[Freeciv-Dev] Re: (PR#4538) [Bug] is_enemy_unit_tile sometimes doesn't work in client.

["Gregory Berkolaiko" <Gregory.Berkolaiko@xxxxxxxxxxxx>]

On Thu, 17 Jul 2003, Jason Short wrote:

> 
> [glip - Fri Jul 11 18:11:05 2003]:
> 
> > If you call, from the client, is_enemy_unit_tile on a tile containing
> > enemy unit in a city, the functio will return NULL.  This so far only
> > causes problems for path finding but will also cause trouble for any
> > future client-side AI.
> > 
> > Same problem applies to all is_*_unit_tile functions.
> > 
> > Here we have a choice:
> >  (1) fix all is_*_unit_tile functions,
> > or
> >  (2) fix the functions which use them.
> > To do (1), the signature of the functions must be changed to bool (not
> > that it's ever used as a pointer).
> > 
> > And a problem: the information given by a "occupied" flag over city is
> > not
> > complete.  Enemy city may contain neutral unit.  And for the fogged
> > city,
> > the occupied flag means nothing.
> 
> Rather than return a punit, they could return a pplayer.  This would
> keep the semantics of the function as far as most users are concerned,
> would continue to work even in the problem case you describe, and would
> allow certain usages (like is_non_allied_unit_tile in unithand.c) to
> continue to work.
> 
> On the other hand, if they are called "is_***" they really should have
> boolean return values...

Return value aside, I think the resolution of this "bug" should be a 
warning before these functions that one should be careful when using them 
in the client.

G.