[Freeciv-Dev] Re: (PR#12313) Battle computation questionable

["Jason Short" <jdorje@xxxxxxxxxxxxxxxxxxxxx>]

<URL: http://bugs.freeciv.org/Ticket/Display.html?id=12313 >

Christian Knoke wrote:
> <URL: http://bugs.freeciv.org/Ticket/Display.html?id=12313 >
> 
> 
> CVS 21 FEB 2005 GTK2 S2_0
> 
> Maybe this is a case where the fight computation is wrong - please check.
> 
> Battle:   Tank attacks stack of 20 howitzers
> Ruleset:  default
> Options:  mostly default (irrelevant AFAICS)
> 
> Details:    Tank     Any of 20 howitzers (identical)
> ----------------------------------------------------
> Attack:     10       12
> Defense:     5        2
> Firepower:   1        2
> Hitpoints:  30       30
> HP (act.):  30       30
> Movepoints:  3        2
> MP (act.):   2       (unknown)
> Veteran:     1.75     1.50
> 
>   Boni:
> fortified:  no       no (F, not X)
> Hills:      --        1.5
> River:      --        1.5
> 
> Attack value (Tank):        10 * 1 * 1.75 = 17.5
> Defense value (Howitzer):   2 * 2 * 1.5 * 1.5 * 1.5 = 13.5
> 
> Winning chance displayed in popup: 45% (!)
> 
> Actual winner in game: Howitzers

Hills are x2, not x1.5.  (This comes from the defense_value of 20 in the 
ruleset.  I guess this is not a per-cent but is a per-deca or something.)

So it is instead (with the "internal" factor of 10 added on):

Attack: 10 * 1 * 1.75 = 175
Defense: 2 * 2 * 2 * 1.5 * 1.5 = 180.

> Statistical test: 
> 
>   Don't know how to do. From savegame reproducible, tank wins with 10 HP
>   remaining. Inserting other battles before, changes this. I didn't check
>   this systematically.
> 
> Note: 
> 
>   I saved exactly before the battle. So why isn't the result the same after
>   loading the savegame?

Saving and loading affects results, because the loaded game doesn't have 
exactly the same state (some values aren't saved) and sometimes loading 
the game itself will invoke new random calls.  This has defied all of 
our attempts to fix it.

-jason