[freeciv-ai] Re: veteran cleanup

[Ross Wetmore <rwetmore@xxxxxxxxxxxx>]

It might be worthwhile doing a quickie RFC on the strategy of such things
and fold the summary results into the code as a comment.

Two reasons for building a barracks and one for not ...

1)  City is under constant attack or heavy nearby threat. The heal factor
    for reconstituting units in an action area is worth the barracks cost.

2)  You only need 1 barracks in every 3 or 4 cities. Most cities should not
    be building military units.

3)  City has poor shield resources. Will take forever to build anything so
    the barracks is not effective. City should build improvements and buy
    them when need is pressing - never buy units because of the cost. The
    reverse of this is actually a reason for building a barracks.

Cheers,
RossW
=====

Gregory Berkolaiko wrote:
> On Sun, 16 Nov 2003, Per I. Mathisen wrote:
>
>
> > I've backported and improved some AI changes from veteran level patch.
> > Makes use of do_make_unit_veteran() where it wasn't. Attached. Committed.
>
> The main change is that we do not assume that barracks will be built 
> before we built any military guys.  While *always* building barracks is 
> not obviously good, changing it might backfire: if attacking with a 
> non-vet unit is futile, the city will never attempt to build mil. units 
> and therefore will never build barracks -- a vicious circle.
> This is my general idea for building barracks:  they should be built if
> (a) veteran unit has higher want even if we take the cost of barracks into 
> account
> or
> (b) the city has been deciding to go for military unit for many turns in 
> recent history -- this can be realised by bumping up the barracks want 
> every time the city decides to built an attacker (and introducing some 
> time-decay for the want).
> G.