[Freeciv-Dev] Re: (PR#3646) freeciv-ai ai_military_findvictim

["ue80@xxxxxxxxxxxxxxxxxxxxx" <ue80@xxxxxxxxxxxxxxxxxxxxx>]

On Fri, Mar 07, 2003 at 09:13:44AM -0800, Gregory Berkolaiko wrote:
> 
> On Fri, 7 Mar 2003, ue80@xxxxxxxxxxxxxxxxxxxxx wrote:
> 
> > On Fri, Mar 07, 2003 at 08:03:55AM -0800, Per I. Mathisen wrote:
> > > 
> > > On Fri, 7 Mar 2003, ue80@xxxxxxxxxxxxxxxxxxxxx wrote:
> > > > When looking at some AI-games (with diplomacy patch) is saw that AI is
> > > > better in landwar thenin seawar. Really bad in expanding, bad in
> > > > cityplacement, not optimal in the building it builds ...
> > > 
> > > True. Mostly because the settler code sucks. Secondly because the ferry
> > > code sucks.
> > 
> > Where is the settler code?
> 
> server/settler.c oddly enough
 
Thx
 
> > You seldem build fighters near enemy cities, they have to "jump" to the
> > front too. And using paras as normal units should be only done when you
> > have to defend. But in the attacking case "jump" step by step to the
> > front. When there is no front which can be reached within 10 fields of
> > your imperium you have to move them with ships or build cities between.
> > 
> > Think the airlift code and the airgoto codo should be a good place
> > to search how to handle the move from paras.
> 
> Yes, but at the current level I feel we should improve the basics: city 
> placement, expansion, defense, rather then venturing into _very_ hard but 
> specialised problems like paradrops.  

I have no problem with serializing issues. I only mentioned it here
because its a way to get a city in 1 turn but the unit isn't adjacent to
the city.

So i've put it into the testsample where we can check future
implementation against my opinion how it should be.

Thomas
-- 
Thomas Strub  ***  eMail ue80@xxxxxxxxxxxxxxxxxxxxx
Wenn Du nicht programmieren kannst und Dir für Arbeit zu schade bist:
Werde Berater, Analyst oder organisiere Kongresse.