it, using the fact that punit->id will be zero if we pass
> a virtual unit to fstk.
> 
> Diff:
> -    if (!HOSTILE_PLAYER(pplayer, ai, aplayer)) {
> +    if ((punit->id == 0 && !HOSTILE_PLAYER(pplayer, ai, aplayer))
> +         || (punit->id != 0 && !pplayers_at_war(pplayer, aplayer))) {
> 
> Up to you, Greg.

The second way is better IMO (with a comment).  Please do the patch, Per.  
I am really busy now.

G.