[Freeciv-Dev] Re: (PR#4095) Allied Victory :-)

["Reinier Post" <rp@xxxxxxxxxx>]

On Mon, Apr 28, 2003 at 12:41:34PM -0700, Per I. Mathisen wrote:

[...]

> You get the (now infamous): allied(A, B), allied(B, C) and war(A, C).
> 
>   - Per

Just a remark from a bystander, but hacking pieces of code where these
states are used (such as the unit movement code) doesn't appear to be
the right solution; instead (as has been suggested) it seems that this
state of affairs should never be allowed to occur.

Mathematically I can see two different issues:

  - the transitivity of allied():

    is there any harm in allowing
       allied(A, B), allied(B, C) without allied(A,C) ?

  - the incompatibility of allied+ and war:

    even if allied isn't actually transitive itself,
    it seems reasonable to impose:

       not (allied+(A,B) and not war(A,B))

    where allied+ is the transitive closure of allied.

-- 
Reinier