[Freeciv-Dev] Re: (PR#4785) disappearing archers

["Jason Short" <jdorje@xxxxxxxxxxxxxxxxxxxxx>]

Michael Mielke wrote:
> On Thu, 7 Aug 2003, Jason Short wrote:
> 
>>Michael Mielke wrote:
>>
>>>On Thu, 7 Aug 2003, Jason Short wrote:
>>>
>>>>jorneg@xxxxxxxxxxx wrote:
>>>>
>>>>>   What has happen is that the two archers attacked the cannon and 
>>>>>loose the combat. And then, the cannon attacked and killed the settler.
>>>>>   And you have reason, there should be a message like in normal combat.
>>>>
>>>>You're saying there's a missing message?
>>>>Even so I find it hard to believe; either one of the settlers should 
>>>>have easily defeated the cannon.
>>>
>>>Perhaps maybe an upkeep message?
>>
>>No, they didn't die for lack of upkeep.  (1) I would have gotten a 
>>message for that.  (2) No other units died from lack of upkeep the 
>>entire game; waht are the odds that these two were the first?  (3) Under 
>>Monarchy few units die from lack of upkeep.
>>Killed while attacking the cannon is very unlikely, but possible.  Of 
>>course this still leaves the question of why there was no message.
>>jason
> 
> 
> I seemed to notice, that often an ai player makes 2 moves in one turn with 
> a unit.

Yes, this is because the unit gets to move and sometimes it's next 
turn's (or was it previous turn's?) goto will also be executed before 
you get to move again.

This is a bug in the end-of-turn code (see PR#725) and the reason I 
originally used goto to try to kill the cannon "before it got to attack 
me" (i.e., this trick can work both ways).  I'm beginning to think a 
partial solution to this problem is that shuffle_players() should only 
be called once, when the game starts - and the order saved in the 
savegame (see PR#4733).  But even so I think there would still be a goto 
issue (goto executed at the wrong time for AI players).

jason